Safety culture and safety management within the Norwegian-controlled shipping industry

Safety culture and safety management within the Norwegian-controlled shipping industry State of art, interrelationships, and influencing factors By

Helle A. Oltedal Thesis submitted in fulfillment of The requirements for the degree of PHILOSOPHIAE DOCTOR (PhD)

Faculty of Social Sciences 2011

University of Stavanger N-4036 Stavanger NORWAY www.uis.no Š2011 Helle A. Oltedal ISBN: 978-82-7644-464-3 ISSN: 1890-1387

ii

Preface My interest in the shipping industry started while working on a project in which I calculated the risk for ships making contact with high-rise constructions located shore side. When searching for background information, I realized that very little research concerning safety related to human factors and seafarers’ welfare existed. I found evidence of seafarers working under unacceptable working conditions, sailing on “rust buckets,” jeopardizing their own safety as well as that of the vessels—in some cases afraid to come forward out of fear of losing their jobs. At the same time, working at sea was ranked as one of the most dangerous occupations in the world. This situation triggered my interest in the field and the research presented in this thesis. I need to thank several people for their contribution during the process of completing this PhD. First of all, I would like to thank my supervisor Ole A. Engen for his encouragement, help, and support throughout this process. I would also like to thank my colleague Doctor David McArthur for helping me with statistical problems as well as useful comments and suggestions while writing the articles. Dean A. Rune Johansen has also been a great support during my research. Thanks to his engagement in the maritime industry, I was able to receive funding for five more years of research—something that I appreciate tremendously. I would like to mention Captain Vigleik Storesund, who has been a great help in enabling me to understand the maritime industry and life at sea. Last, but not least, Chief Officer Johanne Marie Trovåg, Professor Knud Knudsen, and Professor Preben Hempel Lindøe all made tremendous efforts in reading through my final work.

iii

Furthermore, I would like to thank the Norwegian Shipowners’ Association for helping me get in touch with their members. Although they cannot all be mentioned by name, I wish to thank the shipping companies who volunteered to participate and all the seafarers who—despite a hectic working situation— found time to contribute with their views. In addition several companies have enabled me to sail on their vessels and in other ways participate in their daily operations. I would also like to thank Dr. Nick Bailey and Professor Helen Sampson at the Seafarers International Research Centre (SIRC) in Cardiff, Wales, for making my research stay there possible. My apologies to all of you who have not been mentioned. I cannot mention you all, but no one has been forgotten. During my research, I have gained valuable and useful contributions from many people of different nationalities, including academics, practitioners, seafarers, and shore personnel working in all parts of the shipping industry. Finally, I would like to thank my family for their encouragement and support. I also want to apologize to my two young children who have had a mum preoccupied with research for such a long time. I have spent long periods away from home and, when staying at home, I have been mentally absent. I hope I can now make up for the lost time. It would not have been possible to complete this thesis without the support of my incredibly patient and supportive husband Mikal. Thus, I dedicate this thesis to them—Mikal and our two exceptional and wonderful children, Jørgen and Elene. Haugesund, April 2011, Helle A. Oltedal iv

Summary This research focuses attention on safety challenges within the Norwegian shipping industry. A status picture of the shipboard safety culture and the interrelationships with safety management and organizational factors is given. Three research questions are explored: (1) What characterizes safety culture and safety management within the shipping industry? (2) What is the relationship between safety culture and safety performance within the shipping industry? (3) What characterizes shipping companies’ application of the safety management concept? In order to explore these research questions, four aims were defined to guide this work: (1) to outline and discuss the application of safety culture and safety management within merchant shipping; (2) to outline and discuss relevant theories of safety culture and safety management and analyze the relationship between safety culture and safety management; (3) to support the use of a methodological framework for the assessment of safety culture in relation to safety management; and (4) to assess safety culture within merchant shipping and analyze the relationship with safety management and actual performance. The research questions are further examined and specified in six journal articles. The thesis is divided into two main parts. Part I includes the overall framework in relation to research aims. Part II presents the six journal articles. In part I, chapter 1, a general introduction and a status picture of risk, safety management, and safety culture within the shipping industry are presented, which gives reason for the research aims and questions introduced in the chapter. Chapter 2 outlines the safety responsibilities within the industry at the international, national, and company levels. Emphasis is placed on the v

International Safety Management (ISM) Code, which provides the minimum standards and guidelines for operational safety management. Chapter 3 provides theoretical clarification and framing with regard to safety culture and safety management. This chapter also introduces a general working model used in the studies of safety culture and safety management in this thesis. Chapter 4 presents the methodological approach. The thesis builds upon a mixed method approach where both qualitative and quantitative techniques are used. The main results are briefly summarized in Chapter 5, followed by a discussion in Chapter 6 and concluding remarks in Chapter 7. The concluding remarks concern study limitations, implications, and suggestions for future research. The thesis draws upon theory from both the socio-anthropological and organizational psychological directions. In accordance with the organizational psychological perspective, a survey was carried out. A safety culture questionnaire developed by Studio Apertura, a constituent centre of The Norwegian University of Science and Technology (NTNU), in collaboration with the Norwegian DNV and the research institution SINTEF was used. In total, 1,574 questionnaires were distributed to 83 tanker and bulk/dry cargo carriers, with 1,262 being returned from 76 of the vessels. The vessels were initially randomly selected from the Norwegian Shipowners’ Association member list, but as participation was voluntary, some withdrawal occurred. Statistical analysis involves descriptive statistics, factor analysis, regression analysis, and structural equation modeling. The statistical survey results were complemented by qualitative data obtained through document studies, case studies including two tanker companies and two bulk/dry cargo companies,

vi

interviews, participating observations and field studies at sea, and participation in other maritime forums. The study results indicate several deficiencies in all parts of a traditional safety management system defined as: (1) the reporting and collection of experience data from the vessel; (2) data processing, summarizing, and analysis; (3) the development of safety measures; and (4) implementation. The underreporting of experience data is found to be a problem, resulting in limitations related to the data-processing process. Regarding the development of safety measures, it is found that the industry emphasizes the development of standardized safety measures in the form of procedures and checklists. Organizational root causes related to company policies (e.g., crewing policy) is to a lesser degree identified and addressed. The most prominently identified organizational influential factors are the shipping companies crewing policy, which includes rotation systems, crew stability, and contract conditions, and shipboard management. The companies’ orientation toward local management, which includes leadership training, educational, and other managerial support, are also essential. The shore part of the organization is identified as the driving force for development and change in the shipboard safety culture. Thus, safety campaigns should to a larger degree include and be directed toward shore personnel.

vii

List of articles in thesis: Article 1 Oltedal, H. A., & Engen, O. A. (2009). Local management and its impact on safety culture and safety within Norwegian shipping. In S. Martorell, C. Guedes Soares & J. Barnett (Eds.), Safety, Reliability and Risk Analysis: Theory, Methods and Applications (pp. 1423-1430). London: Taylor & Francis Group. Article 2 Oltedal, H. & Wadsworth, E. (2010). Risk perception in the Norwegian shipping industry and identification of influencing factors. Maritime Policy & Management, 37(6), 601-623. Article 3 Oltedal, H. A. (2010). The use of safety management systems within the Norwegian tanker industry—Do they really improve safety? In R. Bris, C. Guedes Soares, & S. Martorell (Eds.), Reliability, Risk and Safety: Theory and Applications (pp. 2355-2362). London: Taylor & Francis Group.

viii

Article 4 Oltedal, H. A., & Engen, O. A. (2010). Tanker versus dry cargo—The use of safety management systems within Norwegian dry cargo shipping. In J.M. Ale, I.A. Papazoglou, & E. Zio (Eds.), Reliability, Risk and Safety (pp. 21182125). London: Taylor & Francis Group. Article 5 Oltedal, H. & McArthur, D. (2010). Reporting practices in merchant shipping, and the identification of influencing factors. Safety Science, 49(2), 331-338. Article 6 Oltedal, H. A., & Engen, O. A. (2010). Safety Management in Shipping— Making Sense of limited Success. Safety Science Monitor, submitted.

ix

Contents Preface ............................................................................................................ iii Summary ..........................................................................................................v List of articles in thesis: ............................................................................... viii Contents ............................................................................................................x Part I .................................................................................................................1 1

2

General background and introduction ..................................................1 1.1

Safety culture and safety management within shipping.................... 2

1.2

Research aims and research questions .............................................. 5

1.3

Aims of articles ................................................................................. 7

Safety responsibilities in maritime industry ........................................11 2.1

The International Safety Management (ISM) Code ........................ 13

2.2

Maritime administrations and responsibilities ................................ 17

The flag state ........................................................................................... 18 The Port State Control ............................................................................ 20 The company and crew management ...................................................... 23 2.3

When the regulatory framework and safety management fails ....... 24

2.4

Safety responsibilities in the maritime industry—a summary ........ 27

x

3

Safety culture and safety management in theory and practice ..........29 3.1

Safety culture as an organizational integrated concept ................... 33

3.2

Organizations, management and cultural change ........................... 37

3.3

Organizational culture and safety management .............................. 41

The Man Made Disaster model ............................................................... 43 Normal Accident Theory ........................................................................ 45 High Reliability Organizations Theory ................................................... 47 Managing risk and safety culture ............................................................ 49 The theory of Practical Drift – why organizations fails .......................... 51 3.4 4

Safety culture and measurable outcome variables .......................... 53

Research methodology ...........................................................................56 4.1

Quantitative research and questionnaire survey.............................. 58

Questionnaire development .................................................................... 60 Survey sample and respondents’ demographics ..................................... 61 Validity and reliability through theoretical conceptualization ................ 62 Validity and reliability through factor analysis....................................... 66 Validity and reliability through scale analysis ........................................ 68 Causal relationship through structural equation modeling ..................... 70

xi

4.2

Qualitative research design ............................................................. 71

Document study ...................................................................................... 71 Case studies............................................................................................. 73 Interviews................................................................................................ 73 Participatory observation ........................................................................ 74 Participation in maritime forums ............................................................ 75 4.3 5

Applied methods and statistics in articles ....................................... 76

Research results .....................................................................................78 5.1

Summary and results of article 1 .................................................... 78

5.2

Summary and results of article 2 .................................................... 80

5.3

Summary and results of article 3 .................................................... 83

5.4

Summary and results of article 4 .................................................... 85

5.5

Summary and results of article 5 .................................................... 87

5.6

Summary and results of article 6 .................................................... 89

5.7

Causal relationships between components of safety culture ........... 91

5.8

Summing up and presentations of main conclusions ...................... 98

6

Discussion ...............................................................................................99

7

Concluding remarks ............................................................................107

xii

8

7.1

Methodological limitations ........................................................... 107

7.2

Theoretical limitations .................................................................. 109

7.3

Future research .............................................................................. 109

7.4

Final remarks ................................................................................ 110

References .............................................................................................112

Part II ............................................................................................................124 Article 1 Article 2 Article 3 Article 4 Article 5 Article 6

xiii

General background and introduction

Part I 1 General background and introduction “Never before have so few done so much for so many.” When opening the Year of the Seafarer in 2010, these brave words—a quotation paraphrased from one of Winston Churchill’s most famous speeches—were similarly strikingly declared by Efthimios E. Mitropoulos, Secretary General of the International Maritime Organization (IMO). Few people seem to understand the importance of seafarers and shipping in our society. The worldwide population of seafarers serving in international trade is estimated to be approximately 1,187,000 people from virtually every nationality. Worldwide, about 50,000 ships carry about 90% of the world’s trade; thus, these more than one million seafarers are transporting goods for the benefit for the world’s population of almost 7 billion. The seafarers’ and shipping industry’s global importance is commonly highlighted by the phrase “without shipping, half the world would starve and the other half would freeze.” The current research is conducted in light of safety challenges within merchant shipping. During the first five years of the previous decade (i.e., 2000 to 2005), an average of 18 ships collided, grounded, or caught fire every single day, and two vessels were sinking every day (Gregory & Shanahan, 2010). Merchant shipping and seafaring are traditionally perceived as a risky industry—a risk partly induced by its situational characteristics. Work at sea is demanding as both work and leisure time happens within a small group, at the same place, for a long period of time and with few possibilities to interact with the surrounding world. The seafarer’s only alternative whereabouts when 1

General background and introduction

at sea is the sea itself, where harsh conditions prevail. Seafarers and their vessels are constantly exposed to forces beyond their control, such as storms, freak waves, and strong currents. Being far from port most of the time, the seafarer must handle critical situations with little or no support from others, with only their own competence and expertise to rely on.

1.1 Safety culture and safety management within shipping Human error is associated with the vast majority of accidents and incidents within shipping. An estimated 75% to 96% of marine casualties are caused— at least in part—by some form of human error (Anderson, 2003; Rothblum, 2000; Wagenaar & Groeneweg, 1987). However, within recent safety management theories, human error is not seen as a cause of accidents and incidents, but rather as something shaped and provoked by upstream organizational factors. Thus, human error is not an explanation per se, but something that needs further explaining (Hollnagel, 2004; Reason, 2001). Possible explanatory factors may be related to seafarers’ cognitive system (e.g., human information processing, training, motivation, and fear), social system (e.g., social pressure, role, and life stress), and situational system (e.g., physical stress, environmental stress, and ergonomic aspects), which are all assumed to be mutually interdependent (Wagenaar & Groeneweg, 1987). It is also widely accepted that individual factors are inextricably linked to organizational factors and decisions (Hollnagel, 2004; Reason, 2001; Schager, 2008).

2

General background and introduction

Safety at sea is regulated by the UN’s agency for maritime affairs, the International Maritime Organization (IMO). From the IMO perspective, safety management and human error are closely intertwined with the industry’s definition and application of the safety culture concept, regulated through the International Safety Management (ISM) Code (IMO, 2010a; Lappalainen, 2008; Mitroussi, 2003). The ISM Code, which became mandatory for all merchant vessels from July 1998 to July 2002, formally introduced the idea of safety culture in shipping: The application of the ISM Code should support and encourage the development of a safety culture in shipping. Success factors for the development of a safety culture are, inter alia, commitment, values and beliefs (IMO, 2010a, p. 35). However, despite the implementation of the ISM Code, recent statistics indicate that losses are continuing to increase, resulting in a heavy loss of life and serious damage to the environment (Soma, 2010). The statistics in Figure 1 illustrate the frequency of navigational accidents (collisions, contacts, and wrecked/stranded vessels) from 1993 to 2009.

3

General background and introduction

Figure 1: Navigational accident frequency in relation to the world fleet size, 1993-2009 (Source: Lloyds’ Fairplay, 2010) Fleet size in number of crude oil tankers over 100,000 dwt, chemical tankers over 10,000 dwt, containers over 20,000 dwt, RoRo cargo over 10,000 dwt, bulk over 50,000 dwt.

As shown in Figure 1, the frequency of serious navigational accidents has increased significantly since 2002. It is also interesting to note that, since the first introduction of the code in 1998, none of the subsequent years show lower accident frequency than before the code was introduced. This statistical trend raises a fundamental question: Why do we have such an increase in the accident rates despite the introduction of the ISM Code, emphasis on safety culture, and lower tolerance for non-conformities? Three possible explanations have been put forth: 1. Shipping companies’ implementation of the ISM Code and understanding of safety management are inadequate. The IMO assessment of the effectiveness of the ISM Code (IMO, 2005) indicates that implementation has resulted in more administrative 4

General background and introduction

work, procedures, checklists, and other means in order to control human behavior. However, is safety best ensured by controlling and restricting human behavior? 2. The ISM Code’s underlying theoretical rationale of linear causality is inadequate. Is it possible to prevent future accidents by learning from past events? Are there any causal links between near misses, minor incidents, and major accidents? When dealing with future events evolving in an unforeseen and complex pattern, are other rationalities more adequate? 3. The ISM Code’s assumption of a relationship between safety culture and actual safety performance and outcome is inadequate. What is organizational safety culture and what determines its relations to safety management, organizational practices, and safety performance? This thesis provides an account of these three possible explanations. Empirical data are collected from the Norwegian controlled liquid and dry cargo shipping industry for this purpose. The data are analyzed and discussed in light of theory on safety management and safety culture. The seafarers’ perspective and their operative experiences are emphasized. The analyses and discussion will be consistent with the scope of the research aims and questions, as formulated in the following sections.

1.2 Research aims and research questions Although shipping is known to be a risky industry, surprisingly little research has been done within this area. In recent years, a few articles and doctoral 5

General background and introduction

theses on maritime safety culture and climate have been published (e.g., Antonsen & Norges teknisk-naturvitenskapelige universitet. Institutt for sosiologi og statsvitenskap, 2009; Christophersen, 2009; Ek, 2006; HĂĽvold & Norges teknisk-naturvitenskapelige universitet. Institutt for industriell økonomi og teknologiledelse, 2007; Lamvik, 2002; Ă˜streng, 2007). In light of the discussed situation, it is important to get a better understanding of what characterizes safety culture within shipping and how shipboard safety culture relates to safety management and human error. Thus, the following four research aims have been developed in order to provide direction for this thesis: 1. To outline and discuss the application of safety culture and safety management within merchant shipping. 2. To outline and discuss relevant theories of safety culture and safety management and analyze the relationship between safety culture and safety management. 3. To give reason for a methodological framework for assessment of safety culture in relation to safety management. 4. To assess safety culture within merchant shipping and analyze the relationship with safety management and actual performance. In order to pursue these aims, a questionnaire survey was carried out within merchant shipping along with field studies, case studies, interviews, and other qualitative methods. The population is defined within the Norwegian controlled shipping industry as liquid tankers and dry cargo carriers above 500 gross ton. Norwegian controlled is defined as vessels owned by

6

General background and introduction

Norwegian parties where the owners’ safety management department is located in Norway. In the study, the seafarers’ perspective is emphasized. Based on the four research aims previously described, three research questions were developed for the purpose of the thesis: 1. What characterizes safety culture and safety management within the shipping industry? 2. What is the relationship between safety culture and safety performance within the shipping industry? 3. What characterizes shipping companies’ application of the safety management concept?

1.3 Aims of articles The thesis includes six separate studies with their own main aims. All aims for each study are discussed in this section. 1. Local management and its impact on safety culture and safety within Norwegian shipping: The first study uses survey data collected in 2006. The data cover all sailing personnel on 76 Norwegian controlled liquid tankers and dry cargo carriers. The aims of the study are to:  Explore and analyze the shipboard characteristics of safety culture;  Identify which factors affect the shipboard safety culture; and  Get results in order to set direction for further studies. 7

General background and introduction

2. Risk

perception

in

the

Norwegian

shipping

industry

and

identification of influencing factors: This second article makes use of the same survey data. In this study, the catering personnel and captains are excluded. Risk perception is used as an indicator for shipboard safety. The aims of the study are to:  Assess the relationship between risk perception and dimensions of safety culture; and  Explore the influence of organizational structural variables. 3. The use of safety management systems within the Norwegian tanker industry and whether they really improve safety: The third study involves both quantitative survey data and qualitative data. The sub-sample dry cargo carriers are excluded. The data and analyses are organized in accordance with the sub-components and information flow of a traditional safety management system. The aims of the study are to:  Describe the status of safety management within the liquid tanker sector; and  Identify organizational structural factors that influence the safety management performance.

8

General background and introduction

4. Tanker versus dry cargo regarding the use of safety management systems within Norwegian dry cargo shipping: The fourth study is a follow-up of the third, and follows a similar structure related to the sub-components and information flow of a traditional safety management system. The study includes both quantitative survey data and qualitative data. The sub-sample liquid cargo carriers are excluded. The aims of the study are to:  Describe the status of safety management within the dry cargo sector;  Identify organizational structural factors that influence the safety management performance; and  Compare current situations between the two sectors—namely, dry and liquid cargo carriers. 5. Reporting practices in merchant shipping and the identification of influencing factors: The fifth study involves quantitative survey data. Both the third and fourth study identified underreporting of experience data as a substantial problem. The reporting of experience data is regarded as a main cornerstone in a safety management system. Thus, the aims of the article are to:  Assess the relationship between reporting practices and the dimensions of safety culture;  Explore the influence of local management; and

9

General background and introduction

ďƒź Further explore differences between the dry and liquid cargo carrier sectors. 6. Safety management in shipping and making sense of limited success: The sixth study involves both quantitative survey data and qualitative data. As the previous studies (i.e., three through five) point to a substantial weakness in current safety management practices, the aim of this study is to: ďƒź Explore and identify reasons for the gaps between safety ambitions inherent in traditional safety management systems and operational practices.

10

Safety responsibilities in maritime industry

2 Safety responsibilities in maritime industry This section is, in accordance with specified research aim 1, formulated as follows: 1.

To outline and discuss the application of safety culture and safety

management within merchant shipping. The development of international trade and shipping in today’s globalized market has to a large degree determined the regulative structure of the industry. The international regulative system is of high importance for the safety of ships and crew sailing the seven seas, as every shipping company is required to relate to this during daily operations. In order to gain proper understanding of safety management within shipping, knowledge of the most important laws and the international regulative framework is necessary. Thus, some of the historical mainlines and the present situation related to safety management and the regulatory system will be further presented. An overview of the international regulatory system, maritime administration, and conventions (conventions in bold) related to safety management are shown in Figure 2 (next page).

11

Safety responsibilities in maritime industry Maritime Administrations (Flag and Costal States)

United Nations (UN)

United Nations Convention on the Law of the Sea (UNCLOS)

International Maritime Organization (IMO)

Convention for the Safety of Life

The International Safety

at Sea (SOLAS)

Management (ISM) Code

Figure 2: The maritime international regulatory system related to safety management

As shown in Figure 2, the maritime administrations’ (flag and coastal states) safety responsibilities are determined by the UN through the Convention on the Law of the Seas (UNCLOS). Although UNCLOS sets the broad regulative framework, the task of developing and maintaining workable regulations on ship safety within this framework is delegated to the UN agency the IMO, which is now responsible for 35 international conventions and agreements. For the purpose of this thesis, the most relevant is the Convention for the Safety of Life at Sea (SOLAS) chapter IX, management for the safe operation of ships, and the guidelines for SOLAS IX—namely, the ISM Code.

12

Safety responsibilities in maritime industry

2.1 The International Safety Management (ISM) Code The ISM Code became mandatory for all merchant vessels above 500 gross tons in two waves, depending upon type of vessel—namely, July 1, 1998, and July 1, 2002. Until the adoption of the ISM Code, IMO had attempted to improve shipping safety largely by improving the hardware of shipping (e.g., the construction of ships and their equipment). By comparison, the ISM Code focuses on the way shipping companies are managed. The ISM Code is the first to provide regulations and guidelines to promote the development of sound management and operating practices in order to ensure crew safety and avoid damage to the environment. The shipping industry is known to have a reactive approach toward safety as the process of regulating the activity has evolved primarily as a response to maritime disasters. Development of the ISM Code was also based upon a growing recognition that loss of life at sea and environmental pollution are influenced by the way in which companies manage their fleets. Table 1 (next page) summarizes some of the accidents’ precursory to the ISM Code (Anderson, 2003).

13

Safety responsibilities in maritime industry Table 1. Accidents Precursory to the ISM Code 1987

Herald of Free Enterprise capsized off Zeebrugge; 190 people lost their lives.

1987

Donna Paz ferry collided with a tanker in the Philippines; an estimated 4,386 people were killed.

1989

Exxon Valdes ran aground off the coast of Alaska, spilling 37,000 tons of oil and causing extensive environmental damage.

1990

Scandinavian Star caught fire; 158 people lost their lives.

1991

Agip Abruzzo, with 80,000 tonnes of light crude on board, was in a collision with the ro-ro ferry Moby Prince off Livorno, Italy. Fire and pollution occurred, and 143 people died.

1991

Have experienced fire and explosion off Genova, spilling 50,000 tons of crude oil; 6 people were killed.

1991

The Egyptian ferry Salem Express struck a reef and sank; 470 people were killed.

1991

Aegean Sea broke in two off La Coruna, Spain, spilling about 74,000 tons of crude oil; extensive pollution occurred.

1993

Braer driven onto the Shetland Island, carrying about 84,700 light crude oil; extensive pollution occurred.

1994

Estonia ro-ro passenger ferry sank after the bow door fell off during heavy weather at sea; 852 people lost their lives.

A common factor appearing in these accidents was human error, which could be traced back to poor safety management and organizational practice. By introducing the ISM Code, IMO intended to adopt a proactive approach

14

Safety responsibilities in maritime industry toward safety, where future accidents should be prevented by learning from and reflecting upon previous mistakes and experiences. The ISM Code requires shipping companies to develop, implement, and maintain a safety management system, which includes the following functional requirements: (1) a safety policy; (2) instructions and procedures to ensure safe operations of ships in compliance with relevant international and flag state legislation; (3) defined levels of authority and lines of communication between and amongst shore and shipboard personnel; (4) procedures for reporting accidents and non-conformities with the provision of the ISM Code; (5) procedures to prepare for and respond to emergency situations; and (6) procedures for internal audits and management reviews. In the code’s guidelines, emphasis is placed on near-miss reporting and how to create an organizational atmosphere in which people are willing to report accidents and non-conformities by developing a just culture. The concept of a just culture is also known to be a fundamental element in James Reason’s theory of safety culture and safety management (Reason, 2001). Moreover, in order to achieve the development of an organizational safety culture, IMO identifies three key elements: (1) recognizing that accidents are preventable by following correct procedures and establishing best practices; (2) constantly thinking about safety; and (3) seeking continuous improvement. IMO’s approach and perspective to safety culture is apparently instrumental, where safety culture is seen as something that may be engineered by an organization’s structures and control systems in order to produce desired behavioral norms and accompanying safety outcomes. Safety management, as described in the ISM Code, is founded on a linear causality, in which future events are attempted, predicted, and prevented by 15

Safety responsibilities in maritime industry analyzing past operational experiences. Thus, a critical system requirement is reliability and accuracy of input data—the experience, near miss, and accident reports ; as long as the input is reliable, the overall system presupposes the possibility of developing efficient standardized measures in order to control operational safety (Kjellen, 2000). One underlying assumption is that serious injuries and accidents may be prevented by learning from and reflecting upon incidents with no injury or damage. This idea is frequently illustrated as a near miss-accident pyramid. However, previous research does not support this theory (Anderson, 2003). Moreover, IMO recognized that near misses are underreported (IMO, 2007c), and the input system requirement is not met. This also provides a reason to question the underlying theory of linear causality, especially if near misses, small-scale accidents, and more serious events have the same causal chain (Rundmo, 1996). In 2005, IMO provided a report assessing the impact and effectiveness of implementation of the ISM Code (IMO, 2005). Based on the data collected, IMO concluded that—when the ISM Code and safety cultural development is embraced as a positive step—tangible positive benefits are evident. It was also recognized that ISM Code compliance could be made easier through a reduction in the administrative processes by, inter alia, the reduction of paperwork, increased reporting of operational experience data, and greater involvement of seafarers in the development of ISM manuals, the procedural system, and checklists. In the industry, it seems to be a common misconception that the ISM Code requires large quantities of paperwork and administration to function and that ticking boxes and checklists would replace good training and seamanship (Anderson, 2003).

16

Safety responsibilities in maritime industry Although the national maritime government is responsible for implementation of the ISM Code, the coastal state is responsible for enforcement of the code, and each shipping company has the primary responsibility for safe operations. However, these responsibilities are challenging as shipping today has become—more than ever before—a globalized industry. For example, a vessel may have owners in one state, be registered in a second state, be chartered by a company from a third state, and be transporting goods whose owners belong to a fourth state. To make it even more complicated, the vessel is sailing between ports in different states and is manned with a multinational and culturally diverse crew, who are managed by a company in yet another state. These sector-related circumstances have resulted in specific challenges with regard to the administration and enforcement of international regulations, as outlined in the following section.

2.2 Maritime administrations and responsibilities A maritime administration may have two different roles: a flag state and a coastal state. The coastal states’ responsibility for the enforcement of international regulations is done through inspections and Port State Control (PSC) of vessels entering their own coastal territorial waters, regardless of which flag the vessel is flying (Stopford, 2009). The coastal state may be the same as the flag state, but this is far from always the rule. Any ship owner is free to register a vessel in any of the world’s flag states.1 The term flag state

1

The definition of a flag state is not straightforward. A myriad of descriptions of flag states have evolved,

including traditional maritime nation, embedded maritime nation, national flag, classis register, open register, opportunist register, international open register, international register, closed register, second register, dependent territory register, offshore register, and flag of convenience (Mansell & SpringerLink,

17

Safety responsibilities in maritime industry (or administration) is used to refer to a country that maintains a vessel’s registry. The flag state has the overall responsibility for ensuring compliance with international regulations. This responsibility encompasses the operation of the ship, the physical status of the ship, the activities of the ship owners, and the working conditions of the seafarers. The flag administration, in the first instance, underwrites the safe operation of those ships under its flag.

The flag state Each flag state may have a national register, second register, and/or open register. A national register is reserved to vessels with national ownership. Second registers, which are additional to national registers, are mostly open registers. In an open register, ships owned by foreign entities may register. The creation of second registers is a response to intensified competition in the market for ship registration. In the early 1980s, the shipping market experienced a severe depression. Since the late 1980s, a number of states have created second registers in addition to their first national register in order to provide some or all of the advantages of an open register as a result of the economic crisis. A common motivation for establishing such second registers are to attract shipowners or prevent shipowners from flagging out by providing other or more relaxed application of the international IMO regulations (Alderton, 2004). Income in the form of tonnage taxation fee is also a motivation for some nations to establish a register when they do not necessarily have the means, will, or competence to meet their responsibilities as a flag state. According to Alderton (2004), low barriers to entry into the

2009). In this thesis, flag state refers to an open international register where any shipping owner is free to register a merchant vessel.

18

Safety responsibilities in maritime industry flag market exist, with minimal start-up cost or time being required. This situation has led to competition among some maritime administrations, which—in order to encourage registration of vessels under their flag—permits less bureaucratic control along with relaxed requirements. International regulations adopted by IMO intend to provide a harmonized set of rules for the industry. The previously described situation has resulted in variations among maritime administrations in performance and application of the international regulations (Alderton, 2004). Alderton (2004) distinguishes between three types of administrations: (1)

Regulatory efficient states in which the state seeks to regulate the full

extent of maritime operations. (2)

Regulatory inefficient states; the main distinction between this

category and the first lies in the treatment of labor issues. (3)

Unregulated states, in which the regulatory environment within these

registers is almost non-existent. Many of today’s safety-related criticisms are related to ship registration and to which flag the vessels fly. Although substandard shipping is mostly associated with regulatory inefficient and unregulated states, even administrations regarded as being regulatory efficient may have defective performance. An audit of the Norwegian Maritime Directorate, which is regarded as being regulatory efficient, revealed that the administration does not have, inter alia, adequate operational control with its own working procedures and that the administrated regulations are not comprehended in unison, which may result in misinterpretations and erroneous decisions (Riksrevisjonen, 2010).

19

Safety responsibilities in maritime industry As long as the flag states benefit from running open registries and shipowners can benefit from it, the situation will most likely never change. Moreover, the lack of flag state control, as evident in several countries, has made PSC even more important.

The Port State Control In the wake of some major maritime disasters in the European area (e.g., accidents with the Erika in 2000 and Prestige in 2002, which both occurred after the implementation of the ISM Code), it was realized that the PSC and ParisMOU2 could and should take a more determined stance against substandard shipping in order to ensure better enforcement of the international regulations. The Prestige was a Greek-operated oil tanker, officially registered in the Bahamas, but with a Liberian corporation registered as the owner. The

2

The ParisMOU (Memorandum of Understanding) on Port State Control is the official document in which

the 27 participating maritime authorities agree to implement a harmonized system of PSC. The MOU consists of a the main body in which the authorities agree on: 1) their commitments and the relevant international conventions, 2) the inspection procedures and the investigation of operational procedures, 3) the exchange of information, and 4) the structure of the organization and amendment procedures. The current member states of the ParisMOU region are, in alphabetical order, Belgium, Bulgaria, Canada, Croatia, Cyprus, Denmark, Estonia, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Latvia, Lithuania, Malta, Netherlands, Norway, Poland, Portugal, Romania, Russian Federation, Slovenia, Spain, Sweden and The United Kingdom. Following the foundation built by the ParisMOU, several other regional MOUs have been signed, including the Tokyo MOU (Pacific Ocean), Acuerdo Latino or Acuerdo de Viña del Mar (South and Central America), the Caribbean MOU, the Mediterranean MOU, the Indian Ocean MOU, the Abuja MOU (West and Central Atlantic Africa), the Black Sea MOU, and the Riyadh MOU (Persian Gulf). In this thesis, due to its geographic area, references will be made to the ParisMOU (http://www.parismou.org/).

20

Safety responsibilities in maritime industry ownership of the Prestige was unclear, and it was difficult to establish responsibility for the accident, which resulted in the spill of more than 60,000 tons of heavy fuel oil. Prior to the accident, the Prestige set sail without being properly inspected, although a previous captain had complained about numerous structural deficiencies. The oil tanker Erika also experienced structural failure. Although the structural failures were visible, the vessel was found to be seaworthy by the classification society. The Erika was sailing under a Maltese flag and chartered by a shipping company registered in the Bahamas on behalf of a French oil company. With regard to the Erika, it was also difficult to establish responsibilities (CPEM, 1999). The Erika accident resulted in a spill of about 19,800 tons of heavy fuel oil. Although both accidents involved structural failures, they can also be characterized as stemming from a non-functional ISM system. In a functional safety management system, such structural failures should have been detected and handled appropriately by the shipping company. In the case of Prestige, the captain had even notified the company about structural deficiencies that had not been handled properly. In the aftermath of these accidents, the response from ParisMOU came in form of developing a harmonized vessel detention policy, guidelines for operational PSC, and others (ParisMOU, 2007). One of the strategies was naming and shaming. Today, all inspection results and detentions, with detailed information about the company, vessel, and flag, are registered in a public database (at parismou.org). Information about banned vessels and “rust buckets� are also made public. A more recent initiative includes a list for the performance of flag states (MARISEC, 2008; MARISEC, 2006; Winchester, Alderton, & Seafarers International Research Centre, 2003). On this list, each flag state is evaluated and ranked based on 21

Safety responsibilities in maritime industry their performance on certain aspects, as PSC records ships flying their flags along with the implementation and enforcement of important international treaties, such as the ISM Code. Flag state performance is then ranked and placed on a black list (poor performance), grey list (mediocre performance), and white list (good performance). The black, grey, and white lists for 2009 included a total number of 82 flags, 24 on the black list, 19 on the grey list, and 39 on the white list (ParisMOU. 2010). Despite ParisMOU’s intention to make shipping safer, the current inspection system has inherent weaknesses. ParisMOU’s target is to inspect 25% of all vessels calling port. The ships are selected based on criteria as previous inspections reports from the MOU region. A vessel flying a poor performance black-listed flag is more likely to be selected for an inspection than others. These criteria for the selection of vessels are understandable; however, they disregard the fact the some of the ships flying a poor performance flag are owned by companies that take their responsibilities for the safe operation and crew welfare seriously. As a result, PSC resources may be used inefficiently. It is widely known that vessels in bad shape continue to be operational, without getting caught by the inspection net (Corbett, 2009; Tradewinds, 2007). In order to improve the current system, the ParisMOU introduced a new inspection system in January 2011, whereby each ship is ranked as high, standard, or low risk; this will determine the frequency of inspections. These changes intend to prevent low-risk vessels from being overly inspected in order to release resources for more frequent inspections of high-risk vessels. With the implementation of this new system, it remains to be seen if it will capture those vessels that deliberately avoid inspections.

22

Safety responsibilities in maritime industry

The company and crew management Although the overall responsibility for ensuring compliance with international regulations belongs to the flag state, each shipping company has the primary responsibility for the safety of their ships and crews. The ISM Code requires shipping companies to develop, implement, and maintain a safety management system for this purpose. However, safety management is not only a system property; system efficiency is determined by its human interrelationships. On board the vessels, the crew is the ultimate asset to ensure safety at sea, which is dependent on their experience and competence. The recession of the 1980s brought about several structural changes apart from flagging out—namely, the establishment of manning and crewing agencies. Crew management normally involves finding, organizing, paying, and training crews. In order to survive financially, some companies turned to managing ships for other owners as a means of utilizing spare management capacity. Others found it necessary to turn to crew managers in order to hire cheaper crew in other and unfamiliar parts of the world. This has resulted in ships being crewed by mixed nationalities working under different contracts and employment terms. The sudden switch to employing seafarers from nations without maritime traditions is claimed to result in a reduction in standards of competence, except from those relatively unusual cases where the shipowners invest in training. In addition, within crew management, there is variability in performance. At one end are those who have become seriously involved in training, some with their own training facilities and with established systems of testing crew competence (Alderton, 2004). On the other end are those who do not—nor intend to—perform such control over crew competence, which is quite a 23

Safety responsibilities in maritime industry problem, as fraudulent certificates of competence are an issue within the industry (IMO STW 41/4, 3, 2009). As pointed out by Anderson (2003), one might question if crew from manning agents takes the companies’ (safety) goals and objectives to heart due to a lack of ownership and short employment. When crew management is carried out by an external party, the shipowner will lose control over assessing and ensuring qualifications, training, and competence. The shipowner then depends on third-party qualifications, thoroughness, and follow-up when providing crew.

2.3 When the regulatory framework and safety management fails Major accidents are valuable sources of information about the regulatory framework, organizational practices, and cultures and in which way these impact safety. In order to illustrate how the regulatory system may fail, the explosion and sinking of the chemical tanker Bow Mariner is further outlined. The Bow Mariner was one out of four3 chemical tankers that exploded during a six-month period between December 2003 and June 2004. The Bow Mariner case is interesting for several reasons. First, the accident occurred after the introduction of the ISM Code. Second, the accident investigation of Bow Mariner indicated that safety culture and poor safety management are explanatory factors of the accidents. Finally, what happened aboard the Bow Mariner is a driving force for further amendments of the international regulations, concerning shipboard leadership and managerial skills (IMO, 2007a; IMO, 2007b). Based on this situation, the Bow Mariner is seen as a

3

The other three were the tankers Sun Venus, Panama Serena, and NCC Mekka.

24

Safety responsibilities in maritime industry case suitable for the purpose of understanding and exemplification. All factual information derives from the official accident investigation report (United States Coast Guard, 2005). On February 28, 2004, the chemical tanker Bow Mariner exploded and sank in the seas outside Virginia, United States, causing the death of 21 crewmembers. The Bow Mariner, owned by the Norwegian company Odfjell Tankers, was flying a Singaporean flag, was operated by the Greek company Ceres, and was manned by Greek officers and Filipino crew. The vessel had a valid Safety Management Certificate (SMC). As part of the investigation, the vessel’s inspection history for a five-year period before the explosion were reviewed and found to be unremarkable. However, during an internal audit in June 2003, 25 observations were recorded, including one pertaining to the failure to complete an enclosed space entry permit and another for failure to record training. These latter non-conformities were also present during the accident and pointed to as possible influencing factors to what happened. The accident investigators point to numerous indications that the ISM Code requirements were not fully implemented or functional aboard the vessel, despite apparent documentation of full compliance with the code. These ISM non-conformities were found to contribute to the accident. Amongst others, no crew familiarization with the vessel was conducted. It is also explicitly stated that the shipboard social culture and safety culture contributed to the occurrence of the accident. This included poor shipboard management. Aboard the Bow Mariner, the Greek captain had—in accordance with the company policy—full authority over all personnel. Such full authority is not unusual aboard a seagoing vessel. However, at the Bow Mariner the distinctions between Greek and Filipino nationality were 25

Safety responsibilities in maritime industry remarkable. Filipino officers did not take their meals in the officers’ mess, and the Filipino crew were given almost no responsibility and were closely supervised in every task. The Filipino crew were simply doing what they were ordered to do. As a result, they gained little knowledge about important aspects of their jobs. The lack of technical knowledge and fear of the Greek senior officers provide an explanation as to why the Filipino crew did not question the masters’ unsafe order to open all the empty tanks, which was a significant breach of normal safe practices for such ships. If the tanks had remained closed, the explosions would not have occurred. However, as stated in the company policy, the captain’s orders should never be questioned, and the failure to obey orders was a reason for disciplinary actions. Investigations of the accident leave no question that such fear of the ship management or senior officers can lead to a shipboard culture where safety takes a backseat to preserving one’s employment. Interviews with crew from another Ceres vessel indicate that this poor culture was the general rule in the entire Ceres company. With the case of Bow Mariner, it is evident that the ISM Code did not generate the intended outcome—namely, safe operations and a good safety culture—as a result of reasons not necessarily related to shortcomings in the ISM Code itself. This situation can also be related to the lack of ability to reveal the onboard conditions, which are related to coastal administration and inspections. Despite documentation and certification confirming full compliance with the code, the accident investigators pointed to numerous indicators that the code was neither fully implemented nor functional aboard the vessel. The onboard situation is created by organizational factors (e.g., crewing and shipboard management policies). The accident investigators also regarded commercial pressure as a contributory factor to what happened. 26

Safety responsibilities in maritime industry

2.4 Safety responsibilities in the maritime industry— a summary Thus far, the safety responsibilities in the maritime industry have been shortly outlined and analyzed. The situation is summarized in the following:  With the UN delegation of authority, IMO is responsible for developing and maintaining workable safety regulations and laws regulating ship safety.  The maritime administrations encompass flag state and coastal state. The coastal state is responsible for enforcing maritime regulations while the flag state is responsible for ensuring compliance with international regulations.  The shipping companies have the primary responsibility for the safe operations of ships and crew safety. Safety management is regulated through the ISM Code, which is developed by IMO.  The crew is the ultimate asset for ensuring safety at sea. Shipping companies may ensure safe operations by investing in crew training and competence and ensuring that crew experience is made use of in the company safety management system. When ensuring operational safety, the crew relates to and is influenced by all these actors and levels of authority. National and international legislations represent minimum standards. Beyond the minimum standards, each shipping company determines the crew’s working conditions. On each vessel, the framework given by the shipping company is moderated by the ship’s

27

Safety responsibilities in maritime industry management. Safety culture and safety management in theory and practice are further elaborated in the following chapter.

28

Safety culture and safety management in theory and practise

3 Safety culture and safety management in theory and practice This section focused on specified research aims 2 and 3, formulated as follows: 2. To outline and discuss relevant theories of safety culture and safety management and analyze the relationship between safety culture and safety management. 3. To give reason for a methodological framework for assessment of safety culture in relation to safety management. The concept of safety culture as a term and an explanatory factor in an accident investigation was first used by the International Atomic Energy Agency (IAEA) International Nuclear Safety Advisory Group (INSAG) following the Chernobyl accident that occurred on April 26, 1986 (IAEA, 1991). The Chernobyl accident occurred while a test was being performed on a turbine generator during a normal, scheduled shutdown of one of the reactors. At the time, written test procedures were unsatisfactory from a safety point of view. In addition, serious violations of basic operating safety rules were present, as the operators deliberately withdrew most control rods from the core and switched off important safety systems (The United Nations Scientific Committee on the Effects of Atomic Radiation, 1988). Both safety management and the interrelationship with the human factors and human error were brought into the safety culture concept and safe operations. Safety culture was defined as both attitudinal as well as structural, relating to both the organizational framework and structures along with the attitude of

29

Safety culture and safety management in theory and practise employees at all levels in responding to and benefitting from the framework (IAEA, 1991). More recently, several diverse definitions of the safety culture concept have abounded in the safety research and organizational literature (Guldenmund, 2000; Sorensen, 2002; Wiegmann, Zang, von Thaden, Sharma, & Mitchell, 2002a). In general, all the conceptual definitions can be placed in two broad categories: the socio-anthropological and the organizational psychology perspective (Wiegmann et al., 2002b). One difference between these perspectives concerns the conceptual definition, which is also reflected in methodology. From the socio-anthropological perspective, it is argued that a superficial research model of culture should be avoided in order to build cultural research on a deeper, more complex anthropological model. From an anthropological perspective, the practice of ethnography and fieldwork, qualitative in-depth studies with data deriving from interviews, observations, and/or participation is commonly accepted as appropriate research methods. Within this scientific direction, culture is described in text with an emphasis on the organizational member’s subjective interpretation and sense making. From the organizational psychology perspective, it is argued that culture can be described with a limited number of dimensions, usually sought through large organization-wide questionnaire surveys. From this latter perspective, the culture concept is assumed to express itself through an organizational climate—a set of perceptually based psychological attributes (Guldenmund, 2000). Another important difference between the two directions concerns their view toward cultural change. The socio-anthropological direction considers organizational culture to be an “evolved construct” deeply rooted in history, 30

Safety culture and safety management in theory and practise collectively held, and sufficiently complex to resist any attempt at direct manipulation (Mearns & Flin, 1999; Wiegmann et al., 2002b). In contrast, the organizational psychologists regard culture as changeable and tend to focus on its functional significance and the means by which it may be manipulated to improve productivity and safety (Wiegmann et al., 2002b). The organizational psychology perspective provides a conceptual bridge between safety culture, safety behavior, and organizational safety management systems, with the aim of controlling, guiding, or directing first-line operators’ attitude and behavior toward safe operations. The concept of safety culture—and climate—has over time been a theme of heated discussion, with little theoretical consensus emerging on the ontological, epistemological, and methodological questions relating to the subject. The main differences in these questions seem to be: (1) What is the scope of safety culture and the relationship between culture and climate? (2) How does the concept relate to other organizational aspects and outcome? (3) Which methods are most suitable for measurement? (Peterson, Ashkanasy, & Wilderom, 2000). These fundamental questions have already been elaborated upon

by

many

researchers

(e.g.,

Antonsen

&

Norges

teknisk-

naturvitenskapelige universitet. Institutt for sosiologi og statsvitenskap, 2009; Cooper, 2000; Glendon & Stanton, 2000; Guldenmund, 2000; Håvold & Nesset, 2009; Olsen, 2009; Sorensen, 2002; Tharaldsen, 2011; Wiegmann et al., 2002a; Wiegmann et al., 2002c; Zhang, Wiegmann, von Thaden, Sharma, & Mitchell, 2002). Based on his review, Guldenmund (2000) pointed out that most of the characteristics given to culture equally apply to climate, and within recent research it is more commonly accepted that climate is a reflection of an underlying culture. Hale (2000) even proposed that one should stop talking about safety culture completely and instead talk about 31

Safety culture and safety management in theory and practise (organizational) cultural influences on safety (Hale, 2000). In order to grasp as many facets as possible of the safety culture concept, a multi-method approach is needed. As the safety culture-climate debate seems to be settling down and has already been thoroughly discussed by many, the concept will in this thesis only be touched upon in brief. A general working model used in the studies of safety culture and safety management in this thesis is shown in Figure 3.

Figure 3: The general working model used in the studies of safety culture and safety management

With reference to Figure 3, organizational culture/climate is seen as an integrated concept subject to change by organizational management practices and structures. It is assumed that both organizational cultural and managerial features influence safety, which is defined as safety culture. As such, organizational safety culture is perceived as a concept with integrated parts from organizational management system practices and organizational culture/climate. The organizational safety culture is assumed to reflect the status of safety in the organization. Organizational safety, or safety culture, is assessed using two measurement outcome variables: risk perception and the 32

Safety culture and safety management in theory and practise state of the safety management system (SMS). The various concepts and relationships shown in Figure 3 are further elaborated upon in the following three sub-chapters.

3.1 Safety culture as an organizational integrated concept Within the field of organizational safety, the climate concept was first introduced by Zohar in 1980 (Zohar, 1980). In more recent publications, Zohar relates safety climate to an overall organizational climate made of shared perceptions among employees concerning the procedures, practices, and kinds of behavior that are rewarded and supported with regard to a specific strategic focus. When the strategic focus involves the performance of high-risk operations, the resultant shared perceptions define safety climate (Zohar, 2010). Although climate and the underlying culture may have a particular referent as safety, they embrace and are influenced by more than a single unit or function in the organization. An organization has multiple goals and multiple policies that are all manifested in organizational behavior and practices. The different goals are often in competing conflict, like profit and safety (Hollnagel, 2004; Hollnagel, 2009). For example, the crew may be expected to cut corners and work faster without getting crossing prevailing rules and regulations or jeopardizing safety. The safety climate concept integrates perceptions toward the organization’s total contexts as a regulatory framework and competitors as well as internal matters as finance, marketing, human resources, control systems, safety management systems, and so on. Consequently, when measuring climate with a particular reference, it is important to embrace the organization in a wider sense in order to reveal conflicting areas and the priority of importance. True priorities at work (e.g., 33

Safety culture and safety management in theory and practise efficiency versus safety) have been shown to provide the strongest prediction of actual behavior (Zohar, 2008). The theoretical roots of the safety culture discussion can be traced back to Barry Turner and the introduction of the manmade disaster model (Pidgeon & O'Leary, 2000; Turner, 1978). In the manmade disaster model, an accident is defined not by its physical impact, but in sociological terms, as a significant disruption or collapse of the existing cultural beliefs and norms regarding hazard. These cultural beliefs and norms are assumed to be formally laid down in rules and procedures or more tacitly taken for granted and embedded within working practices. This is also related to managerial and organizational practices. Andrew Pettigrew (1979), whose background is in anthropology and sociology, relates the cultural concept to the everyday tasks and objectives in organizations as a product of social processes connecting the past, present, and future. In many ways, Pettigrew’s definition encompasses Turner’s definition. According to Pettigrew (1979), culture is related to the less rational and instrumental tasks in an organization as well as the more expressive social tissue that give those tasks meaning, such as the meaning of having a safety policy, if procedures should be followed only when safety inspectors are present, or if efficiency is the real area of priority and the safety-first policy only serves as a function for external stakeholders. It is argued that, in order for people to function within any given setting, they must have a continuing sense of what reality is all about in order to be acted upon. In this setting, culture is the system of such public and collectively accepted meanings with regard to safety, operating for a given group at the time (Pettigrew, 1979).

34

Safety culture and safety management in theory and practise One of the most influential anthropologists of modern time, Clifford Geertz, regards culture as “webs of significance” spun by man and in which man are suspended. According to Geertz (1973), cultural scientists should try to interpret those webs in search of meaning and explanation. Geertz (1973) concluded that culture is most effectively treated as a symbolic system. By isolating the elements of the symbolic system, specifying the internal relationship, the whole system may be characterized in general. Symbols are the surface expression of the underlying cultural structure. Pettigrew (1979) also emphasized symbols, languages, ideologies, beliefs, rituals, and myths as an important part in the codification of meaning and emergence of normative patterns. For example, in accordance with requirements of the ISM Code, shipping companies have safety policies. From a cultural perspective, policies such as safety first are not important in themselves. The importance stems from how such policies, as a symbol, make sense for the organizational members. A safety-first policy may be perceived as a statement aimed to attract customers or directed toward other external stakeholders. In the case of the Bow Mariner, it is likely that the company safety policy would be given meaning as a façade maintained toward external stakeholders, which can explain the actual safety-degrading behavior on board. The vessel did hold a valid ISM certificate and documentation, but during normal operations they were not acted upon and complied with. This understanding of culture also establishes a relation between culture and climate—namely, that cultural beliefs and meaning given to organizational factors are reflected in actual behavior. In this particular case, the symbolic system is explained by the gap among safety policy, guidelines, and actual behavior. One of the most widely used organizational culture frameworks is probably that of Edgar Schein (1992, 2004), a framework that built upon Pettigrew’s 35

Safety culture and safety management in theory and practise cultural theory. This framework explains culture at three levels: (1) artifacts— visible organizational structures and processes that are difficult to measure but are felt and heard by individuals who enter a new culture; (2) espoused values—norms, standards, and moral principles usually measured through questionnaire surveys; and (3) basic underlying assumptions—unconscious taken-for-granted beliefs, perceptions, thoughts, and feelings, which may be understood by ongoing observations and participation. Schein (1992, p. 12) formally defined culture as: a pattern of shared basic assumptions that the group learned as it solved its problems of external adaption and internal integration that has worked well enough to be considered valid and, therefore, to be thought to new members as the correct way to perceive, think, and feel in relation to those problems. Schein regards culture as shared assumptions expressing consistent, clear, and organization-wide consensus. However, in looking for organization-wide consensus, important areas of conflict will be disregarded and lost. What is regarded and learned as valid patterns of shared assumptions depends upon how the group is defined, which in turn opens up for the existence of various groups and subcultures within an organization. Regarding the Bow Mariner, it is natural to assume that the group of Greek officers and the group of Filipino crew differ with regard to what is shared, which also brings about the notion of subcultures. In general, it is assumed that subcultures are found within the shipboard departments’ deck, engine, and galley. Analyses of other levels in the organization may give different results, as the group may be defined as the whole fleet or as all shipping 36

Safety culture and safety management in theory and practise companies. A cultural trait identified at the industry level may be related to manning policies and the extended use of manning agencies and contract employment, which most regard as the only possible solution to manning—a solution that is taken for granted and not questioned by insiders, thereby determining national subcultures. When distinguishing climate from culture, climate is often suggested to arise from individuals whereas culture is suggested to arise from group or interpersonal processes (Dansereasu & Alutto, 1990) Safety culture is assumed to be influenced by and seen as an integrated part of an organizational culture and a product of equal processes. Organizational culture is a relatively stable, multidimensional, holistic construct shared by groups of organizational members that supply a frame of reference. It gives meaning to and/or is typically revealed in certain practices manifested as organizational climate. In the same way, Mearns and Flin (1999) described safety climate as employees’ perceptions, attitudes, and beliefs about risk and safety whereas safety culture is a more complex and enduring trait reflecting fundamental values, norms, assumptions, and expectations. These cultural elements can be seen through safety management practices, which again are reflected in the safety climate and in actual behavior (Mearns & Flin, 1999). In conclusion, most of what is true for safety culture is also considered true for safety climate (Guldenmund, 2000).

3.2 Organizations, management and cultural change Safety management and safety culture are all about change—a change toward enhanced operational safety. Both Pettigrew (1979) and Schein (2004) regarded individuals, including entrepreneurs and leaders, as important in the process of creating and managing an organizational culture. According to 37

Safety culture and safety management in theory and practise Schein, one of the most decisive functions of leadership is the creation, management, and sometimes even the destruction of culture. Although Schein regarded leaders as important in these processes, leaders are not regarded as the only determiner of culture. Schein also stated that culture is a result of complex group of learning processes only partially influenced by leaders. Groups (e.g., departments or vessels that operate within similar situations) may behave very differently from one another as the group dynamics differ. Schein (2004) referred to culture as those elements of a group or an organization that are most stable and least malleable. However, according to Schein (2004), the group itself needs a certain degree of stability. Any group with a stable membership and a history of shared learning will have developed some level of culture, but a group having either a great deal of member turnover and/or a history without any challenging events may lack shared assumptions. Zohar (1980) identified a stable workforce with less turnover and older workers as an organizational characteristic when determining safety climate. A stable workforce is then vital for both climate and culture. Within the segments of shipping that have large turnover and less group stability, it might be questioned if they have developed any shared basic assumptions (culture) or working practices (climate). According to Schein (2004), some barriers to the development of an integrated shared culture exist, including the insufficient stability of group membership and the insufficient shared history of practice or the presence of many subgroups with different kinds of shared experience. This may lead to a situation of ambiguity and conflict. Joanne Martin (Martin, 1992) distinguished three perspectives: (1) integration, (2) differentiation, and (3) fragmentation. Martin’s (1992) recommendation is that an organization be viewed from all three perspectives, as one perspective’s strengths are 38

Safety culture and safety management in theory and practise another’s weaknesses. As a result, a greater understanding of an organization’s culture and how to approach cultural change may be obtained. From each perspective, different aspects of culture and cultural change are captured. This view was also supported by Alvesson (1993). From the integration perspective, culture is described as patterns of manifestations shared by all members of the organization, and the organization’s manager/leader is regarded as the primary source of cultural change. One of the most renowned representatives within the integration perspective is probably Edgar Schein (Frost, 1991; Richter & Koch, 2004; Schein 2004). According to Schein’s theory, the organization’s leader is regarded as being able to create and manage the culture in pre-given directions. The development of subcultures is regarded as undesirable side effects appearing when the organizations grow and mature. In such a situation, Schein argued that the leader’s effort should focus on integrating the variety of subcultures. From the differentiation perspective, cultural manifestations are described as sometimes inconsistent when consensus occurs only within the boundaries of subcultures, which often is regarded as being in conflict with each other. With regard to influence and change, greater importance is assigned to the environment, and teams of leaders are ascribed to have secondary influence to cultural change. Nick Pidgeon (1998) considered the importance of subcultures, questioning whether a unified culture may be designed within a large organization and arguing that existing subcultures should be given attention regarding how they differ in (safety-related) priorities, perceptions, and interpretations of emerging safety problems. How these aspects interact with each other, existing power relations, and the like is of equally great importance (Pidgeon, 1998). 39

Safety culture and safety management in theory and practise While both the integration and differentiation perspectives focus on what is shared and accounts of planned and directed goals change, the fragmentation perspective regards ambiguity as the essence of culture. In this perspective, culture is something that is constantly fluctuating, and no stable organizationwide or subcultural consensus is supposed to exist. Cultural change is seen as being in a constant flux, where the power of change is usually seen as being diffused broadly in the environment and among organizational members. The ambiguity is supposed to emerge from the complexity and unpredictability in the organization and society; the fragmentation perspective does not assume that the organizational members have similar reactions to these ambiguities. Karl Weick and the theories of high reliability organizations (HRO) are well known within this field (Weick & Sutcliffe, 2007). Within

shipping,

subcultures,

conflicts,

ambiguity,

stress,

and

misunderstandings are likely to be present due not only to the lack of stability of membership, but also to an insufficient shared history of practice. This relates to both the specific work situation, but also, amongst others, to nationality. National cultures are known to differ in aspects such as power distance and the degree of human inequality, uncertainty avoidance, and how they adapt to unstructured situations as well as how they integrate into a group with regard to individualism versus collectivism (Hofstede, 2001). In addition, value of life, safety standards, and risk perception are known to differ between nationalities. Geert Hofstede perceived culture as mental programming—a pattern of thinking, feeling, potential acting, and unwritten rules of the social game that distinguishes the members of one group of people from others. A certain culture is learned through the lifetime. What is acquired in early childhood and once established is difficult to change (Hofstede & Hofstede, 2005). Hofstede argued that layers of culture (e.g., 40

Safety culture and safety management in theory and practise organizational culture) acquired later in life tend to be more changeable. Organizational acquirements and practices—the visible part of culture—are regarded as faster and easier to change. Based on the previous discussion, it is concluded that organizational safety culture and behavior are subject to change by organizational practices and structures, which includes safety management systems. Furthermore, individuals in the organization (e.g., their leaders) do have a mediating effect upon the formation of an organizational safety culture.

3.3 Organizational culture and safety management Safety management systems have two interrelated main functions: to avoid accidents and improve safety. Theories of accident causation and safety management have progressed over time (Borys, Dennis & Legget, 2009). H. W. Heinrich is considered to be a pioneer within safety management and accident causation (Heinrich, Roos & Petersen, 1980) from the first age of safety—namely, the technical age. With reference to the shipping industry, the first development of SOLAS belonged to a technical age; the introduction of the ISM Code represented a transition to the age of management systems and culture. The age of human factors occurred in between, in which the view of human error considered the interaction of human and technical factors when exploring the causes of errors and accidents (Borys et al., 2009). As previously highlighted, traditional safety management systems (SMS) described in the ISM Code fall along a linear causality where attempts are made to predict and prevent future incidents by reflecting upon previous experience related to empirical safety control. The traditional SMS contains several subsystems. First, a system for reporting and collecting experience data from the vessel itself is required, followed by a system of data processing 41

Safety culture and safety management in theory and practise (i.e., summarization and analysis) in order to reveal causal factors and perform trend analyses, which form the basis for the development of safety measures. Both identification of causes and remedial actions are closely intertwined

with

how

the

organization

addresses

technological,

organizational, and individual factors. Irrespective of this, one critical system requirement is the reliability and accuracy of input data (i.e., near miss and accidents reports). As long as the input is reliable, the overall system presupposes the possibility of revealing a root cause and develops efficient measures in order to control operational safety (KjellĂŠn, 2000). Although the fundamental rationale of safety management has changed little over the years, the rationale for the understanding of human error and causal factors, root causes, and adequate safety measures has changed. However, quite surprisingly, recent research also points to insufficient scientific evidence on the effectiveness of systematic safety management to make recommendations either in favor of or against them (Robson et al., 2007). Thus, the obvious question is why: Is the reason found within the theoretical rationale of safety management or within organizational understanding and applications of the systems or is the reason found elsewhere? The application of traditional safety management is questioned with a distinction between small-scale accidents and larger more infrequent accidents (Rundmo, 1996). According to Rundmo (1996), the empirical safety control, in which measures are developed through the analysis of past events, is only applicable for frequent and small-scale accidents such as ordinary work accidents. When it comes to medium-size and more infrequent accidents (e.g., groundings and collisions) and large-scale accidents with very low probabilities (e.g., the capsize of the Herald of Free Enterprise and the fire at the Scandinavian Star; see Table 1), traditional safety management is not 42

Safety culture and safety management in theory and practise considered to be applicable. Such accidents are often too unique and complex to grasp, and it will not be possible to single out some isolated underlying causes or develop measures that cover all involved risk—a risk that in the first place is considered too complex to understand. Others, such as Scott Sagan (1993) and Charles Perrow (1999), questioned the possibility of foresight and preventions of accidents through empirical safety management. Within the age of human factors, human error is regarded as primary cause of accidents. In more recent theories and subsequent ages of safety, human error is seen as a consequence of faults deriving from other parts in the organization or environment, complexity, interactions, and/or organizational culture. A big difference between human error as a cause and human error as a consequence is seen in the characteristics of remedial actions. With human error cited as the cause of failure, the tendency for safety measures is to seek to control human behavior with inter alia procedures and checklists. Then, when the real cause is found to lie elsewhere in the organization, such measures may not clearly be the answer to the underlying problem and incidence of failure, and accidents will continue to occur. The following sections shed light on these issues from the theoretical perspective of accident causation and prevention.

The Man Made Disaster model Barry Turner is presumably among the first to regard latent conditions as a primary cause in accident causation. With the development and introduction of the Man Made Disaster model, accidents and disasters are proposed to develop through a long change of events leading back to root causes like lack of information and misperception among individuals (SINTEF, 2003). Turner argued that this is a result of an organizational culture where information and interpretations of hazard signals fail. Thus, a typical accident can be traced 43

Safety culture and safety management in theory and practise back to initial beliefs and norms—culture and climate—that do not comply with existing operational reality. From this perspective, accident development is viewed as a process, often over years, developing from an interaction between human and the organizational arrangements of the socio-technical system (SINTEF, 2003). With reference to accidents such as the capsize of Herald of Free Enterprise, a common understanding of the crew—given by the management of the organization—was a general understanding that the vessel should leave some minutes ahead of schedule, even if it involved putting pressure on those who did not move fast enough. The inherent risks of such a practice were not questioned. From the Man Made Disaster perspective, systematic safety management should deal with these breakdowns in the interpretation of information. For example, are some danger signals or causes systematically disregarded or misunderstood? As with the Bow Mariner, why did none of the involved stakeholders (e.g., charterers, vetters, inspectors, and flag state) manage to reveal the degrading shipboard safety situation? As such, existing cultural beliefs and norms are the essence of the Man Made Disaster theory. According to Turner, such cultural beliefs and norms might be formally laid down in rules and procedures or more tacitly taken for granted and embedded within working practices. An accident is then assumed to occur because of inaccuracy or inadequacy in the accepted norms and beliefs and of a discrepancy between the way the world is thought to operate and the way it really operates (Pidgeon & O'Leary, 2000). When acknowledging the weaknesses in traditional safety management and failure of foresight, these institutional barriers to effective learning should, according to Pidgeon and O’Leary (2000), be addressed. The aim for efficient safety management should be to let all organizational members develop a safety 44

Safety culture and safety management in theory and practise imagination that breaks the pattern of becoming overly fixated with prescribed patterns, simplification, and ignorance. Pidgeon and O’Leary (2000) presented seven guidelines for fostering a safety imagination: (1) attempt to fear the worst, (2) use good management techniques to elicit varied viewpoints; (3) play the “what if” game with potential hazard; (4) allow no worst case situation to go unmentioned; (5) suspend assumptions about how the safety task was completed in the past; (6) approach the edge of a safety issue with a tolerance of ambiguity, as newly emerging safety issues will never be clear; and (7) force oneself to visualize “near miss” situations developing into accidents. It is argued that such safety imagination is a critical facet of organizational learning and an effective safety culture.

Normal Accident Theory The Man Made Disaster theory also highlights how system vulnerability arises from unintended and complex interactions between contributory preconditions (Pidgeon, 2000), which may be linked to Charles Perrow’s theory of normal accidents (NAT). However, in contrast to Barry Turner, Perrow regards some systems to be to complex and interactive to avoid organizational accidents completely. In such systems, safety management is regarded as futile as accidents are doomed to happen due to the system characteristics. The development of NAT started with Perrow’s exploration of the 1979 accident at the Three Mile Island nuclear plant. During this investigation, Perrow was struck by the fact that the present accident literature overwhelmingly blamed the operators (Perrow, 1999). Perrow then looked into accident reports from various industries, such as mining, aircraft, and marine accidents, and evolved the alternative theory that risk is a result of two 45

Safety culture and safety management in theory and practise dimensions—interactions and complexity—rather than human error. It is argued that the operator is free from blame as the overall system, interactions, and interdependencies of events are incomprehensible for a critical period of time. Like Rundmo (1996), Perrow distinguishes between small-scale, frequent personal accidents and other medium-sized accidents and larger-scale accidents. Risk is, according to Perrow (1999), regarded as something designed into organizations in the form of complex systems with tight couplings. The paradox is, when barriers and other safety measures are built into the system to increase safety, often it is the complexity that is increased at the cost of safety. Perrow (1999) regards such error-prone organizations as impossible to manage safely in the long run, thereby resulting in the notion of normal accident. In the maritime setting, Perrow (1999, p. 230) described the risk-inducing complexity as follows: The ship itself, with its power plant explosive mixture, steering apparatus, and draft in shallow channels is important, but so are other ships, the insurance industry, the fragmented shipping industry, attempts are regulation, rules of the road, dangerous cargoes, national jealousies and interests, and, of course the horrendous environmental problems of fog, ice, and storms. In his analysis, Perrow also pointed to conflicting organizational goals, production pressure, and organizational pressure related to risk taking, authoritarian structure on board, and inappropriate leadership. One of Perrow’s (1999) main points was to regard all human constructions (e.g., vetting, class, regulatory bodies, flag state, insurance, manning companies) as systems and not as collections of individuals or representatives of ideologies. Dangerous accidents lie in these systems based on how the different parts fit 46

Safety culture and safety management in theory and practise together and how they interact. From this point of view, safety management is about reducing the system complexity and/or loosening the couplings to reduce interactivity. From the perspective of NAT, Turner’s notion of “safety imagination” is a dead end for organizational safety as long as complexity, incomprehensible interdependencies, and tight couplings are present.

High Reliability Organizations Theory The High Reliability Organizations Theory (HRO) developed as a result or a continuance of Perrow’s rather pessimistic message—namely, that accidents are inevitable in some systems or organizations due to their characteristics (Roberts, 1990). The concept of safety culture constitutes a central difference between NAT and HRO. Whereas NAT argues that in some systems accidents are inevitable, HRO argues that even in the most vulnerable and error-prone systems, safety culture has characteristics that can counteract the inherent system risk. In many ways, HRO is in line with the arguments given by Turner, Pidgeon, and Perrow. HRO recognized that everything that may fail during operations has not yet been experienced; therefore, a system based on experience feedback (as the ISM code) is doomed to fail on its own premises. HRO also recognizes that not all incidents may be deduced to detect all possible failure or error modes. Recognizing that the world is complex, unstable, unknowable, and unpredictable, the HRO perspective maintains reluctance to the simplification inherent in traditional risk assessment. Both procedures and checklists may represent simplification of measures when a complex, unstable, unknowable, and unpredictable working situation is attempted to be controlled by preplanned prescriptions. This view is shared by Perrow, and

47

Safety culture and safety management in theory and practise both NAT and HRO are skeptical towards those who heavily rely on risk assessment. In HRO, more attention is given to the real work going on in frontline operations. By empowering those doing the actual work, operators have the possibility to solve the situations themselves, based on their own experience and knowledge. This is a contrast to traditional safety management from the era of human factors with a focus on the control of human behavior. Within the HRO paradigm of HRO, frontline personnel operate by using pre-planned descriptions, but it is accepted that in real situations deviations will occur. One of the HRO’s key points is mindfulness, which is related to the concept of safety culture and similar principles such as Turner’s safety imagination. Mindfulness is understood as a combination of alertness, sensibility, flexibility, and adaptability. This perspective argues that unexpected events should be handled by creating a mindful infrastructure by following five main principles:

(1)

continuously

tracking

small

failures,

(2)

resisting

oversimplification, (3) being sensitive to operations, (4) maintaining capability for resilience, and (5) monitoring the shifting locations of expertise. The violation of these principles is regarded as a setback toward the more traditional approach, where simple diagnoses are accepted, frontline expertise is overridden by faith in risk analysis, and safety measures are developed detached from operations. The HRO view implies that the operators gain more responsibility, so other parts of the organization have to give them the possibility to act. In other words, control is taken from the upper levels of the organization in favor of lower levels. Within HRO, safety culture is seen as essential in managing risk. All of Johanne Martin’s three perspectives of safety are adopted: “Each form of culture handles ambiguity differently: Integration denies it, differentiation 48

Safety culture and safety management in theory and practise selectively clarifies it, and fragmentation accepts it. In a mindful culture, all three forms of culture are present� (Weick & Sutcliffe, 2007, p. 112). HRO does not reject the fact that organizational members have shared values and beliefs. However, with regard to safety these shared patterns are not regarded as vital for the outcome. The shared orientations are accommodated differently in all situations, and the chain of events and patterns of interactions between people fall under the influence of situational conditions as stress, misunderstandings, interpretation, and others conditions specific to each chain of event. From an HRO perspective, more weight is placed on fragmentation than on differentiation and more on differentiation than on integration.

Managing risk and safety culture James Reason (Reason, 2001) developed a widely used practical definition and approach to safety culture. Barry Turner and the Man Made Disaster model provided much of the conceptual foundation for Reason’s work (Pidgeon & O'Leary, 2000). Reason’s approach to safety management and safety culture is also to a large degree adopted by the HRO perspective (Weick & Sutcliffe, 2007). Based on the organizational culture literature, Reason (2001) differs between two theoretical stands: those who regard culture as something an organization has and those who regard culture as something the organization is. Reason favors the former approach and thus regards culture as something changeable and manageable by organizational practices. Like Turner, NAT, and HRO, Reason is also concerned about the organizational factors that trigger an accident. According to Reason (2001), accidents by their nature are not directly controllable, as many of the causal influencing factors lay outside organizational control and influence. As such, 49

Safety culture and safety management in theory and practise rather than struggling vainly to exercise direct control over behavior, accidents, and incidents, organizational managers should measure and improve the processes of underlying factors, such as training, procedures, planning, budgeting, goal conflicts, and others. Thus, an efficient SMS system should help identify those conditions most needing correction and not be limited to non-compliance of global rules. Attention should also be directed toward the quality of these global rules (e.g., accuracy, relevance, availability, and workability of procedures). The information reported into the system should embrace organizational factors as well as local workplace factors and unsafe acts. The cultural factor is linked to commitment, competence, and cognizance within the organization as a whole. Reason (2001) regards safety culture as a cornerstone in efficient safety management in order to get the needed operational information. He identifies safety culture using four critical subcomponents: (1) a reporting culture, (2) a just culture, (3) a flexible culture, and (4) a learning culture. Together, these interact to create an informed culture. According to Reason, an informed culture is one that collects operational experience data that are characterized by an organization’s climate in which members feel free to report without experiencing negative, unfair, or in other ways meaningless consequences. Attention is also given to how the interpretation of information and outcome is influenced by the overall company policy. Reason is more preoccupied with organizations’ internal organizing than how safety is influenced by other organizations and macroeconomics than by national and international conditions such as politics, laws and regulations.

50

Safety culture and safety management in theory and practise

The theory of Practical Drift – why organizations fails Safety culture in shipping is often depicted side by side with compliance to prevailing procedures and other safety measures. However, if compliance is to be a valid key performance indicator, it is presupposed that these safety measures are appropriate for the actual action. The development of standardized measures fitting all real-life situations are a challenging if not an impossible task considering the complexity and unpredictability of most situations. This is also pointed out by NAT and HRO. The Practical Drift Model (PDM) provides an explanation for how and why organizations experience such gaps among standardized measures, real situations, and actions, referred to as practical drift (Snook, 2000). PDM combines HRO and NAT in two ways, emphasizing how different degrees of mindfulness will depend on different situations and contexts. During their lifetime, organizational systems develop both tight and loose couplings, which they shift in between—tight to loose couplings and back again—as the various sub-units within the operative part of the organization alternate between a low and high degree of interdependence. The model also captures both contextual and temporal factors when explaining why incidents and accidents occur, along with practical drift from the global rules, such as standards, procedures, and checklists. Standardized rules are often designed according to organizational lifecycle, where the governance structure is top-down oriented. Using organizational safety management systems, organizational managers put large effort into developing extensive routines and procedures in order to make the organization robust and resilient against future unforeseen events. When designed, tight couplings and rule-based action of logic is assumed to 51

Safety culture and safety management in theory and practise characterize the organization. This, in contrast to an operational situation, is assumed to be loosely coupled, and the real world does not act in accordance with the organizational design. When organizations experience that unforeseen events do not occur as expected, the attention toward the limitations and inadequacy in routines and procedures become a part of everyday life and practice and are therefore more relaxed. Others (e.g., Hollnagel, 2004) point to real-life work processes that are irregular and unpredictable in contrast to work regulations—either explicitly by procedures and instructions or implicitly by rules, standards, or good practice. Another issue arises when these rules are developed detached from the operational situations (e.g., in shipping by shore personnel); consequently, accuracy, relevance, availability, and workability of the rules may be low. Moreover, company policy may favor efficiency over safety. As a result, the sharp end operators—when aware of this situation—will be able to break the strict rules without fear of sanctions or punishment. On a local basis, breaking strict rules may actually get the job done quicker and more efficiently. The operator may be “rewarded” for achieving additional goals in the organization. During such a process of de-coupling, the organization will become increasingly free from global rationality. Subcultures with their own logic of action based on experiences and tacit knowledge develop, and the operators accordingly drift further away from a rule-based system to a more task-based system. According to PDM, accidents and incidents occur when the system suddenly and stochastically becomes tightly coupled, as with the Bow Mariner, Scandinavian Star, and Herald of Free Enterprise. In such situations, the involved operators are forced to act on the assumption that all others act in accordance with the original rules and procedures initially designed. The 52

Safety culture and safety management in theory and practise operators become trapped in a game in which trusting their own logic of actions is the only solution while they must simultaneously base their decision on the assumption that others are following the general rules. After an unwanted event, the outcome is often even more tightly designed control criteria. James Reason describes such an approach to safety management as a person-oriented approach, which may also be perceived as belonging to a blame culture, implying crew shortcomings as the cause of error. With human error cited as the cause of failure, the tendency is for safety measures to seek to control human behavior by developing more procedures and checklists. When the real cause is found to lie elsewhere in the organization, such measures may clearly not be the answer to the underlying problem and incidence of failure, and unwanted events will continue to occur. Global rules may be perceived as less and less meaningful; thus, constantly relying on them will further undermine the safety system as more procedures are violated and the local units drift even further apart from global rationality. Less reporting of experience data could also be a consequence in the longer run.

3.4 Safety culture and measurable outcome variables The IAEA has built its concept of safety culture upon Edgar H. Schein’s three-level model and regards safety culture as both attitudinal as well as structural relating to both the organizations and the individuals (IAEA, 1991). Based on this conceptual definition, INSAG and IAEA regard systematic safety management, which belongs to the first layer in Schein’s model (i.e., artifacts), as a tool for promoting a strong safety culture and achieving a good safety performance. This may be measured in the second layer, espoused

53

Safety culture and safety management in theory and practise values, using a questionnaire survey. The third layer, basic assumptions, may be captured by observation. Using previous research within the oil industry, Rundmo (1996) demonstrated that risk perception and risk behavior are significantly correlated, but also relatively independent from each other. The association between risk perception and risk behavior is then caused by the fact that the same predictor affects both variables. Crew risk perception and other subjective assessments are suggested as good indicators of the safety level, but not as predictors for risk behavior. It is further suggested that employees’ behavior to a great extent is constrained by the conditions under which they work. When the working conditions are not perceived to be satisfactory, employees know that the occupational risk is higher; they feel more unsafe, which will affect their risk perception (Rundmo, 1996). Risk perception as a measurement variable is supported by Zohar (1980), who used climate research to assume that an individual’s

perception

focuses

on

the

organizational

environment,

organizational control system, and safety management system. Drawing from the Man Made Disaster theory, safety culture is the product of cultural beliefs and behavioral norms regarding hazards, which are laid down in the organizational control system and thus reflected in procedures and guidelines. This is in accordance with Schein (1992, 2004), who regards organizational control systems (e.g., safety management systems) as a manifestation of basic underlying cultural assumptions. In this respect, companies approach to safety management is a manifestation of the underlying beliefs and thoughts on how safety management should be performed. Pettigrew (1979) relates the cultural concept to how everyday tasks and objectives in the organization are expressed as well as their meaning, including how the organizational members comprehend the 54

Safety culture and safety management in theory and practise importance of the organizational control system and derived safety measures in relation to other directions given by the company. This approach aligns with Geertz’s (1973) work referring to the social processes where meaning and sense making arise. Social processes are influenced by crew composition and human resource policies. Personal beliefs and values are also antecedents to behavioral norms. Traditional cultural researchers also regard beliefs and values as less changeable as they are acquired through a lifetime and therefore deeply rooted within the individual. However, Hofstede and Hofstede (2005) argue that beliefs and values learned through an organizational context are more changeable as they are acquired at a later stage of life. The organizational context and organizational control system are assumed to directly impact behavioral norms. When managing behavioral norms by means of an organizational control system, Kjellen (2000) defined the system as one that provides the information needed for safety and signaling related to health and safety matters. In this regard, Reason (2001) regarded safety culture as the cornerstone for ensuring the information flows as needed. Reason also addressed the safety management system as a whole, with regard to how the incoming experience data are analyzed and processed into safety measures. From this, the state of the safety management system is regarded as a measurement outcome variable.

55

Research methodology

4 Research methodology This chapter describes the research methods applied in this thesis. Based on the discussion in Chapter 3, climate will not be separated from culture, but in theory and research methodology the recommendation from both perspectives will be taken into consideration. This synthesis of qualitative and quantitative methods is regarded as important in order to understand how culture is created through social processes while quantitative methods simultaneously say something about how widespread certain patterns of behavior and perceptions are within the industry and statistical associations. The application of a multimethod approach is also supported by others. According to Cooper (2000), the triangulation of different methods allows the researcher to take a multifaceted view of safety culture, so that the interrelationships among psychological, behavioral, and situational factors can be examined with a view to establish antecedents, behaviors, and outcomes within the specific contexts. Triangulation allows the employment of each method’s strengths in order to overcome the others’ weaknesses. Both Rousseau (1990) and Schein (1992, 2004) identified that different layers of culture are amenable to different research methods. For example, the fundamental content of culture is assumed to be unconscious and highly subjective. The organizational members’ basic assumptions, values (what is important), and beliefs (how things work) as well as culture’s social construction are difficult to assess without interactive probing. Moreover, the organizational members’ fears and defenses are difficult to elicit without interaction, which gives reason for an ethnographic methodology. On the other hand, the organizational members’ patterns of behavioral norms (how things are done) are far more accessible to observation from outsiders and respond to structured instruments and quantitative methods. 56

Research methodology According to Geertz (1973), cultural analysis is (or should be) guessing at meanings, assessing guesses, and drawing explanatory conclusions from better guesses. However, the previously described examples illustrate how such guesses might be wrong if not related to individuals’ own experiences and national, situational, and/or historical context. Street Corner Society is a method of participant observation where becoming native, without being too attached, is a part of the research strategy (Whyte, 1991). In accordance with Street Corner Society, the researcher lives with the community in order to understand the nature of the field, learn to understand the group, and build trust and credibility. Whyte also demonstrated how such qualitative studies may be expressed and presented in a more quantitative format. Whyte has become a major spokesman for the advantages of integrating research methods, including those typically associated with quantitative research (Bryman, 1991). As such, there is no adversative relationship between cultural field studies from an anthropological perspective representing qualitative methods and the quantitative methods typically from the psychological perspective. From the author’s own experience, the ethnographic approach has been valuable not only to understand, but to correct misunderstandings. For example, when conducting the first field studies, one Philippine mate discussed her former Norwegian captain, who refused any crew members to whistle on the bridge. Both the Philippine mate and I interpreted this as an indication of authoritarian leadership. However, months later, after getting a better understanding of seamanship in a national historical context, I found that this interpretation was wrong and in fact related to superstition, which is quite common among the older generation of seafarers. In the days of sails, the seafarers needed wind, and they whistled to call for the wind. Nowadays, 57

Research methodology with engine propulsion, wind is no longer wanted as wind causes waves. Thus, whistling is not allowed as it calls for the wind. In other situations, misinformation may be a result of deliberately withholding information as the group being researched does not trust the researcher. In another field study, I observed that all engine crew wore helmets in the engine room, as required. However, after one week at sea, once I had become familiar with the group, I learned that usually no one wore their helmets. They only applied the rules when they had a third party on board—an outsider. Apart from this, also favoring an integrated use of methods, cultural interpretation, and theoretical development is suggested to follow its own unplanned course in the search of grasping and analyzing “the web of significance” and structure of symbols and meaning, plunging more deeply into the same ideas. These theoretical formulations do not make much sense or hold much interest apart from the context of interpretation. Indeed, a safety culture study carried out within the anthropological tradition alone could have brought descriptions, interpretations, and understanding of how safety is interwoven with symbols and cultural elements on a single vessel or in one department. However, the developed theories would not make any sense outside that unit. Thus, within the organizational aim of enhancing safe operations in general, this approach would not be very useful.

4.1 Quantitative research and questionnaire survey A quantitative design incorporating a questionnaire is used to develop an understanding of the manifestation of culture and to give direction for the subsequent qualitative studies. In order to grasp the underlying dimension of safety culture through the use of questionnaire, a high level of instrument quality—namely, reliability and validity—is required. 58

Research methodology The starting point of every questionnaire is item generation, concerning which questions and themes should be included. This stage is also related to content validity. Although Hinkin (1995) suggests a strong theoretical framework as a starting point, followed by a sorting process allowing for the deletion of conceptually inconsistent items, Guldenmund (2007) suggests two different approaches for questionnaire development. First, a descriptive model of the construct can be used as a starting point—namely, a normative or theoretical approach. Second, theories and results of previous research can be used in combination to construct a new questionnaire, which is a more pragmatic approach. However, some inherent difficulties exist in both of these frameworks. First, if starting with a theoretical approach, a lack of theoretical consensus may result in different themes, scales, and items depending upon the researchers’ theoretical stand. Second, as a result of the first point, previous research may be difficult to use due to, inter alia, the variety of themes, scales, and items used. Moreover, when established theory is derived from empirical research, the theory itself may be misleading due to the vast amount of far-fromvalidated measurements in use. Hale (2000) perceived part of the problem as being induced by the tendency for each researcher to start from scratch by developing his or her own instrument. Hardly any scales have been reused in the same form in more than one study, and they can therefore not be systematically refined and improved by combined research efforts across several research groups. It is important to keep in mind that the cultural disagreements and differences related to epistemology, ontology, and methodology, to a great extent, may explain the many different measurements in use.

59

Research methodology Theory plays a key role in how measurement is conceptualized, and the lack of theoretical consensus poses a clear challenge to researchers. Under such circumstances, it is especially important to be mindful of measurement procedures and development. In addition, it is considered impossible to reach theoretical progress without adequate measurement (Hinkin, 1995). As described here, the situation within the field is complicated, and some important questions are raised regarding whether the methodological disagreements derive from a lack of theoretical consensus or if theoretical consensus is lacking due to methodological differences and the use of flawed measures. In order to overcome some of these problems, a previously developed instrument was used in this thesis

Questionnaire development In order to examine safety culture, a questionnaire developed by Studio Apertura, a constituent centre of The Norwegian University of Science and Technology (NTNU), in collaboration with the Norwegian DNV and the research institution SINTEF, was used. Their development was based on a theoretical review and an evaluation of eight preexisting questionnaires—four developed in Norway, two in Denmark, and two in the United Kingdom. The evaluation was carried out according to five criteria: (1) foundation (theoretical foundation, documentation, and premises for application), (2) thematic width, (3) practical experience of use, (4) the ability to describe and measure safety culture, and (5) the ability to be used at multiple levels (individual, group, team, company). A more thorough description of the development is available (Antonsen & Norges teknisk-naturvitenskapelige universitet. Institutt for sosiologi og statsvitenskap, 2009; SINTEF, 2003; Studio Apertura, 2004). The questionnaire was previously used to survey 60

Research methodology safety culture on board supply vessels and found to be acceptable for use within merchant shipping. The full questionnaire and letter of introduction are included in Appendix 1.

Survey sample and respondents’ demographics The research population is Norwegian-controlled dry cargo and liquid carriers above 500 gross tons. A total of 150 target group vessels were randomly selected from the 953 vessels within the Norwegian Shipowners’ Association’s list of members for 2005. The target group of 150 vessels represented approximately 15% of the overall population, which was considered to be large enough to be representative of the population as a whole (Neuman, 2000). A sample of 10% is recommended, but some withdrawals were expected; thus, a 5% margin was included in the original sample. The sample was stratified with regard to status of the vessel’s flag register (white, grey, or black listed flag) and type of vessel (general cargo, bulk carrier, oil tanker, gas tanker, or chemical tanker). Following the initial selection, telephone calls were made to each company to ask for their participation. Thirty-one companies, with a total of 83, vessels agreed to participate while 45 companies with 67 vessels in total declined. Reasons for not participating included:  Being unable to contact the company despite repeated efforts (23 vessels, 16 companies).  The vessel was not owned by a Norwegian party and therefore was not defined as Norwegian controlled (15 vessels, 8 companies).

61

Research methodology ďƒź Ship management was outsourced to a non-Norwegian country and therefore was not defined as Norwegian controlled (14 vessels, 8 companies). ďƒź The company refused to participate (12 vessels, 10 companies). ďƒź The remaining vessels were sold (3 vessels, 3 companies). The population was later redefined, and vessels managed from a nonNorwegian country were not considered to be Norwegian controlled. In total, 1,574 questionnaires were distributed to 83 tankers and bulk/dry cargo carriers; 76 vessels from 29 companies returned a total of 1,262 forms, resulting in an individual response rate of 80.2%, a vessel response rate of 91.5%, and a company response rate of 93.5%. The questionnaires were returned from 40 liquid bulk carriers (liquid tanker) and 36 dry bulk carriers (dry cargo); 63% of the respondents were employed on a liquid tanker and 37% on a dry cargo vessel. Twenty-two nationalities were represented, with the majority from the Philippines (65.5%), followed by Norway (9.2%), Poland (8.1%) and Russia (5.5%). Unfortunately, no company with vessels flying a black-listed flag was willing to participate. The further validation process (results are included in the articles in part II) is based on the following premises and methodological guidelines.

Validity and reliability through theoretical conceptualization The starting point when developing a questionnaire and scales is conceptual definition, which specifies the theoretical basis. A questionnaire is normally comprised of several dimensions or constructs represented by several partly overlapping items, called multidimensional scales. When generating the item 62

Research methodology pool, each item making up a construct should reflect the latent variables underlying the theme (i.e., the different features or dimensions of the safety culture concept). In order to truly reflect the underlying feature, the items in each dimension should ideally have a common cause (i.e., local management) or consequence (i.e., work practices). Thus, an underlying assumption is that the items reflecting one single construct are unidimensional. In other words, within each measured dimension, items are strongly associated with each other while simultaneously representing a single dimension of the concept. Three reasons for using a multi-item measure instead of a single-item measure are noted. First, an individual item is not reliable due to a considerable random measurement error. Second, an individual item lacks precision and can only categorize people into a relatively small number of groups. Third, an individual item lacks scope, and it is very unlikely that a single item may represent a complex theoretical concept (DeVillis, 2003; Gliem & Gliem, 2003; Hair, 1998; Shevlin, Miles & Bunting, 1997; Spector, 1992). Summing up, single items are considered to be less valid, less accurate, and less reliable than multi-item constructions. It is also suggested that a scale should consist of a minimum of three items in order to be robust (Pett, Lackey & Sullivan, 2003). However, due to collinearity, the use of multiple items could represent a problem in regression models and when independent variables are created by summing items in a scale. Such an additional method represents a procedure that does not control for the effect of measurement error. Regression parameter estimates may be attenuated or increased (Shevlin et al., 1997). When ensuring the conceptual definition, the primary concern is content validity, which is a requirement for construct validity (Hinkin, 1995). Content validity is the degree to which elements of the measurement are relevant to 63

Research methodology and representative of the underlying safety culture concept. Determining whether the scale or item-set has good content validity can be done from a number of sources of relevant theory, empirical literature, and expert judgment. Construct validity concerns the degree to which inferences can legitimately be made from the operationalized constructs in the questionnaire to the theoretical concepts on which those operationalizations were based. When using multidimensional scales, both the convergent validity of the respective subscales (i.e., the degree to which the items within a particular subscale measure the same unidimensional construct) and their discriminant validity (i.e., the degree to which the items in different subscales measure different rather than the same construct) need to be considered. Both content and construct validity are concerned with how the measurement fits with the theoretical foundation and power of generalization窶馬amely, external validity. When the objective of a study is to establish a causal relationship (i.e., using regression analysis), internal validity is of particular consideration, referring to the confidence placed on the assessed cause-effect relationship. The internal validity of the conclusions reached depends on the reliability and validity of the questionnaire or scales used (Neuman, 2000; Raubenheimer, 2004). Other aspects that should be taken into consideration are whether the items are measuring a perception or an attitude. Perceptions are considered more volatile and mostly oriented toward the current workplace conditions, whereas attitudes are considered to be less open to change, more durable, and developed through experiences both inside and outside the workplace. Cooper (2000) cautions against the use of measurements that include attitude scales due to the risk of muddying the construct. Previous research has shown that attitudinal questions have more positively skewed responses than the 64

Research methodology perceptional questions and may therefore influence the analytical results. Moreover, Cooper (2000) indicated that the mix of attitudinal-perceptual questions is one explanatory reason that different factor structures emerge across research groups. Measurement errors threaten the validity of the conclusion about the relationship between the constructs. Method bias has both a systematic and a random component, with the systematic error in particular being considered a major problem. One source of method bias is apparent when the same measurement is used for all constructs, making it difficult to assess the strength of the bias. The direction also varies, and the observed relationships may be either inflated or deflated. Potential sources of common method biases are produced by a common source or evaluator (e.g., social desirability, consistency motive acquiescence, or positive and negative affectivity) whereas method effects are caused by an item’s characteristics (i.e., complexity, ambiguity, scale format, and negatively worded items). Method effects are caused by item context produced by the measurement context (Podsakoff et al., 2003). Podsakoff et al. (2003) presented several approaches to addressing this problem. However, others regard the common method biases as an urban legend, claiming that the supposed effect on correlations is overstated (Spector, 2006). Spector (2006) further pointed to the fact that, as long as there is uncertainty related to the presence and size of a possible bias, applying methods, inter alia statistical methods, in order to control the bias effect might produce biases itself, as one might control for something that does not exist. However, being aware of the possible problems makes it easier to consider them during the process of developing a valid measurement.

65

Research methodology

Validity and reliability through factor analysis When data are collected, factor analysis is a common method for validation of the questionnaire’s conceptual definition. Factor analysis defines the underlying structure of the interrelationship (correlation) between the variables in the questionnaire data by defining a set of common underlying dimensions known as factors. However, not only do the constructs share that they are facets of the same concept, but correlations could also be—due to similarities in measurement—common source and/or common method (DeVillis, 2003; Podsakoff et al., 2003). If the data are not biased, the extracted factors comprise internally consistent and correlated items that externally differ from the other factors. Thus, the extracted factors are assumed to have discriminant and convergent validity. The convergent validity is further assessed by the means of scale analysis and inter-item statistics. For the method of factor analysis to be appropriate, a certain sample size is required. Preferably, the sample size should be 100 or larger. As a general rule, it is suggested to have at least five times as many observations as there are variables to be analyzed. Some even propose 20 cases for each variable. Small sample sizes or low variable-case ratio lead to higher chances of “overfitted” data (i.e., deriving factors that are sample specific with little generalizability) (Hair, 1998). Hair (1998) also pointed out that correlations in small samples could be deemed significant and appear in the factor analysis just by chance. In addition, if no items are substantially correlated, factor analysis is not applicable. The method’s applicability is commonly tested by Barlett’s test of sphericity, which should be significant, and Kaiser-MayerOlkin (KMO), which should exceed 0.60 (Hair, 1998).

66

Research methodology Factor analyses can be done from an exploratory or confirmatory perspective. Exploratory techniques are often more useful early in the validation process, while confirmatory techniques are far more common when the instrument in question has been previously validated. The most common method for extracting factors is Principal Component Analysis (PCA), which is considered suitable when the research purpose is data reduction or exploration, but should not be used in causal modeling. When the research purpose is theory confirmation and causal modeling, Common Factor Analysis (CFA) (e.g., Principal Axis Factoring [PAF]) is most suitable (Hair, 1998). An important tool for interpreting factors is factor rotation (Hair, 1998). Varimax rotation is the most common method for interpretation. However, the varimax rotation method, which belongs to the group of orthogonal rotation techniques, may be problematic to use. Orthogonal techniques assume that the underlying factors are independent, but from the theoretical perspective dimensions of, for example, safety culture, they are not regarded as independent but as an integration of various sub-facets. To validate a questionnaire, CFA is recommended with an oblique rotation technique. Oblique techniques allow for correlation between factors and are preferable when the researcher’s aim is to obtain several theoretically meaningful factors or constructs (Field, 2005; Hair, 1998; Pett et al., 2003). Moreover, confirmatory analysis is recommended when the final objective is structural equation modeling (SEM) analysis (Hoyle, 1995). Each item’s, or variable’s, “fit” with the underlying dimension is represented by factor loading. Factor loadings range from 1 to -1; the closer to 1, the better the representation of the underlying dimension. The definition of a significant loading depends upon the sample size. A small sample size requires higher loading than a large sample. In addition, variables’ 67

Research methodology communalities should be assessed. Communality refers to the total amount of variance an original variable shares with all other variables included in the factor analyses. Variables with low loadings and low communalities should be considered for deletion (Hair, 1998). By one rule of thumb in CFA, loadings should be 0.7 or higher to confirm that independent variables identified a priori are represented by a particular factor. However, such high loadings (≼ 0.70) are not typical, and real-life data may not meet this criterion. Thus, some researchers—particularly for exploratory purposes—use a lower level such as 0.4 for the central factor and 0.25 for other factors (Raubenheimer, 2004). On the other hand, factor loadings must be interpreted in light of theory and their practical significance, not by arbitrary cutoff levels alone. In addition, items with multiple significant loadings at various factors should be deleted, as this is a sign of multidimensionality (Hair, 1998) and, thus, not discriminant valid. In a multidimensional scale, it is recommended that a minimum of three items load significantly in each factor. The more items there are per factor, the more likely is it that the factor will replicate as originally constructed (Pett et al., 2003; Raubenheimer, 2004). When using a validated questionnaire, the extracted factors should be similar to the theoretical construct as operationalized. Factor correlation analysis is often done in order to check construct validity, which is viewed as the extent to which an operational measure truly reflects the underlying safety culture concept as well as whether it operates in a consistent manner.

Validity and reliability through scale analysis Various statistics can be selected in order to estimate the reliability of scale and items, including alpha models, split-half models, Guttman models, and 68

Research methodology parallel and strict parallel models. Cronbach’s alpha is extensively reported as the most commonly accepted measure for internally consistency reliability (Hinkin, 1995; Shevlin, Miles, Davies, & Walker, 2000). Internal consistency (convergent validity) is statistically tested by Cronbach’s alpha coefficient and inter-item statistics. Although no consensus exists with regard to the Cronbach’s alpha coefficient, usually a value above 0.7 is considered acceptable, although some advocate for a level of 0.8 or better, especially when a new scale is being evaluated (Netemeyer, Sharma & Bearden, 2003; Raubenheimer, 2004). Others suggest that when dealing with psychological constructs, values below even 0.7 can, realistically, be expected because of the diversity of constructs being measured (Field, 2005). Although high reliability is generally cited as evidence of good psychometric properties of a scale, it is noted that Cronbach’s alpha value on its own should be used with caution (Shevlin et al., 2000). The value depends upon the number of items in the scale and is a function of, inter alia, the inter-item correlation and the item-total correlation. Thus, the inter-item statistics should also be examined. Inter-item statistics, or convergent validity, are related to the extent to which different scale items assumed to represent a construct converge on the same construct. Convergent validity is the degree to which multiple attempts to measure the same concepts agree, which may be tested by the item-total correlations. Rules of thumb suggest that the item-total correlation should exceed 0.5 (Hair, 1998) or 0.4 (Field, 2005) and the interitem correlation should exceed 0.3 (Hair, 1998), but not 0.8. An inter-item correlation exceeding 0.8 suggests that items are duplicates of one another (Pett et al., 2003). Moreover, Shevlin et al. (2000) argued that a high estimate of Cronbach’s alpha may indicate the presence of systematic error, such as scales deviating from unidimensionality. In such cases, extraneous variables 69

Research methodology can make a substantial contribution to inflating the Cronbach’s alpha value rather than the actual dimension being measured. Indeed, when the factor loadings of the dimension being measured are low, the presence of systematic errors can greatly inflate the estimate of Cronbach’s alpha, especially with large sample sizes (Shevlin et al., 2000). These findings substantiate the importance

of

reporting

item

statistics

so

that

the

presence

of

unidimensionality can be evaluated, along with factor loadings and cross loadings.

Causal relationship through structural equation modeling Structural equation modeling (SEM) is applied to test the causal relationship between the components deriving from factor analysis. Through SEM analysis, it is possible to estimate multiple and interrelated dependence relationships. SEM is focused on testing causal processes inherent in theory. Moreover, this method has the ability to represent unobserved concepts, as safety culture, in these relationships and account for measurement error in the estimation process (Hair, 1998). The structural relations between tested variables are specified with both a theoretical and empirical foundation. SEM is an extension of both factor analysis and regression analysis. The method serves purposes similar to multiple regressions, but in a way that takes into account the modeling of interactions, nonlinearities, measurement error, and correlated error terms. SEM also considers when the independent variable(s) as well as the dependent variable are measured with multiple indicators, such as when extracted using factor analysis. Hence, the advantages of SEM compared to multiple regressions include more flexible assumptions. SEM analysis also opens up the possibility to explore multiple relationships simultaneously, where regression analysis only examines a 70

Research methodology single relationship at the time, holding all other variables constant (Hair, 1998). The final model is evaluated with goodness-of-fit criteria assessing the overall model. Assessing the goodness-of-fit is not as straightforward as with other multivariate dependence techniques (e.g., multiple regressions) as no single test best describes the “strength” of the SEM model. Instead, a number of goodness-of-fit measures have been developed; when used in combination, the results are assessed from three perspectives: overall fit, comparative fit to a base model, and model parsimony. However, there is no consensus of what accurate levels of fit are, as none of the measures (except the chi-square statistics) have an associated statistical test. According to Hair (1998), several guidelines have been suggested, but ultimately each researcher must decide whether the model fit is acceptable (Hair, 1998).

4.2 Qualitative research design The term qualitative research refers to any kind of research that produces findings not arrived at by means of statistical procedures or other means of quantification. Although some of the qualitative data may be quantified, as with census data, the analysis itself is a qualitative one (Strauss & Corbin, 1990). The most prominent qualitative research techniques employed in this study are: (1) document study, (2) case studies, (3) interviews, (4) participatory observation, and (5) participation in maritime forums.

Document study Document study is an indirect method of data collection that does not require participation of the subjects involved. Official maritime accident investigation 71

Research methodology reports (CPEM, 1999; Danish Maritime Authority, 2009; Justis-og politidepartementet 1991; National Transportation Safety Board, 1990; United States Coast Guard, 2005) are studied in order to understand how safety culture, safety management, and context interact and influence the course of events. In addition, administrative safety management documentation has been studied in the four different companies selected for the case study (further information about these cases follows). This includes safety meeting minutes, reported events, root cause analyses, procedural manuals and checklists, and other available relevant documentation. Analyses were performed with the intention of understanding the companies’ approaches to safety management and priority areas. Of particular interest were safety information data analyses—namely, how the experience information was categorized and their approach to identifying causes in cases of near-miss, incidents, and accidents. Such analyses occurred with reference to the previous mentioned ages of safety and to which degree technical, personal, and underlying organizational causes were identified. The analyses were also seen in relation to safety measures and changes done upon the processed information—namely, if changes aimed at introducing more control in the form of procedures and checklists or if changes were done in other levels of the organizational structures and policies (e.g., manning policies). This part of the document analysis was complemented with interviews. The document studies have given valuable insights into understanding how safety culture, safety management, and context interact and influence accidents. These studies also provided a better understanding of shipping companies’ approach to safety management. They have enabled the study of 72

Research methodology past events and issues to identify changes over time; however, some limitations exist. The documents are not representative; therefore, findings cannot be generalized. Which documents were made available depended upon each company; thus, a full comparative study between cases could not be performed. In addition, all documentation should be considered as biased as it represents the view of its authors (Sarantakos, 1998)

Case studies Case study research involves studying individual cases—in this case shipping companies—in their natural environment for a long(er) period of time, employing a number of methods of data collections and analyses. Four case studies were carried out in two tanker companies and two dry cargo companies. The statistical results from survey data analyses were used as criteria for selecting cases and focus area—namely, mixed crew nationality and ship management. Thus, a better understanding of the structure, processes, and complexities underlying the statistical results were achieved. Administrative document studies (as previously described), interviews, and participatory observation were performed along with seminar participation. The information gathered is used to illustrate, explain, and expand the quantitative findings. However, some of the drawbacks with case study as a method are poor representativeness and poor replicability (Sarantakos, 1998).

Interviews In contrast to document studies, interviewing requires direct interaction with the respondents and heavily relies on their involvement, participation, and contribution. Both formal and informal interviews were performed. Open formal interviews were carried out with shore-side personnel working in 73

Research methodology selected case companies within the department of safety management and/or manning. When available, top-level management was also interviewed. The interview process had to be adapted to the subjects’ availability; thus, individual interviews were conducted in some cases and group interviews in others. The scope of the interviews also changed over time as a better understanding of the industry was acquired. Therefore, the first interviews are more superficial in character then the last. All interviews were recorded. When interviewing the sailing personnel, a more informal and ethnographic approach was selected. Safety issues, behavioral norms, violation of standards, and the like are sensitive issues. Most crews are not Norwegian and do not enjoy fixed employment; thus, many fear losing their job from being open about the situation. Most interviews conducted in the field studies were done as a part of daily conversations after a trusting relationship had been established. None of these interviews were recorded.

Participatory observation Participatory observation at sea was also carried out. During the study, vessels from various companies sailed for different periods of time, ranging from one to two weeks.4 During the field studies, I participated in daily work activities such as ballast tank cleaning, mooring operation, loading, and acting as watch keeper at night. However, in a 24-hour society, leisure time is also of great importance. During my spare time, I participated in leisure activities and games. I experienced that spending time with the crew was essential to

4

The shipping industry is known to be a highly transparent industry. Thus, to ensure that all participants

remained unidentifiable, the number of vessels sailed is not included.

74

Research methodology gaining their trust, and the first days no questions of a more sensible character were asked. When interacting with the crew, informal interviews were carried out as part of the daily conversation. Both interviews and observations aimed to understand how the ship management influenced the work on board and to understand team and group processes from within the group. Although the crew knew about my research area in more general terms, research questions were never explained in detail to avoid biases. Participatory observation has the advantage that group processes and management practices may be observed in their natural environment. In addition, it is possible to retrieve firsthand information that respondents are unable or unwilling to offer during formal interviews (e.g., deliberate violations of safety standards). However, one limitation is that participatory observation can only be employed with smaller groups; thus, findings cannot be generalized. Although behavior is directly observed, the method does not offer frequency of behavior. The method is also exposed to observer bias, selective perception, and memory and offers no control measure regarding bias, attitude, and opinion of the observer. In particular, the latter were experienced during the field study, especially when working with lower-paid crew from the third world originating from poor conditions. I had much empathy for this group. Being aware of my own bias gave me some control over the situation in order to remain objective.

Participation in maritime forums Participation in maritime forums and seminars as a method was used for two reasons: (1) to gain an understanding of the maritime context and interrelationships and (2) to ensure quality assurance of results. Inspired by William Foote Whyte and the Street Corner Society, an ethnographic 75

Research methodology approach was adopted to gain an understanding of the research area and build up credibility and trustworthiness. As a part of this, the maritime aspects became integrated into my own social life as well as my research. Early in the study, I reallocated myself from my institute’s economical and administrative department, where I was originally employed, to the department of nautical sciences. Along the road, I also became chairman of the Nautical Institute Norway Branch, an international organization working to improve the safety and efficiency of shipping. I then became an official member of the Norwegian delegation participating in IMO meetings. Thus, by carrying out research at the maritime industry, I became a part of the industry itself. Several presentations of my results have been given at national and international conferences, both for governmental and non-governmental industrial stakeholders and company conferences. Feedback from the audience has been used to interpret statistical results and define critical areas for further investigation. As with participatory observation, going native was—and still is—an area of concern.

4.3 Applied methods and statistics in articles An overview of methods for data collection and statistics applied in the different papers is presented in Table 2. Although not explicitly stated in all articles, all quantitative results are interpreted in a qualitative framework.

76

Research methodology Table 2. Overview of Methods for Data Collection and Statistics

1 x

Article Id. 2 3 4 x x x

5 x

6 x

Field study at sea

---

---

x

x

---

x

Shore interviews

---

---

x

x

---

x

Maritime/safety related conference attendance

---

---

---

x

---

x

IMO attendance

---

---

---

---

---

x

Survey comments

x

---

x

---

---

x

Survey quantitative data

x

x

x

x

x

x

Exploratory factor analysis (EFA)

x

x

x

x

x

x

Confirmatory factor analysis (CFA)

x

---

---

---

---

---

Cronbach’s alpha

x

x

---

---

x

---

Correlation (Pearson’s r)

x

x

---

---

x

---

Inter-item, item-total statistics

x

x

---

---

x

---

Analysis of variance (ANOVA)

---

x

---

---

---

---

Linear regression analysis

---

x

---

---

---

---

Logistic ordered regression analysis

---

---

---

---

x

---

x

x

---

---

---

---

Transformed standardized factor scores

---

---

---

---

x

---

Summarized

---

---

x

x

---

x

Data collection Maritime document study

Analysis

Development of factors used in analysis Summarized

items

from extracted

factor

structure items

based

on

theoretical

relationship

77

Research results

5 Research results This section summarizes each article, including the objective, applied method, main results, conclusions, and interrelationships. All articles are related to the three research questions developed for the purpose of this thesis:  What characterizes safety culture and safety management within the shipping industry?  What is the relationship between safety culture and safety performance within the shipping industry?  What characterizes shipping companies’ application of the safety management concept? Finally, a structural model testing the causal relationship among the latent dimensions of safety culture is presented.

5.1 Summary and results of article 1 Oltedal, H. A., & Engen, O. A. (2009). Local management and its impact on safety culture and safety within Norwegian shipping. In S. Martorell, C. Guedes Soares & J. Barnett (Eds.), Safety, Reliability and Risk Analysis: Theory, Methods and Applications (pp. 1423-1430). London: Taylor & Francis Group. The objective of this article was threefold: (1) explore and analyze the shipboard characteristics of safety culture; (2) elaborate upon which factors affect the shipboard safety culture; and (3) use the results to set the direction for future studies. The first two objectives are stated in the article.

78

Research results The article first states the theoretical approach to safety culture, along with the methodological framework. Safety culture is perceived as a reciprocal interrelated fusion of three main elements. The first element is internal psychological factors, including each individual’s attitude and perception toward safety, work situation, and organization. This element was measured using a questionnaire. In addition, formal and informal interviews were performed to get a more comprehensive understanding of what is happening inside people’s heads. The second element is observable safety-related behavior, or what individuals are actually doing on board. Safety-related behavior, which is perceived to be partly determined by the first element (i.e., psychological factors), was also measured by a questionnaire, in which individuals reported their own behavior in working situations. In addition, participatory field studies were conducted to observe actual behaviors. Behavioral patterns are regarded as a manifestation of shipboard culture and of a culture existing on a higher level in the organization (i.e., organizational factors). Finally, the third element is organizational factors (e.g., employment policy, safety management policy, approach towards efficiency versus safety), which were measured by case studies and interviews. However, the questionnaire also comprised questions related to organizational factors. In this approach, the contextual influence is also important. Distinctive characteristics of the shipping context are, inter alia, life and work on board as a total institution and a 24-hour society. A mixed method approach was applied, where retrieved data were integrated into the interpretation, although this was not explicitly stated in the article. Consequently, psychological measures traditionally referred to as safety climate and safety culture traditionally explored by qualitative methods were not distinguished. In this article (and the others), statistical data were all 79

Research results interpreted in a qualitative framework. It was recommended that the questionnaire not be used as a single method; thus, it was referred to as a safety culture questionnaire. The statistical approach was based on a combination of exploratory and confirmatory factor analyses, using the method of principal component. The exploratory factor analyses were carried out with orthogonal varimax rotation. In the confirmatory factor analysis a one-factor solution on each construct was performed. Based on a comparison of the exploratory and confirmatory factor analyses, five factors were found to be valid: (1) crew interaction, (2) reporting practices, (3) competence, (4) local management, and (5) working situation (proactive work practices). The qualitative data indicated that the shipping companies’ crewing strategy, which includes employment terms, rotation systems, and policy toward shipboard management, are interrelated with how safety culture is manifested on board. Considering seafaring as a 24-hour society and the geographical distance between the on-shore organization and the vessel may affect both the quality of those systems and plans developed on shore and their implementation on the vessels. Thus, ship management was identified as a key factor to a sound safety culture along with the on-shore crewing strategy.

5.2 Summary and results of article 2 Oltedal, H., & Wadsworth, E. (2010). Risk perception in the Norwegian shipping industry and identification of influencing factors. Maritime Policy & Management, 37(6), 601-623. Based on the findings from the first article, the second paper aimed to assess the relationship among shipboard safety, safety culture, and shore-based 80

Research results organizational factors, using risk perception as a proxy variable for the general safety level on board. The objective of this article was to assess the relationship between risk perception and the dimensions of safety culture. With regard to assessing the relationship between risk perception and safety culture, the article introduced risk perception as a dependent variable, thereby expanding the trinity methodological framework presented in the first article (i.e., person, behavior, and situation) with a measurable dependent variable. Risk perception has been found to be adequate as a dependent variable, as previous research indicates that risk perception and risk behavior are strongly correlated. It is suggested that risk perception is a good indicator for safety level in general, which is constrained by the situation and context. Explorative principal component analysis with varimax rotation was applied in order to explore the latent dimensions of safety culture. Eight factors were identified, providing a good representation of the concept of shipboard safety culture: (1) competence, (2) interpersonal relationship (crew interaction), (3) shore orientation, (4) ship management (local management), (5) proactive work practices, (6) feedback, (7) demand for efficiency, and (8) reporting practices. Furthermore, a one-way analysis of variance (ANOVA) was carried out to explore any associations between the demographic (nationality, age, department, and vessel type) and organizational (work description) variables as well as both dependent variables (i.e., risk perception and the independent safety culture dimensions). A linear regression analysis (OLS) was subsequently conducted to assess the associations between risk perception and the dimensions of safety culture, controlling for any potentially influential demographic and organizational factors.

81

Research results Based on the results, the overall safety in respondents’ working situations was perceived to be very high, which may indicate a relatively good safety standard. The ANOVA analysis showed significant differences between age groups in the dimensions of interpersonal relationship and shore orientation. The type of vessel indicated significant differences on the dimensions of competence, local management, and feedback. Those working on dry cargo vessels perceived the feedback on reported experience data to be better than those working on liquid tankers. However, those working on liquid tankers had a better perception of their own level of competence and their local management than those working on dry cargo vessels. Finally, work description (i.e., teamwork versus individual) showed significant differences on all dimensions of safety culture. Crews working on a team perceived the dimensions of competence, interpersonal relationship, local management, feedback, and reporting practices to be better than those working on an individual basis. Crews working on a team also felt less demand for efficiency and perceived the shore side of the company to be more safety orientated. The regression analysis indicated that local management, working practices, and reporting practices have a positive association with risk perception while demand for efficiency has a negative association with risk perception. The working situation has a positive association when work is performed on a team in contrast to when work is performed on an individual basis. None of the demographic data were significantly associated with risk perception. For future research, it was suggested to further examine the characteristics of teamwork along with the concept of group identity. It would also be of interest for future research to examine risk perception both in general and in

82

Research results relation to potential differences among nationalities, along with differences among white, grey and black listed flags of registration. Given that the overall research topic in the current thesis concerns safety management on dry cargo vessels and liquid tankers, the remaining articles address safety management within dry cargo vessels and liquid tankers in greater detail, starting with liquid tankers. The third article also suggests differences with regard to safety management within the two sectors.

5.3 Summary and results of article 3 Oltedal, H. A. (2010). The use of safety management systems within the Norwegian tanker industry—Do they really improve safety? In R. Bris, C. Guedes Soares, & S. Martorell (Eds.), Reliability, Risk and Safety: Theory and Applications (pp. 2355-2362). London: Taylor & Francis Group. The aim of this article is twofold: (1) describe safety management within the liquid tanker sector and (2) identify factors that influence safety management performance. The theoretical rationale of traditional safety management systems was introduced for the first time in this article. Traditional safety management was presented as a system containing four sub-systems: (1) reporting and collection of experience data from the vessel; (2) data processing, summarizing, and analysis; (3) development of safety measures; and (4) implementation. The topic of focus was presented in a situational context where safety concerns need balance to ensure profits and economical concerns. Safety management as such is related to the International Safety Management (ISM) code. This article was the first to explicitly state that it 83

Research results adopted a multi-method approach combining surveys, case studies, field studies, interviews and other qualitative information, although all the concerned articles are based on a multi-method approach. With regard to the statistics and factor analysis, the survey items were grouped as they relate to the information flow in a safety management systems—more precisely: (1) crew’s reporting practices, (2) analysis and follow up by shore side, (3) procedures and checklists, and (4) perceived balance between commercial pressure and safety concerns. Explorative principal component analysis was carried out in each group in order to examine the items’ interrelationships. Shore-side development of safety measures was also analyzed, although most findings stemmed from the qualitative data. The results indicated a situation with substantial underreporting of experience data from the vessels. Such underreporting may be explained by the crew’s fear of negative consequences, a complicated reporting system, and a lack of understanding of the overall safety management system. The development of measures tends to focus on controlling human actions, often in the form of excess use of procedures and checklists. This situation was traced back to a person-oriented approach in safety management. Moreover, procedures and checklists are often perceived as being problematic to use in daily shipboard operations. In order to turn such a situation around, it was suggested that the seafarers’ experience be taken seriously, with regard to both reasons for underreporting and their experience with new measures. On board, the ship management was identified as a factor strongly influencing the shipboard situation, and it is suggested that the shore side pay more attention to that

84

Research results element. Other organizational factors that were suggested to influence the situation included employment conditions and crew stability.

5.4 Summary and results of article 4 Oltedal, H. A., & Engen, O. A. (2010). Tanker versus dry cargo—The use of safety management systems within Norwegian dry cargo shipping. In J.M. Ale, I.A. Papazoglou, & E. Zio (Eds.), Reliability, Risk and Safety (pp. 21182125). London: Taylor & Francis Group. As the overall research topic in the current thesis concerns safety management on both dry cargo vessels and liquid tankers, this article focused on safety management within the dry cargo sector in order to compare the findings with the liquid tank sector. The aim of this article was to (1) describe safety management within the dry cargo industry, (2) identify factors that may influence safety management performance, and (3) compare the current situation within the dry cargo to the liquid tanker industry. This was the first article to provide a description of the two sectors (i.e., liquid and dry cargo) with the purpose of conducting comparative exploration and analysis. The two sectors were introduced in order to make each sector’s major safety challenges visible. This article served as a follow-up to the third article and, thus, followed a similar structure related to the information flow of a traditional safety management system. Although a multi-method approach was applied, only areas in which statistical data were available for both sectors were presented and analyzed—namely, (1) crew’s reporting practices, (2) procedures and checklists, and (3) perceived balance between commercial pressure and safety

85

Research results concerns. An explorative principal component analysis was carried out at each group in order to examine the items’ interrelationships. This article pointed to a central custom-related difference between the two sectors. Although customers of dry cargo shipping have fewer safety-related requirements, the tanker sector is extensively embedded in the Norwegian oil industry, which explains the extended focus on safety in general, safety management, attitudes, etc., within the tanker industry compared to dry cargo. Underreporting of minor incidents and near misses are more present within the dry cargo industry. In both sectors, reporting frequency is correlated with feedback given upon reported events. Although both sectors have substantial underreporting, our data indicated that the safety campaigns, which are typical for the oil industry, have some positive effects on reporting practices. However, within the dry cargo industry, such feedback is perceived as better than within the tanker sector. The analysis suggested that the shore tanker organization is not prepared to manage the growing workload that increased reporting brings about. The organization fails to obtain feedback’s motivating effect, which again may counteract the effect of safety campaigns. However, it was suggested that the dry cargo industry could benefit from such campaigns when it comes to increasing crews’ awareness and recognition of a near miss, along with a better understanding of reporting’s importance in safety-management systems. It was recommended that the shore side provide resources for the potential increase in number of reports placed and proper follow-up. When following up the reports, the development of new measures should also be considered carefully, and alternatives to the development of new procedures should be developed. The experiences from the tanker industry suggested that the development of a constantly increasing, detailed, and extensive procedural system may undermine safety. Moreover, the 86

Research results analysis suggested that safety measures be initiated by internal and industrial need, not external demands from the customer. When externally initiated, safety management may be less integrated into the operational part of the organization. It was also suggested that this external demand is related to the existence of a poor procedural system. Both the third and the fourth articles pointed to a situation with substantial underreporting of experience data. Reporting is regarded as a critical cornerstone in formal safety management. Some influencing factors were suggested, and the fifth article further assessed the relations among reporting practices, safety culture, and organizational factors.

5.5 Summary and results of article 5 Oltedal, H., & McArthur, D. (2010). Reporting practices in merchant shipping, and the identification of influencing factors. Safety Science, 49(2), 331-338. This article pursued three aims: (1) assess the relationship between reporting practices and safety culture, (2) explore the influence of familiarity with local managers, and (3) further explore differences between vessel type (i.e., liquid and dry cargo). Reporting practices are regarded as a cornerstone when working with systematic safety management, and this article was the first to statistically explore factors affecting reporting practices. A review of safety research within the maritime sector indicated that—although underreporting of experience data is regarded as a major problem within the industry—little research has been done. With a foundation in both theory and results derived 87

Research results from research within other high-risk sectors, barriers to experience data reporting were summarized in the following categories: (1) fear of disciplinary action or of other people’s reactions; (2) risk acceptance, where incidents are regarded as a part of the job or unpreventable; (3) useless, as reporting does not lead to any changes; and (4) practical reasons like time pressure or a complicated reporting system. In the statistical analysis, explorative principal component analysis with varimax rotation was carried out, followed by a scale reliability analysis. Seven factors were extracted and found to be valid, reflecting crews’ perceptions of (1) their own competence, (2) interpersonal relationships among the crew, (3) shipboard management, (4) work practices, (5) feedback on reported safety information, (6) shore orientation to safety, and (7) perceived demand for efficiency. An ordered logistic regression was then carried out in order to explore the relationships between extracted factors and reporting practices. In the analysis, the dependent variable reporting practices were measured with four possible outcomes: (1) never/seldom, (2) sometimes, (3) often, and (4) always. The results indicated that high competence, a good and open interpersonal relationship among the crew, safety-oriented management, execution of proactive work practices, and feedback on reported events all increase the odds of being in a higher category of the reporting frequency measure. On the other hand, shore orientation downgrading safety and prioritizing efficiency increase the odds of being in a lower category of the reporting frequency measure. With regard to feedback, vessel type, and management, the effect of these variables is dependent on the category of the dependant variable. The effect of both vessel and feedback is larger when moving between the higher 88

Research results categories. Crews who have been working with their closest manager for more than one year tend to report more often. However, the effect of being familiar with one’s superior is larger when moving between the lower categories. None of the identified factors should be addressed in isolation from each other. As followed from the discussion, they are all important and mutually dependent. Thus, the internal relationship between the identified dimensions of safety culture should be further explored using, for example, structural equation modeling and/or path-analysis, as performed in this thesis.

5.6 Summary and results of article 6 Oltedal, H., & Engen, O. A. (2011). Safety Management in Shipping— Making sense of limited success. Safety Science Monitor, submitted This article sought to (1) explore the gaps between the safety ambitions (in the form of, e.g., rules and procedures) and operational practice on board and (2) identify possible pitfalls when relying on safety through a system perspective (as described in articles three through five), without focusing on its human interrelationships. A multi-method approach combining surveys and case studies, including field studies and interviews, was applied. Previous articles (i.e., three through five: Oltedal, 2010; Oltedal & Engen, 2010; Oltedal & McArthur, 2010) all pointed to substantial weaknesses in current safety management; thus, this article explored the theoretical rationale behind traditional safety management. Normal Accident Theory (NAT) and High Reliability Organization Theory (HRO) were outlined, with an emphasis on how they explain and make sense of safety, risks, and accidents. The conceptual framework of the Practical 89

Research results Drift Model (PDM), which seeks to combine NAT and HRO and explain why seemingly well-regulated organizations (e.g., within shipping), develop traits that may evolve into big accidents and disasters, was applied. The PDM model allows us to look for possible new explanations of the success and failure of safety strategies. In an effort to question what the scientific safety literature has offered to make sense of the gaps between safety ambitions and the practical outcome, some examples of gaps from our own data were examined more closely. The structure of the article followed the four stages of the PDM model, with each stage being supported by qualitative and quantitative data. (1) Design refers the stage in which organizational managers or designers develop extensive routines and procedures in order to make the organization robust and resilient against attacks and unforeseen events. (2) Engineered refers to an operational situation in which the routines and procedures are first applied and experienced to not always match the real situations. (3) Applied refers to situations in which the designed control measures are substituted with a logic of action based on individuals’ experiences and tacit knowledge. (4) Fails refers to a situation of change (e.g., major unforeseen event such as ship collisions), when individual units are forced to act on the assumption that all others are acting in accordance with the original rules and procedures as they were initially designed. The actors are then trapped in a game where trusting their own logic of actions is the only solution while they must simultaneously base their decision on the assumption that others are following the general rules. The article concluded that the industry could gain from abandoning the person-oriented approach, where control measures are designed to control human actions, often in the form of the excess use of procedures and 90

Research results checklists.

These

measures

developed

through

a

traditional

safety

management system are standardized to fit all—whether in a fleet of 5 vessels or 100 vessels. This creates a paradox when confronting actual work situations, where operations are never the same. The vessels are different, the people, constellations of people, power figurations, weather and so on. A standardized measurement will therefore never align with reality. Yet human actions and deviations are compared to the standard and found to be erroneous. In an attempt to gain control, new and even more detailed measurements may be developed, thereby creating a vicious cycle resulting from the anxiety of not being in control. Organizations should abandon such person-oriented approaches in their search for causal and influencing factors.

5.7 Causal relationships between components of safety culture In order to test the multidimensionality of the theoretical safety culture construct, a first-order confirmatory factor analysis (CFA) was carried out, resulting in six dimensions that were found to be a reliable reflection of the safety culture concept: (1) company orientation, (2) local management, (3) crew interaction, (4) competence, (5) proactive working practices, and (6) risk perception. All dimensions and their interrelationships were further discussed in articles number 1 (Oltedal & Engen, 2009), number 2 (Oltedal & Wadsworth, 2010) and number 5 (Oltedal & McArthur, 2010) and are briefly presented below. Company orientation reflects the crew’s perception of the shore organization. A high score indicates that the company has a reactive approach and statistics are a major concern, whereupon the crew perceives the safety work to be a façade. A low score indicates that the company is proactive and cares about 91

Research results the human consequences of hazardous situations, whereupon the crew perceives the company’s safety work to be a real priority area. Competence reflects the crew’s perception of their own training and ability to work safely and handle critical and hazardous situations. A high score indicates that crews see themselves as having a high level of competence in these areas; a low score indicates a low competence level. Local management reflects crew’s perception of their closest manager as a role model as well as the manager’s engagement and interest in ensuring safety in work operations. A high score indicates good safety management; a low score indicates poor safety management. Proactive work practices reflect performance of proactive activities, such as safe job analysis and hazard identification, as well as how safety is prioritized in daily operations. A high score indicates that proactive work practices are performed on a regular basis, whereas a low score indicates the infrequent performance of proactive work practices. Group interaction reflects the relationship amongst the crew, including their problem-solving abilities, form of communication, and sharing of safety information. A high score indicates the presence of a good interpersonal relationship amongst the crew, whereas a low score indicates that the onboard group interaction is poor. Risk perception is an indicator related to the onboard safety and the crew’s own assessment of the probability that they or any other crewmembers will have an accident on board the vessel during the next 12 months. A high score indicates a low probability of an accident to occur; a low score indicates a high probability that an accident will occur. 92

Research results In the SEM structural model (see Figure 4) the latent dimensions of safety culture are tested for causal relationships. Postulated causal relationships are grounded in both theory and empirical results. The hypothesis is that the shipping company’s orientations toward safety at the shore side of the organization influence the safety culture on board the vessel; thus, company orientation towards safety is set as an exogenous variable. The outcome measurement variable is represented by the crew’s overall risk perception. Each path was analyzed and evaluated; paths with no significant effect were removed from the model.

Figure 4: Structural model testing for validity of causal structure

All paths in Figure 4 are reasonable and consistent with the theoretical construct. The model shows a good fit (RMR=0.044, CFI=0.957, and RMSEA=0.052). The root mean square residual (RMR) represents the average residual value derived from the fitting of the variance-covariance matrix for the tested hypothesized model. In a well-fitting model, this value will be small (0.5 or less). The Comparative Fit Index (CFI) is a measure of 93

Research results complete covariation in the data. CFI ranges from zero to 1.00, with values close to 0.95 being advised. The root mean square error of approximation (RMSEA) takes the error of approximation in the population into account. RMSEA values of less than 0.05 are indicative of a good fit between the hypothesized model and the observed data. RMSEA values as high as 0.08 represent reasonable errors of approximation (Byrne, 2010). The standardized regressions weighted (Stand. Reg.) together with significance value (P) are presented in Table 3. Table 3. Standardized Regression Weights and Significance Value of Structural Model Dimension of safety culture

Stand. Reg.

P

Local management.

<--- Company orientation

-.390

***

Crew interaction.

<--- Local management.

.494

***

Competence

<--- Local management

.169

***

Competence.

<--- Crew interaction.

.575

***

Proactive work

<--- Local management

.300

***

Risk perception

<--- Company orientation

-.256

***

Risk perception

<--- Crew interaction.

.131

***

Risk perception

<--- Local management

.122

.002

Proactive work

<--- Competence

.233

***

i Proactive work

<--- Crew interaction.

.194

***

Standardized total effect of each latent variable is further presented in Table 4. Standardized total effect is the sum of each latent dimensionâ&#x20AC;&#x2122;s direct and mediated effect. 94

Research results

Table 4.

Risk perception

.000

.000

.000

.000

.152

Crew interaction

-.193

.494

.000

.000

.000

.000

.244

Competence

-.177

.453

.575

.000

.000

.000

.456

Risk perception

-.329

.187

.131

.000

.000

.000

.151

-.195

.501

.328

.233

.000

.000

.362

Proactive practices

work

Correlations

Competence

.000

Squared Multiple

Crew interaction

-.390

practices

Local management

Local management

Proactive work

Company orient.

Standardized Total Effects

The results from the SEM analyses indicate that company orientation has a direct effect upon the performance of local management on board the vessel (-.390) as well as a direct effect on the general risk perception (-.256). When safety work in the company is not perceived as a genuine effort, local management is perceived as being less safety oriented while the risk level on the vessel is generally perceived to be higher. Management/leadership style, commitment, and visibility are also among the most commonly measured 95

Research results dimensions in safety research in general (Flin, Mearns, O'Connor, & Bryden, 2000). Within the maritime sector, research initiated by the Maritime and Coastguard Agency in the United Kingdom identified various core leadership qualities necessary for effective safety leadership (Maritime and Coastguard Agency, 2004). On board, these qualities were primarily geared toward the captain as a key leader of safety, but also toward lower ranks with leadership responsibilities. Local management demonstrated a direct effect upon the competence dimension (.169), proactive work practices (.300), risk perception (.122), and crew interaction (.494), suggesting that local management plays a major role in the development of safety culture on board. Proactive work practices increase the chances of revealing potentially dangerous situations; thus, preventive measures should be introduced in advance of operations. Error detection and correction are also assumed to be stimulated by teamwork (Kontogiannis & Malakis, 2009). The relationship between local management and company orientation (-.390) indicates that the company is the driving force for how management is performed. The relationship between company orientation and local management is one directional, indicating that experiences from shipboard management do not influence the overall company policy. Likewise, management-relationships are one directional, indicating that the management is not adapted to the shipboard situationâ&#x20AC;&#x201D;an individual is either a good manager or he/she is not. Traditionally, no requirements concerning formal management competence or training have been established within shipping. Such requirements were recently adopted by amendments of the STCW convention, which will put into force on January 1, 2012.

96

Research results Furthermore, competence has a direct effect on proactive work practices (.233), indicating that well-informed crews trained to handle risk are better in performing proactive work practices. However, quite surprisingly, proactive work practices have no direct effect on the general level of risk perception. Competence includes both training acquired on board the vessel and training acquired on shore. Also included are the orientation new crew members receive when joining the vessel, which is often referred to as familiarization and safety-related drills carried out on board. Within safety research, competence is among the top five most commonly measured themes (Flin et al., 2000). Minimum training requirements are covered by international conventions and regulations developed by IMO, with parts are required to be performed onboard. For example, every crew member must participate in at least one abandon ship and one fire drill every month. These drills should, as far as practicable, be conducted as if there were an actual emergency (International Convention for the Safety of Life as Sea, 1974, International Maritime Organization, 2009). On board, the captain and the ship management are ultimately responsible for how such drillsâ&#x20AC;&#x201D;as well as other onboard training arrangementsâ&#x20AC;&#x201D;are carried out. Finally, crew interaction has a direct effect upon the competence dimension (.575), proactive work practices (.194), and risk perception (.131). This relationship is also one-directional; thus, crew interaction is suggested to be a driving force when it comes to how onboard competence-increasing activities are performed as well as proactive work practices. The relationship between local management and crew interaction is one directional, suggesting that management plays a major role in how crewmembers interact, but management does not adjust its style to the crewâ&#x20AC;&#x2122;s characteristics and dynamics. A trusting relationship is regarded as a key factor in forming a 97

Research results safety culture (Reason, 2001). The importance of group interaction and teamwork has also been suggested as influencing accidents to a greater extent than individual unsafe acts (Barnett, Gatfield & Habberley, 2010; Mitropoulos & Cupido, 2009).

5.8 Summing up and presentations of main conclusions The main conclusions drawn from the research indicate that the shore side of the company is the driving force for onboard safety culture. Thus, undesirable onboard working practices may be traced back to shore-side organizational decisions. The onboard conditions are influenced by the shore side in three ways: (1) the organization’s manning policy (e.g., contract arrangements, mix of nationalities, and rotation systems), which establish the premises for the cultural development; (2) the approach to safety management, which lay the foundation for the reporting culture and system efficiency; the industry to a large degree is person oriented, with excessive use of standardized measures; and (3) shipboard management has a major mediating influence as the shoreside organization’s contribution as a support function sets the stage for shipboard management performance. All three areas are further discussed in the following chapter.

98

Discussion

6 Discussion This chapter presents a discussion of the overall results from all articles included in this thesis. The discussion is related to the research questions given in the introduction: 1. What characterizes safety culture and safety management within the shipping industry? 2. What is the relationship between safety culture and safety performance within the shipping industry? 3. What characterizes shipping companiesâ&#x20AC;&#x2122; application of the safety management concept? At the international level, safety management within shipping is carried out through the delegation of authority from the UN to IMO, delegating responsibility for developing and maintaining workable safety regulations. Each coastal state is responsible for the enforcement of maritime regulations, while each flag state is responsible for ensuring compliance with international regulations. The results of this study indicate that the structure of the administrative authorities has some deficiencies as some maritime administrations may be involved in shipping activities without fulfilling their safety responsibilities. In many instances, they are unable to detect real safety threats on board when performing inspections. However, the shipping companies have the primary responsibility for the safe operations of ships and the welfare of the crew. Safety management is regulated through the ISM Code, which was developed by the IMO. One of the intentions behind the ISM Code has been to develop a safety culture within shipping. Study results indicate several deficiencies related to both the industryâ&#x20AC;&#x2122;s application of safety 99

Discussion management and the underlying theoretical rationale. It is important to emphasize that there are variations within companies. Articles 3 and 4 indicated that the application of safety management in relation to culture is associated with two areas. First, safety culture is associated with a reporting culture. A vessel with a high reporting frequency is regarded as having a good safety culture. Reporting emphasizes what has or could have gone wrong, while experience data and suggestions for improvement are less emphasized. This restricts the crewâ&#x20AC;&#x2122;s ability to learn and improve. The qualitative results point to several situations where less serious episodes (e.g., a misallocated knife) has been reported while more serious episodes (e.g., gas on deck) have not. Reasons for not reporting include fear of being blamed by the ship management or shore organization. The review of investigation reports shows that human error is still identified as the major cause of maritime accidents, which impedes the possibility of revealing underlying organizational causes. An inherent intention of the ISM Code was to move the shipping industry application of safety management from the age of human factors to the age of organizational factors. This study indicates that the application of the code still operates within the era of human errors, resulting in safety measures aiming to control human performance through procedures and checklists. Second, a safety culture is associated with compliance with the developed safety measures, but without questioning whether or not the given safety measures are adequate for the situation. Depending upon the type of operation, the crew is expected to relate to safety directives given not only by their own company, but also by the operator, charterer, and customer. In many instances, these directives are not applicable, are incompatible, or do not reflect the onboard operation. Better integration of operating personnelâ&#x20AC;&#x2122;s experience and expertise could improve this situation as 100

Discussion safety measures in some cases are developed by people with no seagoing experience. The overall study results indicate that such weaknesses are related to the shore side of the organization. The SEM analyses support the relationship between shore-side and shipboard practices. In his thesis, Soma (2005) found that safety is a quality of the ship owner rather than the vessel. Within the industry—especially within the liquid tanker sector—reporting and compliance are attempted by safety campaigns aimed at altering the crew’s attitudes. However, according to the theoretical discussion, personal attitudes are less amendable by such means. Despite the campaigns, the results indicate that a substantial underreporting of experience data occurs. It could be time to focus attention on shore-side personnel. Launching safety campaigns aimed at shore-side personnel would enable them to gain a better understanding of their own roles as a support function and how their practice affects operational practice—not only with regard to safety management per se, but also concerning the influence of other parts of the organization (e.g., commercial pressure and crew resource management). Another drawback is standardization. Measures are generally standardized to fit all vessels within a company fleet. The shore personnel fail to understand the diversity created by situational circumstances in operations, crew constellations, the vessel’s technical condition, and so on. Accordingly, the crew bypasses the problem with standardization and develops their own deviating working practice. In particular, article 6 dealt with this subject. These processes may be seen in relation to Snook’s theory of practical drift (Snook, 2000). Formal instructions given by the company are complied with on paper, but onboard the vessels working practices are adapted to the situation. However, as the crew at this point uses their experience and tactical

101

Discussion knowledge to ensure safe operations, deviations from standardized procedures should not uncritically be put on par with unsafe operations. The standardization of safety measures is necessary in relation to crew organization and lack of crew stability. Life and work on board are known to be a highly formalized, hierarchical, and authoritarian organization, with a clear chain of command, clear communication lines, levels of authority, and clearly defined tasks and activities that are more or less the same on all vessel. Such organization makes it possible to handle crew as changeable components. Indeed, within the industry, such an organizational structure is necessary in order to handle the lack of crew stability. However, the lack of both crew stability and standardization may be problematic for other safety reasons as well. A lack of crew stability may be a barrier to change, and safety management is all about change—changes of work practices and attitudes and others; thus, crew may not see the benefit of their efforts. By the time necessary changes are highlighted, crew have most likely signed onto another vessel, where they might experience the same problems. Shipboard management may be problematic as the ship’s management is not familiar with each crewmember’s capabilities and limitations; by the time they finally get to know them, the crewmember changes vessel. Finally, crewmembers lack familiarity with the specific ship, the rest of the crew, and the management. The influence and importance of crew stability is discussed in articles 1, 3, 5, and 6. Crew stability also relates to team dynamics; article 2 indicates that the onboard safety level is perceived as better when work is performed as a team. The overall study results further identified the onboard management as a driving force. Management and leadership as mediating factors have support in climate research from Zohar (1980), who found management’s attitude to 102

Discussion be of major importance in fostering a safe work environment. This is also supported by cultural research done by Pettigrew (1979) and Schein (1992, 2004), who both regard individuals (entrepreneurs and leaders, respectively) as important in the processes of creating and managing an organizational culture. In articles 2 and 5, ship management was statistically associated with risk perception and reporting practices. The shipboard management is responsible for implementing prevailing regulations and company policies. The SEM analyses demonstrated that local management has a direct effect upon all shipboard dimensions of safety culture and that the quality of the shipboard management is influenced by the shore side of the company. Evidence indicates that the ship officers are getting dual instructions regarding safety performance as well as demand for efficiency from other parts of the organizations. Thus, safety and efficiency are balanced in daily operations, where efficiency and corner-cutting activities are rewarded in absence of accidents, but are simultaneously identified as causal factors when accidents occur. Ambiguous instructions from shore should be counteracted by a strong shipboard management. However, according to the results, few ship managers have formal management and leadership education. In addition, shipboard management is not evaluated by crewmembers on a regular basis. By recognizing the importance of ship management, the company should provide support in order to ensure adequate onboard management and leadershipâ&#x20AC;&#x201D;for both the crew and the shore side. Shipping companiesâ&#x20AC;&#x2122; understanding of safety management has also resulted in more administrative work, which makes less time available for attention to operations. The constant development of new procedures increases the systemâ&#x20AC;&#x2122;s complexity as defined by Perrow (1999), resulting in the industry becoming increasingly prone to accidents. The work situation is less easy to 103

Discussion understand when the procedural framework does not fit the situation and when each procedure is not internally coordinated with the overall framework. With reference to Perrow (1999), fewer procedures may simplify the system characteristics and increase resilience. To this end, an understanding of when standardized measures and procedures are appropriate should be reached. This is not to suggest that standardized measures and procedures should be completely abandoned. However, the crew should be able to trust their relevance and applicability, which relates to the casual rationality inherent in safety management systems along with limitations of foresight. Safety management as described by the ISM code is based upon causal rationality, where attempts are made to prevent future events by reflecting upon previous experiences. However, insufficient scientific evidence exists on the effectiveness of systematic safety management to make recommendations either in favor of or against them (Robson et al., 2007), which may be related to their application. As pointed out by Rundmo (1996), the application of traditional safety management is questioned with a distinction between smallscale accidents and larger and more infrequent accidents (Rundmo, 1996). Measures developed through the analysis of past events are in theory only applicable for frequent and small-scale accidents, such as ordinary work accidents. When it comes to medium-size and larger and more infrequent accidents, traditional safety management is not considered to be applicable. Such accidents are often too unique and complex to grasp or to single out some isolated underlying causes. When the course of event is unclear, it is difficult to develop measures that cover all involved risk as the risk in the first place is considered too complex to be fully understood. This highlights the second point: the limitation of the foresight of future events. The study results

104

Discussion indicated that the shipping industry does not differ sufficiently between these types of events and, thus, applies the same logic independent of type of event. The results from the study point to a situation in which serious accidents are on the rise, despite the introduction of the ISM Code and systematic safety management. This can also be seen in relation to the different ages of safety and the recent development of an adaptive age. The adaptive age of safety is characterized by a shift from reliance on systems supported by safety culture to operations. The focus on operations is also in accordance with the theories of HRO and mindfulness. When recognizing the limitations of safety management systems and safety rules, which attempt to control human behavior, it is proposed that adaptive cultures should be embraced (Borys et al., 2009). Consequently, resilience engineering requires a change in perspective from human variability as a liability and in need of control to human variability as an asset in a situation getting out of control, thereby making it important for safe operations. Embracing variability as an asset challenges the comfort of management and, thus, may meet resistance from the industry. However, with reference to the standardization and development of global rules, the industry could ask whether they are made for comfort and to simplify the work for shore personnel or to support safety in daily operations. A consequence of safety management is to shift the focus to how crew is coping in daily operations under constantly shifting circumstances as well as learn from their adaption processes. This should also be seen in relation to the fact that most of the time, when crewmembers adapt, the operations are still performed in accordance with the super-eminent objectives (Hollnagel, 2009). In order to follow such an approach, the crew should be able to trust in the applicability of procedures and the safety measures that ought to be in place. Second, crewmembers should have the possibility to 105

Discussion develop skills, competences, and tactical knowledge in order to handle any unexpected, infrequent situations that cannot be prevented through traditional safety management and standardized measures. Third, when things do go wrong, organizations should remove themselves from a person-orientated approach in which operators are blamed. Safety culture encompasses not only what is done at the operational level, but also at all levels in the company. In order to improve safety, companies should look for influencing factors derived from organizational structures and policies. The shipping industry is known to employ crewmembers from various nationalities through manning agents located in each country, without giving the crewmembers a fixed company employment, thereby resulting in instability. With this in mind, one might question if some of the vessels do have a (safety) culture at all as they lack stability in the group. Without structural stability, they have not been given the possibility to develop the deeper and less conscious levels of cultural patterns and sense making; nor do they have the possibility to develop the experience and tactical skills necessary to handle new and emerging unexpected situations as their efforts are orientated toward making sense of what is unfamiliar窶馬amely, new crew, new power relations, new social constructions, and the like. The study results indicate that loose conditions of employment affect the overall safety management system (e.g., in the form of more underreporting of experience data due to fear of negative consequences and a lack of trust). By changing such a manning strategy, the company could have better possibilities to create a positive safety culture and build competence so that the crewmembers have better premises for handling unsafe situations.

106

Concluding remarks

7 Concluding remarks This chapter includes a discussion of the major limitations for the thesis with regard to methods and measurements. Suggestions for future research and implications are also addressed.

7.1 Methodological limitations The major methodological limitations concern the validity of the questionnaire and sample characteristics. The questionnaire did not show the ability to cover all aspects of the safety culture concept, which is a limitation affecting the overall validity of the study. Due to low reliability, several items and constructs were excluded from the further statistical analysis. One explanation is poor representative reliability across subpopulations or groups of people (Hair, 1998; Neuman, 2000). It may be fair to assume that some groups (e.g., Norwegian employees or senior officers) are better informed about their company’s strategic and tactical management and operations and, therefore, are better placed to answer some of the questions related to the company. A second issue concerns the constructs itself. DeVillis (2003) focused on the fact that the items constituting a construct or dimension should share a common cause or consequence. Some of the questionnaire constructs did not meet this latter requirement. Biases could also be produced by national differences, languages, and response style. In a cross-national study, Harzing (2005) found that English language survey versions tended to be more homogenized, potentially obscuring

cross-national

differences.

McCrae

(2001),

who

studied

Norwegians and Filipinos, did not find such differences. Given that questionnaire language could potentially bias the results, the questionnaire was also made available in Norwegian, Polish, and Tagalog. Taking this into 107

Concluding remarks consideration, cross-cultural comparisons of results are not performed in the current study. In addition, indicated differences at the organizational level (e.g., between type of vessel and employment terms) should not be overestimated. The survey data are representative of vessels flying a white and grey flag only, as those registered under a black listed flag did not want to participate. As participation was voluntary on behalf of the company, it is assumed that those participating do, in general, emphasize safety in their operations; thus, the results are biased in a positive direction. Moreover, the survey data are only representative of members of the Norwegian Shipownersâ&#x20AC;&#x2122; Association. The possibility exists that the results are subject to the common method bias (Podsakoff et al., 2003) due to the data deriving from a common source (e.g., a common scale for different questions). Potential statistical remedies have been suggested. Spector (2006) is skeptical of the merits of such approaches. He argued thatâ&#x20AC;&#x201D;given that it is not possible to know the existence or extent of any possible biasâ&#x20AC;&#x201D;treating it could in fact introduce more bias than what existed in the first place. He recommended using a multi-method strategy so that results do not rely exclusively on the results of one questionnaire. In the current research, case studies, interviews, participatory, and field studies were used to validate the data. As with the survey data, a question of validity arises to the qualitative data. In order to be certain that the elements of culture identified by qualitative methods, Hopkins (2006) recommended consulting the members of the culture. If the members of the culture fail to recognize the description of their culture, the description must be called into question. Accordingly, all results in this thesis were also presented to several people working within the 108

Concluding remarks industry; they expressed that they believe the results to be giving an accurate representation of the situation.

7.2 Theoretical limitations The principal objective of this thesis has been to examine the role of safety culture for safety management and vice versa. Limitations also follow from the theoretical stand and research perspective. By focusing on cultural influences on safety management, other areas of equal importance give way. Research with other perspectives (e.g., professional culture, national culture, or a sociotechnical approach) would bring about different results. For example, technological changes have unquestionably left their mark on both operational safety and the organizational structure of the industry. The shipping industry has, since the early 1960s, steadily adopted the automation and integration of new technology (Alderton, 2004). Yet despite the introduction of new technology partly intended to increase safety by, for example, reducing human error, new technology may also be the cause of new and emerging risk (Schager, 2008). This could be a mismatch between ergonomic aspects and the human information processing system, overreliance in technology that may fail, loss of operational skills and experience necessary for handle critical and unexpected situations, or changes in the social and organizational system.

7.3 Future research With estimation that 75% to 96% of marine casualties are caused by some form of human error (Anderson, 2003; Rothblum, 2000; Wagenaar & Groeneweg, 1987), human error is possibly overemphasized as a causal explanation for accidents at sea. Research on organizational and structural factors in shipping accidents indicates that the human element is identified as 109

Concluding remarks a causal factor without addressing the relationship to underlying organizational and structural factors. Still, the need exists to trace the human factors to conditions resulting from decisions taken at higher organizational levels (IMO, 2010b). A new investigation into accident reports can possibly identify other organizational and structural factors related to shipping accidents. If reviewed and analyzed according to, for example, the accident model developed by James Reason (2001), latent organizational influences, local workplace factors, preconditions for unsafe acts, and unsafe acts can be identified. With new findings discussed in light of theories developed by Snook (2000) and Hollnagel (2009), among others, alternative explanations can be put forward, such as how frontline personnel make sense of organizational safety communications and adapt their work practices through social relations and psychological mechanisms, thereby moving safety management and research into the new era of the adaptive age (Borys et al., 2009; Hollnagel, 2009; Snook, 2000). The adaptive age embraces adaptive cultures and resilience engineering and requires a change in perspective from human variability as a liability and in need of control to human variability as an asset and important for safety. Efforts could also be made to better identify and measure the social processes among workers, along with further exploration of the relationship among management, leadership, and safetyrelated matters Thus, better insights into how behavioral norms interact with and are formed by the social life on board could be achieved.

7.4 Final remarks This thesis has explored the safety culture and safety management within shipping in relation to current theories of safety management and safety culture. The major limitations of the research along with implications for

110

Concluding remarks safety practitioners and researchersâ&#x20AC;&#x201D;previously addressed in this thesisâ&#x20AC;&#x201D;can be summarized as follows. Survey: Parts of the applied questionnaire showed several deficiencies, and results may be biased due to common method, psychometric properties, language, and characteristic with the sample, which may affect the validity of the conclusions. Future research should strive to develop an instrument in order to reduce such biases. Research model: The strengths and limitations of both qualitative and quantitative research should be acknowledged, and future research should be open to a multi-method approach. Safety researcher: As the theories of safety management are developing over time, safety researchers should strive to develop a better understanding of the limitations of current safety management systems and be open to research within the prevailing adaptive age. Safety practitioners: In practical applications of safety management, one should rely less on safety through standardized measures and experience data. This includes understanding the difference between events where such measures are applicable and unexpected events where it is adequate to support competence-promoting activities so that the operators have the ability to adapt their behavior to new situations. The human inferential capacity in handling unexpected situations should not be underestimated in relation to technology.

111

References

8 References Alderton, T. (2004). The Global seafarer: Living and working conditions in a globalized industry. Geneva: International Labour Office. Alvesson, M. (1993). Cultural perspectives on organizations. New York: Cambridge University Press. Anderson, P. (2003). Cracking the code: The relevance of the ISM code and its impact on shipping practices. London: Nautical Institute. Antonsen, S. & Norges teknisk-naturvitenskapelige universitet. Institutt for sosiologi og statsvitenskap. (2009). Safety culture theory, method and improvement. 2009, 47. Barnett, M., Gatfield, D., & Habberley, J. (2010). Shipboard Crisis Management: A Case Study [Online]. Retrieved from: http://disruption.solent.ac.uk/mhfr/resources/RINA%202002.pdf Borys, D., Dennis, E., & Legget, S. (2009). The fifth age of safety: The adaptive age. Journal of Health & Safety Research & Practice, 1(1), 19-27. Bryman, A. (1991). Street Corner Society as a Model for Research into Organizational Culture. In P. J. Frost, L. F. Moore, M. R. Louis, C. C. Lundberg, & J. Martin (Eds.), Reframing organizational culture (p. 205). Newbury Park, CA: Sage. Byrne, B. M. (2010). Structural equation modeling with AMOS: Basic concepts, applications, and programming. New York: Taylor & Francis group. Christophersen, J. G. (2009). Sikkerhetsstyring i skipsfarten 1998-2008: Bakgrunnsfaktorer for reguleringsmessig etterlevelse og overtredelse av ISM-koden. Det juridiske fakultet: Universitetet i Oslo. 2009, 22 Cooper, M. D. (2000). Towards a model of safety culture. Safety Science, 36(2), 111-136. 112

References Corbett, A. (2009). Thousands of ships slip through inspection net. Tradewinds. 9 January 2009. CPEM. (1999). Report of the enquiry into the sinking of ERIKA of the coasts of Brittany on 12 December 1999. Danish Maritime Authority. (2009). The grounding of MIRABELLE on 16 December 2008. Denmark: Division for Investigation of Maritime Accidents. Dansereasu, F. J., & Alutto, J. A. (1990). Level-of-analysis Issues. In B. Schneider (Ed.), Organizational climate and culture (pp. 193-236),. San Francisco, CA: Jossey-Bass. DeVillis, R. F. (2003). Scale Development, Theory and Applications.London: Sage Publications. Ek, Ă&#x2026;. (2006). Safety culture at Sea and Aviation Transport (Doctoral Thesis, Lund University). Field, A. (2005). Discovering statistics using SPSS (and sex, drugs and rock 'n 'roll) (2nd ed.). London: Sage. Flin, R., Mearns, K., O'Connor, P., & Bryden, R. (2000). Measuring safety climate: Identifying the common features. Safety Science, 34(1-3), 177-192. Frost, P. J. (1991). Reframing organizational culture. Newbury Park, CA: Sage. Geertz, C. (1973). The interpretation of cultures. New York: Basic Books. Glendon, A. I., & Stanton, N. A. (2000). Perspectives on Safety Culture. London: Brunel University Research Archive. Gliem, J. A., & Gliem, R. R. (2003). Calculating, Interpreting, and Reporting Cronbach's Alpha Reliability Coefficient for Likert-Type Scales. Midwest Research to Practice Conference in Adult, Continuing, and Community Education. Columbus, OH. 113

References Gregory, D., & Shanahan, P. (2010). The Human Element: A Guide to Human Behaviour in the Shipping Industry, The Stationery Office (TSO). Guldenmund, F. W. (2000). The nature of safety culture: A review of theory and research. Safety Science, 34(1-3), 215-257. Hair, J. F. (1998). Multivariate data analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall. Hale, A. R. (2000). Culture's confusions. Safety Science, 34(1-3), 1-14. Harzing, A. W. (2005). Does the use of English-language questionnaires in crossnational research obscure national differences? International Journal of Cross Cultural Management, 5(2), 213-224. Håvold, J. I., & Nesset, E. (2009). From safety culture to safety orientation: Validation and simplification of a safety orientation scale using a sample of seafarers working for Norwegian ship owners. Safety Science, 47(3), 305-326. Håvold, J. I., & Norges teknisk-naturvitenskapelige universitet. Institutt for industriell økonomi og teknologiledelse. (2007). From safety culture to safety orientation: Developing a tool to measure safety in shipping. Norges teknisk-naturvitenskapelige universitet. Institutt for industriell økonomi og teknologiledelse. 2007, 180. Heinrich, H. W., Roos, N., & Petersen, D. (1980). Industrial accident prevention: A safety management approach (5th ed.). New York: McGraw-Hill. Hinkin, T. R. (1995). A review of scale development practices in the study of organizations. Journal of Management, 21(5), 967-988. Hofstede, G. (2001). Culture's consequences: Comparing values, behaviors, institutions, and organizations across nations (2nd ed.). Thousand Oaks, CA: Sage. Hofstede, G., & Hofstede, G. J. (2005). Cultures and organizations: Software of the mind (2nd rev.). New York: McGraw-Hill.

114

References Hollnagel, E. (2009). The ETTO Principle: Efficiency-Thoroughness TradeOff Why Things That Go Right Sometimes Go Wrong. Aldershot: Ashgate. Hollnagel, E. (2004). Barriers and accident prevention. Aldershot: Ashgate. Hopkins, A. (2006). Studying organizational cultures and their effects on safety. Safety Science, 44, 875-889. Hoyle, R. H. (1995). Structural equation modeling: Concepts, issues, and applications. Thousand Oaks, CA: Sage. International Atomic Energy Agency (IAEA). (1991). Safety Culture. Vienna: IAEA. IMO STW 41/4, 3. (2009, July). Unlawful Practices Associated with Certificate Competency. Sub-Committee on Standards of Training and Watchkeeping 41st Session. International Convention for the Safety of Life at Sea (1974), & International Maritime Organization. (2010). SOLAS : Amendments 2008 and 2009. (pp. 78). London: International Maritime Organization. International Maritime Organization (IMO). (2010a). ISM code: International safety management code and guidelines on implementation of the ISM code (3rd ed.). London: International Maritime Organization. International Maritime Organization (IMO). (2010b). Study on human and organizational factors by WMU. London: Subcommittee on flag state implementation.

115

References International Maritime Organization (IMO). (2007a). Comprehensive Review of the STCW Convention and the STCW Code—Communication and leadership skills (STW/39/3). London: Sub-Committee on Standards of Training and Watchkeeping 39st Session. International Maritime Organization (IMO). (2007b). Comprehensive Review of the STCW Convention and the STCW Code—Leadership and managerial skills (STW 41/7/13). London: Sub-Committee on Standards of Training and Watchkeeping 41st Session. International Maritime Organization (IMO). (2007c). Role of the human element: Near miss information. London: International Maritime Organization. International Maritime Organization (IMO). (2005). Role of the Human Element—Assessment of the impact and effectiveness of implementation of the ISM Code (MSC 81/17/1). London: International Maritime Organization. Justis-og politidepartementet. (1991). Scandinavian Star-ulykken, 7 april 1990. Oslo: Norges Offentlige Utredning. Kjellen, U. (2000). Prevention of Accidents through Experience Feedback. London: Taylor & Francis. Kontogiannis, T., & Malakis, S. (2009). A proactive approach to human error detection and identification in aviation and air traffic control. Safety Science, 47(5), 693-706. Lamvik, G. M. (2002). The Filipino seafarer: A life between sacrifice and shopping. Trondheim: Dept. of Social Anthropology, Norwegian University of Science and Technology. Lappalainen, J. (2008). Transforming Maritime Safety Culture: Evaluation of the impacts of the ISM Code on maritime safety culture in Finland. Finland: Centre for Maritime Studies, University of Turku. Mansell, J. N. K., & SpringerLink. (2009). Flag State Responsibility. Berlin, Heidelberg: Springer-Verlag Berlin Heidelberg. 116

References MARISEC. (2008). Shipping Industry Flag State Performance Table. London: Maritime International Secretariat Services Ltd (Marisec) MARISEC. (2006). Shipping Industry Guidelines on Flag State Performance. London: Maritime International Secretariat Services Ltd (Marisec). Maritime and Coastguard Agency. (2004). Driving Safety Culture: Identification of Leadership Qualities for Effective Safety Management. Aberdeen: Maritime and Coastguard Agency. Martin, J. (1992). Cultures in organizations: Three perspectives. New York: Oxford University Press. McCrae, R. R. (2001). Cross-cultural research on the five-factor model of personality. In W. J. Lonner, D. L. Dinnel, S. A. Hayes, & D. N. Satler (Eds.), Online Readings in Psychology and Culture (Unit 6, Chapter 1). Washington: Center for Cross-Cultural Research, Western Washington. Retrieved from http://www.wwu.edu/~culture Mearns, K. J., & Flin, R. (1999). Assessing the State of Organizational Safetyâ&#x20AC;&#x201D;Culture or Climate? Current Psychology, 18(1), 5. Mitropoulos, P., & Cupido, G. (2009). The role of production and teamwork practices in construction safety: A cognitive model and an empirical case study. Journal of Safety Research, 40(4), 265-275. Mitroussi, K. (2003). The evolution of the safety culture of IMO: A case of organisational culture change. Disaster Prevention and Management, 12(1), 16-33. National Transportation Safety Board. (1990). Grounding of U.S. Tankship EXXON VALDEZ on Bligh Reef, Prince William Sound Near Valdez, AK March 24, 1989. Netemeyer, R.G., Sharma, S. & Bearden, W.O. (2003). Scaling procedures : issues and applications, Sage Publications, Thousand Oaks, Calif. Neuman, W. L. (2000). Social research methods: Qualitative and quantitative approaches (4th ed.). Boston: Allyn and Bacon.

117

References Olsen, E. (2009). Safety climate and safety culture in health care and the petroleum industry: Psychometric quality, longitudinal change, and structural models. Stavanger: University of Stavanger, Faculty of Social Sciences. Oltedal, H. (2010). The use of safety management systems within the Norwegian tanker industry—do they really improve safety? In R. Bris, C. Guedes Soares, & S. Martorell (Eds.), Reliability, Risk and Safety: Theory and Applications (pp. 2355-2362). London: Taylor & Francis Group. Oltedal, H., & McArthur, D. (2010). Reporting practices in merchant shipping, and the identification of influencing factors. Safety Science, 49(2), 331-338 Oltedal, H., & Wadsworth, E. (2010). Risk perception in the Norwegian shipping industry and identification of influencing factors. Maritime Policy & Management, 37(6), 601-623. Oltedal, H. A., & Engen, O. A. (2010). Tanker versus dry cargo—The use of safety management systems within Norwegian dry cargo shipping. In J.M. Ale, I.A. Papazoglou, & E. Zio (Eds.), Reliability, Risk and Safety (pp. 2118-2125). London: Taylor & Francis Group Østreng, D. (2007). I samme båt-Forholdet mellom sjøfolk på multinasjonale skip. Oslo: Universitetet i Oslo, Institutt for sosiologi og samfunnsgeografi. Paris MOU. (2010). Port State Control—In the offing. Annual Report 2009. ParisMou. Paris MOU. (2007). On course for safer shipping–25 years. ParisMou. Perrow, C. (1999). Normal accidents: Living with high-risk technologies. Princeton, NJ: Princeton University Press. Peterson, M. F., Ashkanasy, N. M., & Wilderom, C. P. M. (2000). Handbook of organizational culture & climate. Thousand Oaks, CA: Sage.

118

References Pett, M. A., Lackey, N. R., & Sullivan, J. J. (2003). Making sense of factor analysis: The use of factor analysis for instrument development in health care research. Thousand Oaks, CA: Sage. Pettigrew, A. M. (1979). On Studying Organizational Cultures. Administrative Science Quarterly, 24(4), 570-581. Pidgeon, N. (1998). Safety Culture: Key theoretical issues. Work & Stress, 12(3), 202-216. Pidgeon, N., & O'Leary, M. (2000). Man-made disasters: Why technology and organizations (sometimes) fail. Safety Science, 34(1-3), 15-30. Podsakoff, P. M., MacKenzie, S. B., Lee, J., & Podsakoff, N. P. (2003). Common Method Biases in Behavioral Research: A Critical Review of the Literature and Recommended Remedies. Journal of Applied Psychology, 88(5), 879. Raubenheimer, J. E. (2004). An item selection procedure to maximize scale reliability and validity. South African Journal of Industrial Psychology, 30(4), 59-64. Reason, J. (2001). Managing the risks of organizational accidents. Aldershot: Ashgate. Richter, A., & Koch, C. (2004). Integration, differentiation and ambiguity in safety cultures. Safety Science, 42(8), 703-722. Riksrevisjonen. (2010). Riksrevisjonens undersøkelse av Sjøfartsdirektoratets saksbehandling ved dokumentkontroll av fartøy og sjøfolk. Oslo: Nærings-og handelsdepartementet. Roberts, K. H. (1990). Some Characteristics of one Type of High Reliability Organization. Organization Science, 1(2), 160-176. Robson, L. S., Clarke, J. A., Cullen, K., Bielecky, A., Severin, C., Bigelow, P. L., Irvin, E., Culyer, A., & Mahood, Q. (2007). The effectiveness of occupational health and safety management system interventions: A systematic review. Safety Science, 45(3), 329-353. 119

References Rousseau, D. M. (1990). Assesing organizational culture: The case for multiple methods. In B. Schneider (Ed.), Organizational climate and culture (pp. 153). San Francisco, Calif.: Jossey-Bass. Rothblum, A. R. (2000). Human Error and Marine Safety. Paper presented at the National Safety Council Congress and Expo, Orlando, October 13-20. Rundmo, T. (1996). Associations between risk perception and safety. Safety Science, 24(3), 197-209. Sagan, S. D. (1993). The limits of safety: Organizations, accidents, and nuclear weapons. Princeton, NJ: Princeton University Press. Sarantakos, S. (1998). Social research (2nd ed.). Houndmills: Macmillan. Schein, E. H. (1992). Organizational culture and leadership (2nd ed.). San Francisco: Jossey-Bass. Schein, E. H. (2004)., Organizational culture and leadership (3rd ed.). San Francisco: Jossey-Bass. Schager, B. (2008). Human error in the maritime industry: How to understand, detect and cope. Sweden: Marine Profile Sweden AB. Shea, I. P. (2005). The organizational culture of a ship: A description and some possible effects it has on accidents and lessons for seafaring leadership (PhD dissertation, University of Tasmania). Shevlin, M., Miles, J. N. V., & Bunting, B. P. (1997). Summated rating scales. A Monte Carlo investigation of the effects of reliability and collinearity in regression models. Personality and Individual Differences, 23(4), 665-676. Shevlin, M., Miles, J. N. V., Davies, M. N. O., & Walker, S. (2000). Coefficient alpha: A useful indicator of reliability? Personality and Individual Differences, 28(2), 229-237. SINTEF. (2003). Sikkerhetskulturer i transport: En kunnskapsoversikt. 120

References Snook, S. A. (2000). Friendly fire: The accidental shootdown of U.S. Black Hawks over Northern Iraq. Princeton, NJ: Princeton University Press. Soma, T. (2010). It takes management to reduce accidents [Online]. Retrieved from http://www.dnv.com/industry/maritime/publicationsanddownloads/pu blications/dnvtankerupdate/2008/no12008/ittakesmanagementcommit menttoreduceaccidents.asp Soma, T. (2005). Blue-chip or sub-standard - a data interrogation approach to identify safety characteristics of shipping organisations (DrIng thesis, NTNU Trondheim). Sorensen, J. N. (2002). Safety culture: A survey of the state-of-the-art. Reliability Engineering & System Safety, 76(2), 189-204. Spector, P. E. (2006). Method Variance in Organizational Research: Truth or Urban Legend? Organizational Research Methods, 9(2), 221. Spector, P. E. (1992). Summated rating scale construction: An introduction. Newbury Park, CA: Sage Publications. Stopford, M. (2009). Maritime economics (3rd ed.). London: Routledge. Strauss, A. L., & Corbin, J. M. (1990). Basics of qualitative research: Grounded theory procedures and techniques. Newbury Park, CA: Sage. Studio Apertura. (2004). Sikkerhetskulturer i transportsektoren. Metoder for kartlegging av sikkerhetskultur: Evaluering av noen eksisterende verktøy. Trondheim: Studio Apertura. Tharaldsen, J. (2011). "In safety we trust": Safety, risk and trust in the offshore petroleum industry. (DrPhil thesis, UiS, Stavanger) The United Nations Scientific Committee on the Effects of Atomic Radiation. (1988). UNSCEAR 1988 Report Annex D Exposures from the Chernobyl accident.Vienna: The United Nations Scientific Committee on the Effects of Atomic Radiation.

121

References Tradewinds. (2007). "Evangelia" slips through the cracks.Tradewinds. Turner, B. A. (1978). Man-made disasters. no. 53, New York : Crane, Russak, Wykeham science series United States Coast Guard. (2005). Investigation into the explosion and sinking of the chemical tanker Bow Mariner in the Atlantic Ocean on February 28, 2004, with loss of life and pollution. Washington, DC: United States Coast Guard. Wagenaar, W. A., & Groeneweg, J. (1987). Accidents at sea: Multiple causes and impossible consequences. International Journal of Man-Machine Studies, 27(5-6), 587-598. Weick, K. E., & Sutcliffe, K. M. (2007). Managing the unexpected (2nd ed.). San Francisco, CA: Jossey-Bass. Whyte, W. F. (1991). Street Corner Society: Excerpts from the Appendix to the 1955 Edition. In P. J. Frost, L. F. Moore, M. R. Louis, C. C. Lundberg, & J. Martin (Eds.), Reframing organizational culture (p. 173). Newbury Park, CA: Sage. Wiegmann, A. D., Zang, H., von Thaden, T., Sharma, G., & Mitchell, A. (2002a). Safety Culture: A Review. Illinois: Aviation Research Lab Institute of Aviation, University of Illinois. Wiegmann, A. D., Zang, H., von Thaden, T., Sharma, G., & Mitchell, A. (2002b). A Synthesis of Safety Culture and Safety Climate Research. Illinois: Aviation Research Lab Institute of Aviation, University of Illinois. Wiegmann, A. D., Zang, H., von Thaden, T., Sharma, G., & Mitchell, A. (2002c). Safety Culture: A concept in chaos? To appear in the Proceedings of the 46th Annual Meeting of the Human Factors and Ergonomics Society. Santa Monica, Human Factors and Ergonomics Society. Winchester, N., Alderton, T., & Seafarers International Research Centre. (2003). Flag state audit 2003: Introduction & appendices. Cardiff: Seafarers International Research Centre. 122

References Zhang, H., Wiegmann, D. A., von Thaden, T. L., Sharma, G., & Mitchell, A. A. (2002). Safety culture: a concept in chaos? Conference Proceedings. Zohar, D. (1980). Safety Climate in Industrial-Organizationsâ&#x20AC;&#x201D;Theoretical and Applied Implications. Journal of Applied Psychology, 65(1), 96102. Zohar, D. (2008). Safety climate and beyond: A multi-level multi-climate framework. Safety Science, 46(3), 376-387. Zohar, D. (2010). Thirty years of safety climate research: Reflections and future directions. Accident Analysis & Prevention, 42(5), 1517-1522.

123

Part II Article 1 Oltedal, H. A., & Engen, O. A. (2009). Local management and its impact on safety culture and safety within Norwegian shipping. In S. Martorell, C. Guedes Soares & J. Barnett (Eds.), Safety, Reliability and Risk Analysis: Theory, Methods and Applications (pp. 1423-1430). London: Taylor & Francis Group.

Article 2 Oltedal, H. & Wadsworth, E. (2010). Risk perception in the Norwegian shipping industry and identification of influencing factors. Maritime Policy & Management, 37(6), 601-623.

Article 3 Oltedal, H. A. (2010). The use of safety management systems within the Norwegian tanker industry—Do they really improve safety? In R. Bris, C. Guedes Soares, & S. Martorell (Eds.), Reliability, Risk and Safety: Theory and Applications (pp. 2355-2362). London: Taylor & Francis Group.

Article 4 Oltedal, H. A., & Engen, O. A. (2010). Tanker versus dry cargo—The use of safety management systems within Norwegian dry cargo shipping. In J.M. Ale, I.A. Papazoglou, & E. Zio (Eds.), Reliability, Risk and Safety (pp. 21182125). London: Taylor & Francis Group.

124

Article 5 Oltedal, H. & McArthur, D. (2010). Reporting practices in merchant shipping, and the identification of influencing factors. Safety Science, 49(2), 331-338.

Article 6 Oltedal, H. A., & Engen, O. A. (2010). Safety Management in Shipping— Making Sense of limited Success. Safety Science Monitor, submitted.

125

Article 1 Oltedal, H. A., & Engen, O. A. (2009). Local management and its impact on safety culture and safety within Norwegian shipping. In S. Martorell, C. Guedes Soares & J. Barnett (Eds.), Safety, Reliability and Risk Analysis: Theory, Methods and Applications (pp. 1423-1430). London: Taylor & Francis Group.

Safety, Reliability and Risk Analysis: Theory, Methods and Applications – Martorell et al. (eds) © 2009 Taylor & Francis Group, London, ISBN 978-0-415-48513-5

Local management and its impact on safety culture and safety within Norwegian shipping H.A. Oltedal University College Stord/Haugesund, Norway

O.A. Engen University of Stavanger, Norway

ABSTRACT: This paper addresses safety culture on tankers and bulk carriers and which factors affect the safety culture onboard vessels. The empirical setting for the study is the Norwegian shipping industry. Safety management is a challenging issue within shipping for several reasons. First of all, life and work onboard a vessel is a 24 hour activity and the crew has few possibilities of interacting with the surrounding society. Secondly the geographical distance between the on-shore organization and the vessel may affect both the quality of those systems and plans developed on shore and their implementation on the vessels. The ship management is thus identified as a key factor to a sound safety culture along with the on shore crewing strategy.

1

INTRODUCTION

In this paper we will discuss the safety culture within the Norwegian shipping industry with tankers and bulk carriers, and identify which organizational factors may affect this particular safety culture. In Norway, shipping has for several centuries been the principal trade, and Norway as a maritime nation has roots way back in the Viking age. Today Norway is one of the five largest shipping nations in the world, after Greece, Japan, Germany and China. In the third quarter of 2007 the Norwegian foreign-going fleet comprised 1,795 ships, the highest number ever in Norwegian history, of which about 49 percent are flying under the Norwegian flag (Nærings- og handelsdepartementet 2007). The remaining 51 percent may register in any of the world’s more than 150 flag states. Norwegian shipping companies employ some 57,000 seamen from more than 60 different nationalities and of which about 30 percent are Norwegian Nationals (Norwegian Shipowners’ Association). The crew may be recruited and managed by the shipping company itself, or by one of the world’s many professional crew hiring companies. Within the Norwegian fleet, most sailors are contract-employees working on different vessels during each enrolment, which results in continually shifting working groups. The situation today is a result of a structural change dating back to the 60s and 70s when technical development allowed for bigger vessels with more automation and monitoring,

along with the need for reorganization to improve efficiency. This resulted in a cut in the crewing level. Later in the 80s a global recession caused further structural changes; flagging-out, use of external crewing agencies and signing on crew from developing countries and lower wages (Bakka, Sjøfartsdirektoratet 2004). However, the shipping industry is today facing new manning related challenges as there is a global shortage of manpower, this is due to three main challenges: First, it is less attractive nowadays to work in the shipping industry. Second, the recruitment for ship crews has been slow. This has resulted in the third situation where the liquefied natural gas (LNG) shipping sector is drawing crew from the tanker industry, and the tanker industry in turn is drawing people from the dry bulk sector. In 1894 the British Board of trade carried out a study which showed that seafaring was one of the world’s most dangerous occupations, and it still is (Li, Shiping 2002). Regulations in order to reduce the risk at sea were introduced about 150 years ago. These regulations initially encompassed measures to rescue shipwrecked sailors, and further requirements for life-saving equipment, seaworthiness and human working conditions. Traditionally the safety work has focused on technical regulations and solutions even though experience and accident statistics indicate that most of the accidents at sea somehow were related to human performance (Bakka, Sjøfartsdirektoratet 2004). However, a few very serious accidents at seathat

1423

occurred in the late 80’s resulted in a change towards how safety was organised, and more focus was given to the human barriers and how the seafarers’ working conditions were affected by organisational and managerial factors—both on shore and at sea. Along with this the term safety culture started to gain a foothold also within shipping. The idea of safety culture within shipping was officially introduced on the 4th November 1993 by the adoption of a new resolution, the present SOLAS Convention 1974 Chapter IX, entitled ‘‘Management for the Safe Operation of Ships and for Pollution Prevention’’, also known as the International Safety Management Code (ISM Code) (Le Meur 2003). Hence, the main purpose of this paper is to elaborate the following questions:

Person Safety Climate

CONTEXT External Observable Factors

Situation Organisational l Factors

2

APPROACH TOWARDS SAFETY CULTURE

There seems to be no clear consensus concerning the ontological, epistemological, and methodological questions related to the topic of safety culture. The main differences seem to be 1. Definition of the scope of safety culture and the relationship between culture and climate. 2. Which methods are regarded as most suitable for measurement. 3. The relationship to other organisational (safetyrelated) aspects (Cooper 2000, Guldenmund 2000, Neal, Griffin & Hart 2000, Peterson, Ashkanasy & Wilderom 2000, Sorensen 2002, Yule 2003). However, it is not the scope of this paper to problematise the concept of safety culture. As a point of departure we will apply Schein’s definition of organisational culture: ‘‘A pattern of shared basic assumptions that the group learned as it solved its problems of external

Behavior Safety Behavior

Figure 1. Reciprocal safety culture model (adopted from Cooper, 2000).

– What characterises safety culture on tankers and bulk carriers? – Which factors affect the safety culture on board vessels? With reference to shipping, this article will more concretely analyse crewing strategies such as outsourcing of crewing management and the extended use of contract employment instead of permanent employment. Our hypothesis is that these conditions may contribute to an unfavourable and error-inducing working environment, i.e. poor communication between shore management and the ship management and the remaining crew, unworkable procedures, lack of loyalty to the organisation, dysfunctional interaction, fear of reprisals, which again counteract the development of a safety culture.

Internal Psychological Factors

adoption and internal integration, that has worked well enough to be considered valid and, therefore, to be taught to new members as the correct way to perceive, think and feel in relation to those problems’’(Schein 2004). Further we have decided to use a methodological framework presented by Cooper (2000), and the application of this framework will be discussed below. Cooper (2000) introduces a reciprocal model of safety culture that allows the multi-faceted and holistic nature of the concept to be fully examined by using a triangular methodology approach, depicted in Figure 1. Cooper’s (2000) model contains three elements: 1. The subjective internal psychological factors i.e. attitude and perceptions. 2. Observable on-going safety related behaviour. 3. The organisational situational features. According to Cooper (2000) these elements reflect those accident causation relationships found by a number of researchers, such as John Adams, Herbert William Heinrich and James Reason. The investigation of several serious shipping accidents such as Herold Free Enterprize (Department of Transport 1987), Exxon Valdes (National Transportation Safety Board 1990) and Scandinavian Star (Justis- og politidepartementet 1991) is also congruent with their findings. The Herold Free Enterprize accident was partly caused by members of the crew not following best practice, but was also due to managerial pressure from the organization’s upper level to sail as early as possible, along with other mistakes made by the on-shore management. Two years later when the US tanker ‘‘Exxon Valdes’’ grounded, the accident investigation determined several probable causes linked to human errors induced by managerial faults in the upper levels of the organisation. At the vessel, the third mate failed to properly manoeuvre the vessel, possibly due to fatigue and excessive workload. The master failed

1424

to provide a proper navigation watch, possibly due to impairment from alcohol. At the onshore part of the organisation, the shipping company fails to supervise the master and provide a rested and sufficient crew for the ‘‘Exxon Valdez’’. In addition to this effective pilot and escort services were lacking. The following year, in 1990, there was a fire on the passenger liner ‘‘Scandinavian Star’’. In the aftermath of this accident the investigation brought into focus organisational and managerial faults with regard to a lack of competence and training, but also weaknesses in the wider social-technical system. These weaknesses consisted of ownership separated from management, unsatisfactory control routines by the flag state and, in general, an overall maritime system with a lack of transparency. Further, Cooper’s (2000) three elements will be outlined more in detail, starting with safety related behaviour. 2.1

The importance of safety related behaviour

Herbert William Heinrich work (published in 1931 Industrial Accident Prevention) is the basis for the theory of Behaviour-Based Safety (BBS), which holds that as many as 80–90 percent of all workplace accidents are caused by human error and unsafe acts (Tinmannsvik, 2008). Schein’s (2004) definition of culture does not clearly address observable behaviour patterns, but behaviour is regarded to be partly determined by a person’s perceptions, feelings and thoughts. However, Schein (2004) regards behavioural patterns as a manifestation of a culture existing at a higher level in the organisation, and not as culture itself. When it comes to BBS, the current theories posit that safety culture, and a reduction of accidents may be achieved through continuous attention to three domains: 1. Environmental factors such as equipment, tools, physical layout procedures, standards, and temperature. 2. Person factors such as people’s attitudes, beliefs, and personalities. 3. Behaviour factors, such as safe and at-risk work practices, referred to as the Safety Triad (Geller 2001). When adopting this approach humans are seen as a cause of accidents, whereupon interventions to enhance safety are aimed at changing attitude or behavior (i.e. poster campaigns, training, procedures and so on, or changing the technology they operate). This orientation towards risk and safety management has traditionally been and still is adopted from the majority of the shipping companies. The BBSapproach has been criticised for placing too much responsibility on the people operating the systems, assuming that they are responsible for the outcome

of their actions (Dekker & Dekker 2006). An alternative view is to recognise human error not as a cause of accidents, but as a consequence or symptom of organisational trouble deeper within the organisation, arising from strategic or other top level decisions. This includes resource allocation, crewing strategy and contracting (Dekker, Dekker 2006, Reason 2001, Reason, Hobbs 2003). An organisation is a complex system balancing different, and often also conflicting, goals towards safety and production in an aggressive and competitive environment (Rasmussen 1997), a situation that to a large extent is current within shipping. The BBS approach towards safety often implies that more automation and tighter procedures should be added in order to control the human actions. However, the result may be that more complexity is added to the system. This in combination with the organisation’s struggle to survive in the competitive environment, leads to the system becoming even more prone to accidents (Perrow 1999) (Dekker & Dekker 2006, Reason 2001, Reason & Hobbs 2003). However, the concept of focusing on the human side of safety is not wrong. After all, the technology and production systems are operated, maintained and managed by humans, and as the final barrier towards accidents and incidents they are most of the time directly involved. The proponents of the BBS approach argue that behaviour control and modification may bring a shift in an organisation’s safety culture, also at the upper level, but this is most likely if the focus is not exclusively addressing observed deficiencies at the organisation’s lower levels (DeJoy 2005). DeJoy (2005) calls attention to three apparent weaknesses related to the BBS approach: 1. By focusing on human error it can lead to victimblaming. 2. It minimises the effect of the organisational environment in which a person acts. 3. Focusing on immediate causes hinders unveiling the basic causes, which often reside in the organisational environment. Due to this, we will also include the organisational environment in the safety culture concept, as proposed by Cooper (2000). 2.2

The relation to organisational factors

When human error is not seen only as a cause of accidents, but as a symptom and consequence of problems deeper inside the organisation, or what Reason (2001, 2003) refers to as latent organisational factors, emphasis is placed on weaknesses in strategic decisions made at the top level in the organisation. These strategic decisions may reflect an underlying assumption about the best way to adapt to external factors and to achieve internal integration, and if they are common for most shipping companies an organisational culture may also

1425

be revealed (Schein 2004). Schein (2004) also stresses the importance of leadership. The top management influences the culture as only they have the possibility of creating groups and organisations through their strategic decisions. And when a group is formed, they set the criteria for leadership and how the organisation will support and follow up their leaders. The leaders, at all levels on shore and at the vessel, are also key figures in the development of a safety culture. It is their task to detect the functional and dysfunctional elements of the existing culture, and to channel this to the upper levels of the organisation. In return, the upper levels of the organisation should give their leaders the support necessary in order to develop the desired culture. 3

METHODOLOGICAL IMPLICATIONS

Development of questionnaire items

The survey instrument was developed by Studio Apertura in collaboration with DNV and SINTEF. The development was based on an evaluation of seven already existing questionnaires in comparison with various theoretical views of the safety culture concept (Studio Apertura 2004). Minor adjustments were made after a pilot for use within the tanker and bulk carrier sector. This resulted in a questionnaire with constructs and accompanying number of items as presented in table 1. All items were measured on a 5 point likert scale ranging from strongly disagree to strongly agree, or very seldom/never to very often/always. 3.2

Questionnaire constructs and number of items.

Construct

Number of items

Top management’s safety priorities Local management Procedures & guidelines Interaction Work situation Competence Responsibility & sanctions Working environment Reporting practices

3 7 7 18 8 5 7 9 10

response rate of 80% and a vessel response rate of 91.5%. The survey was carried out in 2006.

Cooper’s (2000) framework put forward the importance of methodological triangulation in order to grasp all facets of the cultural concept. The internal psychological factors are most often assessed via safety climate questionnaires. Our approach is to start with such a survey in order to gain insight into the seafarer’s perceptions and attitudes related to safety, along with self-reported work behaviour related to risk taking, rule violation and accident reporting. The survey also includes questions related to crewing strategy, which opens up the possibility of assessing the relationship between the organisational situation and actual behaviour. The survey results are used to determine which organisational factors are most likely affect the safety culture, and to define research areas for a further qualitative study. 3.1

Table 1.

Questionnaire sample

A total of 1574 questionnaires were distributed to 83 randomly selected Norwegian controlled tankers and bulk carriers. All vessels were flying a flag on the Paris MOU white or grey list. 76 vessels returned a total of 1262 completed forms, which gives an individual

3.3

Statistical analysis

The Statistical Package for the Social Sciences (SPSS) v.15.0 was used to perform all of the analysis, which included descriptive statistics, exploratory factor analysis (EFA), confirmatory factor analysis (CFA) and bivariate correlation analysis. With regard to the EFA, the principal component analysis with Varimax rotation was carried out. The factors were extracted based on the three following analytical criteria: (1) Pairwise deletion, (2) Eigen value more than 1, and (3) factor loading more than 0.50. Of the extracted factors, all factors with 2 or fewer items were removed, based on the notion that a factor should be comprised of at least three items to be robust (Pett, Lackey & Sullivan 2003). A confirmatory factor analysis (CFA), using a one-factor solution for each construct, has also been performed. The advantage of the CFA is that it allows for more precision in evaluating the measurement model (Hinkin 1995), and the results were compared with the EFA for providing validity evidence based on the hypothesis that a valid instrument should produce similar results. Each factor was then evaluated using the KaiserMeyerto-Olkin (KMO) parameter, and only factors with KMO value at 0.60 or above were included in the further analysis (Hair 1998). This was followed by a scale-reliability test. For that purpose, the Cronbach’s Alpha coefficient of internal consistency was calculated, and evaluated along with inter-item statistics. Cronbach’s Alpha is a measure of scale reliability concerned with the proportion of a scale’s total variance that is attributable to a common source, presumed to be the true score of the latent construct being measured. In our case that will be the safety culture. Usually a value above 0.7 is considered acceptable, although some advocate an alpha level of 0.8 or better (Netemeyer, Sharma & Bearden 2003). As the alpha value is a function of, inter alia, the average inter-item correlation; the inter-item

1426

correlation and item-total statistics have also been evaluated. Rules of thumb suggest that the item-total correlation should exceed .50, and the inter-item correlation should exceed .30 (Hair 1998), but it should not exceed .80. An inter-item correlation exceeding .80 suggests that items are duplicates of one another (Pett, Lackey & Sullivan 2003). Then the remaining items went through a last CFA, a five-factor solution, in order to provide each factor’s explained variance. Finally, correlation analysis has been carried out in order to evaluate the construct validity, which is viewed as the extent to which an operational measure truly reflects the underlying safety culture concept, and if they operate in a consistent manner. Based on this analytical process, five factors (1) interaction, (2) reporting practices, (3) competence, (4) local management, and (5) work situation were found to be reliable and valid. The aforementioned factors are presented further in detail in the next section. 4 4.1

RESULTS Results from descriptive analysis

Regarding demographics, 21 different nationalities are represented. The Filipino contingent forms the largest group constituting 63% of the sample, followed by the Norwegian group with almost 11%, and the Polish which represents 9%. The last major group was the Russians with 6%. The other remaining 17 nationalities were represented in a range from 3% to 1%. There is also great variation with regard to employment conditions. All in all, 12% of the sample consists of permanent employees, of whom 80% are Norwegian and 16% from the European Union. 91% of the Norwegians are permanent employee. The remaining 9% are apprentices, substitutes or newly employed on probation. Only 3% of the non Norwegian sailors are permanent employees. With regard to the Filipino seafarers, the largest nationality, 99.6% are contract employees, most on 9 month contracts (62%), followed by 6 month contracts (27%). The extended use of contract employment is reflected in their experience. All in all, 85% had three years or more experience within shipping in general. However, 69% of the sample had worked on the current vessel for only 1 year or less. The employment terms were in general different for the captains. The captains normally do not have sailing periods that exceed 6 months. The most typical sailing period for the captains is 3 months or less. 4.2

Results from factor analyses

From the 9 theoretical safety culture constructs, a five factor solution was derived, (1) interaction, (2)

reporting practices, (3) competence, (4) local management, and (5) work situation. With regard to the ‘‘local management’’, ‘‘competence’’ and ‘‘work situation’’ factor both EFA and CFA result in final solutions consisting of the same items, but with minor differences in factor loading. The CFA included three more items in the ‘‘interaction’’ factor than the EFA, and the final factor, ‘‘reporting practices’’ resulted from only the CFA. Four of the constructs did not pass the reliability tests. The first, ‘‘top management’s safety priorities’’, was excluded due to low representative reliability across subpopulations. This construct also consisted of too few items. The remaining three constructs, ‘‘procedures and guidelines’’, ‘‘responsibility and sanctions’’ and ‘‘working environment’’ were excluded due to low validity, mostly resulting from poor theoretical relationship within the items of each construct. For the further analysis the results from the CFA are used. The 5 factors in question are presented in Table 2 along with number of items and explained variance. Each factors Cronbach’s alpha value and inter item statistics is presented in table 3. The alpha values range from .808 to .878, and the internal item statistics are all within the recommended levels. The five factors are therefore considered to be a reliable and valid reflection of the underlying safety culture concept. Further, table 4 presents the correlation coefficients between the factors, or safety culture dimensions. All correlations are significant at the 0.01 level (2-tailed)

Table 2. Final factors, number of items and explained variance. Factor

Number of items

Explained variance

Interaction Reporting practices Competence Local management Work situation

8 5 4 3 3

35.63% 9.77% 7.12% 5.96% 5.08%

Table 3. Final factors, Cronbach’s alpha and inter-item statistics. Factor

Alpha

Inter-item range

Item-total range

Interaction Reporting practices Competence Local Management Work situation

.878 .808 .839 .866 .817

.360–.606 .335–.761 .497–682 .692–.716 .512–.749

.520–.724 .491–.668 .628–.712 .724–774 .554–.739

1427

Table 4.

Factor correlation matrix. Pearson’s r. F1

F1: Interaction F2: Reporting practices F3: Competence F4: Local management F5: Work situation

F2

F3

1 .323

1

F4

F5

1 .352 .639 .474

.362

.367

1

.494

.322

.441

.444

1

The five safety culture dimensions correlate in a positive direction, which is consistent with the theoretical concept, and they are therefore considered to be a valid reflection of the underlying safety culture construct.

5

DISCUSSION

All three constructs have a good alpha level, and as the alpha levels are concerned with the variance that is common among the items, these constructs also reflect the areas where it is possible to speak about safety culture. With reference to Cooper’s framework towards safety culture, we will further discuss how the organisation’s factors such as crewing strategy, witch includes employment terms, rotations system and policy towards the on board shipping management, may affect the on board safety culture and climate represented by the identified dimensions. The organisation’s structural factors are all to be found within Cooper’s element of situation, while the identified safety culture dimensions are to be found within the elements of person and behaviour. Interaction is the dimension accounting for the largest proportion of the total explained variance, with 35.63%, meaning that with regard to safety culture most of the variance in the original data is explained by this dimension. When taking into account how distinctive a ship is as a work place, this is no surprise. A ship may be characterised as a total institution since both work and leisure time happen at the same place and with few possibilities to interact with the surrounding world (Goffman 1968). In such a setting the crew members are socialised into a common culture and rules of interaction. Schein (2004) refers to this as internal integration. The interaction climate is characterised by lack of stability within the crew due to different terms of employment. First of all, permanent employment seems to be reserved for the Norwegian sailors. Sailors of other nationalities are almost all contract employees. In addition, the length of contract varies and all crew members have different

dates for signing on and signing off. Schein (2004) points out that lack of stability may be a threat to the possibility of developing a culture: ‘‘( . . . ) there must be a history of shared experience, which in turn implies some stability of membership in the group.’’ Even if the crew as a group is in constant change, they all have common history as seafarers. So even if lack of stability within the group indicates that a common culture should not develop on the ship, a common culture of how to act and interact may have developed amongst the seafarers, and when a new crewmember is signed on a new vessel, he knows what is expected from him. However, the question is if such a culture is a safe culture? Reason (2001, 2003) emphasize that to reach a safe culture, the organisation should strive for an informed culture where those who manage and operate the system, both on board and on shore, have current knowledge about the factors that determine the safety of the system as a whole, which again depends on that the crew on board are prepared to report their errors and near misses, and the reporting practice is one of the dimensions deriving from the analyses, explaining 9.77% of the variance. This dimension also includes feedback on reported events. In order to attain good reporting practices, the organisation should strive to create an atmosphere of openness, trust and loyalty. Integrating into the group is also a survival mechanism, and every crewmember will most likely make an effort to integrate. If not, he would most likely have a hard time during his contract period with no possibility to leave the vessel and the other crewmembers. However, to compromise oneself and be open about one’s own mistakes is not always an easy task, especially not in an unknown working environment. Something that may reinforce the crewmembers’ fear of reporting their own mistakes is the ongoing practice that each crew member is evaluated by their senior officer/captain, and based on this report get recommended or not recommended for re-hire. Interviews have revealed that this evaluation practice differs. Some practise an open evaluation where all parts are involved, with focus on how to improve the evaluated crew’s shortcomings, and where the shore organisation seeks to ensure that the evaluation is conducted in as objective a way as possible. At other vessels, the evaluation is closed for insight by the evaluated and may also be highly subjective. Some of the respondents have expressed that by reporting, their next contract may be at stake, or they may meet with other negative consequences. So, lack of stability and constantly changing working groups may sacrifice a trusting and open environment, and thus also the sailors’ commitment to safety. A crew committed to safety is essential, but not enough. Lack of competence may cause a situation where the crew do not identify potential dangerous situations, or create them. Competence, which accounts

1428

for 7.12% of the total variance, is in this setting comprised of activities performed on board the vessel, and is all under the control of the captain, training, drills and familiarisation when signing on. Also, the competence dimension does correlate strongly with the interaction dimension with a correlation coefficient at .639. This indicates that a situation when the sailors are feeling confident with the nature of their task also results in a better interaction climate where conflicts are more likely to be absent. As with the interaction climate, competence will also be affected by the crew stability. A crew member that is constantly signing on new vessels and that has to interact with new crew members and leaders, uses more effort adapting to the new situation, working out how things are done at that specific vessel, the informal structure onboard and so on. When more stability is provided, more effort may be placed on upgrading their competence, and the competence will be kept within the vessel and organisation. Both the training activities and crewing strategy may be controlled by the ship management, and thus these safety culture dimensions are also, to a certain degree, controllable. The dimension of work situation consists of proactive activities as Safe Job Analysis (SJA), safety evaluations and the possibility they have to prioritize safety in their daily work. So how may the organisation affect this? For one, they may supply sufficient crew. Today many vessels are sailing with a smaller crew at the same time as new standards and resolutions like the ISM-code increase the amount of paperwork to be done. Both own observations and interviews reveal that inter alia check lists and SJA are done in a mechanical manner. This may originate from various reasons such as an overload of work, no understanding of the importance of those activities, lack of feedback or improper planning by the local management. The local management dimension, accounts for 5.96% of the explained variance, and the direct effect of local management is relatively small. However, local management is considered to have an indirect effect on the safety climate through the managers, or senior officers, affect on the interaction climate, competence and training activities, reporting practices and the work situation. Again we wish to focus on the importance of stability within the work group. Most captains have a sailing period of 3 months or less, while most of the non Norwegian ratings have a sailing period of 9 months. Most senior officers also have a shorter sailing period then an ordinary rating. Then a rating possibly has to deal with several different leaders during his stay. And each captain and department manager’s leadership style may vary, and are sometimes even destructive, as shown by following comment from a Pilipino engineer. ‘‘The only problem on board is the treatment of senior officers to the lowest rank. (. . . ) There are some senior officers

who are always very insulting towards jr. officers and rating.’’ Schein (2004) regards the leader as an important key figure in the cultural development. At sea the captain holds a key role. The captain is the one in command at every vessel, and according to Schein (2004) the captain’s orientation will affect the working climate, which precedes the existence of a culture. So, in a situation where lack of crew stability impedes the development of a safety culture, the role of the captain is even more vital. Also, it is important to take into account that the leadership style that is practised on board not only affect the sphere of work, but also time off. However the Captains themselves may not be aware of their own importance, or how they affect safety. Most Captains, or other department leaders for that matter, do not have managerial training or education. When adopting a cultural view towards safety, as in this research, in as opposed to a behaviour based view, more emphasis is placed on organisational factors. Decisions regarding crewing strategy, employment terms and managerial development programmes are all strategic decisions made on shore. With reference to Schein’s culture definition, we will argue that the safety culture originates within the organisation on shore. Based on Scheins’ definitions of culture there ought to exist a pattern of shared basic assumptions that may solve the problems the shipping industry is facing. Our case however has revealed an offshore practise characterised by extended use of contract employment, lack of stable working conditions on board the vessels, and little or no use of managerial training and development. This practice does not promote a good safety culture and is considered to has a negative effect on the overall safety level. 6

CONCLUSIONS

The aim of this paper was to analyse the characteristics of the safety culture on Norwegian tankers and bulk carriers, and identify what organisational factors may affect the safety culture on board vessels. Statistical analysis identified five safety related dimensions on board the vessels: interaction climate, reporting practices, competence, local management and work situation. Within shipping the interaction climate is characterised by unstable working conditions. Under such conditions it is difficult to achieve and maintain a stable crew, and proper management becomes even more important. Also the Captain has a vital role, as he has the possibility to directly affect all the other safety related aspects through his own leadership style. The Captains, officers and ratings normally have different employment terms and shift terms. This may jeopardise the development of a sound safety culture as the crew has a poor possibility of developing common behaviour practices and a mutual understanding

1429

of how to do things right. As neither the Captains nor the officers normally have any managerial training, their leadership styles often affect the safety in a negative direction. The on board situation is to a large extend considered to be created by the on-shore crewing strategy and management policy. In order to develop a sound safety culture on-board, the shipping companies should go in new directions and pursue a crewing strategy which offers more favourable employment terms and fixed shifts for all nationalities, and strive for a more stable workforce. Another measure would be to accept the Captain’s and department managers’ roles as leaders, and offer managerial development. A final measure will be to develop a policy and system that ensure proper onboard management. REFERENCES Bakka, D. & Sjøfartsdirektoratet 2004, Hundre år for sikkerhet til sjøs: Sjøfartsdirektoratet 1903–2003, Direktoratet, Oslo. Cooper, M.D. 2000, ‘‘Towards a model of safety culture’’, Safety Science, vol. 36, no. 2, pp. 111–136. DeJoy, D.M. 2005, ‘‘Behavior change versus culture change: Divergent approaches to managing workplace safety’’, Safety Science, vol. 43, no. 2, pp. 105–129. Dekker, S. & Dekker, S. 2006, The Field guide to understanding human error, Ashgate, Aldershot. Department of Transport 1987, mv Herald of Free Enterprise. Geller, E.S. 2001, The psychology of safety handbook, Lewis Publishers, Boca Raton, FL. Goffman, E. 1968, Asylums: essays on the social situation of mental patients and other inmates, Penguin, Harmondsworth. Guldenmund, F.W. 2000, ‘‘The nature of safety culture: a review of theory and research’’, Safety Science, vol. 34, no. 1–3, pp. 215–257. Hair, J.F. 1998, Multivariate data analysis, 5th edn, Prentice Hall, Upper Saddle River, N.J. Hinkin, T.R. 1995, ‘‘A review of scale development practices in the study of organizations’’, Journal of Management, vol. 21, no. 5, pp. 967–988. Justis- og politidepartementet 1991, Scandinavian Star— ulykken, 7 april 1990, Justis- og politidepartementet, Oslo.

Le Meur, C. 2003, Maritime Safety Culture. Li, K.X. & Shiping, Z. 2002, ‘‘Maritime professional safety: prevention and legislation on personal injuries on board ships’’. Nærings- og handelsdepartementet 2007, Stø kurs: regjeringens strategi for miljøvennlig vekst i de maritime næringer, Nærings- og handelsdepartementet, Oslo. National Transportation Safety Board 1990, Grounding of U.S. Tankship EXXON VALDEZ on Bligh Reef, Prince William Sound Near Valdez, AK March 24, 1989. Neal, A., Griffin, M.A. & Hart, P.M. 2000, ‘‘The impact of organizational climate on safety climate and individual behavior’’, Safety Science, vol. 34, no. 1–3, pp. 99–109. Netemeyer, R.G., Sharma, S. & Bearden, W.O. 2003, Scaling procedures: issues and applications, Sage Publications, Thousand Oaks, Calif. Norwegian Shipowners’ Association, www.rederi.no. Perrow, C. 1999, Normal accidents: living with high-risk technologies, Princeton University Press, Princeton, N.J. Peterson, M.F., Ashkanasy, N.M. & Wilderom, C.P.M. 2000, ‘‘Handbook of organizational culture & climate’’ in Sage, Thousand Oaks, Calif., pp. 193. Pett, M.A., Lackey, N.R. & Sullivan, J.J. 2003, Making sense of factor analysis: the use of factor analysis for instrument development in health care research, Sage, Thousand Oaks, Calif. Rasmussen, J. 1997, ‘‘Risk management in a dynamic society: a modelling problem’’, Safety Science, vol. 27, no. 2–3, pp. 183–213. Reason, J. 2001, Managing the risks of organizational accidents, Ashgate, Aldershot. Reason, J. & Hobbs, A. 2003, Managing maintenance error: a practical guide, Ashgate, Aldershot. Schein, E.H. 2004, Organizational culture and leadership, 3rd edn, Jossey-Bass, San Francisco. Sorensen, J.N. 2002, ‘‘Safety culture: a survey of the stateof-the-art’’, Reliability Engineering & System Safety, vol. 76, no. 2, pp. 189–204. Studio Apertura 2004, Sikkerhetskulturer i transportsektoren. Metoder for kartlegging av sikkerhetskultur: Evaluering av noen eksisterende verktøy. Tinmannsvik, R.K. 2008 Robust arbeidspraksis: hvorfor skjer det ikke flere ulykker på sokkelen? Tapir akademisk forl., Trondheim. Yule, S.J. 2003, ‘‘Senior manager’s transformational leadership behaviours for safety’’. Australian Journal of Psychology, vol. 55, no. Supplement.

1430

Article 2 Oltedal, H. & Wadsworth, E. (2010). Risk perception in the Norwegian shipping industry and identification of influencing factors. Maritime Policy & Management, 37(6), 601-623.

MARIT. POL. MGMT., NOVEMBER VOL.

37,

NO.

2010,

6, 601–623

Risk perception in the Norwegian shipping industry and identification of influencing factors HELLE OLTEDAL*y and EMMA WADSWORTHz

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

yFaculty of Technology, Business and Maritime Education, Stord/Haugesund University College (HSH), Bjørnsonsgate 45, Haugesund 5528, Norway zSeafarers International Research Centre (SIRC), Cardiff University, 52 Park Place, Cardiff CF10 3AT, UK In the recent years, safety culture in relation to shipboard safety and organizational factors has received increasing interest within the shipping industry. This study examines this relationship, where risk perception is an indicator for shipboard safety. Data was derived from a survey carried out in 2006, where 1262 questionnaires were collected from 76 vessels. Explorative factor analyses were used to extract factors of safety culture. Analysis of variance was used to assess the associations between the safety cultural factors and demographic and organizational variables. Finally, linear regression analysis was carried out to assess the association between risk perception and safety culture, controlling for the influence of demographic and organizational variables; age, department, vessel type and nature of work. The results suggest that safety-oriented shipboard management style, performance of proactive working practices and good reporting practices all contribute to a better perception of shipboard safety, while a high demand for efficiency contributes to a more negative perception of the safety level. Further, safety is perceived as better when work is performed as a team. To gain a better understanding of risk perception and safety at sea, it would be helpful to further examine the characteristics and influences of teamwork.

1. Introduction The term ‘safety culture’ first appeared after the Chernobyl accident in 1986 [1, 2]. Since then there has been considerable interest in the concept and how it relates to other safety related organizational, sociological and individual conditions, with some researchers suggesting that it may be the main recent issue in organizational safety [3]. However, there is still considerable debate about safety culture with no clear consensus yet on a number of fundamental aspects, including: its definition and scope; the theoretical issues underpinning it; its relationship to safety climate and to other organizational factors and characteristics; and how best to measure it [4–9]. Nevertheless, safety culture has been implicated by a number of recent industrial accident investigations (see, e.g. [1, 10–12]) and it is widely accepted that organizations with a strong safety culture are more effective at preventing workplace accidents and injuries [13]. Broadly, it is assumed that an organization’s strategic approach towards business, operational management and safety is reflected in the local working situation in the form of patterns of staff safety related behaviours, *To whom correspondence should be addressed. E-mail: helle.oltedal@hsh.no Maritime Policy & Management ISSN 0308–8839 print/ISSN 1464–5254 online ß 2010 Taylor & Francis http://www.tandf.co.uk/journals DOI: 10.1080/03088839.2010.514954

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

602

H. Oltedal and E. Wadsworth

perceptions and attitudes, and this is what is commonly referred to as the organization’s ‘safety culture’ [6, 7, 9, 14, 15]. In this article, risk perception is used as an indicator of the actual shipboard safety level, as previous research indicates that risk perception and risk behaviour are strongly correlated [16]. Also, recent research within the maritime industry has suggested risk perception to be a good indicator for safety level in general [17]. An important conclusion of Rundmo’s study [16] was the indication that behaviour is largely constrained by working conditions. When these conditions are not judged to be at a satisfactory level of safety, employees know that the risk of suffering an accident is enhanced and consequently they feel unsafe. Perceived risk, therefore, may be directly affected by an organization’s safety management, demands for efficiency and other local workplace and organizational factors. Reason [5] has identified three organizational driving forces for safety; commitment, competence and cognizance. Commitment refers to motivation and resources. Organizational leaders, such as shipboard management, are important motivational driving forces. Management at local level often reflects the organization’s strategic orientation, whether to pursue commercial fortunes or good safety records and practice. A highly committed organization also makes available sufficient recourses, such as time, equipment and people. Competence is related to the organization’s safety management systems; whether they get in the right experience data, the quality of data, how the organization acts upon it in the form of feedback, training and development of other safety measures. However, neither commitment nor competence is enough if the organization does not have cognizance, or awareness, of the dangers in their operation. Such awareness may be nurtured by active discussions within working groups, proactive activities such as safe job analysis and other activities which give safety and risk continuous attention. These three Cs, commitment, competence and cognizance, may be referred to as cornerstones in safety culture [5]. Seafaring is widely acknowledged to be, comparatively, a very high risk occupation [18–21]. An examination of the statistics for loss of life and personal injuries has ranked seafaring as the second most dangerous occupation in the world [20]. Fatal accident rates among British seafarers, for example, are over 26 times higher than those of other workers [19]. Within the maritime setting, it is traditionally estimated that about 80% of accidents (i.e. individual injury and navigational accidents such as groundings and collisions) are attributable to human error [22, 23]. Such a ‘person oriented’ focus is not unique to the maritime industry [24, 25]. As a result of such a person-oriented focus, procedures and standards are in place to control the behaviour of seafarers, although with limited success, as accidents involving human error seem to be on the rise [26]. However, there has been a recently increasing interest in other potentially influential factors which are found in the organizational and situational context, inducing human error through unfavourable working conditions. This is a move away from blaming individuals towards a more holistic understanding of accidents and their causes [15, 22]. The influence of organizational issues may appear in the form of productivity pressure, poor training or inadequate manning, which may result in unfavourable working conditions, such as increased demand for efficiency and work stress and lack of both resources and competence [5, 23]. Hetherington et al. [23] reviewed the current status of safety in the maritime industry, and the contribution of human factors in the causal chain of events in shipping accidents. She indicates that a particular combination of low manning level, fatigue, stress, work pressure, communication,

Risk perception in the Norwegian shipping industry

603

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

environmental factors, and long periods of time away from home are workplace dangers. When moving away from a person-oriented focus, blaming the individuals, it is such organizational factors that ought to be brought to light and reformed. Within the maritime setting, the notion of safety culture and its relation to safety was first introduced by the International Safety Management (ISM) code [15, 27]. Until recently, however, safety culture has been given relatively little attention within the shipping industry [23, 28]. The aim of this study, therefore, is to explore further the concept of safety culture and its relationship to safety by using risk perception as a proxy variable. Our contribution to the domain of maritime research is to bring forward an estimation of the functional relationship between risk perception and safety culture by using regression analysis.

2. Method 2.1. The Norwegian shipping industry On 1 January 2010, the Norwegian-controlled fleet totalled 40.4 million dwt. Since 1 January 2009 the fleet has decreased by 3%. The number of ships in the fleet stands at 1836. The total fleet has decreased by 40 ships since January last year. The number of ships flying the Norwegian flag has decreased by 48 during the same period, and 46% of the ships now fly the Norwegian flag (NOR, The Norwegian Ordinary Ship Register or NIS, The Norwegian International Ship Register) [29]. The other 54% may be registered in any of the world’s 183 ship registers [30]. On an international basis, the various flag states performances are assessed on standards of safety, environmental and social performance and are maintained and enforced by flag states, in full compliance with international maritime regulations. Based on their performance, the flag states are then categorized into three subgroupings; the white list (good performance), the grey list (mediocre performance) and the black list (poor performance) [31]. The Norwegian-controlled foreign-going fleet and distribution of type of vessel and flag are given in Table 1 (29). On 1 January 2010, seafarers of the nationalities given in Table 2 were employed on Norwegian controlled ships (29).

Table 1.

Norwegian-controlled foreign-going fleet, 1 January 2010: type of vessel and registration.

Type of vessel Passenger vessels Gas tankers Chemical tankers Shuttle/storage tankers Other oil tankers Combined carriers Bulk carriers Other dry cargo vessels Offshore service Total

Number of NOR

Number of NIS

Foreign flag

Total

– – 41 204

9 54 139 8 50 12 58 154 100

11 68 167 37 32 – 93 398 186

26 122 307 51 84 12 151 593 490

260

584

992

1836

6 – 1 6 2

604

H. Oltedal and E. Wadsworth Table 2.

Country of domicile Philippines Norway India Russia Poland Ukraine Latvia Romania China Croatia

Crew by country of domicile: employed on ships. Number of people 21 570 10 370 6135 4020 2505 1140 960 920 660 580

Country of domicile Sweden UK Bangladesh Estonia Brazil Canada Portugal Spain Other countries

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Total

Number of people 490 455 380 320 285 190 160 150 1960 53 250

Statistics from the Norwegian Maritime Directorate, up to 2008, indicate a reduction in most types of accidents except for groundings which has been increasing since 2005, with a marginal decrease in 2008—80, compared to 90 groundings in 2007. In addition, 23 foreign vessels grounded in Norwegian waters in 2008. The Norwegian Maritime Directorate points to causes related to navigation errors, deficient watch keeping, inattention and use of non-authorized electronic equipment. A frequent common underlying cause of groundings is fatigue. However, work related and individual accidents/injuries have shown a pronounced decrease, with a total of 451 registered personal accidents/injuries on Norwegian vessels in 2008, a decrease from more than 1000 in 2003 [32]. 2.2. Questionnaire development In order to examine safety culture, a questionnaire developed by Studio Apertura, a constituent centre of The Norwegian University of Science and Technology (NTNU), in collaboration with the Norwegian Det Norske Veritas (DNV) and the research institution SINTEF (Selskapet for INdustriell og TEknisk Forskning) was used. The development was based on a theoretical review and an evaluation of the eight pre-existing questionnaires, four developed in Norway, two in Denmark and two in the UK [33, 34]. The evaluation was carried out according to the following five criteria: (1) foundation (theoretical foundation, documentation and premises for application), (2) thematic width, (3) practical experience of use, (4) the ability to describe and measure safety culture, and (5) the ability to be used at multiple levels (individual, group, team, company). The questionnaire was made up of 12 sections; background, demographic information and risk perception, with the remaining 10 sections representing dimensions of safety culture; top management’s safety priorities, local management, procedures and guidelines, interaction, work situation, competence, responsibility and sanctions, working environment, learning from incidents and description of the organization. Except for the dependent variable risk perception, all constructs were measured on five-point Likert scale ranging from ‘strongly disagree’ to ‘strongly agree’, or from ‘very seldom/never’ to ‘very often/ always’. Risk perception was measured on a scale ranging from one to ten, with 1 indicating ‘very bad’ and 10 indicating ‘very good’.

605

Risk perception in the Norwegian shipping industry Table 3.

Original population, redefined population and number of participating vessels. Number of vessels in NSA memberlist

Approximate number of Norwegiancontrolled vessels

Number of responding vessels

Total response, % of Norwegian controlled

Bulk carrier Chemical tanker Combined carrier Gas carrier General cargo Oil tanker Shuttle tanker

172 257 20 121 255 79 49

127 171 13 102 215 30 43

20 17 2 6 19 5 7

16 10 15 6 9 16 16

Total

953

702

76

11

White flag Grey flag Black flag

799 91 63

589 67 46

66 10 0

11 15 –

Total

953

702

76

11

Norwegian flag Other flags

492 461

362 340

39 37

11 11

Total

953

702

76

11

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Original population

2.3. Survey sample A total of 150 vessels were randomly selected from the 953 vessels within the Norwegian Shipowners’ Association’s (NSA) list of members for 2005 for the target group of liquid tankers, general cargo and bulk carriers. This number represented approximately 15% of the overall population, which was considered to be large enough to be representative of the population as a whole [35]. A sample of 10% is recommended, but some withdrawals were expected, and thus a 5% margin was included in the original sample. The sample was stratified with regard to status of the vessel’s flag register (white, grey or black listed flag) [31, 36] and type of vessel (general cargo, bulk carrier, oil tanker, gas tanker and chemical tanker). Following the initial selection, telephone calls were made to each company to ask for their participation. Thirty-one companies with 83 vessels agreed to participate, and 45 companies with 67 vessels declined. Reasons for not participating included the following: . Being unable to contact the company despite repeated efforts (23 vessels, 16 companies). . The vessel was not owned by a Norwegian party, and therefore not defined as Norwegian controlled (15 vessels, 8 companies). . Ship management was outsourced to a non-Norwegian country, and therefore not defined as Norwegian controlled (14 vessels, 8 companies). . The company refused to participate (12 vessels, 10 companies). . The remaining vessels were sold (3 vessels, 3 companies). The population was redefined, and vessels managed from a non-Norwegian country were not considered Norwegian controlled. The original and redefined populations (Norwegian controlled) are given in Table 3, along with the number and percentage of responding vessels in each category. In total, 1574 questionnaires were distributed to 83 tankers and bulk carriers. Seventy-six vessels from 29 companies

606

H. Oltedal and E. Wadsworth

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

returned a total of 1262 forms, which gives an individual response rate of 80.2%, a vessel response rate of 91.5% and a company response rate of 93.5%. Unfortunately, no company with vessels flying a black listed flag was willing to participate. 2.4. Statistical analyses The Statistical Package for the Social Sciences (SPSS) v.16.0 was used to perform all the analyses, which included descriptive statistics, exploratory factor analysis (EFA), correlation analysis, analysis of variance (ANOVA) and linear regression analysis (LRA). With regard to the EFA, principal component analysis with Varimax rotation was carried out, in order to explore the latent underlying dimensions reflecting the concept of safety culture on board. Factors were extracted based on the three following analytical criteria: (1) pairwise deletion, (2) eigenvalue more than 1.0 and (3) factor loading more than 0.50. This was followed by a scale reliability test evaluated upon the following three criteria: (1) Cronbach’s alpha coefficient 40.70, (2) item-total correlation40.40 and (3) inter-item correlation40.30 and50.80 [37, 38]. Correlation analysis was performed to evaluate the construct validity. To be a true reflection of the underlying safety culture dimension, factors should correlate in a consistent manner with the theoretical concept. One-way ANOVA was carried out to explore any associations between the demographic (nationality, age, department and vessel type) and organizational (work description) variables and both the dependent variable risk perception and the independent safety culture dimensions. Finally, LRA was carried out in order to assess the associations between risk perception and the dimensions of safety culture, controlling for any potentially influential demographic and organizational factors. Forward stepwise regression analysis was carried out in one block (entry p50.05).

3. Results 3.1. Demographics Of the 1262 returned questionnaires, the following three subgroups were excluded from the analyses: the captains, whose closest managers are shore-based and not shipboard personnel (n ¼ 66); catering personnel (n ¼ 126), who have fewer maritime education requirements than deck and engineering personnel; and a group of questionnaires forwarded by a Philippine crewing agency (n ¼ 81). The latter group were excluded as there was no way to control whether the employing company was within the target group (i.e. Norwegian controlled). The remaining 989 responses (78%) were used in the analyses presented below. See Table 4 for demographical distribution. Only one of the respondents was female, and the rest male. Filipino seafarers dominate the sample, followed by Polish and Norwegian. 3.2. Dependent variable The dependent variable, risk perception, was measured on a 10-point scale (from 1 ‘very bad’ to 10 ‘very good’) with the question ‘All in all, how would you assess the safety in your working situation?’ Overall safety level was perceived as relatively good, with a mean at 8 (standard deviation (SD) ¼ 2).

607

Risk perception in the Norwegian shipping industry Table 4.

Demographic profiles of respondents. N

Age 531 years 31–40 years 41–50 years 51–60 years 460 years Missing

30.7% 28.2% 28.8% 10.4% 0.7%

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Total

304 279 285 103 7 11

N Gender Male Female Missing Total

99.9% 0.0%

973 1 15 989

63.0% 8.4% 8.2% 5.9% 3.4% 2.8% 2.7% 5.6%

623 83 81 58 34 28 27 34 21

989

Vessel type General cargo/bulk Liquid tanker Missing Total

43.1% 56.9%

426 563 – 989

Nationalities Filipino Polish Norwegian Russian Indian Romanian Chinese Others (13 nations) Missing Total

Table 5.

Age 531 years 31–40 years 41–50 years 51–60 years 460 years Vessel type General cargo/bulk Liquid tanker Work description Individual Team

989

ANOVA analysis between risk perception and groups. N

Mean

SD

SE

df

F

Sig.

823 263 229 234 91 6 831 360 471 736 158 578

8.09 8.06 7.96 8.12 8.47 7.33 8.09 8.06 8.11 8.07 7.24 8.30

1.714 1.669 1.778 1.765 1.537 0.816 1.710 1.772 1.662 1.713 2.210 1.710

0.060 0.103 0.117 0.115 0.161 0.333 0.063 0.093 0.077 0.063 0.176 0.063

4

1.796

0.128

1

0.221

0.638

1

50.799

0.000

Note: SE, standard error; df, degrees of freedom.

The results from ANOVAs assessing differences in risk perception by the demographic (age and vessel type) and organizational (team or individual work) factors are presented in Table 5. The results show no significant differences between age groups or vessel types However, there were significant differences between those who work on an individual basis and those who work in a team. When work is on an individual basis, risk perception is higher (indicated by a lower mean score corresponding to poorer perception of safety (i.e. higher perception of risk)).

608

H. Oltedal and E. Wadsworth Table 6.

Factor

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Competence Interpersonal relation Shore orientation Ship management Work practices Feedback Demand for efficiency Reporting practices

Factor reliability test.

Cronbach’s alpha

Inter-item range

Item-total range

0.907 0.856 0.745 0.868 0.788 0.780 0.754 0.842

0.549–0.785 0.369–0.718 0.313–0.422 0.654–0.730 0.345–0.735 0.416–0.761 0.368–0.723 0.736–0.736

0.646–0.820 0.543–0.710 0.470–0.530 0.721–0.777 0.450–0.702 0.449–0.722 0.417–0.692 0.736–0.736

3.3. Factor analysis Explorative factor analysis of the items in the questionnaire resulted in an eight factor solution, explaining 67.10% of the total variance. The Kaiser–Meyer–Olkin (KMO) statistic was very high (0.888) indicating the existence of a strong relationship between the items. The factor solution is presented in the Appendix along with each item’s factor loading. The resulting factors, or dimensions of safety culture, were as follows: (1) (2) (3) (4) (5) (6) (7) (8)

Competence level—training and education. Interpersonal relationship among the crew. Shore orientation towards risk and safety management. The ship management’s safety engagement. Proactive work practices. Feedback from reported conditions. Demand for efficiency. Reporting practices of near miss and incidents.

All factors returned a scale reliability estimate (Cronbach’s alpha) above 0.70. All inter-item statistics are also between the recommended levels of 0.30 and 0.80, and item-total above 0.40. As given in Table 6, the overall evaluation of the scales demonstrates a good internal consistency, and therefore also a good representation of their underlying safety culture dimensions. As given in Table 7, all the eight safety culture dimensions correlate in consistence with the theoretical concept, both internally and with the dependent variable of risk perception (DV). They are therefore considered to be a valid reflection of the underlying safety culture construct. Table 7 presents the correlation coefficients between the factors (corr.), along with significance level (sig.) and number of respondents (N). All correlations are significant at the 0.001 level (two-tailed). 3.4. Univariate analyses The results from the analyses of variance testing for differences on the safety culture dimensions by demographic (age and vessel type) and organizational (team or individual work) factors are presented in Table 8. When significant differences are found, number of respondents, mean and SD for each subcategory are also presented. Summarized results from the ANOVA are given in Table 9. The detailed results in Table 8, which are summarized in Table 9, show significant differences between age groups on the dimensions of interpersonal relationship and

Interpersonal relations

Shore orientation

Local management

Working practices

Feedback

Demand for efficiency

Reporting practices

Risk perception

F2

F3

F4

F5

F6

F7

F8

DV

Corr. Sig. N Corr. Sig. N Corr. Sig. N Corr. Sig. N Corr. Sig. N Corr. Sig. N Corr. Sig. N Corr. Sig. N Corr. Sig. N

Note: **denotes significance at 0.001 level.

Competence

F1

0.569** 0.000 884 0.283** 0.000 800 0.331** 0.000 935 0.432** 0.000 933 0.238** 0.000 933 0.251** 0.000 907 0.298** 0.000 831 0.277** 0.000 815

1

F1

0.412** 0.000 764 0.407** 0.000 880 0.490** 0.000 874 0.243 0.000 752 0.351** 0.000 853 0.321** 0.000 782 0.321** 0.000 759

1

F2

Table 7.

0.354** 0.000 791 0.279** 0.000 788 0.251** 0.000 714 0.507** 0.000 771 0.329 0.000 782 0.338** 0.000 691

1

F3

0.473** 0.000 922 0.280** 0.000 788 0.344** 0.000 898 0.346** 0.000 819 0.406** 0.000 804

1

F4

Factor correlation matrix.

0.273** 0.000 790 0.309** 0.000 901 0.372** 0.000 817 0.374** 0.000 800

1

F5

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

0.245** 0.000 775 0.469** 0.000 742 0.207** 0.000 692

1

F6

0.318** 0.000 801 0.358** 0.000 784

1

F7

0.316** 0.000 724

F8

Risk perception in the Norwegian shipping industry 609

610

H. Oltedal and E. Wadsworth Table 8.

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Dimension (in bold)

ANOVA analysis between factors and groups.

N

Mean

SD

SE

df

F

Sig.

Competence Age 531 years 31–40 years 41–50 years 51–60 years 460 years Vessel type Liquid tanker Dry cargo Work description Individual Team

952 297 273 276 99 7 963 551 412 733 156 577

21.39 21.28 20.97 21.65 22.15 20.71 21.41 21.58 21.18 21.37 20.67 21.56

0.75 2.74 2.91 2.54 2.50 4.39 1.75 2.71 2.79 2.73 3.07 2.60

0.08899

4

4.384

0.002

0.08853 0.11527 0.13743 0.10074 0.24561 0.10074

1

4.835

0.028

1

13.497

0.000

Interpersonal relations Age 531 years 31–40 years 41–50 years 51–60 years 460 years Vessel type Work description Individual Team

888 279 257 257 90 5 898 677 145 532

25.03 24.81 24.46 25.46 26.07 25.40 25.14 24.91 23.52 25.28

3.18 3.36 3.20 2.96 2.83 3.21 3.18 3.17 3.61 2.94

0.10684 0.20089 0.19955 0.18475 0.29848 1.43527 0.10616 0.12197 0.29992 0.12725

4

6.072

0.000

1 1

0.995 37.201

0.322 0.000

Shore orientation Age 531 years 31–40 years 41–50 years 51–60 years 460 years Vessel type Work description Individual Team

797 256 229 228 78 6 805 622 126 496

12.57 12.85 13.00 11.82 12.60 12.33 12.56 12.48 13.71 12.17

3.67 3.73 3.67 3.43 3.95 3.01 3.70 3.69 3.31 3.72

0.13004 0.23336 0.24241 0.22703 0.44712 1.22927 0.13076 0.14785 0.29451 0.16692

4

3.583

0.007

1 1

1.022 17.885

0.312 0.000

Local management Age Vessel type Liquid tanker Dry cargo Work description Individual Team

942 953 546 407 722 153 569

12.70 12.70 12.86 12.49 12.75 11.87 12.98

2.26 2.27 2.11 2.46 2.33 2.70 2.16

0.07364 0.07369 0.09047 0.12201 0.08661 0.21802 0.09055

4 1

2.141 5.999

0.074 0.014

1

28.646

0.000

Work practices Age Vessel type Work description Individual Team

937 948 722 154 565

17.39 17.39 12.75 16.36 17.65

0.28 2.27 2.33 2.44 2.16

0.07433 0.07378 0.08661 0.19689 0.09082

4 1 1

1.848 0.004 28.646

0.118 0.950 0.000

(continued )

611

Risk perception in the Norwegian shipping industry Table 8. Continued.

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Dimension (in bold)

N

Mean

SD

SE

df

F

Sig.

Feedback Age Vessel type Liquid tanker Dry cargo Work description Individual Team

800 808 465 343 614 126 488

11.03 11.03 10.78 11.35 11.31 10.24 11.58

2.93 2.93 2.90 2.92 2.90 3.04 2.80

0.10347 0.10285 0.13470 0.15777 0.11696 0.27055 0.11696

4 1

1.398 7.429

0.233 0.007

1

22.350

0.000

Demand for efficiency Age Vessel type Work description Individual Team

912 922 705 144 561

7.14 7.13 7.03 7.76 6.85

2.64 2.64 2.63 2.61 2.63

0.08726 0.08692 0.09899 0.21709 0.09899

4 1 1

1.208 0.243 13.942

0.306 0.622 0.000

Reporting practices Age Vessel type Work description Individual Team

912 922 705 136 509

7.14 7.13 7.03 6.76 7.42

2.64 2.64 2.63 2.07 2.03

0.08726 0.08692 0.09899 0.17781 0.09017

4 1 1

1.208 0.243 13.942

0.306 0.622 0.000

Note: SE, standard error; df, degrees of freedom.

shore orientation. Those of age 41 and above perceive interpersonal relationships to be better than those of age 40 and below. Those between 31 and 40 years perceive the shore organization to be less safety orientated than the other groups. The group between 41 and 50 years had the best perception of the shore organizationâ&#x20AC;&#x2122;s safety orientation. Type of vessel showed significant differences on the dimensions of competence, local management and feedback. Those working on dry cargo vessels perceive the feedback on reported experience data to be better than those working on liquid tankers. However, those working on liquid tankers had better perception towards level of competence and their local management than those working at dry cargo vessels. Work description, teamwork versus individual, showed significant differences on all dimensions of safety culture. Those working in a team perceived the dimensions of competence, interpersonal relationship, local management, feedback and reporting practices to be better than those working on an individual basis. Those working in a team also felt less demand for efficiency and perceived the shoreside company to be more safety orientated. 3.5. Results from regression analysis Table 10 presents the results from the regression analysis. The regression results indicate that local management, working practices and reporting practices have a positive association with risk perception, and that demand for efficiency has a negative association with risk perception. Moreover, working situation has a positive association when work is performed as a team in contrast to

Group Age Vessel type Work description 0.002 0.028 0.000

Competence

Table 9.

0.000 0.322 0.000

Interpersonal relations 0.007 0.312 0.000

Shore orientation 0.074 0.014 0.000

Local management 0.118 0.950 0.000

Working practices 0.233 0.007 0.000

Feedback

Summarized results: ANOVA analysis between groups and safety culture dimensions.

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

0.306 0.622 0.000

Demand for efficiency

0.306 0.622 0.000

Reporting practices

612 H. Oltedal and E. Wadsworth

613

Risk perception in the Norwegian shipping industry Table 10.

Constant Local management Demand for efficiency Working practices Reporting practices Work description

Results from regression analysis. R2 change

B

SE B

0.184 0.086 0.051 0.031 0.016

3.126 0.145 0.146 0.131 0.139 0.491

0.687 0.033 0.027 0.033 0.035 0.163

0.200* 0.236* 0.173* 0.169* 0.119*

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Note: R2 ¼ 0.35, *p50.001.

when work is performed on an individual basis. None of the demographic data were significantly associated with risk perception. Risk perception and significant safety culture dimensions will be further discussed in the following section.

4. Discussion 4.1. Risk perception Overall safety in the respondents’ working situation is perceived as very high, which may indicate a relatively good safety standard. This result is surprising as seafaring is considered to be a high-risk occupation [19, 20, 39]. One explanation for the good risk perception is that the sample is representative only for vessels flying a white or grey flag, which is presumed to perform better on safety. A second explanation is based upon voluntary participation, that the companies that do not prioritize safety may be those who did not participate. A third explanation for the high score is negligence of danger as a coping strategy in a high-risk environment. Similar high scores on safety are also seen among fishermen and employees of offshore service vessels, even though both are considered high-risk industries [40]. A fifth explanation is that risk perception is not a good indicator of ‘objective’ risk levels after all when it comes to the maritime offshore industry. A final explanation may be related to the implementation of many policies focusing on safety, such as the ISM Code and International Ship and Port Facility Security (ISPS) Code, which may have improved the overall standard of safety on board. However, there is also another side to the picture of feeling safe. Perception is assumed to influence behaviour in such a way that more risky actions are taken by those who feel at greater risk, while fewer are taken by those who feel safer [41]. Hence, the perception of safety per se, may induce accidents. Similar situations are also suggested at the organizational level, in that a low accident rate gives an impression of safety, which may or may not reflect the reality, resulting in safety being down prioritized, and thus the organization becoming more prone to accidents [5]. The regression analysis suggests that variance in risk perception is associated with local management, working practices, demand for efficiency, reporting practices and work description (i.e. if work is normally done on an individual basis or in a team).

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

614

H. Oltedal and E. Wadsworth

4.2. The importance of working in a team Whether or not a person works in team or on an individual basis is an organizational decision. The results suggest that when working in a team, the perception of safety is better than when working on an individual basis. It is of particular interest that team versus individual work resulted in significant differences on all the eight dimensions of safety culture, as well as the risk perception variable, suggesting that teamwork is associated with better safety culture overall. When working alone, one has to rely on oneâ&#x20AC;&#x2122;s own experience and competence alone, both during normal work duties and if something unexpected should occur. In teamwork, however, each individual contributes with unique experience and competence, which may contribute to enhanced situational awareness and handling of unexpected situations. Competence is related to training and education, which along with experience are important elements to shape skill-, rule- and knowledge-based activities and/or decisions [5, 14, 42]. Crew members with high levels of experience, education and training are expected to be better able to recognize, handle and control hazardous shipboard situations [43]. Inadequate competence related conditions, with consequent lack of risk awareness, have also been found to be an important influencing factor in collisions and groundings [22], along with inadequate teamwork [43]. The results also suggest a strong relationship between the interpersonal relationship dimension and competence. When competence relates to skill-, ruleand knowledge-based activities and/or decisions, safety in practice relates to people questioning the established mind set, work plans and the like. Such behaviour is assumed to not only increase situational awareness, but also lead to increased competence. The interpersonal relationships amongst crew are assumed to be a determinant factor for creating an open and trusting relationship facilitating such behaviour [5]. Also working in a team and on-the-job training have been found to be important within navigation, as not all skills can be learned through formal education and training or simulation [44]. Research initiated in 1990 by the US Office of Naval Research, with scenario-based training being a core concept, also points to the importance of such training for, inter alia, adequate measurement of team performance, possibilities for diagnosing team performance and knowledge about how to remediate poor performance (both process and outcome measures) [45]. This research [45] has also identified four dimensions underlying good teamwork: (1) information exchange, (2) communication, (3) supporting behaviour and (4) team leadership. The importance of teamwork has received attention from others. For example, Mitropoulos and Cupido [46] found that accidents are more influenced by social relations in teamwork rather than unsafe acts. 4.3. The importance of management Perception of management, i.e. management commitment, management style, management visibility, etc. is the most commonly measured dimension in safety research in general [42] and also the most replicated across studies [47]. Our results also suggest an association between management and risk perception. By sincere safety engagement, involvement and follow-up, the management style at the vessels is assumed to convey to the crew the importance of safety in their work at the vessel, and for the organization as a whole. However, at sea, most department leaders, or the captain for that matter, do not have managerial training or education. Onboard management style is therefore left to each individual, and may vary substantially

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Risk perception in the Norwegian shipping industry

615

from vessel to vessel or from one sailing period to another. The management orientation will affect the shipboard safety by creating a shipboard atmosphere, i.e. whether it is possible to report all kinds of experience data without getting sanctions [15]. Across other studies, two critical functions of management have been found; to help the team to compete a task, and to keep members maintained and function as, inter alia, developing a positive culture [48]. From the construction industry, also considered a highrisk occupation, it has been found that successful managers use the following management practice: conduct new workers orientation, watch out for vulnerable crew members, analyse productivity problems with the crew, respond to good work, and create a calm and friendly job atmosphere [46]. It is also suggested that effective management should be adjusted to the situation. Some studies indicate that when situations are highly standardized or when the team members are competent, directive leadership becomes unnecessary [48]. When leading and working in a team, the managementâ&#x20AC;&#x2122;s shifting role may be challenging. However, team leaders can be trained to shift their role from: being a supervisor; to being a team member engaging in command-and-control activities; or to being a facilitator who helps teams diagnose and discuss their own performance in an open, trusting and constructive manner [45]. Also, how to manage depends on the situation, be it normal operations or an emergency or crisis [43]. The ship managementâ&#x20AC;&#x2122;s importance for safety within the shipping industry is also suggested in other research [15, 23, 49]. 4.4. The importance of proactive work practices Proactive work practices are activities where safety is discussed with others, or by means of, inter alia, safe job analysis and safety evaluation. The intention is to identify weaknesses and to take remedial actions before something fails or becomes critical, and such activities may also be performed by single individuals [5, 50]. Our results suggest that when proactive activities are performed, the working environment is perceived as safer. Our results also indicate that proactive activities are more frequently performed when working in a team. These activities are performed most efficiently when the working team are actively in dialogue, exploring all possibilities of modes of failure with no restrictions. As discussed above, when a team is composed of members with a variety of experience and competence, the possibility for error detection increases, as long as the interpersonal relationship among the team members allows it. For example, other research points to a situation in a harbour, where each of the team members committed at least one error. However, the error was detected and corrected by another team member [44] and thus possible failure avoided. Seifert and Hutchins [44] also suggest that group identity appeared to provide motivation for individuals to participate in minimizing error. Each person in the team feels responsible not only for his own job, but for all parts and processes. Hence proactive work practices are not limited to formal activities done before each specific work operation, such as safe job analysis and the use of pre-printed error detection forms and tools, but happen continuously during each work operation. One pitfall is to become overly fixed within prescribed patterns of thinking [51]. Other more recent research also points to the problem of error detection in planning, as opposed to on-the-job error detection, as this requires people to strive to see things in a new way each time [52]. The same research [52] also favours on-the-job error detection and correction stimulated by teamwork. When it comes to the possibility of foresight, another distinction may be made between emergency and crisis [43]. An emergency is defined as a situation outside normal operating parameters, and the

616

H. Oltedal and E. Wadsworth

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

necessary corrective actions are more often based on documented procedures. Crises, on the other hand, differ as they might be more difficult to successfully handle based on pre-planned procedures. These situations are often more complex and unpredictable than emergencies, and thus require a creative and flexible response [43]. It has been observed that within shipping, the more formal proactive approaches sometimes are performed in a mechanical manner, and not as intended [15]. Thus, these may also be less efficient than the error detection and prevention that happens during operations on an informal basis. 4.5. The importance of demand for efficiency Work pressure is also noted to be one of the most common themes in safety surveys [42]. As shipping is an international and global activity, increased competitiveness, cost reduction and demand for efficiency will always, to a certain degree, be present. When, and if, the company communicates this through the organization, the demand for efficiency may result in people crossing the boundaries for safe behaviour [53]. The origin of demand for efficiency may also be traced back to organizational design. For example, error diagnosis and safety may improve by including more crew members in the form of expertise in redundant units within the system. However, this also increases overall manning costs due to frequently underutilized expertise and training [44]. If a specific area or department is constantly double manned in order to handle unplanned and critical events better, the extra personnel may be redundant most of the time, i.e. during ‘normal’ operations. The result is that organizations may choose to have a manning level designed for normal operations, presupposing that everything occurs as planned. Thus, the race for cost-effectiveness directly affects thoroughness when situations are deviating from the pre-planned, or in more hectic operations. The resulting high workload will then be followed by work stress, perceived demand for efficiency and trespassing of safety boundaries. We suggest that there are parallels between demand for efficiency due to economic concerns and responding to emergencies and crises. The common factor is decision making under stress. When a situation is deviating from normal operations, i.e. an escalating marine emergency, it is suggested that two fundamental and interrelated skills are required: (1) situation assessment (what is the problem) and (2) decision making (what shall I do). Moreover, it is noted that experienced people, as opposed to novices, are generally those making the decisions. Also, if decisions need to be made quickly, time pressure becomes an additional stressor [43]. From other research, stressors have been divided into the following three categories: (1) characteristics of the environment (e.g. noise, heat, cold, vibration, altitude) that could have a direct physical impact on operators; (2) task complexity or individual-level stressors (e.g. time pressure, workload, information uncertainty and negative feedback) that occur as a result of the functions required in the process of individual performance; and (3) teamwork factors (e.g. lack of cohesion, communication delays, intra-group conflict and role ambiguity) that could act either as ‘stressors’ or as ‘stress moderators’. Team characteristics affect the operator’s perception of resources and may either exacerbate or alleviate the effects of individual-level stressors [54]. Our results suggest that when working in a team, the crew felt less demand for efficiency, and teamwork may thus be a counteracting buffer.

Risk perception in the Norwegian shipping industry

617

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

Nevertheless, both on the individual and organizational levels, the trade-off should be balanced so that efficiency or thoroughness does not happen at the cost of acceptable safety. Other research also points to demand for efficiency as one of the most frequent reasons for violations of procedures and checklists [55], and an important influencing factor in collisions and groundings [22] as within the offshore oil industry [44]. Research from Wright (1986), cited by Mitropoulos and Cupido (2009 [40]), highlights production pressure and the consequent development of shortcuts, suggesting that while this did not present a problem for regular employees who were familiar with work conditions and processes, the shortcuts presented a more hazardous problem for contract employees less familiar with the production system [46]. 4.6. The importance of reporting practices A near miss and incidents are situations that could have serious outcomes under slightly different circumstances. They are more frequent than accidents, and regarded as useful information for preventing accidents. A good reporting culture where people are willing and motivated to report their own near misses and mistakes is regarded as a cornerstone in order to achieve a safe culture. Interpersonal relationships, i.e. openness, communication, problem solving and appreciation, lay the foundations for the possibility of such reporting without the risk of negative consequences or reactions such as blame [5]. This is supported by our results, which indicate that interpersonal relationship is associated with reporting frequency. Our results also suggest that when working in a team, the crew members are more likely to place a report than when working on individual basis. One reason could be positive group dynamics emphasizing safety and the importance of reporting. Another reason could be that the near miss or incidents have been observed, and therefore are more difficult to cover up. Reporting practices were also negatively associated with demand for pressure, which suggests that when having high work load, the crew find less time to place a report. A good reporting culture is the basis for an organization to be informed about the risks in their working situation. When managers are informed, necessary precautions may be taken, which is assumed to bring increased safety. Previous research from the shipping industry suggests that reporting practices are influenced and characterized by the following: (1) lack of feedback given upon the reportsâ&#x20AC;&#x201D;with feedback perceived to be inadequate, too general and delayed, (2) lack of understanding of the importance of the reports as preventive measures, (3) difficulties with defining a near miss, which result in fewer reports, (4) fear of being blamed and negative consequences; and finally (5) a complicated reporting system which is difficult to relate to and time consuming [55].

5. Conclusion, limitations and suggestion for future research The objective of this article was to assess the relationship between risk perception and safety culture in the Norwegian-controlled shipping industry, and to identify which factors affect risk perception. Risk perception has been used as a proxy measure of the overall safety level on board vessels. The results suggest that a safety-oriented shipboard management style, performance of proactive working practices and good reporting practices, all contribute to

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

618

H. Oltedal and E. Wadsworth

a better perception of the on-board safety level, while a high demand for efficiency contributes to a poorer perception of the safety level. Description of work—team versus individual work situation—had significant association with risk perception. The results suggest that safety is perceived as better when work is performed as part of a team, than on an individual basis. Also, when working in a team, management was perceived to be more safety oriented, proactive working practices were carried out more frequently and near misses and incident reports were reported to be placed more frequently. Moreover, when working in groups, the perception of efficiency pressure was less. However, none of these factors should be addressed in isolation from the others. As followed from the discussion, they are all important and mutually dependent. The sample is representative of vessels flying a white and grey flag only, as those registered under a black listed flag did not want to participate, As participation was voluntary on behalf of the company, we assume that those participating do emphasize safety, in general, and the development of a sound safety culture. Thus, it is not possible to generalize the results to the overall population. To understand risk perception and safety at sea, it would be helpful to further examine the characteristics of teamwork along with the concept of group identity. Shipping is distinctive among industry sectors in this respect in part because its teams are often international and less stable due to different employment terms, and in addition because the vessel is a 24 h society influencing all aspects of life on board, not simply working processes. It would also be of interest to examine risk perception both in general and per factor among the different nationalities of the respondents. We also suggest that future research should address differences between white, grey and black listed flags of registration.

Acknowledgements The authors thank Neil Ellis at Seafarers International Research Centre (SIRC) for the useful comments and suggestions.

References 1. INTERNATIONAL ATOMIC ENERGY AGENCY, 1991, Safety Culture (Vienna: IAEA). 2. INTERNATIONAL ATOMIC ENERGY AGENCY, 1992, The Chernobyl Accident: Updating of INSAG-1: INSAG-7 (Vienna: IAEA). 3. TURNER, B. A. and PIDGEON, N. F., 1997, Man-Made Disasters, 2nd ed. (Oxford, UK: Butterworth–Heinemann). 4. CHOUDHRY, R. M., FANG, D. and MOHAMED, S., 2007, The nature of safety culture: A survey of the state-of-the-art. Safety Science, 45(10), 993–1012. 5. REASON, J., 2001, Managing the Risks of Organizational Accidents (Aldershot, England; Burlington, VT: Ashgate). 6. COOPER, M. D., 2000, Towards a model of safety culture. Safety Science, 36(2), 111–136. 7. GULDENMUND, F. W., 2000, The nature of safety culture: A review of theory and research. Safety Science, 34(1–3), 215–257. 8. NEAL, A., GRIFFIN, M. A. and HART, P. M., 2000, The impact of organizational climate on safety climate and individual behavior. Safety Science, 34(1–3), 99–109. 9. PETERSON, M. F., ASHKANASY, N. M. and WILDEROM, C. P. M., 2000, Handbook of Organizational Culture & Climate (Thousand Oaks, CA: Sage). 10. CAIB, 2003, Columbia Accident Investigation Board Report. http://caib.nasa.gov/news/ report/volume1/default.html, retrieved 10 March 2010.

Risk perception in the Norwegian shipping industry 11. 12. 13. 14. 15.

16.

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35.

619

CULLEN, W. D., 2001, The Ladbroke Grove Rail Inquiry. Part 2 Report. http:// www.railwaysarchive.co.uk/docSummary.php?docID=48, retrieved 10 March 2010. HIDDEN, A., 1989, Investigation into the Clapham Junction Railway Accident. http:// www.railwaysarchive.co.uk/docSummary.php?docID=36, retrieved 10 March 2010. BARAM, M. and SCHOEBEL, M., 2007, Safety culture and behavioral change at the workplace. Safety Science, 45(6), 631–636. GULDENMUND, F. W., 2007, The use of questionnaires in safety culture research – an evaluation. Safety Science, 45(6), 723–743. OLTEDAL, H. A. and ENGEN, O. A., 2008, Local management and its impact on safety culture and safety within Norwegian shipping. In: Safety, Reliability and Risk Analysis Theory, Methods and Applications, edited by M. Artorel, C. Guedes and J. Barnett (Valencia, Spain: European Safety and Reliability Association), pp. 1423–1430. RUNDMO, T., 1996, Associations between risk perception and safety. Safety Science, 24(3), 197–209. GRABOWSKI, M., AYYALASOMAYAJULA, P., MERRICK, J. and MCCAFFERTY, D., 2007, Accident precursors and safety nets: Leading indicators of tanker operations safety. Maritime Policy and Management, 34(5), 405–425. HA˚VOLD, J. I., 2005, Safety-culture in a Norwegian shipping company. Journal of Safety Research, 36(5), 441–458. ROBERTS, S. E. and MARLOW, P. B., 2005, Traumatic work related mortality among seafarers employed in British merchant shipping 1976–2002. Occupational and Environmental Medicine, 62(3), 172–180. ANDERSON, P., 2003, Cracking the Code: The Relevance of the ISM Code and its Impact on Shipping Practices (London: Nautical Institute). HANSEN, H. L. and PEDERSEN, G., 1996, Influence of occupational accidents and deaths related to lifestyle on mortality among merchant seafarers. International Journal of Epidemiology, 25(6), 1237–1243. MACRAE, C., 2009, Human factors at sea: Common patterns of error in groundings and collision. Maritime Policy and Management, 36(1), 21–38. HETHERINGTON, C., FLIN, R. and MEARNS, K., 2006, Safety in shipping: The human element. Journal of Safety Research, 37(4), 401–411. CRICHTON, M., 2005, Attitudes to teamwork, leadership, and stress in oil industry drilling teams. Safety Science, 43(9), 679–696. HEINRICH, H. W., ROOS, N. and PETERSEN, D., 1980, Industrial Accident Prevention: A Safety Management Approach, 5th ed. (New York: McGraw-Hill). FRANK, J. and OSLER, D., 1998, Accidents double on fleet boom, officer shortage, Lloyd’s List, 21 February. LE MEUR, C., 2003, Maritime Safety Culture. http://www.ismcode.net/research_projects/ chrystelle_safety_paper.pdf, retrieved 10 March 2010. HA˚VOLD, J. I., 2000, Culture in maritime safety. Maritime Policy and Management, 27(1), 79–88. NORWEGIAN SHIPOWNERS’ ASSOCIATION, 2010, Quarterly Information – Shipping and Offshore Activities, No. 1 2010. http://www.rederi.no/default.asp?V_ITEM_ID= 502&AID=6874&TEMPORARY_TEMPLATE=79, retrieved 10 March 2010. LLOYD’S REGISTER FAIRPLAY, 2008, World Fleet Statistics, retrieved from International Maritime Organization upon request. MARISEC, 2008, Shipping Industry Flag State Performance Table. http://www.marisec. org/flag-performance/FlagStatePerformanceTable09.pdf, retrieved 10 March 2010. NORWEGIAN MARITIME DIRECTORATE, 2009, Annual Report 2008. http://www.sdir.no/ upload/Om%20sj%C3%B8fartsdirektoratet/%C3%85rsmeldinger/%C3%A5rsmelding %202008%20til%20nett.pdf, retrieved 10 March 2010. SINTEF, 2003, Sikkerhetskulturer i transport: En kunnskapsoversikt. Report No. STF22 A03300, ISBN: 82-14-02830-2. STUDIO APERTURA, 2004, Sikkerhetskulturer i transportsektoren. Metoder for kartlegging av sikkerhetskultur: Evaluering av noen eksisterende verktøy. Arbeidsrapport 1b/2004. NEUMAN, W. L., 2000, Social Research Methods: Qualitative and Quantitative Approaches, 4th ed. (Boston: Allyn and Bacon).

620 36. 37. 38. 39. 40. 41. 42. Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55.

H. Oltedal and E. Wadsworth MARISEC, 2006, Shipping Industry Guidelines on Flag State Performance. http:// www.marisec.org/flag-performance/, retrieved 10 March 2010. HAIR, J. F., 1998, Multivariate Data Analysis, 5th ed. (Upper Saddle River, NJ: Prentice Hall). FIELD, A., 2005, Discovering Statistics Using SPSS: (and sex, drugs and Rock ’n’ Roll), 2nd ed. (London: Sage). HA˚VOLD, J. I., 2005, Safety-culture in a Norwegian shipping company. Journal of Safety Research, 36(5), 441–458. BYE, R. and LAMVIK, G., 2007, Professional culture and risk perception: Coping with danger on board small fishing boats and offshore service vessels. Reliability Engineering and System Safety, 92(12), 1756–1763. MEARNS, K. and FLIN, R., 1995, Risk perception and attitudes to safety by personnel in the offshore oil and gas industry: A review. Journal of Loss Prevention in the Process Industries, 8(5), 299–305. FLIN, R., MEARNS, K., O’CONNOR, P. and BRYDEN, R., 2000, Measuring safety climate: Identifying the common features. Safety Science, 34(1–3), 177–192. BARNETT, M., GATFIELD, D. and HABBERLEY, J., 2010, Shipboard Crisis Management: A Case Study. http://disruption.solent.ac.uk/mhfr/resources/RINA%202002.pdf, retrieved 10 March 2010. SEIFERT, C. M. and HUTCHINS, E. L., 1992, Error as opportunity: Learning in a cooperative task. Human-Computer Interaction, 7(4), 409–435. SALAS, E. and CANNON-BOWERS, J. A., 1998, Making Decisions Under Stress: Implications for Individual and Team Training (Washington, DC: American Psychological Association). MITROPOULOS, P. and CUPIDO, G., 2009, The role of production and teamwork practices in construction safety: A cognitive model and an empirical case study. Journal of Safety Research, 40(4), 265–275. MEARNS, K., WHITAKER, S. M. and FLIN, R., 2003, Safety climate, safety management practice and safety performance in offshore environments. Safety Science, 41, 640–680. KU¨NZLE, B., KOLBE, M. and GROTE, G., 2010, Ensuring patient safety through effective leadership behaviour: A literature review. Safety Science, 48(1), 1–17. HA˚VOLD, J. I., 2010, Safety culture and safety management aboard tankers. Reliability Engineering and System Safety, 95(5), 511–519. KJELLE´N, U., 2000, Prevention of Accidents Through Experience Feedback (London: Taylor & Francis). PIDGEON, N. and O’LEARY, M., 2000, Man-made disasters: Why technology and organizations (sometimes) fail. Safety Science, 34(1–3), 15–30. KONTOGIANNIS, T. and MALAKIS, S., 2009, A proactive approach to human error detection and identification in aviation and air traffic control. Safety Science, 47(5), 693–706. RASMUSSEN, J., 1997, Risk management in a dynamic society: A modelling problem. Safety Science, 27(2–3), 183–213. KONTOGIANNIS, T. and KOSSIAVELOU, Z., 1999, Stress and team performance: Principles and challenges for intelligent decision aids. Safety Science, 33(3), 103–128. OLTEDAL, H., 2009, The use of safety management systems within the Norwegian tanker industry – do they really improve safety? In: Reliability, Risk, and Safety: Theory and Applications, edited by R. Bris, C. Guedes Soares and S. Martorell (London: Taylor & Francis), pp. 61–64.

I have received the education that is necessary in order to handle critical or hazardous situations I have received the education that is necessary in order to work safely I have received the training that is necessary in order to handle critical or hazardous situations I have received the training that is necessary in order to work safely New crew members get a thorough introduction to safety related issues I feel appreciated by my co-workers I feel appreciated by the ship management My co-workers do their jobs in a way that make me feel safe We solve problems and conflicts in a good manner My co-workers can communicate effectively in English The working environment is characterized by openness and dialogue I feel that the work we do onboard is too little appreciated by the company

Item description

Appendix Eight factor rotated solution

0.176

0.049

0.721 0.670 0.591 0.563 0.231

0.272 0.268 0.321 0.207 0.043

0.704

0.061

0.162

0.072 0.114

0.095 0.127

0.788 0.753

0.152 0.152

0.159

0.124

0.095

0.239

0.650

0.152

0.093

0.112

0.274

0.797

0.024

0.079

0.069

0.198

0.837

0.234

0.069

0.067

0.246

0.846

0.115

0.092

0.057

Local management

Shore orientation

0.141

Interpersonal relations

0.861

Competence

Factors

0.104

0.126

0.135

0.203

0.120

0.164 0.094

0.197

0.079

0.143

0.095

0.161

Working practices

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

0.007

0.093

0.012

0.012

0.090

0.065 0.073

0.072

0.083

0.067

0.072

0.080

Feedback

(continued )

0.132

0.104

0.181 0.045

0.064

0.061

0.160

0.121 0.140

0.093

0.053

0.059

0.054

0.047

Reporting practices

0.042

0.134

0.066

0.042 0.150

0.092

0.023

0.044

0.022

0.101

Demand for efficiency

Risk perception in the Norwegian shipping industry 621

What we learn in courses is not relevant in practice Reporting itself takes too much time In my company, they are more preoccupied with the statistics than the human consequences of an incident When an undesirable incident has occurred, people are more preoccupied with placing blame than finding the cause of the incident Does your closest superior follow up to ensure that all work on board is done in a safe manner? Is your closest superior clear in his engagement to ensuring his co-workers safety? Is your closest superior a good role model when it comes to attending to his own and others safety? Do you carry out a ‘safe job analysis’/ ‘risk analysis’ before high-risk operations?

Item description

Appendix. Continued.

Interpersonal relations 0.049 0.093 0.024 0.283

0.085

0.163

0.220

0.114

Competence 0.073 0.094 0.076 0.069

0.140

0.131

0.100

0.139

0.207

0.826

0.823

0.802

0.0.73

0.140 0.099 0.114 0.058

0.067 0.162

0.022

Local management

0.535

0.676 0.640

0.689

Shore orientation

Factors

0.845

0.106

0.178

0.237

0.032

0.095 0.022

0.092

Working practices

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

0.015

0.086

0.071

0.105

0.012

0.006 0.163

0.104

Feedback

0.100

0.100

0.100

0.106

0.237

0.077 0.272

0.087

Demand for efficiency

0.083

0.088

0.097

0.090

0.134

0.091 0.025

0.079

Reporting practices

622 H. Oltedal and E. Wadsworth

0.061 0.235 0.120 0.123 2.645 8.533

0.065 0.013 0.109 0.138 9.257 29.861

1.803 5.815

0.112 0.168

0.138 0.092 2.101 6.776

0.160

0.017

0.153

0.244

0.123

0.031

0.067

0.101 0.109

0.102 0.122 0.166

0.173

0.006

0.060

0.073

0.098

0.136

0.038

0.093

0.232

0.039

0.232

0.162

0.222

0.081

0.290

0.179

0.132

0.133

0.147

0.195

Notes: KMO ¼ 0.888. Loadings in bold indicate the factor in which the item was placed.

Do you carry out safety evaluation before new working methods, tools or routines are introduced? Do you have the possibility to prioritize safety first in your daily work? Do you discuss safety issues with your co-workers? Do you receive constructive feedback from the captain on the conditions you report? Do you receive constructive feedback from the captain on the conditions you report? Do you get information from incidents/ accidents on other vessels? Due to captain’s demand for efficiency we sometimes have to violate procedures Due to the company’s demand for efficiency we sometimes have to violate procedures Do you ever feel forced to continue your work even if safety may be threatened? Do close calls get reported in writing? Do minor incidents get reported in writing? Eigenvalue Explained variance 1.444 4.656

0.069 0.161

0.124

0.122

0.012

0.123

0.060

0.050

0.501

0.619

0.819

Downloaded By: [Swets Content Distribution] At: 11:23 1 November 2010

1.324 4.241

0.284 0.212

0.135

0.024

0.069

0.635

0.858

0.874

0.213

0.073

0.074

1.197 3.862

0.114 0.139

0.521

0.846

0.858

0.019

0.108

0.075

0.040

0.108

0.090

1.028 3.318

0.833 0.819

0.102

0.086

0.062

0.217

0.106

0.091

0.042

0.037

0.094

Risk perception in the Norwegian shipping industry 623

Article 3 Oltedal, H. A. (2010). The use of safety management systems within the Norwegian tanker industry—Do they really improve safety? In R. Bris, C. Guedes Soares, & S. Martorell (Eds.), Reliability, Risk and Safety: Theory and Applications (pp. 2355-2362). London: Taylor & Francis Group.

Article 4 Oltedal, H. A., & Engen, O. A. (2010). Tanker versus dry cargo—The use of safety management systems within Norwegian dry cargo shipping. In J.M. Ale, I.A. Papazoglou, & E. Zio (Eds.), Reliability, Risk and Safety (pp. 2118-2125). London: Taylor & Francis Group.

Reliability, Risk and Safety – Ale, Papazoglou & Zio (eds) © 2010 Taylor & Francis Group, London, ISBN 978-0-415-60427-7

Tanker versus dry cargo—The use of safety management systems within Norwegian dry cargo shipping H.A. Oltedal Stord/Haugesund University College, Haugesund, Norway

O.A. Engen University of Stavanger, Stavanger, Norway

ABSTRACT: In this article, we explore how the Norwegian controlled dry cargo shipping industry compares to the tanker industry regarding reporting practices, procedures and checklists, and how the industry balances commercial pressure and safety concerns. The article presents several differences between the two sectors, i.e. underreporting of minor incidents and close calls are present to a larger degree within the dry cargo sector than within tank. Also within the dry cargo industry, the procedural system is experienced as being more useful in daily operations. The statistical data derive from a survey carried out in 2006, supported by qualitative information from two case studies. The analysis suggests that the differences originate partly from each sector’s relationship with their respective customers. Customers of dry cargo shipping demand fewer requirements with regard to safety management and liquid tankers are to a larger degree embedded in the oil industry, with a more mature safety management system. However, it is suggested that this relationship between the liquid tanker industry and their customers, brings about changes based on external demands, in contrast to internal needs. This may explain both better reporting practices and the development of an extensive procedural system within liquid tankers. 1

INTRODUCTION

During the second half of the 1980s and early 1990s, there seemed to be a significant increase of maritime accidents. Just to mention a few examples: In 1980 the ore/bulk/oil carrier MV Derbyshire was lost at sea with all 42 crew and two supernumerary wives on board during a typhoon off Japan. In 1989 the oil tanker Exxon Valdes ran aground off the coast of Alaska causing extensive environmental damage and in 1992 the ore carrier Aegean Sea broke in two at La Coruna. From 1990 to mid-May 1997 alone, a total of 99 bulk carriers were lost at sea, with the death of 654 people. It became apparent that the international shipping industry was perhaps no longer capable of regulating itself and action was needed to reverse the downward spiral of maritime calamity. Studies collected and reviewed by the International Maritime Organization (IMO), showed that maritime accidents involved people making mistakes, mistakes which may be traced back to on-shore decisions (IMO 1999). This triggered the development of the International Safety Management (ISM) code. The ISM codes’ intention is to make the shore side management responsible for effective safety management and the on board safety situation, and

2118

it was formally adopted by IMO November 1993. For oil tankers, chemical tankers, gas carriers, bulk carriers and cargo high speed craft of 5000 gross tonnage and upwards, implementation became mandatory no later than 1th of July 1998, and extended to remaining categories of ships in 2002. Seaborne world trade may be divided into two main groups such as tank and dry cargo. Tank, or transportation by tankers, refers to cargo in liquid form, as oil products, chemicals, vegetable oil and wine, transported in tanks. Dry cargo refers to all cargo which is transported in dry form, traditionally split into two groups; general cargo and bulk. General cargo refers to cargo that is susceptible for loading in general, nonspecialized stowage areas or standard shipping containers; e.g., boxes, barrels, bales, crates. The bulk carrier carries dry cargoes which are shipped in large quantities and do not need to be carried in packaged form. The principal bulk cargoes are grains, such as wheat, coal, iron ore, bauxite, phosphate and nitrate. (IMO 1999, Skipsfartens bok 2008). The academic literature dealing with human factors, safety and safety management systems within merchant shipping is found to be limited, which is also noted by other studies, see Hetherington, Flin & Mearns (2006), Christophersen (2009), Håvold (2005). It has been even more difficult to

retrieve academic literature dealing with differences of safety management between liquid and dry cargo shipping, However, differences in reporting practices between dry cargo and tanker vessels are suggested by Ellis et al. (2010), Oltedal et al. (2010). Safety management through the ISM-Code empathize organizational learning and improved safety through collection of experience data, e.g., by means of self reported accidents, incidents and near-misses. The information is processed and used for development of remedial actions to prevent future individual incidents and accidents, and total loss of vessel and cargo (Kjellén 2000, Anderson 2003). However, up to present time, far more effort is given to technical safety than human factors, management and organizational factors. This misalignment in safety work may explain why maritime accidents are still on the rise (Christophersen 2009). Underpinning the ISM-Code and the safety management system is the idea of a safety culture being developed in the industry, and incorporation of the human elements. Research carried out in the aftermath of ISMCode implementation suggests industrial opposition towards implementation, increased workload of paperwork, and the potential situation of self incrimination by producing documentation which in turn may be used as evidence (Christophersen 2009, Anderson 2003, Oltedal 2010). This article explores three main research questions:

The customers/charterers of liquid tankers are known to emphasize safety, both environmental and individual, when shipping their cargo. One of the most significant safety initiatives was introduced by the Oil Companies’ International Maritime Forum (OCIMF) in 1993, when launching a Ship Inspection Report Programme (SIRE). This programme was originally launched to specifically address concerns about sub-standard shipping, and is today a unique tanker risk assessment tool of value to charterers, ship operators, terminal operators and government bodies concerned with ship safety (SIRE 2010). Despite of this, recent research into the Norwegian controlled tanker industry’s safety management system indicates a situation with substantial under-reporting of experience data, such as self experiences, near misses and incidents. Moreover, reports that were received by shore side appeared to have been “massaged”, to present a better state of affairs than was in fact the case. The situation is suggested induced by a blame culture, where the seafarers fear for the consequences of reporting. Other factors counteracting reporting are suggested to be a complicated and paper producing reporting system and inadequate feedback upon the reports. Also, although safety is the oil companies’ official priority, the sailors often experienced commercial pressure and efficiency to be the shore sides’ priority (Oltedal 2010).

1. What is the present safety management situation within the dry cargo industry? 2. What are the main factors explaining the safety situations on dry cargo? 3. Finally, what is the current safety situation within dry cargo compared to the liquid cargo industry?

2.2

2

SHIPPING DRY VERSUS LIQUID CARGO

Although both dry and liquid cargo face similar challenges, some are more characteristic for each group. For dry cargo it is of fundamental importance that the cargo is properly distributed throughout the ship so that the structure is not overstressed and the ship has an adequate standard of stability (Skipsfartens bok 2008). A challenge typical for liquid tankers is emission of cargo and environmental pollutions. A shipping accident involving spill and environmental pollution is apparently more often subject to public opinion and criticism. 2.1 Description of the tanker industry In volume, the tanker industry accounts for 35%–40% of world seaborne trade (Skipsfartens bok 2008).

Description of the dry cargo industry

In volume, the bulk industry accounts for 50% of world seaborne trade (Skipsfartens bok 2008). Customers/charterers of dry cargo shipping are as varied as their cargo, which includes everything from bananas to clothing and timber. According to our sources (see methodology) the typical dry cargo customers/charterers do not place the same emphasis on safety, more importance is placed on price and time. Dry cargo has been especially exposed to accidents, and after the increase in accident rates in the 90s, IMO worked with new standards concerning technical and structural issues. However, it was noted that a number of accidents involving bulk carriers have occurred as a result of inadequate loading and unloading and that safe practices could prevent such accidents in future. On some occasions inadequate loading could be traced back to a situation where terminals, which were often owned by the cargo owners or charterers of the ship, were putting pressure upon the ships to amend their loading plans or to load cargo to suit them, with little consideration about the overall safety of the ship, Thus safety was threatened by commercial concerns (IMO 1999). Own field data indicates that this description also depicts today’s situation.

2119

3

METHODOLOGY

A multi-method approach combining surveys and case studies, including interviews, is adopted. The survey aims to indicate the interrelated patterns of the seafarers’ perceptions of and attitudes towards safety and issues related to the formal safety management system. The qualitative results aim to give a more thorough understanding of the overall situation, underlying processes and the seafarers’ experiences concerning own practice and safety management. 3.1 Survey A total of 1574 questionnaires were distributed to 83 randomly selected Norwegian controlled liquid tankers and dry cargo carriers. 76 vessels returned a total of 1262 completed forms. 455 of these respondents worked on 35 dry cargo vessels. The survey forms part of a major safety culture survey carried out in 2006, performed with a validated instrument (Oltedal, Engen 2009). The results from liquid tanker industry are previously published in Oltedal (2010). 3.2 Case studies Two dry cargo companies were studied in 2008 and 2009. In order to ensure the companies’ anonymity, company specific information is retained. In both companies data are retrieved from participation in captains’ conferences where safety was an issue, and both formal and informal interviews with seafaring captains and shore-based personnel involved in safety related matters. 3.3 Statistical analysis All statistical analysis is performed using SPSS version 16.0. Principal Component Analysis (PCA) with Varimax rotation and Pairwise deletion is carried out in order to examine the survey items’ interrelationship, and in terms of their common underlying dimension. The extracted rotated component matrix with factor structure and loadings is presented. The loading represents the correlation between the variable and the extracted factor(s), with estimates ranging from 0 to 1.00. Items that load strongest on a given factor are considered to be most like the factor, and thus the underlying latent dimension (Hair 1998, Pett, Lackey & Sullivan 2003). Descriptive statistics for each item, including percentage frequency distribution, mean and standard deviation are presented. 4

PRESENTATION OF RESULTS

The results are presented in sections as follows; (1) crews’ reporting practices, (2) procedures and

checklists, and (3) perceived balance between commercial pressures and safety concerns. For facilitating comparison, statistical results from the tanker industry are presented in brackets, however only when relevant to discussion and conclusion. Each section is brought to a close with the discussion of the presented data and results. The qualitative information is used as a framework for interpretation of the quantitative results. 4.1 Reporting practices on board Survey questions related to reporting practices are presented in Table 1, together with question reference (ref) and number of respondents (n). The descriptive statistics, together with mean and standard deviation, are presented in Table 2. Explanation for interrelationship between survey questions (Oltedal 2010): Reporting practices (R1, R2 & R5) are assumed to improve if the crew gets feedback from the report follow up (R3), and when the importance of reporting and its role in safety management is comprehended (R4). A blame policy (R6) is supposed to reduce reporting (R1 & R2) and increase the probability of a cover up of own mistakes (R5) (Reason 2001). A blame policy is Table 1.

Reporting practices—survey questions.

Ref

n

R1 Do minor incidents get reported in writing? R2 Do close calls get reported in writing? R3 Do you receive constructive feedback from the company on the conditions you report? R4 Reporting is important to prevent the recurrence of accidents and incidents. R5 Are reports of undesirable incidents ever “fixed up” to cover mistakes? R6 When an undesirable incident has occurred, people are more preoccupied with placing blame than finding the cause of the incident.

406 405 391

Table 2.

451 383 444

Reporting practices—descriptive statistic.

Ref

1 (%)

2 (%)

3 (%)

4 (%)

5 (%)

Mean

Std

R1 R2 R3 R4 R5 R6

7.1 6.2 10.0 0.2 42.0 14.6

10.6 12.3 8.2 0.2 22.2 49.1

22.9 23.0 18.7 2.0 14.9 12.4

30.8 37.0 31.7 59.0 13.3 20.9

28.6 21.5 31.5 38.6 7.6 2.9

3.63 3.55 3.66 4.35 2.22 2.48

1.20 1.14 1.72 0.56 1.32 1.07

R1 through R3 and R5: 1 very seldom/never, 2 seldom, 3 sometime, 4 often and 5 very often/always. R4 and R6: 1 strongly disagree, 2 disagree, 3 not sure, 4 agree and 5 strongly agree.

2120

when people are blamed for an unwanted situation rather than the situational circumstances. The descriptive data in table 2 point to a situation of substantial underreporting. A total of 40.6% of the respondents state that they never or only sometimes report minor incidents (tank: 35.4%) and 41.5% that they never or only sometimes report close calls (tank: 36.3%). In addition, 36.9% state that they never or only sometimes get constructive feedback upon reported events (tank: 48.1%). However, as few as 0.4% of the respondents disagree with the statement that reporting is important to prevent the recurrence of accidents and incidents. A total of 35.8% of the respondents admit to sometimes, often or always fixing up the reports to cover up own mistakes (tank: 36.5%), and 23.8% perceive that a blame culture is present in the organization (tank: 25.7%). The respondents were also asked to list their reasons for not reporting incidents. Each respondent could mark up to three of eleven pre-specified alternatives, or write any other reason if not listed. The three major reasons are listed in Table 3. In order to report an incident or miss, the situation has to be recognized as reportable. In both companies, both seafaring and shore personnel were presented to various near miss, incidents and accidents scenarios (all similar but escalating situations) and asked to define each situation as reportable or not, and categorize its accident potential. The results indicate that both shore personnel and to a larger degree seafarers have problems recognizing a near miss. This was reflected in both companies’ safety management system with few reports. The factor structure matrix and loadings are presented in Table 4. Table 4 indicates two underlying reporting dimensions. One reflecting desired reporting practice (factor 1) and one reflecting undesired reporting practices (factor 2). 4.2

Section discussion

The factor structure (factor 1) indicates relationship between the tendency to report and feedback given by the company on reported events. An interrelationship with the crews’ recognition of reporting’s importance Table 3. survey.

Reasons for not reporting an incident—from Response (%)

The incident did not have any serious consequence There could be negative reactions from my co-workers I am afraid that the information will be used against me.

35.7 29.1 20.6

Table 4. Reporting practices—factor structure and loadings. Ref R1 R2 R3 R4 R5 R6

Fac. 1

Fac. 2

Do minor incidents get 0.854 .226 reported in writing? Do close calls get reported in 0.845 .209 writing? Do you receive constructive 0.782 0.230 feedback from the company on the conditions you report? Reporting is important to 0.433 .141 prevent the recurrence of accidents or incidents. Are reports of undesirable 0.054 0 .900 incidents ever “fixed up” to cover mistakes? When an undesirable incident 0.600 .375 has occurred, people are more preoccupied with placing blame than finding the cause of the incident.

*KMO 0.672. *Underlined loadings indicate the factor on which the item was placed.

in prevention of future incidents and accidents is also indicated. Within the liquid tanker sector, the latter item, reporting’s importance, was negatively associated with undesired reporting practices. The descriptive results imply that underreporting of close calls and minor incidents are present in a larger degree within the dry cargo sector than within the tank. Both sectors have equal ranking of reasons for not reporting an incident (table 3), however within dry cargo 5% more of the respondents have listed “the incidents did not have any serious consequence” as the primary reason for not reporting an incident. Both second and third reason listed for not reporting incidents are related to fear for negative reactions and blame. The factor structure (factor 2) indicates relationship between an organizational blame culture and the tendency to alter the relation of how things happened in the reports, in order to cover up own or others’ mistakes. The results from factor 1 suggest that reporting frequency may increase if feedback upon reported events is improved. Moreover, if the crewmembers understand the importance of reporting to prevent future events, the results suggest that reporting frequency may increase. However, the respondents report less than the respondents within the tank sector, but are at the same time more satisfied with feedback given on the reported conditions. We suggest two possible explanations for this relationship and sector differences. (1) Within dry cargo fewer reports are placed, which gives less

2121

workload on shore personnel handling incoming reports and thus better feedback may be provided. If this hypothesis should be true it also suggests that the liquid tanker industry has not provided sufficient resources/personnel at shore in order to handle incoming reports. (2) The factor structure for dry cargo and liquid cargo differed at one item, which reflects the respondents’ understanding of reporting’s role in preventing future incidents and accidents. This suggests that within dry cargo, increased reporting may be achieved by measures aiming at improving the crews’ understanding of the interrelationship between experience information and prevention. Within the tanker sector, safety campaigns towards sailing personnel are quite common, and if the latter suggestion should be true, it also suggests that these campaigns should have some effect upon increased reporting practices. 4.3 Procedures and checklists Survey questions related to the perception of procedures are presented in Table 5, with question reference (ref) and number of respondents (n). The descriptive statistics, together with mean and standard deviation, are presented in Table 6. Explanation for interrelationship between survey questions (Oltedal 2010): Good procedures are supposed to be helpful in the work (P1). It is assumed that procedures are more easily put into operational practice if the user has received training in how to understand and apply the procedures, and Table 5.

Procedures—survey questions.

Ref

n

P1 P2

The procedures are helpful in my work. I have received good training in the company’s procedures. We have the opportunity to influence and form the procedures. I feel that it is difficult to know which procedures are applicable. The procedures are difficult to understand or are poorly written.

P3 P4 P5

451 450 431 439 447

the safety role they play (P2). They are also assumed to be more workable if they reflect the reality of the working process. That is ensured through involving the crew in the development of procedures, and by paying attention to crews’ experience (P3). If these factors (P2 & P3) are not present, the procedural system may be perceived as confusing and difficult to relate to (P4 & P5) (Reason 2001). Factor analysis also included item R5 “When an undesirable incident has occurred, people are more preoccupied with placing blame than finding the cause of the incident” Our hypothesis is that organizations inclined to focus on the human as a cause, as in a person model (Reason 2001), are also more inclined to react by adding more procedures, which may result in a procedural system difficult to relate to. The descriptive data in Table 6 show that 94% of the respondents regard the procedures as helpful in their work (tank: 93%), and 91.1% states to have received good training in the companies’ procedures (tank: 90.3%). 68.9% state that they have got the opportunity to influence and form the procedures (tank: 67%). A total of 19.1% feel it is difficult to know which procedures are applicable (tank: 22.1%), and 10% that procedures are difficult to understand or poorly written (tank:17.3%). The respondents were also asked to list their reasons for not following procedures. Each respondent could mark up to three of seven pre specified options, or write any other reason if not listed. The three most frequent reasons are listed in Table 7. Results from the qualitative studies indicate that procedures and checklists are used in a less extent within dry cargo than within the tanker sector. The interviewees’ explanation is that their customers (dry cargo) do not make demands for specific use and development of such. According to the interviewees, this is resulting in a procedural system more understandable, simplified and easy to follow. Factor structure matrix and loadings are presented in Table 8. Table 8 indicates two underlying dimensions of procedures. One reflecting helpful procedures (factor 1) and one reflecting poor procedures (factor 2). 4.4 Section discussion

Table 6.

Procedures—descriptive statistics.

The factor structure (factor 1) indicates that procedures are perceived as helpful when sufficient

Ref

1 (%)

2 (%)

3 (%)

4 (%)

5 (%)

Mean

Std

P1 P2 P3 P4 P5

0.00 0.7 3.2 13.0 14.8

1.6 2.4 13.0 49.9 62.2

4.4 5.8 14.8 18.0 13.0

61.6 63.3 57.3 18.2 8.7

32.4 27.8 11.6 0.9 1.3

4.25 4.15 3.61 2.44 2.20

0.61 0.69 0.963 0.96 0.84

Table 7.

P1 through P5: 1 strongly disagree, 2 disagree, 3 not sure, 4 agree and 5 strongly agree.

Reasons for not following procedures. Response (%)

The work will be done faster. The procedures do not work as intended. I feel pressured because I am overloaded with work.

2122

47.6 39.7 35.6

Table 8.

Procedures—factor structure and loadings.

Ref P1 P2 P3 P4 P5 R5

The procedures are helpful in my work. I have received good training in the company’s procedures. We have the opportunity to influence and form the procedures. I feel that it is difficult to know which procedures are applicable. The procedures are difficult to understand or are poorly written. When an undesirable incident has occurred, people are more preoccupied with placing blame than finding the cause of the incident.

Fac. 1

Fac. 2

0.835

.185

0.772

216

0.720

0.019

0.001

0.853

.100

0.807

.300

0.574

and driven by internal need. Our hypothesis is that when the procedural systems are formed by external influence, in contrast to internal need, it is manifested in a procedural system perceived as less suitable, adequate and helpful in daily operations. 4.5

Balance between commercial pressures and safety concerns

Survey questions related to crew’s perception of their priority of safety versus efficiency are presented in Table 9, together with question reference (ref) and number of respondents (n). The descriptive statistics, together with mean and standard deviation, are presented in Table 10. Explanation for interrelationship between survey questions (Oltedal 2010): The priority of safety versus efficiency is supposed to be communicated from the organization’s top management (SE2) down to the lower levels of the organization (SE1), which are,

*KMO 0.634. *Underlined loadings indicate the factor on which the item was placed.

training in using the procedures are provided, and when the users get involved in the development process. On the other hand, the procedural system is perceived as less helpful in companies that are perceived as blame oriented (factor 2). The item relationships in factor 2 suggest that poor procedures and company blame policy do have a common underlying organizational factor, which is reflected in their safety management orientation. In the factor structure representing dry cargo, most variance was explained by factor 1, helpful procedures. Within liquid cargo, most variance was explained by poor procedures. The first extracted factor always explains the largest amount variance, and this difference suggests that the dry cargo sector to a greater degree are represented by helpful procedures, and the liquid cargo sector to a larger degree are represented by a poor procedural system. It is also noted a difference among stated top three reasons for not following procedures. In dry cargo, pressure due to work overload were stated among the top three, and within liquid tanker poor procedures were stated among the top three. The other reasons were ranked equally. This difference also suggests that poor procedures are more representative for the liquid tanker sector. Although both sectors do have to relate to minimum standards given by IMO and flag state administration, within the liquid tanker sector, the procedural system is to a larger degree formed and driven by customer demands. Within dry cargo the procedural system is to a larger degree formed

Table 9. Balance between commercial pressures and safety concerns—survey questions. Ref

n

SE1 I am confident that my company always prioritizes the crew’s safety. SE2 The on-shore top management in my company prioritizes safety before economy. SE3 Do you have the possibility to prioritize safety first in your daily work? SE4 Do you ever feel forced to continue your work even if safety may be threatened? SE5 The number of crewmembers is not sufficient to ensure safety on board. SE6 The management doesn’t care how we do our work as long as the work gets done. SE7 I experience that safety is more a façade than a real priority area.

448 450 459 454 447 445 442

Table 10. Balance between commercial pressures and safety concerns—descriptive results. Ref

1 (%)

2 (%)

3 (%)

4 (%)

5 (%)

Mean

Std

SE1 SE2 SE3 SE4 SE5 SE6 SE7

1.6 2.4 0.7 44.7 17.0 20.7 17.2

2.5 3.2 1.5 20.0 37.8 45.6 39.4

10.5 10.9 4.8 28.2 13.6 16.4 10.0

46.2 55.1 38.1 6.8 26.2 13.7 26.2

39.3 28.4 54.9 0.2 5.4 3.6 7.2

4.19 4.04 4.45 1.98 2.65 2.34 2.67

0.84 0.86 0.72 1.01 1.19 1.06 1.24

SE1, SE2 & SE5 through SE7: 1 strongly disagree, 2 disagree, 3 not sure, 4 agree and 5 strongly agree. SE3 & SE4:1 very seldom/never, 2 seldom, 3 sometime, 4 often and 5 very often/always.

2123

ideally, aligned (Reason 2001). However, deviation from official policies and goals may be experienced (SE6, SE7).The level of stress experienced is interrelated with crewing level. Crewing constitutes a major part of operational expenditures, and how safety is prioritized may be reflected in the on board crewing level (SE5), which again influences the crew’s opportunity to prioritize safety in their daily work (SE3, SE4). The descriptive statistics show that 85.5% of the respondents are confident that their company does prioritize the crews’ safety (tank: 90.6%). A little less, 83.5% do feel that the top management prioritizes safety before economy (tank: 83.8). 93% states to often or always having the possibility to prioritize safety first (tank: 94.6%). However, 35.2% do sometimes or often feel forced to continue work even if safety should be threatened (tank: 32.5%). Moreover, 17.3% agree with the statement that the management does not care how the work is done, as long as the work is done (tank: 13%). 33.4% percent experience safety to be a façade more than a real priority area (tank: 35.6%), and 31.6% that the number of crewmembers is not sufficient to ensure safety (tank: 27.5%). The results from case studies indicate that the sailing officers experience an efficiency pressure due to commercial concerns. However, in one of the companies, the perceived pressure was apparently more prominent than real pressure. From discussions between officers and commercial agents it was apparent that perceived pressure originated from misunderstanding in communication and lack of understanding of the other parts’ work tasks and responsibilities. In the other company, demand for more efficient operations and turnover were clearly expressed by the top management. However, in the latter company the case study was carried out in the period of current financial crises, with low freight rates and earnings, and thus demand for efficiency could be regarded necessary in order to survive as a company. It was also found that safety and attitudes campaigns were implemented in a lesser degree than within the tanker sector. Factor structure matrix and loadings are presented in Table 11. Table 11 indicates two underlying reporting dimensions. One reflecting when safety is perceived to be a priority (factor 1) and one reflecting when demand for efficiency is perceived to be a priority (factor 2). 4.6 Section discussion The factor structure (factor 1) indicates that when top management and the company is perceived to prioritize the crew’s safety, the crew also perceive to have the possibility to prioritize safety in their daily work. No business company exists with the primary objective to be safe. It is also indicated that when the safety priority is perceived as a façade,

Table 11. Balance between commercial pressures and safety concerns—factor structure and loadings. Ref SE1 SE2 SE3 SE4 SE5 SE6

SE7

I am confident that my company always prioritizes the crew’s safety. The on-shore top management in my company prioritizes safety before economy. Do you have the possibility to prioritize safety first in your daily work? Do you ever feel forced to continue your work even if safety may be threatened? The number of crewmembers is not sufficient to ensure safety on board. The management doesn’t care how we do our work as long as the work gets done. I experience that safety is more a façade than a real priority area.

Fac. 1

Fac. 2

0.819

.223

0.767

.106

0.710

.113

.512

0.455

.128

0.745

.363

0.721

.050

0.685

*KMO 784. *Underlined loadings indicate the factor on which the item was placed.

the crew experience management to disregard how work is done together with lack of sufficient crew in order to ensure safety (factor 2). The factor structure differs from the tanker industry in one aspect; within dry cargo most of the variance is explained by safety perceived as priority (factor 1): Within the tanker industry most of the variance was explained by safety being a façade and with efficiency being the real priority. The results from the case studies suggest that demand for efficiency is a reality also within dry cargo. However, within dry cargo the crew is to a lesser degree exposed towards official safety programs, initiated by external partners. It seems that external safety initiatives are potentially less integrated in the organization, compared to when initiated by internal needs. However, the safety initiative within the tanker sector may also have given the crew higher expectations regarding safety priority. Thus the discrepancy may also be explained by differences in expectations, as well as differences in work practices 5

CONCLUSION

The tanker sector is extensively embedded in the Norwegian oil industry. The Norwegian oil industry is dominated by a limited number of big enterprises

2124

with “machine bureaucratic structures” well suited for co-operation with legal authorities and their rules and regulations (Lindøe et al. 2010). This explains the extended focus on safety, safety management, attitudes etc. within the tanker industry compared to dry cargo. Although both sectors do have substantial underreporting, our data indicate that the safety campaigns, as typical for the oil industry, have some positive effects upon reporting practices. However, within the dry cargo industry the feedback upon reported events are perceived as better than within the tanker sector. One hypothesis is that the shore tanker organization is not prepared to manage the increased work load that increased reporting brings along. Feedback is assumed to motivate reporting, and when the shore organization is not prepared to handle increased reporting, this may counteract with the effect of safety campaigns. However, the dry cargo industry could have a positive effect from such campaigns, in increasing the crews’ awareness and recognition of a near miss, along with better understanding of reporting’s importance in safety management systems. The shore side is thus recommended to have resources available for potential increase in number of reports placed, and proper follow up. When following up the reports, the development of new measures should also be considered carefully, and alternatives to development of new procedures developed. The experiences from the tanker industry suggest that development of a constantly increasing, detailed and extensive procedural system might undermine safety. Moreover, the analysis suggests that safety measures should be initiated by internal and industrial need, not external demands from customer. When externally initiated, safety management might be less integrated in the operational part of the organization. It is also suggested that this external demand is related to the existence of a poor procedural system. REFERENCES Anderson, P. 2003, Cracking the code: the relevance of the ISM code and its impact on shipping practices, Nautical Institute, London.

Christophersen, J.G. 2009, “Sikkerhetsstyring i skipsfarten 1998–2008: bakgrunnsfaktorer for reguleringsmessig etterlevelse og overtredelse av ISM-koden”, vol. no. 22, pp. 699. Ellis, N. & Bloor, Michael, Sampson, Helen 2010, “Patterns of seafarer injuries”, Maritime Policy & Management, vol. 37, no. 2, pp. 121–128. Hair, J.F. 1998, Multivariate data analysis, 5th edn, Prentice Hall, Upper Saddle River, N.J. Håvold, J.I. 2005, “Safety-culture in a Norwegian shipping company”, Journal of Safety Research, vol. 36, no. 5, pp. 441–458. Hetherington, C., Flin, R. & Mearns, K. 2006, “Safety in shipping: The human element”, Journal of Safety Research, vol. 37, no. 4, pp. 401–411. IMO and the safety of bulk carriers, International Maritime Organization, 1999, Retrieved 14 March 2010 at http://www.imo.org/includes/blastDataOnly.asp/ data_id%3D7987/BULK99.FIN.pdf. Kjellén, U. 2000, Prevention of accidents through experience feedback, Taylor & Francis, London. Lindøe, P.H., Engen, O.A. & Olsen, O.E. 2010, “Responses to accidents in different industrial sectors.” Safety Science. Forthcoming. Oltedal, H.A. & Engen, O.A. 2009, “Local management and its impact on safety culture and safety within Norwegian Shipping” in Safety, Reliability and Risk Analysis Theory, Methods and Applications, eds. M. Artorel, C. Guedes & J. Barnett, 2008th edn, European Safety and Reliability Association. Oltedal, H. 2010, “The use of safety management systems within the Norwegian tanker industry—do they really improve safety?” in Reliability, Risk, and Safety: Theory and Applications, ed. Bris, Guedes Soares & Martorell, Taylor & Francis Group. Oltedal, H. & McArthur, D. “Reporting practices in merchant shipping, and identification of influencing factors”. Submittet in Safety Science, April 2010. Pett, M.A., Lackey, N.R. & Sullivan, J.J. 2003, Making sense of factor analysis: the use of factor analysis for instrument development in health care research, Sage, Thousand Oaks, Calif. Reason, J. 2001, Managing the risks of organizational accidents, Ashgate, Aldershot. SIRE, 2010, Retrieved 15 March 2010 at http://www. ocimf.com/pages.cfm?action sire_introduction2. Skipsfartens bok: Norges maritime næringer, 2008, Breakwater Publishing.

2125

Article 5 Oltedal, H. & McArthur, D. (2010). Reporting practices in merchant shipping, and the identification of influencing factors. Safety Science, 49(2), 331-338.

Author's personal copy

Safety Science 49 (2011) 331–338

Contents lists available at ScienceDirect

Safety Science journal homepage: www.elsevier.com/locate/ssci

Reporting practices in merchant shipping, and the identification of influencing factors H.A. Oltedal ⇑, D.P. McArthur Faculty of Technology, Business and Maritime Education, Stord/Haugesund University College, N-5528 Haugesund, Norway

a r t i c l e

i n f o

Article history: Received 29 April 2010 Received in revised form 16 August 2010 Accepted 14 September 2010

Keywords: Reporting practices Safety management Safety culture Ordered logistic regression Shipping Seafaring

a b s t r a c t The objective of this paper is to identify the factors determining the reporting frequency of experience data e.g. incidents and accidents. The empirical setting is the Norwegian controlled merchant fleet. Data were collected from a survey carried out in 2006, where 1262 questionnaires were gathered from 76 vessels. The data were subjected to explorative factor analysis, method of principal component and varimax rotation. Seven factors, representing latent dimensions of safety culture, were extracted. Internal consistency (Cronbach Alpha) and scale reliability were found to be acceptable. The factor scores were used in an ordered logistic regression to examine the factors’ relationships to reporting frequency. The results show that enhanced safety related training, a trusting and open relationship among the crew, safety oriented ship management, performance of pro-active risk identification activities and feedback on reported events all are significantly related to higher reporting frequency. On the other hand, demand for efficiency and lack of attention to safety from shore personnel, are significantly related to lower reporting frequency. The results also show a significantly lower reporting frequency among those who have worked with their local manager less than 1 year. Bulk and dry cargo vessels also show significantly lower reporting frequency than those working on liquid bulk carriers. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Information about incidents, near-misses, operational failures and successes are crucial when taking a pro-active approach to safety. Such information is often collected and processed into basic data for remedial actions through the use of a formal safety management system, which constitutes a cornerstone of organizational learning (Kjellén, 2000; Reason, 2001). Within the maritime industry, at the international administrative level, guidelines for formal safety assessment systems have already been developed. In these, identification of potential hazards is the first step (International Maritime Organization, 2007a). At company level, safety management is regulated by the International Safety Management (ISM) code, which requires shipping companies to have a system of reporting and collecting experience data (International Maritime Organization, 2002). Although various systems for safety management do exist, safety management per se is not only a system property. It is not sufficient simply to have a system for the collection of safety related data. The system efficiency is determined by its human interrelationships (crew, shore personnel, analysts and others). A fundamental pillar of safety management is that information reported into the system is reliable and reflects the actual ⇑ Corresponding author. Tel.: +47 20 70 26 44; fax: +47 938 26 187. E-mail address: helle.oltedal@hsh.no (H.A. Oltedal). 0925-7535/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ssci.2010.09.011

situation in working operations. Thus, under-reporting of safety related events constitutes a major threat to the efficiency and utility of a safety management system. There is an extensive literature considering reporting practices and the factors which influence these. Particular interest has been shown in high-risk sectors such as civil aviation and nuclear power plants. Attention has also been given to road and rail transportation, health care and oil & gas related activities. Although the importance of reporting practices in determining the outcome of a safety management system has been acknowledged in the literature, little attention has been given to the topic within merchant shipping. This is somewhat surprising given that seafaring is regarded as a high-risk occupation (Anderson, 2003; Hansen and Pedersen, 1996; Håvold, 2005; Roberts and Marlow, 2005). A few studies which examine merchant shipping have been retrieved by searching through the Science Direct database for peer reviewed publications. The findings indicate a culture of underreporting of safety information, which represents a shortcoming in the their safety management (Psarros et al., 2010; Oltedal, 2010; International Maritime Organization, 2007b; Ellis et al., 2010). An analysis of accident data from the Lloyd’s Register FairPlay (LRFP) and the Norwegian Maritime Directorate (NMD) for vessels registered in Norway, suggests that only 30% of accidents experienced were reported (Psarros et al., 2010). Own research indicates that 36% of shipboard tanker crew never or only

Author's personal copy

332

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338

sometimes report near-misses or minor incidents (Oltedal, 2010). Within the dry cargo sector, around 40% state that they never or only sometime report a minor incident or near-miss (Oltedal and Engen, 2010). Differences in reporting practices between dry cargo and tanker vessels are also suggested by others (Ellis et al., 2010). The International Maritime Organization (IMO) expresses an awareness of the under-reporting of safety information, and highlights that this must be improved (International Maritime Organization, 2007b). 1.1. Theoretical foundation Although various safety management models exist (see Kjellén, 2000), of fundamental importance in all of them is the collection of safety information from the operational productions system, used with the intention of preventing future accidents and other unwanted episodes. A model of a safety management system is illustrated in Fig. 1. Fig. 1 shows the flow of information in a safety management system. The shaded areas represent the safety information system, that is, the part of the safety management system which provides information needed for decisions and signalling related to safety matters. Reporting and the collection of data on, for instance, operational accident risk by means of self reported accidents, incidents and near-misses is regarded as a critical function of the overall system. Information processing and the development of remedial actions all depend on the reliability and accuracy of the reported information (Kjellén, 2000). When it comes to the under-reporting of safety related events, Kjellén (2000) uses behavioural theory to explain different propensities to report. Behavioural theory is primarily used to explain why people deliberately violate safety rules. It focuses on the consequences of behaviour and how these consequences affect people’s judgments in relation to recurring situations, and their ingrained action patterns in these situations. For instance, when performing an operation, people may have two action alternatives to choose between: one which is considered safe and one considered unsafe. Experience shows that people are inclined to choose the unsafe alternative when this alternative provides a consequence regarded as positive, such as saving time or other resources. When people in turn get positive feedback for their efficiency, deliberately violating safety rules becomes a valuable skill, and over time a part of the organizational culture. This safety-efficiency trade-off has also been noted by other researchers, for example (Hollnagel,

Fig. 1. Safety management system (adapted from Kjellén, 2000).

2009). The reporting of an incident or accident is often a time-consuming activity, and it is therefore reasonable to expect that demands for efficiency will lead to less reporting, as well as an increase in situations that should have been reported in the first place. Another factor influencing people’s propensity to report is how the organization handles blame and punishment (Reason, 2001). Reason (2001) emphasises the importance of organizations developing a just culture. In a just culture, there is an atmosphere of trust in which people are encouraged or even rewarded for providing essential safety-related information, but in which there is also a clear line between acceptable and unacceptable behaviour. For example, if efficiency by unsafe acts is knowingly accepted during normal operations, the operator should not be blamed and punished when that very same action is a causal factor in an unwanted event. Feedback of the results to the reporter is considered a motivating factor, as the individuals understand the relevance of reporting and see that is taken seriously. However, feedback on reported events in the form of constantly developing new procedures may also be perceived as blame, as it is signalling that the operator did something wrong. Such a person oriented approach also has another drawback. When searching for human error, the organization often does not look behind the operators when looking for causal factors, which could be found elsewhere in the organization such as, for instance, a pressure for efficiency. Reporting practices, how the organization handles blame, punishment and feedback, and the general functionality of the organization’s safety management, are regarded as key elements in an organization’s safety culture (Reason, 2001). Management characteristics and leadership, at all levels in the organization both shoreside and shipboard, are regarded as major enablers and barriers with respect to the development of an efficient safety culture (Maritime and Coastguard Agency, 2004), and thus also an adequate reporting culture. 1.2. Previous research The under-reporting of safety-related information has been identified as a problem in several industries. For example, it is suggested that up to 68% of all workplace accidents and injuries are not captured in the national injury surveillance systems, which were set up by the Occupational Safety and Health Administration (OSHA) and the Bureau of Labor Statistics (BLS) in the United States (US) (Rosenman et al., 2006). Others suggest that 81% of injuries experienced remain unreported (Probst et al., 2008). Within the health industry, barriers against reporting include fear of reprisals, lack of confidentiality, time constraints and lack of post-reporting feedback (Espin et al., 2007). Although Psarros et al. (2010) concludes that under-reporting constitutes a major problem within the maritime industry, the reasons for non-reporting are not addressed (Psarros et al., 2010). Ellis et al. (2010) suggest that under-reporting is more frequent on general cargo vessels compared to others. However, these conclusions are drawn based on differences in aggregate data in administrations, and not reporting practices directly. Barriers against reporting are not reliably identified, however differences in national/ cultural risk perception have been suggested. A recent cross-industrial review indicates that a fear of blame and punishment (legal, organizational or from co-workers) is the most commonly cited barrier to reporting (van der Schaaf and Kanse, 2004). The International Maritime Organization also addresses fear of being thought blameworthy or of being disciplined, embarrassment, fear of legal liability, and so on, as the main barriers towards reporting (International Maritime Organization, 2007b). van der Schaaf (2004) has also identified other common

Author's personal copy

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338

barriers such as: a lack of follow-up, managerial issues such as no commitment or distrust, time demands, a paper producing reporting system and a lack of understanding of what constitutes a nearmiss and incident. Own research indicates that all of these factors are present within the shipping industry (Oltedal, 2010). Moreover, van der Schaaf (2004) reports that a ‘macho’ environment may discourage reporting, and that injury in such environments is regarded as a part of the work. Based on his review, van der Schaaf (2004) identifies four main categories of reporting barriers: (1) fear of disciplinary action (as a result of a ‘‘blame culture’’ where those who commit an error are punished) or of other people’s reactions (embarrassment); (2) risk acceptance (incidents are part of the job, cannot be prevented, the ‘macho’ perspective); (3) useless (perceived attitudes of management taking no notice, not likely to do anything about it) and (4) practical reasons (too time-consuming, too difficult). 1.3. The objective of the paper It is evident that the under-reporting of safety information and risk, is a problem within the merchant shipping industry, yet we have not been able to find any research which empirically explores which factors affect the on-board reporting practices. As stated earlier, a considerable amount of research has been undertaken looking at reporting practices and the factors influencing it within other industries. However, working within merchant shipping is in many ways different from shore-based activities. Life and work onboard is a 24-h-a-day activity, where the crew has little opportunity to interact with the surrounding society. Most of the sailors are contract employees, comprised of multiple nationalities, with a typical sailing period of 9 months at a time. Also, due to its global nature, the industry is highly exposed to competition. (Oltedal and Engen, 2009). We have therefore identified a need for more research on the factors influencing reporting practices in the setting of merchant shipping. The objectives of the study are: (1) To determine the structure of safety culture in Norwegian controlled merchant shipping by liquid tankers and bulk carriers using exploratory factor analysis. (2) Explore the relationship between on-board reporting practices and safety culture factors. (3) Check for differences in reporting practices and familiarity with their local manager. (4) Check for differences in reporting practices between type of vessel; liquid tanker versus bulk / dry cargo carrier. 2. Method 2.1. Survey samples and administration The survey forms part of a PhD-research project which explores the relationship between shipboard safety culture and management on Norwegian controlled liquid- and drybulk cargo vessels, with a second goal of looking for differences between the two vessel classes. The sample was randomly selected from the Norwegian Shipowners’ Association’s list of members for 2005, and the sample was stratified with regard to type of vessel (general cargo, bulk carrier, oil tanker, gas tanker and chemical tanker). Initially, 150 vessels were selected, which constitutes about 15% of the population. Following the initial selection, telephone calls were made to each company to ask for their participation. Thirty-one companies agreed to participate with 83 vessels. Forty-five companies with 67 vessels declined to participate. Reasons for not participating included:

333

Being unable to contact the company despite repeated efforts (23 vessels, 16 companies). The vessel was not owned by a Norwegian party, and therefore not defined as Norwegian controlled (15 vessels, 8 companies). Ship management was outsourced to a non-Norwegian country, and therefore not defined as Norwegian controlled (14 vessels, 8 companies). The company refused to participate (12 vessels, 10 companies). The remaining vessels were sold (3 vessels, 3 companies). The population was redefined, and vessels managed from a nonNorwegian country were not considered Norwegian controlled. On an international basis, the various flag states’ performances are assessed on standards of safety, environment and social performance and are maintained and enforced by flag states, in full compliance with international maritime regulations. Based on their performance, the flag states are then categorized into three sub groupings; the white list (good performance), the grey list (mediocre performance) and the black list (poor performance) (MARISEC, 2008). It turned out that companies operating vessels flying a black listed flag declined to participate in the study. Thus, the sample is representative only for vessels flying a white and grey listed flag. The results of the study were based on self-completed questionnaire data. In total, 1574 questionnaires were distributed to 83 tankers and bulk carriers. 76 vessels from 29 companies returned a total of 1262 forms, which gives an individual response rate of 80.2%, a vessel response rate of 91.5% and a company response rate of 93.5%. Each vessel received a package with individual questionnaires and a sealable return envelope. On each vessel, the safety delegate received instructions regarding administration, purpose and anonymity. Vessels not returning any questionnaires were reminded up to four times. The survey was administrated during the spring/summer of 2006. 2.2. Questionnaire development The questionnaire was developed by Studio Apertura (2004) (a constituent centre of The Norwegian University of Science and Technology (NTNU), in collaboration with the Norwegian DNV and the research institution SINTEF (2003)) as a part of a programme for research in risk and safety in transport (RISIT) founded by The Research Council of Norway. The main part of the questionnaire was made up of 10 sections representing the following dimensions of safety culture: top management’s safety priorities, local management, procedures and guidelines, interaction, work situation, competence, responsibility and sanctions, working environment, learning from incidents, and description of the organization. All these constructs were measured on five point Likert scales ranging from ‘strongly disagree’ to ‘strongly agree’, or from ‘very seldom/never’ to ‘very often/always’. Håvold and Nesset (2009) discuss the issue of language and response style. They proceed using only an English and Norwegian version of their questionnaire. McCrae (2001) studied Norwegians and Filipinos who completed questionnaires both in their own language (Norwegian/Tagalog) as well as in English. He found no significant differences in the mean responses. However, in a cross-national study, Harzing (2005) found that English language versions tended to be more homogenized, potentially obscuring cross-national differences. Given that forcing the respondents to answer in the official working language of English could potentially bias the results, the questionnaire was also made available in Norwegian, Polish and Tagalog. This covered the main languages in use on-board. In addition, during the fieldwork it became obvious that there was significant variation in the English-language abilities of the crew. This provides further support for the approach adopted here. All

Author's personal copy

334

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338

participants were issued with questionnaires in their own national language and English. They were free to choose which version to return. 2.3. Demographics Questionnaires were returned from 40 liquid bulk carriers (liquid tanker) and 36 dry bulk carriers (dry cargo). 63% of the respondents were employed on a liquid tanker and 37% at a dry cargo vessel. The sample was male dominated (92.5% of the respondents). 22 nationalities were represented. The majority from the Philippines (65.5%), followed by Norway (9.2%), Poland (8.1%) and Russia (5.5%). Just over 56% of respondents were under the age of 40. 3. Results The Statistical Package for the Social Sciences (SPSS) v.16.0 and STATA v. 10.1 were used for the data analysis. 3.1. Dependent variable: reporting frequency The dependent variable, reporting frequency, was measured by the item: ‘‘Do minor incidents get reported in writing”. A total of 13% stated ‘never or seldom report’ (N = 164), 20.7% stated that they sometimes did report (N = 261), 31.8% that they often reported N = 401) and 23.5% that they always reported minor incidents. The remaining 11.1% were missing (N = 140).

Table 1 Factor scale-reliability test, number of items and explained variance. Factor ID

N items

Explained variance

Cronbach’s Alpha

Inter-item range

Item-total range

F1 F2 F3 F4 F5 F6 F7

4 4 3 3 3 4 2

30.621 10.102 7.917 7.296 5.612 5.391 4.366

.914 .825 .867 .819 .790 .669 .852

.651–.806 .453–.680 .647–.710 .527–.741 .440–.767 .262–.405 .743–.743

.781–.821 .604–.673 .731–.779 .564–.733 .470–.732 .410–.527 .743–.743

Factor 2; interpersonal, reflecting the relationship amongst the crew, problem solving abilities, form of communication and sharing of safety information. Factor 3; management, reflecting the crews’ perception of their closest manager as a role model, engagement and interest in ensuring safety in work operations. Factor 4; work practices, reflecting performance of pro-active activities as safe job analysis and hazard identification, and how safety is prioritised in daily operations. Factor 5; feedback, reflecting feedback given to crew on reported safety information, and experience reports from other vessels. Factor 6; shore orientation, reflecting the shore-side part of organisation’s attitude towards and prioritising of safety. Factor 7; efficiency, reflecting the relationship between perceived demand for efficiency and safe working practices. 3.3. Ordered logistic regression

3.2. Factor analysis All 1262 responses were submitted to explorative principal component factor analysis with Varimax rotation, in order to identify the latent underlying dimensions of safety culture. The data were deemed appropriate for analysis, according to the Kaiser– Meyer-Olkin measure of sampling value of .858, and significant Barlett’s test. Factors were extracted based on the following three analytical criteria: (1) Pairwise deletion, (2) Eigenvalue more than 1.0 and (3) factor loading more than 0.50. Items that failed to attain minimum loading of 0.5, or which loaded significantly on more than one factor, were omitted. This resulted in the extraction of 7 factors, explaining 71.305% of the total variance. This was followed by a scale-reliability test. Each factor was evaluated based on the following three criteria: (1) Cronbach’s Alpha coefficient > 0.70 (2) item-total correlation > 0.40, (3) inter-item correlation > 0.30, <0.80. However, these cut-off points are rules of thumb, and no clear consensus with regard to where the cut-off points exists (Hair, 1998; Field, 2005; Pett et al., 2003). Each item’s theoretical significance was also taken into account. Factor correlations analysis was carried out to evaluate the construct validity, which concerns the theoretical relationship between the factors. (Hair, 1998; Field, 2005). All extracted factors were found to be a valid and reliable representation of the underlying safety culture construct. Each factor’s explained variance, Cronbarch’s Alpha value, interitem range and item-total range are presented in Table 1. Extracted factor structure, item description and loadings are presented in Table 2. Loadings in bold indicate the factor onto which the item was placed. The seven extracted factors in Table 2, representing the onboard safety culture, were labelled as followed (label in italic): Factor 1; competence, reflecting the crews’ perception of their own training and education, in order to work safely and handle critical and hazardous situations.

The dependent variable ‘‘reporting frequency” has an ordinal nature, with four possible outcomes (1 = never/seldom, 2 = sometimes, 3 = often and, 3 = always). It is possible to handle such data in a number of ways. Ordinary least squares is seldom appropriate for such data since it requires a continuous dependant variable. To convert our dependant variable to a continuous variable would require an assumption about the distance between each of the categories. Rather than impose such an assumption, we choose an alternative technique: the ordered logistic regression model. The ordered logit regression model can be seen as an extension of standard logistic regression. Standard logistic regression utilises a dichotomous dependant variable. For a polytomous variable, this can be extended to a multinomial logit model. However, this neglects the fact that the variables are ordinal, and assumes that they are simply nominal. The ordinal logistic regression model utilises this important information. In the model, m 1 equation are simultaneously estimated, where m is the number of categories of the dependant variable. The equations are formed by pooling the data i.e. category 1 versus categories 2, 3 and 4, categories 1 and 2 versus categories 3 and 4 and then categories 1, 2 and 3 versus category 4. The assumption made is that the effect of the independent variables on the dependant variable is independent of the category. This assumption may be referred to as the proportional-odds assumption or the parallel regressions assumption. Like any assumption, it is not always met in practice. It can be tested by estimating a generalised ordered logit regression (which does not impose coefficient equality across the equations) and comparing it to the ordered logit regression using a likelihood ration test. We performed this test and obtained a p-value of 0.036. This indicates that at the 5% level of significance, the assumption of parallel regressions is not met. There are a number of options available at this stage. We can assume that the failure of the assumption is due to sampling variability and proceed with the standard logistic regression. We could

Author's personal copy

335

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338 Table 2 Seven factor rotated solution with factor loadings, and explained variance.

I have received the training that is necessary in order to handle critical or hazardous situations I have received the education that is necessary in order to handle critical or hazardous situations I have received the education that is necessary in order to work safely I have received the training that is necessary in order to work safely We receive sufficient safety-related information when we sign on/start a new sailing period We receive sufficient safety-related information when we start a new watch We solve problems and conflicts in a good manner The working environment on-board is characterized by openness and dialog Does your closest superior follow-up to ensure that all work on-board is done in a safe manner Is your closest superior a good role model when it comes to attending to his own and others’ safety Is your closest superior clear in his engagement to ensuring his co-workers’ safety Do you carry out a ‘‘Safe Job Analysis”/”Risk Analysis” before high-risk operations Do you carry out a safety evaluation before new working methods, tools, or routines are introduced Do you have the possibility to prioritize safety first in your daily work Do you receive constructive feedback from the company on the conditions you report Do you receive constructive feedback from the captain on the conditions you report? Do you get information from incidents/accidents on other vessels? In my company, they are more preoccupied with the statistics than the human consequences of an incident Reporting in itself take too much time I experience that safety work is more a facade than a real priority area When an undesirable incident has occurred, people are more preoccupied with placing blame than finding the cause of the incident Due to the captain’s demand for efficiency we sometimes have to violate procedures Due to the company’s demand for efficiency we sometimes have to violate procedures

also move to a multinomial logit and disregard the ordinal nature of the data. However, more insight can be gained by proceeding with the generalised logistic regression. This model allows the parameter estimates to vary by category. However, it may be the case that the parallel regression assumption is violated only with some of the included regressors. It is possible to test for this by imposing equality restrictions across equations on parameter estimates and testing the validity of these constraints using a Wald test. If there are good reasons, a priori, to select such constraints then these can be tested. Otherwise, a stepwise regression approach can be used to test which variables violate the assumption. In our case, there was no reason to suspect that some variables were more likely than others to violate the assumption. For this reason, we chose the stepwise approach. The resulting model is far more parsimonious than the unconstrained generalised logistic regression. The results are presented in Table 3. The results are presented as a set of ‘base’ coefficients and then deviations from proportionality. These deviations are calculated by taking the ratio of coefficients between equations (since the model is presented using odds ratios). For example, the parameters given under the heading ‘‘Deviations from Proportionality 2” are obtained by dividing equation 1 by equation 2. So, for example, the odds ratio for category 3 for the variable ‘Vessel’ can be calculated as 1.6483 0.5328 = 0.8782. Parameters which are constrained to be equal across equations are not shown. The results from the model are in line with expectations. The odds ratios are in the expected order of magnitude and are jointly significant. The factors competence, interpersonal, management, work practices and feedback all increase the odds of being in a higher category of the reporting frequency measure. The two factors reflecting shore orientation and efficiency significantly decrease the odds of being in a higher category of the reporting frequency measure. Crews on tanker vessels are more likely to have a lower reporting frequency than those on bulk vessels. Crews who have been working with their closest manager for more than one year tend to report more often. The three variables which did not meet the parallel regressions assumption were FAC5 Feedback, Vessel and Management. This means that the effect of these variables is dependent on the category of the dependant variable.

F1

F2

F3

F4

F5

F6

F7

.856 .855 .849 .813 .191 .239 .232 .157 .115 .093 .141 .130 .185 .190 .087 .078 .066 .089

.165 .152 .246 .270 .749 .746 .745 .730 .142 .212 .163 .131 .151 .288 .063 .041 .089 .021

.111 .117 .063 .097 .137 .151 .143 .124 .848 .821 .817 .136 .169 .202 .103 .152 .020 .098

.163 .168 .103 .087 .153 .139 .173 .087 .193 .104 .182 .871 .840 .622 .029 .056 .120 .030

.053 .078 .089 .082 .042 .106 .016 .091 .105 .079 .072 .048 .115 .112 .883 .875 .684 .154

.015 .034 .096 .102 .138 .097 .080 .120 .120 .139 .140 .063 .108 .051 .089 .022 .101 .784

.055 .070 .031 .037 .039 .016 .098 .127 .080 .059 .076 .075 .036 .101 .041 .077 .022 .091

.115 .070 .085

.078 .104 .249

.059 .063 .204

.177 .024 .027

.031 .110 .016

.709 .606 .560

.029 .222 .193

.093 .066

.134 .093

.130 .059

.039 .146

.091 .052

.228 .253

.878 .865

Table 3 Results from the constrained generalised ordered logistic regression analysis with factor scores (FAC). Odds ratios are presented. Beta

Odds ratio

Std. err.

z

P > |z|

FAC1 competence FAC2 interpersonal FAC3 management FAC4 work practices FAC5 feedback FAC6 shore orientation FAC7 efficiency Vessel (tank = 1) Management (exp. > 1 year = 1)

1.4949 1.1971 1.5573 1.3319 1.8090 0.6094 0.7581 0.5328 2.9817

0.1387 0.1095 0.1489 0.1262 0.2943 0.0594 0.0942 0.3104 0.3215

4.33 1.97 4.63 3.02 3.64 5.08 2.94 2.03 3.40

0.0000 0.0490 0.0000 0.0020 0.0000 0.0000 0.0030 0.0430 0.0010

Deviations from Proportionality 2 FAC5 Feedback Vessel (Tank = 1) Management (exp. > 1 year = 1)

1.4424 0.8117 0.9719

0.1578 0.2967 0.3092

2.32 0.70 0.09

0.0200 0.4820 0.9260

Deviations from Proportionality 3 FAC5 Feedback Vessel (Tank = 1) Management (exp. > 1 year = 1)

1.6840 1.6483 0.4595

0.2202 0.3701 0.3915

2.37 1.35 1.99

0.0180 0.1770 0.0470

1.6840 1.6483 0.4595

0.2792 0.2153 0.2634

6.05 0.89 6.54

0.0000 0.3750 0.0000

Constants CONS1 CONS2 CONS3 n = 473 ‘ ¼ 526:055 Pseudo R2 = 0.1561 Wald v2 (15) = 154.99

p-value = 0.0000

3.4. Validity and limitations There is a possibility that some of the relationships reported in this section are subject to the common method bias (Campell and Fiske, 1959; Podsakoff et al., 2003). This results from the fact that the data come from a common source. For example, a common scale to the different questions. Potential statistical remedies have been suggested. Spector (2006) is sceptical of the merits of such approaches. He argues that given that it is not possible to know the existence or extent of any possible bias, treating it could in fact introduce more bias than existed in the first instance. He

Author's personal copy

336

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338

recommends using a multi-method strategy so that results do not only rely on the results of one questionnaire. In this research, case studies, interviews, participatory and field studies have been used to validate the data. The results have also been presented to people working within the industry who have expressed that they believe the results to give an accurate representation of the true situation.

4. Discussion The results suggest that there is a positive association between the respondent’s perception of their local manager’s leadership skills and reporting frequency (P > 0.001, odds ratio = 1.5573). A leadership style where the manager is perceived as a good role model ensures and follows up that all work on-board is done in a safe manner, were positively related to increased reporting frequency. This relationship is also supported by other research. Perception of management, such as management/leadership style, commitment and visibility, is the most commonly measured dimension in safety research in general (Flin et al., 2000). Within the maritime sector, research initiated by the Maritime and Coastguard Agency in the United Kingdom identified various core leadership qualities as being necessary for effective safety leadership (Maritime and Coastguard Agency, 2004). Shipboard, it was found that these qualities were primarily geared towards the captain as a key leader for safety, however also for lower ranks with leadership responsibilities. However, perceived gaps between desirable leadership qualities and what is currently being exhibited were also identified. With reference to the explosion and sinking of the chemical tanker ‘‘Bow Mariner” (United States Coast Guard, 2005), poor leadership skills were on the agenda of the IMO (International Maritime Organization 31 October, 2007). The results also suggest increased reporting frequency for those who have worked with their closest superior/manager for one year or more. This ‘‘leadership-familiarity” variable did not meet the parallel assumptions, and thus it is indicated that the effect is dependent on the category. Apparently the effect of being familiar with your superior is larger when moving from the category ‘‘never or seldom” to ‘‘sometimes” report (P > 0.001, odds ratio = 2.9817), and the effect decrease slightly when moving to the higher categories of ‘‘sometimes” to ‘‘often”, and ‘‘often” to ‘‘always”. Leadership and management do have various facets, and may be seen as being both social and cognitive in nature. Social skills include things such as team building, consideration of others, conflict resolution etc. We would suggest that such social skills are of particular importance within this maritime setting. Work at sea may be characterized as a total institution, as defined by Goffman (1968), where both work and leisure time happen at the same time, with few and limited possibilities to interact with the surrounding world. In such settings, leadership and management style influences work and social life in a more all-embracing manner, including the interpersonal relationship among crew members. Although the research regarding managements importance for safety within the off-shore shipping industry is scarce, research from other industries has established a relationship between management, leadership and safety related matters (Geldart et al., 2010; Wu et al., 2008; Vredenburgh, 2002; Zohar, 1980), thus management does not only have a direct effect upon safety and reporting practices, but also an indirect effect by influencing the other factors in the model, further discussed below. The results also indicate that the interpersonal relationship among the crew influences reporting practices (odds ratio: 1.1971, P > 0.0490). With regard to reporting practices, interpersonal relationships relate to, inter alia, the degree of trust and open communication amongst the crewmembers. Reason (2001) regards a trusting relationship as a key factor in getting individuals to re-

port their own mistakes and experiences. Interpersonal relationships amongst crew, in practice, also reflects the degree to which the crew shares safety-related information when changing shifts, and more informal processes of sharing safety-related information during operations. For both subordinate and superior and managerial positions, additional challenges may arise in relation to multinational-crew and unstable crewing with low stability within teams. When signing on a new ship, new crew will be unfamiliar with the ship management’s and closest superior’s management style as well as fellow crewmembers and the on-board working climate. The seafarers require time to familiarize and adjust to the new situation. For instance, if the ship management on the seafarers’ previous vessel were blame oriented, this seafarer will most likely join the new vessel with this latest experience in mind, and be cautious about reporting their own mistakes for fear of being blamed or sanctioned. In time, the seafarer will learn how the management is oriented on that particular vessel. The problem is even more pronounced when the seafarer is constantly changing vessel, with new management each sailing period, and thus has to go through this familiarization process each time. The management style is known to vary within the sector, and poor shipboard management and leadership are identified in other research (Oltedal and Engen, 2010; Knudsen, 2003). The results indicate a positive relationship between competence and reporting frequency (odds ratio: 1.4949, P > 0.0000). Competence is among the top five most commonly measured themes within safety research (Flin et al. 2000), and refers to the perception of own skills and ability to handle critical and hazardous situations, and their ability to perform their work in a safe manner. In our analysis, competence is comprised of two sub facets, formal education and training. Minimum training requirements are covered by international conventions and regulations developed by the IMO, where parts are required to be performed on-board. For example, it is required that every crew member participates in at least one abandon ship and one fire drill every month. Also that these drills, as far as practicable, be conducted as if there were an actual emergency (International Convention for the Safety of Life at Sea (1974), International Maritime Organization 2009). On-board, the captain and ship management are ultimately responsible for how such drills, and other on-board training arrangements, are carried out. Experience from the field shows that in this area there are large variations. On some vessels, if performed, the drills are arranged as mustering, while others are arranged for realism. Variation in on-board training efficiency may be a result of various situations such as a lack of time due to demand for efficiency. The results indicate a negative relationship between demand for efficiency and reporting practices (odds ratio: 0.7581, P > 0.0030). Demand for efficiency, caused by inter alia commercial pressure, has been one of the most frequent reasons for violations of procedures and checklists (Oltedal, 2010), as well as an important influencing factor for collisions and groundings (MacRae, 2009). However, with regard to competence, we suggest that one significant cause could be that the minimum requirements for leadership and managerial skills stated in the international conventions and regulations are inadequate (also noted by IMO in (International Maritime Organization 31 October, 2007)). It is therefore up to each maritime educational establishment to decide to what degree managerial and leadership should be covered, or to each shipping company with regard to the provision of further education. Insufficient managerial and leadership skills could also result in time pressure and demand for efficiency through inadequate planning and resource management. We also suggest that the same relationships exist with regard to the factor concerning pro-active work practices (leadership, managerial skills, time available and competence interrelationship).

Author's personal copy

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338

Work practices refers to the degree to which the crew perform proactive activities like safe job analyses and safety analysis before risk activities, and to which degree they have the opportunity to prioritize safety in their daily work. Such activities increase the chances of revealing potentially dangerous situations, and thus increases the amount of reportable safety information (Kjellén, 2000). The results indicate that pro-active work practices have a positive relationship with reporting frequency (odds ratio: 1.3319, P > 0.0020). Feedback on reported events is held to stimulate organizational learning, and thus better premises for safety improvements, by, inter alia, sharing of experience of near-misses and incidents, as guidelines for corrective actions as well as being a motivator for increased reporting (Reason, 2001). A positive relationship between reporting frequency and feedback is also shown in our data (odds ratio: 1.8090, P > 0.0000). However, the relationship did deviate from proportionality meaning that the effect of feedback is dependent on the category of the dependant variable reporting frequency. Thus, the effect from feedback is larger when moving between the higher categories ‘‘sometimes” to ‘‘often”, and ‘‘often” to ‘‘always”, than when mowing between the lower categories ‘‘never or seldom” to ‘‘sometimes” Treatment of non-conformance and development or remedial actions is normally done by shore personnel. To what degree seafaring personnel are involved in these processes varies within companies. Shipboard feedback is given by the captain and/or shipboard management, who in turn receive the information from the shore-side of the company. Thus the quality of feedback given is not only dependent upon the ship management, but also the shore-side’s orientation towards safety. This relationship between shipboard reporting practices and shore-side safety orientation is also indicated by the results (odds ratio: 0.6094, P > 0.0000). When the shore-side’s safety effort is perceived as a facade and person orientated, it is reflected in lower reporting frequency. Finally, it is indicated that the reporting frequency is lower on dry cargo vessels than on tanker vessels. The relationship deviates from proportionality, meaning that the effect of type of vessel is larger when moving between the higher categories ‘‘often” to ‘‘always”, than when mowing between the lower categories ‘‘never or seldom” to ‘‘sometimes”. Differences between the liquid and dry cargo sector are further discussed in Oltedal and Engen (2009). 5. Conclusion, limitations and suggestions for future research This article has studied the factors which influence the frequency of reporting of experience data such as data on accidents/ incidents. The previous research outlined in the paper has shown that under-reporting is a significant problem within the merchant shipping industry. Under-reporting undermines the foundations on which any safety management system is constructed. If accidents/incidents are not reported then past mistakes cannot be learned from and the probability of future accidents/incidents cannot be reduced. In particular, the analysis presented in this paper is important since it not only identifies the significance of potentially influential factors, but also quantifies the relative strength of these factors. This allows a better targeting of budgetary resources to improve safety. In particular, the objective of this article was to assess the relationship between reporting practices and dimensions of safety culture, management and vessel in the Norwegian controlled shipping industry. The results indicated that high competence, a good and open interpersonal relationship among the crew, a safety oriented management, execution of pro-active work practices and feedback upon reported events all increase reporting frequency. The two dimensions reflecting when shore orientation downgrade safety

337

and when efficiency is given importance decreases reporting frequency. The three variables which did not meet the parallel regressions assumption were Feedback, Vessel and Management. This means that the effect of these variables is dependent on the category of the dependant variable. Crews who have been working with their closest manager for more than one year tend to report more often. However, none of the identified factors should be addressed in isolation from each other. As the discussion made clear, they are all important and mutually dependent. It would therefore be of further interest to explore the internal relationships between the identified dimensions of safety culture, for example with structural equation modelling and/or path-analysis. The data are representative of vessels flying a white and grey flag only, as those registered under a black listed flag did not want to participate. As participation was voluntary on the behalf of the company, we assume that those participating do, in general, emphasise safety in their operations, and the development of a sound reporting culture. However, it would be of further interest to address potential differences between flag of registration and safety. It would also be of interest to further explore the difference with regard to type of vessel. The differences in reporting frequency could be a result of other variables related to type of vessel, as for example customer specific. References Anderson, P., 2003. Cracking the Code: The Relevance of the ISM Code and its Impact on Shipping Practices. Nautical Institute, London. Campell, D.R., Fiske, D.W., 1959. Convergent and discriminant validation by multitrait-multimethod matrix. Psychological Bulletin 56 (2), 81–105. Ellis, N., Bloor, Michael, Sampson, Helen, 2010. Patterns of seafarer injuries. Maritime Policy & Management 37 (2), 121–128. Espin, S., Regehr, G., Levinson, W., Baker, G.R., Biancucci, C., Lingard, L., 2007. Factors influencing perioperative nurses’ error reporting preferences. AORN 85 (3), 527–528 (pp. 530–532, 534–536, 539–543). Field, A., 2005. Discovering statistics using SPSS: (and sex, drugs and rock ‘n ‘roll), second ed. Sage, London. Flin, R., Mearns, K., O’Connor, P., Bryden, R., 2000. Measuring safety climate: identifying the common features. Safety Science 34 (1–3), 177–192. Geldart, S., Smith, C.A., Shannon, H.S., Lohfeld, L., 2010. Organizational practices and workplace health and safety: a cross-sectional study in manufacturing companies. Safety Science 48 (5), 562–569. Goffman, E., 1968. Asylums: essays on the social situation of mental patients and other inmates. Penguin, Harmondsworth. Hair, J.F., 1998. Multivariate data analysis, fifth ed. springer, berlin. Hansen, H.L., Pedersen, G., 1996. Influence of occupational accidents and deaths related to lifestyle on mortality among merchant seafarers. International Journal of Epidemiology 25 (6), 1237–1243. Harzing, A.W., 2005. Does the use of english-language questionnaires in crossnational research obscure national differences? International Journal of Cross Cultural Management 5 (2), 213–224. Håvold, J.I., 2005. Safety-culture in a Norwegian shipping company. Journal of Safety Research 36 (5), 441–458. Håvold, J.I., Nesset, E., 2009. From safety culture to safety orientation: validation and simplification of a safety orientation scale using a sample of seafarers working for Norwegian ship owners. Safety Science 47 (3), 305–326. Hollnagel, E. 2009. The ETTO Principle: Efficiency-Thoroughness Trade-Off Why Things That Go Right Sometimes Go Wrong, Ashgate. International Convention for the Safety of Life at Sea (1974) & International Maritime Organization 2009, SOLAS on CD-ROM + demo, IMO Publ., London. International Maritime Organization 31 October 2007, Comprehensive Review of the STCW Convention and the STCW Code – Communication and leadership skills (STW/39/3), International Maritime Organization, London. International Maritime Organization, 2007a. Formal Safety Assessment: Consilidatated text of the Guidelines for Formal Safety Assessment (FSA) for use in the IMP rule-making process (MSC/Circ.1023-MEPC/Corc.392). International Maritime Organization, London. International Maritime Organization, 2007b. Role of the human element: Near miss information. International Maritime Organization, London. International Maritime Organization, 2002. International safety management code: ISM code: and revised guidelines on implementation of the ISM code by administrations, second ed. International Maritime Organization, London. Kjellén, U., 2000. Prevention of accidents through experience feedback. Taylor & Francis, London. Knudsen, F., 2003. If you are a good leader I am a good follower. MacRae, C., 2009. Human factors at sea: common patterns of error in groundings and collision. Maritime Policy & Management 36 (1), 21–38.

Author's personal copy

338

H.A. Oltedal, D.P. McArthur / Safety Science 49 (2011) 331–338

MARISEC, 2008. Shipping Industry Flag State Performance Table. Maritime International Secretariat Services Ltd (Marisec), London. Maritime and Coastguard Agency, 2004. Driving Safety Culture: Identification of Leadership Qualities for Effective Safety Management. McCrae, R.R., 2001. Cross-cultural research on the five-factor model of personality. In: Lonner, W.J., Dinnel, D.L., Hayes, S.A., Satler, D.N., (Eds.), Online Readings in Psychology and Culture (Unit 6, Chapter 1), Center for Cross-Cultural Research, Western Washington. <http://www.wwu.edu/~culture>. Oltedal, H.A., Engen, O.A., 2009. Local management and its impact on safety culture and safety within Norwegian shipping. In: Artorel, M., Guedes, C., Barnett Safety, J. (Eds.), Reliability and Risk Analysis Theory, Methods and Applications, 2008th ed. European Safety and Reliability Association. Oltedal, H., 2010. The use of safety management systems within the Norwegian tanker industry – do they really improve safety? In: Bris, R., Guedes Soares, C., Martorell, S. (Eds.), Reliability Risk and Safety: Theory and Applications. Taylor & Francis Group. Oltedal, H.A., Engen, O.A., 2010. Tanker versus dry cargo – The use of safety management systems within Norwegian dry cargo shipping. In: Ale, B.J.M., Papazoglou, J.A., Zio, E. (Eds.), Reliability, Risk and Safety. Taylor & Francis Group, London. Pett, M.A., Lackey, N.R., Sullivan, J.J., 2003. Making sense of factor analysis: the use of factor analysis for instrument development in health care research. Sage, Thousand Oaks, Calif. Podsakoff, P.M., MacKenzie, S.B., Lee, J., Podsakoff, N.P., 2003. Common method biases in behavioral research: a critical review of the literature and recommended remedies. Journal of Applied Psychology 88 (5), 879. Probst, T.M., Brubaker, T.L., Barsotti, A., 2008. Organizational injury rate underreporting: the moderating effect of organizational safety climate. Journal of Applied Psychology 93 (5), 1147–1154.

Psarros, G., Skjong, R., Eide, M.S., 2010. Under-reporting of maritime accidents. Accident Analysis & Prevention 42 (2), 619–625. Reason, J., 2001. Managing the risks of organizational accidents. Ashgate, Aldershot. Roberts, S.E., Marlow, P.B., 2005. Traumatic work related mortality among seafarers emplyed in British merchant shipping 1976–2002. Occupational and Environmental Medicine 62, 172–180. Rosenman, K.D., Kalush, A., Reilly, M.J., Gardiner, J.C., Reeves, M., Luo, Z., 2006. How much work-related injury and illness is missed by the current national surveillance system? [Article]. Journal of Occupational & Environmental Medicine 48 (4), 357–365. SINTEF, 2003. Sikkerhetskulturer i transport: En kunnskapsoversikt. (STF 22A03300). Spector, P.E., 2006. Method variance in organizational research: truth or urban legend? Organizational Research Methods 9 (2), 221. Studio Apertura. 2004. Sikkerhetskulturer i transportsektoren. Metoder for kartlegging av sikkerhetskultur: Evaluering av noen eksisterende verktøy. United States Coast Guard, 2005. Investigation into the explotion and sinking of the chemical tanker Bow Mariner in the Atlantic Ocean in February 28 2004 with loss of life and pollution. United States Coast Guard, Washingthon. van der Schaaf, T., Kanse, L., 2004. Biases in incident reporting databases: an empirical study in the chemical process industry. Safety Science 42 (1), 57–67. Vredenburgh, A.G., 2002. Organizational safety: Which management practices are most effective in reducing employee injury rates? Journal of Safety Research 33 (2), 259–276. Wu, T., Chen, C., Li, C., 2008. A correlation among safety leadership, safety climate and safety performance. Journal of Loss Prevention in the Process Industries 21 (3), 307–318. Zohar, D., 1980. Safety climate in industrial-organizations – theoretical and applied implications. Journal of Applied Psychology 65 (1), 96–102.

Article 6 Oltedal, H. A., & Engen, O. A. (2010). Safety Management in Shipping—Making Sense of limited Success. Safety Science Monitor, submitted.

Safety Management in Shipping: Making sense of limited success H.A. OLTEDAL*, O.A. ENGENh *Faculty of Technology, Business and Maritime Education, Stord/Haugesund University College (HSH), Haugesund, Norway. h SEROS-Centre for Risk Management and Societal Safety, University of Stavanger, Stavanger, Norway.

Abstract This article addresses the gaps between regulative safety ambitions and operational results and practice within the maritime sector. In particular, the article focuses on the gaps within safety ambitions in the form of, for example, rules and procedures and practise, exploring possible pitfalls when relying on safety from a system perspective without paying attention to its human interrelationship. This examination applies the Practical Drift Model (PDM) conceptual framework developed by Scott A. Snook (2000). PDM combines the Normal Accident Theory (NAT) and High Reliability Organization Theory (HRO) to explain why seemingly well-controlled and sophisticated organisational systems may in reality increase their risk potential. In fact, existing scientific safety literature has not offered much space in which to make sense of the gaps between theory and practice. Hence, we more closely examine certain examples of such gaps from our own data collected from the Norwegian controlled shipping industry. Applying a multi-method approach combining surveys and case studies, including field studies and interviews, we conclude that maritime accidents to a large degree derive from a mismatch between the local demands of the situation and global rules designed by planners. 1 Introduction Shipping is a global industry as most vessels are affected by legislation in many countries and abide international rules; in addition, owners can choose flag states and labour markets in which to recruit crews. Such interdependencies also make shipping a highly complex business. In the safety area, many stakeholders (e.g., crew, shipping companies, unions, industrial bodies, national and international regulators) constantly act and react to internal and external changes. Although safety is to a large extent regulated and procedures and guidelines for best practices are in place, severe accidents and incidents still frequently happen. 1 International safety regulations stem from the International Maritime Organization (IMO) in the form of conventions, protocols, and resolutions. At other levels, safety is addressed by regional, flag state, and port state regulations (Kuo, 2007), which are transformed into shipping companiesâ&#x20AC;&#x2122; safety management systems in the form of procedures and standards. Companiesâ&#x20AC;&#x2122; safety activities are regulated by the International Safety Management (ISM) 1 Two of particular note are the Norwegian ship Langeland, which sank in a storm off the coast of Sweden (all six crewmen were presumed dead0, and Full City, which went aground and polluted the southern Norwegian coastline. Both accidents happened during the same storm on the night of 31 July 2009.

1

Code, which was fully implemented on 1 July 2002 (719 International Maritime Organization, 2010). Although procedures and standards are in place to control risk and seafarers’ behaviour, accidents involving human error also seem to be on the rise (Soma, 2008). Statistics from Lloyds’ Fairplay (2010), shown in Figure 1, indicate the navigational accident frequency (collisions, contacts, and wrecked/stranded vessels) from 1993 to 2009 in relation to the world’s fleet size. Figure 1: Navigational accident frequency 1993-2009 (Source: Lloyds' Fairplay, 2010)

Navigational accident frequency 0,0200 0,0150 0,0100 0,0050 0,0000 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Navigational accident frequency

Fleet size in number of: Crude oil tankers over 100,000 dwt, Chemical tankers over 10,000 dwt, Containers over 20,000 dwt, RoRo cargo over 10,000 dwt, Bulk over 50,000 dwt.

As Figure 1 indicates, the frequency of serious accidents—especially navigational accidents— has been increasing since 2002, despite the introduction of the ISM Code. The European Maritime Safety Agency (EMSA) has suggested links among accident numbers, loss of life, and economic activity, especially when accidents occur because ships and seafarers are being worked harder (European Maritime Safety Agency, 2010). Surprisingly, the trend indicates that the number of accidents is higher during economic upturns. According to EMSA (2010), during economic downturns, supply overcapacity, high levels of ship scrapping, lower operating speeds, and generally less pressure to meet tight deadlines are seen to be the main reasons for reductions in accidents; the opposite is the case during economic upturns. In addition, our own research demonstrates how well-intended pressures to induce safer practices can at times be either ineffective or directly counterproductive. Furthermore, EMSA (2010) noted that the great majority of accidents include a human error component; seafarers often make mistakes under difficult circumstances, such as bad weather, fatigue, task overload, and training shortcomings. Thus, we would stress the importance of including and paying attention to this human interrelationship when working with safety. 1.1 Objective of the Article This article addresses the gaps between safety ambitions in the form of establishing, for example, rules and procedures and practise, where we also explore possible pitfalls when relying on safety through a system perspective without paying attention to its human interrelationship, especially those in the sharp end—namely, the seafarers themselves. In order to accomplish such objectives, we will consider how the Normal Accident Theory (NAT) and High Reliability Organization Theory (HRO) explain and make sense of safety, 2

risks, and accidents. We will also show how these theories do not sufficiently explain why seemingly well-controlled and sophisticated organisational systems in certain contexts and situations tend to increase their risk potential. In order to make sense of the gaps between safety ambitions and the practical outcome, we will examine data collected from the Norwegian controlled shipping industry. By reconsidering the applicability of HRO and NAT, we apply the Practical Drift Model (PDM) conceptual framework developed by Scott A. Snook (2000). PDM aims to combine NAT and HRO and presumably better explain why seemingly well-regulated organisations develop traits that may evolve into significant accidents and disasters. As such, the PDM model enables us to look for possible new explanations of the success and failure of safety strategies. The article is organised as follows. In the next section, we briefly present our theoretical framework by introducing NAT and HRO. Thereafter, we introduce PDM and explain the relevance of the model. The following section presents relevant data from the survey of Norwegian-controlled shipping industry as well as the chosen methodological framework. The findings and discussion follow the logic of the PDM, which organises the presentation of the qualitative data. Finally, we summarise existing theoretical explanations and suggestions, supplementing them with some of our own new findings to formulate hypotheses for future theoretical research. 2 Theoretical Approaches to Safety Management According to Scott and Davies (2007), NAT belongs to the open system paradigm and thus includes organisational membersâ&#x20AC;&#x2122; intentions, power, loyalty, and identity along with the influence of the environmental context outside the organisation itself. Systems (organisations and their surroundings) are characterised as error avoiding, error neutral, and error inducing, based upon a number of organisational and contextual features. NAT brings forward the notions of interactions and couplings, whereupon systems are placed in a matrix depending upon the characteristics of their interactions and couplingsâ&#x20AC;&#x201D;namely, linear or complex interactions and tight or loose couplings. Accidents are explained by the complexity in the interactions and tight couplings. Therefore, error-inducing systems should strive to reduce the complexity and/or loosen up the system couplings. In order to handle the risk inherent in such systems, some managerial advice has been given. According to Perrow (1999), a system with linear interactions, loose couplings, and few complex interactions may handle both centralised and decentralised (component) failure. A linear and tightly coupled system is expected to be best managed in a centralised manner, as (component) interaction is visible and expected. Such a linear system is also fit for standardised measures and procedures, as there is little variation in the production system. Meanwhile, in a complex and loosely coupled system, decentralisation is desirable in order to cope with unexpected complex interactions that enable people to develop substitutions and alternative paths. In such a system, standardised measures are less fit, as the complexity brings along more variation in operations. Risk in these three mentioned combinations of systems is supposed to be manageable; however, Perrow (1999) pointed out a problem with managing systems that are both tightly coupled and complex. In such systems, centralisation is desirable for coping with tight couplings while decentralisation is desired for coping with the unplanned interactions of failure. According to Perrow (1999), an organisation cannot be simultaneously centralised and decentralised. The essence of NAT is that, no matter how hard we try, these kinds of systems are ultimately bound to fail. Although NAT is built strongly upon a system perspective, the premise is not traditional system theory (see Bertalanffy, 1971, for traditional definitions). In NATâ&#x20AC;&#x2122;s understanding of 3

system, the aim is to understand specific events, such as the failure of an operator to close a valve and examine it in the context of the failure (Perrow, 2004). This specific context could be the operators’ mental models, which again are the result of training, company ideology, experience, intentions, and power figurations. Each element—industry structure, organisational structure, group structure, and the like—should again be examined in its own context. Although NAT apprehends accidents as inevitable in interactively complex and tightly coupled systems, HRO applies a contrary strategy in order to develop nearly error-free organisations (Weick and Sutcliffe, 2007). HRO advocates pay more attention to internal organisational processes than system characteristics, yet they also assume that the research objects all are interactively complex and tightly coupled. Recognising that the world they face is complex, unstable, unknowable, and unpredictable, the HRO perspective is reluctant to accept the simplification and standardisation inherent in traditional risk assessment. This view is shared by Perrow (1999). Both NAT and HRO are sceptical towards those who rely highly on risk assessment when developing preventive safety measures. In HRO, more attention is given to the real work going on in frontline operations. By empowering the persons doing the actual work, operators are enabled to solve the situations themselves, based upon their own experience and knowledge. They operate using pre-planned descriptions, but it is accepted that real situations will include deviations from standardised measures as procedures. One of the HRO’s key points is “mindfulness”. We understand mindfulness as a combination of alertness, sensibility, flexibility, and adaptability. The perspective argues that unexpected events should be handled by creating a mindful infrastructure by following five main principles: (1) continually track small failures, (2) resist oversimplification, (3) be sensitive to operations, (4) maintain capability for resilience, and (5) monitor shifting locations of expertise. The violation of these principles is regarded as a step backwards towards the more traditional approach, where simple diagnoses are accepted and frontline expertise is overridden by faith in risk analysis and developed measures detached from the operations. The HRO view implies that operators gain more responsibility, resulting in the need for other parts of the organisation to give them the possibility to act. In other words, control is taken from the upper levels of the organisation in favour for lower levels. However, the definitions of both complexity and interaction are relative. A new system may be perceived as more complex than a system that has existed for some time. Interactions may be unknown and not comprehensible in the early dawn of new technologies; through experience, the interactions become more understandable and visible. A system may be perceived as complex for a new operator, but as simple for an experienced one. A system may also be perceived as simple in certain situations, but turn complex if changes occur in its context. Therefore, we would like to incorporate a new dimension, acquaintance, representing the knowledge and experience one has with the system and context. Acquaintance complements the concept of mindfulness, particularly when incorporating time as an important factor in the analysis. PDM (Snook, 2000) combines HRO and NAT in two ways. First, it emphasises how different degrees of mindfulness depend on different situations and contexts and how organisational systems during their lifetime develop both tight and loose couplings. The main objective of the model is to capture both contextual and temporal factors when explaining why incidents and accidents occur. More precisely, PDM consists of three dimensions: (1) situational couplings, (2) different logic of actions, and (3) time. The first dimension, situational coupling, refers to Karl Weick’s statement that it is not the existence or nonexistence of loose couplings that is of a crucial determinant of organisational functioning over time, but rather the patterning of loose and tight couplings (Weick, 1976). This statement underlines how organisations shift from tight to loose couplings and back again as the various 4

sub-units within the taskforce alternate between low and high degrees of interdependence. The second dimension, logic of action, refers to what “new institutionalism” defines as contextually dependent mindsets or frames that influence peoples’ behaviour (DiMaggio, 1994). Here, mindsets and frames refer to norms, scripts, routines, or habits possessed by the actors within an organisational setting. Snook’s (2000) premise in PDM is that organisational members shift back and forth between rule- and task-based logics of action depending on shifting contextual factors. The third dimension, time, refers to the lifetime of an organisation, which is characterised by shifting periods of tight and loose couplings and different kinds of logic of actions. The dynamics of PDM are driven by the twin muscles of behaviourally anchored logic of action and stochastically determined situational coupling (Snook, 2000). This duality implies that periods of stabilisation with ruled-based logic of action will be replaced by periods of instability characterised by a more task-based logic of action. The latter invites a higher risk potential and greater possibilities for accidents.

1) Design

4) Failed

Tight ly coupled Ruled based logic of action

Tightly coupled Task based logic of action

2) Engineered

3) Applied

Loosely coupled Rule based logic of action

Loosely coupled Task based logic of action

Unstable

Stable

Figure 2: The Practical Drift Model (Based on Snook, 2000)

The different boxes in Figure 2 refer to various phases of the organisational development and different characteristics of situational couplings and actions of logic. Box 1 in the model is denoted as “design” and refers to the stage of an organisational lifecycle in which the governance structure is top-down oriented. The organisational managers or designers have put significant efforts into developing extensive routines and procedures in order to make the organisation robust and resilient against attacks and unforeseen events. This implies a technocratic character of the organisation, which also assumes tight couplings and a rulebased action of logic. Box 2 is denoted as “engineered” and refers to an operational situation that is more loosely coupled. During operations, the rules of logic, which are designed for tight couplings, do not always match the situations. When organisations based on strict, high reliability principles experience that the unforeseen attacks and events do not occur as expected, the attention focused on the routines and procedures become part of everyday life and practice

5

and, consequently, more relaxed. The logic of action based on the rules of “continuously red alert” tends to normalise, which results in a tendency for de-coupling organisational systems. The real world does not act in accordance with organisational design; as actors become aware of this, they will be able to break the strict rules without any fear of sanctions or punishment. On a local basis, breaking the strict rules may actually get the job done more quickly and efficiently. The accumulated experiences from the real world of everyday practice will thus push the organisation away from the originally designed strict rules and create unstable situations. Such instability generates further pressure for change, as evident in the shift towards box 3 in the model, which is denoted as “applied”. The shift from “engineered” to “applied” involves a change of mentality among the actors and a transition from tight to loosely coupled organisational systems. This process is what Snook (2000) calls “practical drift”. During such a process of de-coupling, the organisation becomes increasingly free from the global rationality that characterises “design” and develops sub-units with their own defined rationality and internal governance structures. Such sub-units are autonomous entities whose logic of action is based on experiences and tacit knowledge. Accordingly, the actors drift further away from a rule-based to a more taskbased system. The locally independent sub-units solve everyday problems in an applied taskbased manner. The problem-solving processes become pragmatic responses to what the single units (e.g., work groups or vessels) justify as their own understanding of “the rules”. Box 4 “fails” indicates a situation in which the system suddenly and stochastically becomes tightly coupled. The circumstances may include an incident in which single units confront one another (e.g., two ships on a collision course). In such situations, the actors in the individual units are forced to act on the assumption that all others act in accordance with the original rules and initially designed procedures. This may create extreme instability and, in the worst case, a catastrophe. The actors are actually trapped in a game where trusting their own logic of actions is the only solution, yet they must simultaneously base their decisions on the assumption that others are following the general rules. From “failed”, the organisation once again enters the phase of “design”. In “redesign”, the organisation tries to control for the experienced unwanted outcome, often in the form of even tighter designed control criteria. Thus, the model indicates a recovery phase in which actors learn from mistakes and reintroduce a top-down-based governance structure with tight couplings and logic of action based on generally accepted rules. Real-life accidents often induce such processes. Therefore, the model indicates certain determinism—namely, “peaceful” organisations without repeated redesign processes inevitably drift towards ruin. On the one hand, this is the old story of “the unrocked boat” (Reason, 2001). On the other hand, the model indicates how structural constraints or enforcements also become influenced by external contexts and either fortify the lack of mindfulness or increase mindfulness based on locally based logic of actions. Both ways of practical drift may be equally dangerous. The practical drift model thus emphasises the organisational contradiction between decentralisation and control, where decentralised decision making is needed to permit flexible responses to surprises while increasing the degree of complexity requires centralisation and discipline. Bridging the gap between safety goals and actual performance aims to reduce this kind of contradiction and create organisational designs that may simultaneously inhibit the logic of hierarchical control and decentralised decision making. 3 Methodological Considerations This article adopts a multi-method approach combining surveys and case studies, including field studies and interviews. Safety management and its interrelationship with 6

human behaviour are, in accordance with current accident and safety management theories (e.g., Cooper, 2000; Reason, 2001; Turner and Pidgeon, 1997; Weick and Sutcliffe, 2007), understood as a multi-layer construction comprising organisational members’ fundamental safety-related assumptions and values (what is important), beliefs (how things work), and patterns of behavioural norms (how things are done). The fundamental assumptions that individual group members have towards safety management and the explanation of actual behaviours at operational level are difficult to understand and explain without interactive probing and qualitative methods. Fundamental assumptions are assumed to be reflected in behavioural norms and perceptions, which are far more accessible through quantitative methods. Once integrated, these factors are commonly referred to as safety culture (Reason and James, 2001; Reichers and Arnon, 1990) The current study was carried out in two phases. The first phase was a self-completion questionnaire survey of crewmen working on Norwegian controlled liquid tankers and dry cargo vessels. A random sample was drawn from the Norwegian Shipowners’ Association’s list of membership. The second phase involved a sub-sample and case studies selected from the survey sample group. This second phase employed semi-structured interviews with crewmen and shore personnel along with participatory observations. Thus, the quantitative results are representative of the selected population and are applicable for generalisation. Although the narratives presented in the articles are not applicable for generalisation, they provide examples and understanding of how the gap between safety goals and actual performance is created through social processes. In addition, some characteristics of the interaction between the various organisational members and the existing safety control systems are brought forward with the qualitative results. Furthermore, a multi-method approach is applied so that the interrelationships between the elements may be examined in order to establish antecedents, performance, and outcomes (Bergman, 2008; Brannen, 2005; Cooper, 2000). More precisely, the survey aims to indicate seafarers’ perceptions of and attitudes towards safety management and risk. The qualitative results aim to provide a more thorough understanding of the overall situation, underlying processes, and seafarers’ own situational experiences. A theoretical and practical framework for the assessment of safety culture open to a multi-method approach is provided by Cooper (2000). This frameworks includes (1) subjective internal psychological factors (i.e., attitude and perceptions); (2) observable ongoing safety-related behaviour; and (3) organisational situational features. According to Cooper (2000), these elements reflect those accident-causation relationships previously found by a number of researchers, such as John Adams (1995), Herbert William Heinrich (1980) and James Reason (2001). The investigation of several serious shipping accidents, such as the Herald Free Enterprise (Department of Transport, 1987), the Exxon Valdez (National Transportation Safety Board, 1990), and the Scandinavian Star (Justis-og politidepartementet, 1991), is also congruent with their findings. 3.1 Questionnaire Development The survey instrument was developed and validated by Studio Apertura in collaboration with the Norwegian DNV and the SINTEF research institution (Studio Apertura, 2004). The main part of the questionnaire comprises 10 sections representing specific safety-related dimensions: (1) top management’s safety priorities, (2) local management, (3) procedures and guidelines, (4) interaction, (5) work situation, (6) competence, (7) responsibility and sanctions, (8) working environment, (9) learning from incidents, and (10) description of the organisation. This article analyses the results from the third section on procedures and guidelines. All items were measured on five-point Likert scales ranging from “strongly 7

disagree” to “strongly agree”. The respondents were also given the possibility to comment in a specific situation. 3.2 Survey Samples and Administration In total, 1,574 questionnaires were distributed to 83 tankers and bulk carriers. Each vessel received a package with individual questionnaires and a sealable return envelope. On each vessel, the safety delegate received instructions related to administration, purpose, and anonymity. Vessels not returning any questionnaires were reminded up to four times. The survey was administrated during the spring/summer of 2006. A total of 76 vessels from 29 companies returned 1,262 forms, providing an individual response rate of 80.2%, a vessel response rate of 91.5%, and a company response rate of 93.5%. In addition, 297 respondents provided written comments. For further information regarding survey sample and administration, see Oltedal and Wadsworth (2010). 3.3 Survey Demographics Questionnaires were returned from 40 liquid bulk carriers (liquid tanker) and 36 dry bulk carriers (dry cargo); 63% of the respondents were employed on a liquid tanker and 37% on a dry cargo vessel. The sample was male dominated (92.5% of the respondents), and 22 nationalities were represented. The majority of respondents were from the Philippines (65.5%), followed by Norway (9.2%), Poland (8.1%), and Russia (5.5%). Just over 56% of the respondents were under the age of 40. Only a few (11.5%) of the respondents (mostly Norwegian nationals) had fixed employment within the shipping company. The remaining 88,5% were contract employees—43.6% with 9-month contracts, 21,5% with 6-month contracts, and 4,5% with 3-month contracts. The remaining 18.9% had a different length of contract, none of which was more than 1 year. Finally, when signing on a new vessel, 15.2% of the respondents stated seldom or never signing on to the same vessel, 39.9% stated staying on the same vessel sometimes, and the remaining 44.8% often or always stay on the same vessel. On seven of the vessels, all respondents who returned the questionnaire stated that they often or always sailed on the same vessel. 3.4 Case Studies Four shipping companies—two tankers and two dry cargo companies—were approached about conducting case studies from 2007 to 2009. In order to ensure the companies’ anonymity, company-specific information was retained. In all four companies, the HSQ manager and Crewing Manager were interviewed. In one company, the safety management system (SMS) data system was examined. In the other companies, all available statistics, experience feedback from reported cases, and safety bulletins were examined. Information was also retrieved through participatory observations at sea from two field studies, participation in captains’ conferences where safety was an issue, and both formal and informal interviews with seafaring captains and shore-based personnel involved in safetyrelated matters. The qualitative information aims to give a more thorough understanding of the processes underlying the presented statistics. The narratives serve two important complementary purposes. First, they contextualise the data in a broader and deeper sense than the quantitative studies are able to offer. Second, they personalise the relationships between regulatory regimes and their subordinates, illustrating what kind of meaning the agents put into their action when deciding to follow or not to follow the prevailing regulatory rules. This article emphasises seafarers’ view. 8

3.5 Statistical Analysis and Presentation of Results All statistical analyses were performed using PASW (by SPSS) statistics version 18.0. Descriptive statistics for each item, including the percentage of frequency distribution, mean, and standard deviation, are presented. Principal Component Analysis (PCA) with Varimax rotation and Pairwise deletion was carried out in order to examine survey items’ interrelationships and their common underlying dimensions. The extracted rotated component matrix and loadings were presented. The loading represents the correlation between the variable and the extracted factor(s), with estimates ranging from 0 to 1,00. Items that load strongest on a given component are considered the most like the underlying latent dimension (Hair, 1998; Pett, Lackey and Sullivan, 2003). 4 Rationale of Procedures in Maritime Safety Management and Research Findings This section presents data related to operational procedures. The use of procedures in error management involves two components: error reduction and error containment (Reason, 2001). According to Reason (2001), some problems are associated with the existing form of error management, including the fact that error management is generally not informed by current knowledge related to error and accident causation. Moreover, error management tends to focus on personal and active failures rather than latent conditions and the situational contributions to making errors. With reference to this article, latent organisational and situational conditions are defined as (1) incompatible goals that include (a) group goal conflicts when the informal norms of a work group are incompatible with the safety goals of the organisation and (b) conflicts at the organisational level in which there is incompatibility between safety and productivity goals; (2) procedures in relation to their quality, accuracy, relevance, availability, and workability; and (3) training and problems that include the failure to understand training requirements, the downgrading of training relative to operations, poor task analysis, and the inadequate definition of competence requirements (Reason. 2001). In the following discussion, our results related to these areas—namely, incompatible goals, procedures, and training—will be further addressed in relation to the industries’ approach to safety management. In order to overcome human error as a cause in maritime accidents, the International Safety Management (ISM) code (Resolution A.741 (18)) was adapted by the International Maritime Organization (IMO) in 1994, with part of the objectives ensuring safe practices on board ships as well as the establishment of safeguards against all identified risks (Anderson, 2003). The code was initiated after the Herald of Free Enterprise accident, when underlying contributory causes to the accident were found to be poor safety management and failure on the part of the shore side to give the proper and clear directions (Department of Transport, 1987). By introducing the code, operational safety is promoted through a formal SMS. A SMS contains several sub-systems. First, a system for reporting and collecting experience data from the vessel itself is required. This is followed by a system of data processing— namely, summarisation and analysis in order to reveal causal factors and perform trend analysis, which forms the basis of the development of safety measures and subsequently often on forms related to areas such as standardised procedures and checklists (Oltedal, 2010). Standardised procedures and checklists are also particularly aimed at controlling human error, which is associated with the vast majority of accidents. An estimated 75-96% of marine casualties are caused—at least in part—by some form of human error (Rothblum, 2000). Although systems for safety management and development of measurements are in place, accidents caused by human error still occur, procedures are violated, and best practices are not followed. According to Standard P&I Club, aspects of the ISM/SMS system are not 9

effectively implemented (e.g., no coherent system to ensure that lessons are learnt from mistakes), which impacts the effectiveness of safety measures (Standard P&I Club, 2010). When it comes to the ineffectiveness of ISM/SMS systems, one particular issue will be put forward herein: Although various systems for safety management do exist, safety management per se is not only a system property. It is not sufficient to have a system for the collection of safety-related data. The system’s efficiency is determined by its human interrelationships (crew, shore personnel, analysts, and others). Thus, we should stress the importance of including and paying attention to this human interrelationship when working with safety. Survey questions related to the perception of procedures are presented in Table 1, with descriptive statistics included in percentages, together with mean and standard deviation. Scale information and coding are 1=strongly disagree, 2=disagree, 3=not sure, 4= agree, and 5=strongly agree. Table 1. Descriptive Statistics, Procedures, and Checklists

Survey item description Due to the company’s demand for efficiency, we sometimes have to violate procedures Due to the captain’s demand for efficiency, we sometimes have to violate procedures I have received good training in the company’s procedures I feel that it is difficult to know which procedures are applicable The procedures are helpful in my work The procedures are difficult to understand or are poorly written We have the opportunity to influence and form the procedures

1 2 3 4 5 14.9 41.4 12.8 26.6 4.4

Mean 2.64

Std 1.151

18.4 42.1 14.3 21.9 3.2

2.49

1.117

58.9 31.5 4.18

0.728

0.9

2.6

6.1

14.4 47.0 17.3 19.0 2.3

2.48

1.028

0.6 1.4 4.8 55.6 37.6 4.28 15.3 57.3 12.6 12.8 2.0 2.29

0.678 0.941

3.4

0.978

13.1 15.5 55.3 12.8 3.61

The descriptive statistics summarised in Table 1 show that 31% of the respondents violate procedures due to the company’s demand for efficiency while approximately 26% do so due to the captain’s demand for efficiency. In addition, about 90% of the respondents are satisfied with the training received in their company’s procedures. Moreover, approximately 93% perceive procedures as helpful in their work, indicating that the respondent’s attitudes towards procedures per se are good. Thus, when procedures are breached, it is more likely due to reasons other than poor attitude. About 21% feel that it is difficult to know which procedures are applicable, which indicates a procedural system not adequate for the situation and, as a result, weaknesses throughout the overall safety management system. In addition, approximately 15% find the procedures difficult to understand and poorly written. Finally, about 68% stated that they have the opportunity to influence and form the procedures. The interrelationship among the items is presented in Table 2, along with items’ component loadings. Based on the analysis, two components (i.e., factors) are extracted. The matrix contains factor loadings for each variable (i.e., item) at each factor. Factor loading is the means of interpreting the role each variable plays in defining each factor and represents the correlation of each variable and factor. Higher loadings (e.g., > 0.3 or higher) indicate the variable representative of the factor and underlying dimensions. Within each factor, interrelated variables are clustered at the same component and are the key to understanding 10

the nature of each particular factor (Hair, 1998). The interrelationships are subject to the following theoretical discussion in order to address their meaningfulness. Table 2 Items’ Interrelationships

I1 I2 I3 I4 I5 I6 I7

Due to the captain’s demand for efficiency, we sometimes have to violate procedures Due to the company’s demand for efficiency, we sometimes have to violate procedures I feel that it is difficult to know which procedures are applicable The procedures are difficult to understand or are poorly written The procedures are helpful in my work I have received good training in the company’s procedures We have the opportunity to influence and form the procedures

Loadings Comp. Comp. 1 2 -0.027 0.845 0.815

-0.077

0.697 0.639 -0.158 -0.186 0.014

-0.093 -0.212 0.793 0.760 0.686

*Comp. = component

The extracted rotated component matrix presented in Table 2 indicates two underlying dimensions: component 1 and component 2. Component 1 reflects an inadequate procedural system and approach (I1 through I4). Component 2 indicates an adequate procedural system and approach (I5 through I7). The items interrelated in component 1 indicate that shipping companies that exert commercial pressure (I1 and I2) are also recognised by an overly complex procedural system (I3) along with procedures that are poorly formulated (I4). High factor loadings on all items included in component 1 denote that these features are likely to be present simultaneously in these organisations. In addition, features denoted in items belonging to component 2 (I5 through I7) are likely to be absent as they have low and negative loadings at component 1. The items interrelated in component 2, indicating companies where the procedural system is perceived to be useful in daily operations (I5), also relate to good training in these procedures (I6) and allow the crew to influence and form the procedures (I7). Companies with these features are also less likely to serve as pressure to ensure efficiency as these items (I1 and I2) have low and negative loadings at component 2. For the same reason, these companies are denoted by a procedural system perceived to be easier to relate to in daily operations. Overall, 297 respondents provided written comments on the questionnaire; 62 made written comments related to procedures and checklists. The comments were categorised into the following three sub–groupings: (1) procedure quality and usability (n=33), (2) violation of procedures due to commercial pressure (n=11), and (3) others (n=18). Comments were provided on a volunteer basis and, thus, not representative of the population. However, they provide valuable insights into characteristics perceived to be problematic. The results are summarised in Table 3. Table 3 Survey Comments Related to Procedures and Checklist

33 respondents

11 respondents 18 respondents

Procedures and checklists are not applicable and do not reflect the situation on board: too detailed, too many, and look like they have been developed by people with no sea-going experience. Procedures are being breached due to commercial pressure. Other, such as the relevance of training and work-specific situations.

11

Respondents were also asked to list their reasons for not following procedures. Each respondent could mark up to three of seven pre-specified options or write any other reason, if not listed. The results are listed in Table 4, with the number of respondents ticking each reason (N), percentage (N%) and valid percentage (N Valid %). When calculating valid percentages, missing cases are omitted. Table 4 Reasons for Violating Procedures Listed by Frequency

1 2 3 4 5 6 7 8

Reason for not following procedures The work will be done faster The procedures do not work as intended There are too many procedures I feel pressured because I am overloaded with work It improved the quality of my work I am not familiar with the applicable procedures The rest of the crew does it Others

N 538 488 394 365 312 227 203 35

N% 42.6 38.7 31.2 28.9 24.7 18.0 16.1 2.8

N Valid % 47.2 42.8 34.6 32.0 27.4 19.9 17.8 3.1

Both Table 3 and Table 4 indicate that procedures are violated and checklists are not followed due to high workloads and commercial pressures. This is supported by the other survey comments, which are presented in Table 5. Of the 297 respondents who provided written comments, 108 comments were related to workload and commercial pressure. The comments were categorised into the following four sub–groupings: (1) high demand for efficiency and time pressure (n=35), (2) low crewing level related to work (n=33), (3) violation of rest hours due to high workload (n=30), and (4) others (n=10). Survey comments related to workload and commercial pressure are presented in Table 5. Table 5 Survey Comments Related to Workload and Commercial Pressure

35 respondents 33 respondents 30 respondents 10 respondents

High demand for efficiency and time pressure, especially when arriving in and leaving port. The number of crewmembers is too low compared to work tasks, which are constantly increasing in quantity—especially administrative Rest hours are not followed, mostly due to low crewing level and high workload. Not possible to categorise, comments as “sorry, I am tired” and “I fell asleep at watch”.

The qualitative results from case studies and interviews indicate that the extensive use of procedures and checklists, including those provided by the shipping company itself as well as by third parties as charterers and customers. Other shipping-related research also identifies an increasing volume of regulations, controls, and administrative work as the main factor negatively affecting on-board safety (Knudsen, 2009). Based on Dreyfus and Dreyfus’ expert model, Knudsen (2009) also pointed out that rules and procedures are justified for inexperienced people. However, research also indicates that use of such standardised measures is more widespread within the tanker sector (Oltedal, 2010) than within bulk and dry cargo (Oltedal and Engen, 2010). These differences are assumed to originate partly from each sector’s relationship with their respective customers. Customers of dry cargo shipping demand fewer requirements with regard to safety management while liquid tankers are to a larger degree embedded in the oil industry, with a more mature safety management system. 12

However, it is suggested that this relationship between the liquid tanker industry and their customers will bring about changes based on external demands in contrast to internal needs for safety reasons. Although some sector differences are evident, the crew customarily deals with procedures and checklists from own their company, charterer, customers, and oil installation—of which all are different but at the same time standardised to fit all. Case studies also indicate that bridge officers regard procedures and checklists as valuable for safety reasons, but within certain limits. Procedures and checklists are also seen as problematic as there are too many of them and they are too detailed and too standardised. The crew experienced less standardisation and determined that the possibility to accommodate procedures and checklists in accordance with the ship-specific situation would improve safety more. Problems with completing checklists and following procedures mostly occur during hectic operations, such as visiting and leaving ports. 5 Discussion and Narratives The discussion and narratives are presented as they relate to Snook’s (2000) four phases of (1) design, (2) engineered, (3) applied, and (4) failed, which includes moving from the first phase (design) through engineered, applied, and into redesign. This section’s discussion supports the statistical results with interview narratives, illustrating the situation in depth. Although the statistical data reveal discrepancies between formal procedures and actual performance, the narratives illuminate why such discrepancies come into existence and how they are characterised. From the methodological point of view, they complement the survey and give us a better supplementary understanding of what is actually taking place in the sharp end of the maritime industry. 5.1 Design Although progress has been made on international cooperation, the late 1980s and early 1990s experienced a series of maritime disasters (Anderson, 2003; International Maritime Organization, 2010), such as the Herald of Free Enterprise in 1987, the Exxon Valdez in 1989, the Scandinavian Star in 1990, and the Estonia in 1994. International conventions alone did not seem to produce the intended levels of safety. Thus, international regulatory bodies began looking for ways to revise and improve safety regimes. Within the industry, SMS came to play an important role in achieving and maintaining high levels of safety and reducing losses resulting from accidents and incidents. Faith in the value of such systems became widespread, and they were made mandatory through the International Safety Management (ISM) Code (International Maritime Organization, 2005). The ISM code requires all companies to develop, implement, and maintain an SMS, which includes the functional requirements of a safety policy as well as instructions and procedures to ensure the safe operation of ships in compliance with relevant international conventions and flag state legislation. Within these requirements, each shipping company is free to find a functional solution best fitted to its own organisation and operations. However, previous research has shown that actual work and practice often do not reflect what is stated in the shipping companies’ overall safety management policy and functional requirements (Oltedal, 2010; Oltedal and Engen, 2010). Statistical results related to the design phase (presented in Table 1 and Table 2) indicate that the procedural system is perceived to be more helpful in daily operations when the crew has the opportunity to influence the procedures during the design phase. The following narrative (Narrative 1), given by a Norwegian captain, illustrated a (common) impression of a 13

safety management system designed in such a way that it is detached from daily operations and does not pay attention to its human interrelationship. Narrative 1: Norwegian captain about safety management system and procedures. “(...) my impression is that the aim is to belch forth as much paperwork as possible. And some of the things, procedures and checklists are not possible to relate to. They are useless crap developed by someone who has never set foot on a vessel. Just look at this (referring to pre-port arrival checklist) ‘Electronic positioning instruments checked–positions verified’. Of course we verify our position. We do that all the time when manoeuvring. We do not need a checklist to remind us to check where we are. And this ‘Pilot and Port Control have been given proper ETA (Estimated Time of Arrival) notice’—and what do they mean by proper ETA. You can be sure that the agent has been at the phone for the last days asking for ETA, and you can never be 100% sure of your arrival. Things happen, and we are not the only vessel entering the port. And all the other points are more or less the same (researcher’s comment: more than 20 items to be checked). But what these checklists really do is draw our attention away from what we are supposed to do: manoeuvre the vessel. We feel like we are being treated like children; don’t they think that we know how to do our job? We even have checklists for the checklists. We spend our days more or less filling in checklists, and for what purpose? To have someone to blame if anything goes wrong?” This story was shared with several other informants, who concurred that it depicted reality on board their vessels. Narrative 1, supported by survey comments presented in Table 3, also illustrates two points brought forward by Snook (2000). First, planners—not operators—must design a system in which they would never have to work. Second, planners are writing for future work situations and, as such, have limited information to draw upon. In this, it is important to keep in mind how procedures are (normally) dealt with within the framework of a traditional SMS as well as its relationship to risk assessment. An SMS consists of several sub-systems. First, a system of reporting and collecting experience data from the vessel itself is required. This is followed by a system of data processing—namely, the summarisation and analysis in order to reveal causal factors and perform trend analysis, which forms the basis for the development of safety measures. One critical system requirement is the reliability and accuracy of input data (i.e., near miss and accidents reports). As long as the input is reliable, the overall system presupposes the possibility of developing efficient measures in order to control operational safety (Kjellén, 2000). Given the arguments related to human rationality, Perrow (1999) has been reluctant to support the usability of risk analysis. In the absence of absolute rationality, during the SMS processes, some risks are minimised and other maximised while information is categorised and simplified in order to facilitate processing. Perrow (1999) suggested that availability heuristics are used when examining all existing cases of a phenomenon, then basing their judgment on all this experience as people tend to judge a situation in terms of the most readily available case—namely, the one most easily remembered. Other issues hampering the applicability of an SMS include planners’ lack of knowledge of the operating system, planners taking risk but not facing the consequences of their own decisions in running operational risk, a lack of education, and a lack of training in probabilities and statistics (Perrow, 1999). Thus, in these processes, crucial information may be lost. According to Perrow (1999), heuristics appear to work because in reality our world is loosely coupled and, thus, has a lot of slack and buffers in it that allow for approximations rather than complete accuracy. Another drawback in the applied SMS, further contributing to uncertainty, is the underreporting of experience data. Other research results indicate that approximately 35% of seafarers working on Norwegian controlled tanker vessels state that 14

they never or only sometimes report minor incidents. About 36% state that they sometimes or always alter the reports submitted in order to cover up mistakes (Oltedal, 2010). Within the dry cargo sector, approximately 40% of the respondents state that they never or only sometimes report minor incidents while about 36% indicate that they sometimes or always alter the reports submitted (Oltedal and Engen, 2010). Under such circumstances, where experience data are missing or incorrect, it is even more challenging to reveal underlying causes and influencing factors. Moreover, when developing measures for a future situation, there is uncertainty with regard to what that situation will display. The complexity of influencing factors and patterns of actions makes it difficult to categorise events into pre-fixed schemes, such as tools for root cause analysis and safe job analysis. In addition, new measures themselves may have unintended consequences. Moreover, the analysis of real and potential accidents often takes place within a blameoriented environment, with a tendency to point to causes of human error; this may hinder revealing other underlying causes (Oltedal, 2010). Such a situation is illustrated in Narrative 2. Narrative 2: Norwegian mate concerning a not-applicable procedure. “(…) in one situation, one of the able bodies cut himself with a knife. It was a small finger cut, and nothing serious. The incident was reported, as we are required to, whereupon we were instructed that it was no longer permitted to carry knives. In situations where knives were required, a safe job analysis and risk assessment were to be performed first. However, what they [people ashore working with SMS] do not understand is that not wearing a knife may involve greater risk. What if someone gets entangled in a hawser, with a risk of being dragged overboard, a situation where a knife would be quickly required to cut loose, what shall we do? Run to the bridge and carry out a risk assessment first? However, we bypassed this new regulation and did a general ‘safe job analysis’, which concluded that all able bodies could wear knives on a general basis.” When it comes to near misses and experience data, HRO argues that mindfulness and preoccupations with small failures may improve the reliability of the operational system. An HRO approach implies, inter alia, that the organisation is sensitive to operations and first-line experiences, encourages alternative frames of reference, and creates an error-friendly learning culture in which people seek feedback, share information, ask for help, and talk about error and experiment (Weick and Sutcliffe, 2007). However, if mindfulness is to be possible, it requires support from the top management and throughout the entire organisation. Mindfulness, as a proactive activity, is resource demanding. Mindfulness implies that every operator thinks differently about success and the possibility for failure along with the use of creativity in order to imagine what can go wrong and how. Mindfulness implies that every operator—if in any doubt at all—may stop operations and, if so, with support from the organisation and co-workers, even if the situation ultimately turned out to be safe every time. We do question if any organisations exposed to competition and constantly facing decisions of safety versus efficiency are capable of having all organisational members in a mindful mind mode for a longer period of time before efficiency comes to dominate safety. After all, no organisation exposed to competition exists with the primary goal of being safe; they need to become competitive in order to avoid bankruptcy. Both Narratives 1 and 2 depict situations in which measures have been developed detached from the operational system, with the consequence that the measures are experienced as difficult to relate to in operations. Narrative 2 also effectively depicts where first-line operators (seafarers) bypass the shore sides’ efforts in the design phase in adjusting to an applied mode. Planning has a symbolic as well as functional aspect (Clarke, 1999). 15

According to Clarke (1999), organisations and experts use plans as a form of rhetoric—tools designed to convince audiences that they ought to believe what the organisation says. In particular, some plans have so little instrumental utility in them that they warrant the label “fantasy document”. The usual presumption in social science is that the first step in an adequate planning process is to assess fairly completely what the problem is; the second step is to write a plan that addresses the problem, and the final step is to implement the plan. However, just like Perrow (1999), both Snook (2000) and Clarke (1999) are sceptical towards the utility of planning, especially in relation to future complex situations with a high degree of uncertainty. The uncertainty planners face when working with SMS may very well result in the development of overly burdensome rules or rules that are too detailed in their efforts to reduce uncertainty by controlling (most) human actions. Such a reality exists within shipping, as supported by the majority of survey comments presented in Table 3. Moreover, the results in Table 4 indicate that about 43% of the survey respondents breach procedures because they do not work as intended, about 35% because there are too many procedures, and about 27% because deviation from procedures will improve the quality of their work. Similar to Perrow (1999), Snook (2000) identified part of the problem as inherent in the rationale of the design phase in a SMS. More precisely, the assumption is that the operational system is tightly coupled, yet most of the time the organisational sub-units are a loosely coupled system. When a global rule is violated, operations usually go on as normal. On the contrary, breaking global rules may be perceived as being more efficient and even rewarded by the organisational management. Thus, a mismatch exists between the rule-based logic of action and the presumed tight situational coupling, which brings us into the next phase: “engineered”. 5.2 Engineered “Engineered” is defined by the interaction of the rule-based logic of actions and loose couplings. Although measures are originally designed for a tightly coupled reality, crews do experience a loosely coupled reality as nothing happens when the rules are broken. When such behaviours are not followed by some kind of reprimand, punishment, or any dangerous situation, their substantive risk-reducing reason may be questioned by operators (Snook 2000). Moreover, the presence of real-world constraints of actions may be overlooked by planners, including commercial pressures from the shore. The results in Table 4 show that the top reason for violating procedures is the need to work faster, which was ticked by about 47% of the respondents. Perceived pressure due to the overload of work is among the top four reasons for violating procedures, with approximately 32% of survey respondents concurring. Commercial pressure is also perceived to be present on board, as supported by survey comments in Table 5, case studies, and previous research carried out within both the liquid tanker sector (Oltedal, 2010) and the dry cargo sector (Oltedal and Engen, 2010). For example, it is well known among seafarers that hours of work are regulated primarily in theory only. Comments in Table 5 substantiate that rest hours are not followed due to high workloads, among other issues. Numerous anecdotal accounts were presented of not only some seafarers’ excessive hours, but also the manipulation of the hours for work/rest records in an attempt to conceal the truth and suggest compliance. One narrative depicting such a situation follows (Narrative 3). Narrative 3: Norwegian captain about rest-hour regulations “Rest-hour regulations? There is no such thing as rest-hour regulations! The regulations say that we are supposed to have 8 hours of rest during the day, but when calling in port after 16

passing through the English Channel, it is not unusual to be at the bridge for up to 18 to 20 hours. When you sail from Ushant to Rotterdam in about 40 hours! Nowadays with crew shortages, the bridge officers are getting promoted faster and are more inexperienced, and it may be difficult to leave them alone on the bridge with all the responsibility. As the master, I am in command of the vessel and responsible for safe navigation. Even though it is not explicitly stated in IMO regulations that I should be at the bridge at all times when passing through the channel, most companies have a standing order that the captain has to be on the bridge in congested waters, such as the Malacca Strait, Singapore Strait, and the English Channel. I know who they would blame if something happened—if the vessel grounded or something—and I was not present. In addition, I am expected to send reports to the company and charterers and so on and thereby have to leave the wheelhouse, contrary to standing orders. If it is for the immediate safety of the vessel, it is okay to depart from the regulations. And that is what we do. When we are short of manpower to do anything like maintenance, painting, or anything, I make the remark that it is for safety reasons, even if it is not, and then it is sometimes okay. Anyway, after passing through the channel, preparations for port arrival have to be done. When the pilot is boarding, I have to be present at all times. When calling in to the port, I may not have slept at all for 30 hours! The first people to meet me upon arrival is the port state control, flag state control, classification, QA control, and vetting inspectors from customers, charterers, or our own company. Sometimes they queue up and everyone expects to be first in line. Each vetting lasts for about 6-8 hours and requires a lot of the crew to be available—mostly officers; the most important people are kept awake for the longest periods. The first thing I am asked is if I have had 8 hours of rest. I have to say ‘yes’; if not, I am in trouble. The inspector knows that I am lying, and I know that the inspector knows that I am lying, but they do not care as long as they can tick of the right box as okay. Apart from this, I have to handle provisions, crew changes, ballasting, loading, and discharging in the shortest time possible. And the return keeps you awake for another few hours. The situation is impossible, and everybody knows that the regulations and ‘safety first’ slogans are all a charade, but nobody cares as long as the paperwork is okay and they have someone to blame if something should happen”. The demand to maximise profit may induce management to promote efficient and unsafe behaviour, which may result in a reduced error margin or the overstepping of the boundary for functionally acceptable performance. Management pressure for efficiency may also result in reduced crew level. The majority of the total operation expenses may be broken down into crewing, insurance, repairs and maintenance, stores and lubrications, and management and operations. Crewing costs typically amount to 30 to 65% of the total, depending on crew nationality, and are the easiest-to-reduce expenses in order to avoid failure or increase profit. Especially in light of the current economic recession, with the accompanying decline in the freight rates, the search for cost-cutting initiatives is a motivator that ensures profits and continued operations. Crew reductions have historically been the main instrument for cutting costs (e.g., the shipping crisis in the late 1970s and early 1980s) and still are a likely area for cost reductions (MacDonald, 2006). The appropriate crew level also depends upon other areas, such as operations, trading areas, and frequency of calling in ports. Too few crew members are more notable when calling in and leaving port—operations that have a higher demand for efficiency (Oltedal, 2010; Oltedal and Engen, 2009). When arriving, staying in, or leaving port, many things happen in a short period of time that are not specified in the principles for safe manning (International Maritime Organization [IMO], 1999), but affect the workload. This could be the need for piloting, discharging, loading, crew change, provisioning, bunkering, inspections, and the like. Fatigue is also considered to be influenced by the demand for efficiency and fast 17

turnover rate (Smith, 2007). Under such fluctuating work pressure, it is even less likely that a standardised global measure is appropriate for the actual situation. Our data also suggest that the industry is well aware of the situation. The following comments on rest-hour regulations were made at a conference for operators, ship management, charterers, and inspectors. “(…) I uphold that everybody is well aware that rest-hour regulations are not complied with. Even when doing the planning, we know it is impossible to comply with the regulations. But there is not that much we can do. If we plan for regulation compliance, we might lose contracts”. This comment was followed by a question: “But if we agree that the violation of rest-hour regulations is a near miss (researcher’s comment: everyone agreed). Should not such violations be reported as near misses? And if you did, what would happen?” The following answer was provided: “If we openly admit that the rest-hour regulations are violated frequently, we will not get the next contract”. Undoubtedly, these IMO conventions (e.g., rest-hour regulations) have improved safety in many areas. Nevertheless, companies and crew find ways to bypass the regulations if needed, and the control systems intercept only some of the concerning actors. However, what is striking in this situation is organisations’ awareness of the problem and that they—with full awareness—design plans that are not in accordance with reality. This disconnection between the real world of everyday practice and design is the driving force of the “practical drift” (Snook, 2000). Snook (2000) defines “practical drift” as a phenomenon resulting from the mismatch between the local demands of the situation and those of global design rules. As such, the pervasive demands of day-to-day practice inevitably shift the logic of action from one based primarily on formal rules to one driven more tightly by the task, loading schedules, port arrivals, and efficiency. Subsequently, over time, pragmatic practice loosens the grip of even the most rational and well-designed formal procedure, which brings us to the next phase: “applied”. 5.3 Applied When moving into the “applied” quadrant, rule-based logic of action yields to task-based logic. The net effect of this practical drift is a general loosening of globally defined rationality, in which task-based logic of action now matches the loosely coupled situation. In this phase, the crew is no longer operating according to globally designed rationality. Rather, it is replaced with the rationality of each local unit (e.g., a vessel, department, or team). However, although the workforce now operates according to multiple, incrementally emergent sets of procedures, each borne out of unique sub-unit logics grounded in the day-today pragmatics or loosely coupled words, the operational system still remains resilient as long as the system remains loosely coupled. As a result, the working environment may be very different within each local unit (e.g., a vessel, department, or team). Crew and team instability constitute additional challenges in this setting. As presented in section 3.3, survey demographics, most non-Norwegian crew members are contract employees hired through crewing agencies and do not work within a stable team. The demographics further show that 18

44.8% often or always sign in the same vessel each sailing period. However, about 45% of these are spread out over various vessels, and most vessels experience various degrees of instability within teams. Previous research also suggests that working within a wellestablished team does have a positive effect upon safety culture in general as well as interpersonal relationships (Snook, 2000; Oltedal and Wadsworth, 2010). Interpersonal relationships relate to, inter alia, the degree of trust and open communication amongst crewmembers. Reason (2001) regards a trusting relationship as a cornerstone for getting individuals to report their own mistakes and experiences, which is fundamental for the basic rationality of an SMS. For both subordinate and superior/managerial positions, additional challenges may arise in relation to the multinational crew and unstable crewing with low stability within teams. When signing on a new ship, new crew is unfamiliar with the ship’s on-board management style as well as fellow crewmembers and the on-board working climate. The seafarers do need some time to familiarise themselves with and adjust to the new situation. For instance, if the ship management at the seafarers’ previous vessel was blame oriented, the seafarer will most likely sign on the new vessel with the latest experience in mind, thereby being cautious about reporting his own mistakes for fear of being blamed or sanctioned. Over time, each seafarer learns how management is oriented at the particular vessel. In other words, seafarers become assimilated into the applied logic of their current unit. The problem is even more pronounced when the seafarer is constantly changing vessels, working with new management during each sailing period and experiencing this familiarisation process each time. The management style is known to vary within the sector. Poor shipboard management and leadership have also been identified in previous research (Oltedal and Engen, 2009; Knudsen, 2004). Narrative 4, from the comments on a survey question, illustrates one example of what happens when moving into the “applied” quadrant. Narrative 4: Filipino mate concerning rest-hour registration “(…) I once experienced being ordered to adjust my work and rest-period registration in a way so as to comply with the regulations. To be open with this or to react in disagreement may somehow jeopardise your next employment”. Interpersonal relationships amongst crew, in practice, also reflect the degree to which the crew shares safety-related information when changing shifts as well as more informal processes of sharing safety-related information during operations; as such, these relationships serve as a premise for the HRO principle of mindfulness. However, another issue not taken into consideration by HRO is relations of power. When entering a relationship with others, we constrain and are constrained by others as well as enabling and being enabled by others (Stacey, 2007). As Narrative 4 indicates, the power balance favours the individual giving the order. However, such power balances are also constantly changing, shifting in favour of some while going against others depending on the relative need they have for each other (Stacey, 2005). The Filipino mate who shared the narrative is in need of employment; as a contract worker, the power balance is against him. In this particular situation, regulations, restrictions, and control systems are of little use. The HRO principles of mindfulness are also of little use as the premises for mindfulness are not present. In other words, no premises exist for a trusting and open relationship. However, by changing the employment terms, the power balance could change, and the situation could become more favourable for compliance with external and internal requirements. Both NAT and HRO consider interaction to be a prime mover constituting networks or systems at different levels that further interact with each other as the system influences the 19

individual’s behaviour within the organisation and in safety-related issues. Where NAT emphasises system characteristics and power constellations as the main factors, HRO is preoccupied with the internal structure, coordination, and goal specification. However, according to Stacey (2005), these so-called forces (established premises) are taken as powerful and stable conditions, arising outside of those who relate to them in daily operations, in their own direct experience and influence. The individuals are supposed to comply with global rules; however, they may feel alienated when relating with them and thus make a local interpretation and adjustments of the standardised rules. Making a departure from Stacey’s (2005) theories, international and national conventions and legislations, such as work-hour regulations, should not be regarded as forces to which all must comply, but rather a result of complex responsive processes of interaction. The point of departure is how the various groupings of local interaction respond to the work-hour regulation and how the regulation is expressed differently in many local situations. From our own research, we find that the responses are expressed in numerous ways. Some have already been mentioned as false registrations of work hours (see Table 5), erroneous justification of a situation by an invented safety reason, ignorance, and so on. From the system perspective, deviations in actual behaviour are controlled by feedback through control systems, such as SMS. However, these are also responses that are adopted differently in each local interaction (i.e., a practical drift in various directions). Developing new conventions for vetting is the way IMO normally responds to risk. Each of these changes is happening in small groups of local interactions among (small) groups of individuals. Each group incorporates different responses, logic, interpretations, and situational sense-making. Hence, we cannot talk about different systems, entities, or levels that are detached from one another; rather, we should focus on changes that occur through continually responsive processes of interaction. The result is that each operational unit over time drifts further away from the globally established standards, and local responsive processes result in the development of new local rationality and rules of engagement. However, although the workforce operates in conflict with global rules, it is still assumed that people in other units behave in accordance with the original set of established rules. Problems then occur when either party miscodes the situation or fails to act according to the shared global rules or when the situation again becomes tightly coupled. 5.4 Practical Drift from Applied to Failed When operating in applied mode, each subgroup follows its own unique path of practical drift. Each uneventful day that passes reinforces a steadily growing false sense of confidence that everything is all right. However, in one rare stochastic fit, the system then becomes tightly coupled, and applied, as the “normal” praxis becomes the cause of the accident (Snook, 2000), as illustrated in the follows example. In December 2008, the Mirabelle, a Maltese flag Norwegian operated general cargo ship, grounded shortly after departure from port. The captain was alone at the bridge when grounding, which was identified as the main cause of grounding. As he was alone, he handled the manoeuvring, the navigation, the lookout, and the handling of the searchlight, thereby trying to do four jobs at once (Danish Maritime Authority, 2009). One month later, on January 16, the Mirabelle grounded once again, this time along the coast of Norway in the early morning. The navigator was alone at the bridge and had fallen asleep on watch (O'Cinneide, 2009). In this case, both accidents were in some way related to crewing level, resulting in the first situation in work stress and in the second situation in fatigue. With reference to the Mirabelle, an important factor is time. These factors with low crewing level had been present for some time, although not intercepted in the ISM/SMS system for remedies. The longer the 20

passage of time between design and failure, the less the possibility of detecting the “real” underlying causes, as crew may be undermanned and fatigued for a long time without any mishaps. Groundings, along with ship collisions, are known to be the most typical type of accidents at sea. Between 1996 and 2003, 652 groundings and collisions involving vessels over 500 gross tonnes were reported to the Marine Accident Investigation Branch (MAIB) under the United Kingdom’s Merchant Shipping Regulations alone (MAIB, 2004). The MAIB investigations showed that a small number of causal factors are common to nearly all bridge watch-keeping accidents and highlighted the following three principal areas of concern: (1) fatigue and groundings: a third of all groundings involved a fatigued officer alone on the bridge at night; (2) lookout and collision: two thirds of all vessels involved in collisions were not keeping proper lookout; and (3) safe manning and role of the master: a third of all the accidents that occurred at night involved a sole watch keeper on the bridge. Similarly, another study undertaken by the Swedish Maritime Administration (SMA) found that, in 84% of 32 reported collisions and groundings occurring between 1997 and 2002, the accident occurred between 2300 and 0800 hours. The study concluded that fatigue-related problems affecting bridge watch keepers are, with very great probability, occurring more frequently than initially believed (IMO sub-committee on standards of training and watch keeping). Although fatigue, rest-hour regulation, and safe manning are recognised as causal factors, accidents still happen due to these factors. An important mechanism to ensure compliance with IMO requirements is ship inspections. However, the flag states are not alone in conducting such inspections. Port states, classification societies, customers, charterers, cargo owners, operators, and others also conduct inspections. Each year, flag states are evaluated and ranked based on their effective enforcement of international rules, whereupon they are placed on a “black list”, “grey list”, or “white list” (MARISEC, 2008). Some flags do not manage or intend to uphold their responsibilities. The drawbacks with flag state control can in theory be balanced by port state control (PSC). However, the PSC also faces problems. In 2007, the European PSC, Paris Memorandum of Understanding (ParisMOU), performed 22,877 inspections on 14,182 individual ships registered with 113 flags, which resulted in 1,250 detentions (ParisMOU, 2010). The ParisMOU is an open register, and results from inspections, target areas and so on are public information as well as important parameters towards customers. The ParisMOU strategy is blaming and shaming. However, the latest review from Paris-based Equasis, which provides PSC data to the shipping industry, reveals that in the same period (2007) no record of inspection existed for some 17,910 ships. These ships avoid inspections and slip through the PSC net. The most common types of vessels slipping through are old general cargo ships, many of which have no classification record and are registered with flag states with a poor safety record (Corbett, 2009). These ships may very well be the ones most in need of an inspection. We also suspect that vessels that are inspected and found to be operating according to regulations are not necessarily doing this. Data from our own research indicate that some companies or vessels deliberately try to avoid the regulations (e.g., comments in Table 5). Vessels have work, overtime, and rest-hour registrations that do not reflect reality. Inspectors may on some occasions reveal the problem by comparing different registrations (i.e., if the registration indicates that the vessels are both bunkering and simultaneously have an engine crew that is off duty). However, we assume that the inspectors, for various reasons, do not manage to reveal all cases that are not in accordance with the regulations. Fatigue, lookout, and safe manning are highly interrelated with one another as well as with safety in general. IMO, in addition to the industry in general, is familiar with this problem. In 1993, a joint ILO/IMO group of experts on fatigue drafted a report on fatigue and its contribution to maritime casualty and accidents (International Maritime Organization). In 21

order to improve overall safety, such experience data are fed back into the original quadrant of “design”, resulting in the “redesign” phase. Based on our data, two different approaches towards redesign were identified: (1) organisations where global rules are redesigned detached from operational systems and (2) organisations where global rules are redesigned by frontline personnel. Research identifies shipping as an industry with an extensive use of procedures and checklists; when redesigning the SMS system, new procedures or checklists are likely to be added (Oltedal, 2010). However, data suggest that crews are more satisfied with operational SMS, procedures, and checklists when given the opportunity to influence them and when sharp end experiences are taken seriously. Snook (2000) also pointed out that, for organisational leaders, it is easy to get carried away on control trips with the misuse of procedures and checklists. When real situations reflect the fact that control is not efficient, one solution might be to control even more. Sometimes such well-intentioned efforts to prevent failure through tight control often produce just the opposite effect. The original problem of practical drift became a mismatch between situational couplings and rationale of logic, with rules being designed for a tightly coupled situation. As a result of this mismatch, we argue that a better solution is to acknowledge the world as loosely coupled, thereby minimising the use of global rules and being open for local adaption. 6 Conclusion A paradox is inherent in safety management by local control strategies. New safety measurements are often a result of deviations (errors, incidents, or accidents) resulting in some disturbance of the production process—namely, material damage or injury (Oltedal and Engen, 2009; Rasmussen, 1997). This implies that, in order to enhance safety, an unsafe deviation has to happen first. Thus, to be safe, you need to be unsafe. Moreover, resulting safety measures are then used as a standard for finding explanatory cause(s) towards future similar deviations. Reality is compared with the organisational safety standards and deviations attributed to human error. This is analogous to having a nautical chart that does not show the sandbank the ship has just grounded on and then blaming the seabed for not corresponding to the nautical chart. As previously mentioned, the shipping industry’s approach is often person orientated. Safety measures tend to aim at controlling human actions, often in the form of excess use of procedures and checklists. These measures are standardised to fit all; it could be a fleet of 5 vessels or 100 vessels. However, in actual work, no operation is ever the same. The vessels are different as well as the people, constellations of people, power figurations, weather, and so on. A standardised measurement will therefore never align with reality. Despite this, human actions and deviations are compared to the standard and found to be erroneous. In the search for control, new and even more detailed measurements may be developed. This is a vicious cycle created by the anxiety of not being in control. Snook (2000), Reason (2001), Hollnagel (2009), and others have all suggested that organisations should leave such person-oriented approach in their search for causal and influencing factors. Reason (2001) also warned about what he calls “anxiety-avoidance”, which fits the previous narrative. Anxiety-avoidance describes an organisation that has discovered a technique to reduce risk and repeats it over and over again regardless of its effectiveness, like constantly adding yet another procedure in response to unwanted incidents. Thus, the organisations working towards safety in this manner do not use the safety system theory as suggested, but abuse it due to the anxiety of not being in control or for other reasons. Reason (2001) and Stacey (2007) further mentioned anxiety related to not being in control as an important reason for overreliance on structure and systems. Moreover, Reason (2001) 22

emphasised taking local experience into account through a safety information system and reporting. During the research, we also encountered companies that let the crew have a major control over the development of the safety system, with the shore managementâ&#x20AC;&#x2122;s views being overridden by the sailors. In these situations, the sailors demonstrated a more positive attitude towards, and experience with, the procedural system. Thus, when the SMS is perceived as relevant, when applied to the premises for practical drift, it is attenuated. However, when it comes to actual behaviour, Reason (2001) recognised that both human attitude and behaviour are extremely difficult to change. This shifts the focus towards organisational structures and practices, is supposed to be more controllable, and should guide and arrange for a certain kind of behaviour through conditions at the local workplace. Furthermore, human error can never be eliminated, and the future is unpredictable. It might be provoking for some, but we suggest that organisations, managers, and planners accept that work and life imply risks and uncertainties; no matter how much we try, unwanted events of some kind will always happen. The real difference in safety management is how we approach such events. Staying with human error as the main explanatory factor may result in greater efforts in order to control all human actions. Alternatively, it is important to understand the underlying processes of practical drift and its interrelationship with mechanisms of control, situation, and human rationality. Yet however much we try, accidents like the Herald of Free Enterprise in 1987, the Exxon Valdez in 1989, the Scandinavian Star in 1990, and the Estonia in 1999â&#x20AC;&#x201D;and moreâ&#x20AC;&#x201D;will always happen. Snook (2000) suggested that any organisation, when working with safety management, should pay attention to three areas: (1) looking beyond individual error by framing puzzling behaviour in complex organisations as individuals struggling to make sense; (2) following the basic design principles of high performance teams and thinking twice about chasing the advantages of social redundancy; and (3) treating organisational states of integration and reliability with chronic suspicion. The important thing is to recognise them for what they are: constant outcomes of dynamic systems and ongoing accomplishments that require active preventive maintenance. This article has also underlined how the individual character of single units must keep a certain degree of flexibility while simultaneously being guided by superior regulations. If a complex and heterogeneous sector such as international shipping is overloaded with standardised regulations, local actions will be hampered in solving everyday incidents. The consequence will be a lid on local flexibility, which may increase risk rather than reduce it. The challenge is to strike a balance between a centrally developed design that is accepted and followed by the local actors and an environment in which these regulations can open up for pragmatic solutions, when required. Such a regulatory framework requires intelligent designers and continuous revision when confronted with an endless stream of real incidents. An effective feedback loop of information between the local actors and the central designers may thus create a pathway between central design and local pragmatism that can prevent the system from moving from practical drift into failure and catastrophe. References Adams, J., 1995. Risk, UCL Press, London. Anderson, P., 2003. Cracking the code: The relevance of the ISM code and its impact on shipping practices, Nautical Institute, London. Bertalanffy, L.v., 1971. General system theory: Foundations, development, applications, Allen Lane, London. 23

Clarke, L.B., 1999. Mission improbable: using fantasy documents to tame disaster, University of Chicago Press, Chicago. Cooper Ph.D., M.D., 2000. Towards a model of safety culture. Safety Science 36 (2), 111136. Corbett, A., 9 January 2009. Thousands of ships slip through inspection net. Tradewinds. Danish Maritime Authority, 2009. The grounding of MIRABELLE on 16 Desember 2008, The division for Investigation of Maritime Accidents, Denmark. Department of Transport, 1987, Report of Court No. 8074. mv Herald of Free Enterprise. Formal investigation. DiMaggio, P., 1994. Culture and Economy. In: Smelser, N.J., Swedberg, R. (Eds.), The Handbook of economic sociology. Princeton University Press, Princeton, pp. 835. European Maritime Safety Agency (EMSA), 2010. Maritime Accident Review 2009, Hair, J.F., 1998. Multivariate data analysis, Prentice Hall, Upper Saddle River, N.J. Heinrich, H.W., Roos, N., Petersen, D., 1980. Industrial accident prevention: a safety management approach, New York. Hollnagel, E., 2009. The ETTO Principle: Efficiency-Thoroughness Trade-Off Why Things That Go Right Sometimes Go Wrong, Ashgate. International Maritime Organization (IMO), 2006. IMO Sub-Committee on Standards of Training and Watchkeeoing (STW) 38/13/7, International Maritime Organization, London. International Maritime Organization (IMO), 2005. Study on the impact of the ISM Code and its effectiveness in the enhancement of safety of life at sea and. protection of the marine environment, International Maritime Organization, London. International Maritime Organization (IMO), MSC/Circ.621, International Maritime Organization, London. International Maritime Organization (IMO), 1999. Resolution A.890(21): Principles of Safe Manning, International Maritime Organization, London.. Justis- og politidepartementet, 1991. Scandinavian Star - ulykken, 7 april 1990, Norway, Oslo. KjellĂŠn, U., 2000. Prevention of accidents through experience feedback, Taylor & Francis, London. Knudsen, F., 2004. If you are a good leader I am a good follower. Research Unit of Maritime Medicine, Denmark. Kuo, C., Nautical Institute, 2007. Safety management and its maritime application 2, 288. 24

MacDonald, R., 2006. Safe Manning of Ships - Yesterday, Today and Tomorrow, WMU Journal of Maritime Affairs 5 (2), 143-151. Marine Accident Investigation Branch (MAIB), 2004. Bridge Watchkeeping Safety Study, United Kingdom. Maritime International Secretariat Services Ltd (MARISEC), 2008. Shipping Industry Flag State Performance Table, United Kingdom, London. National Transportation Safety Board, 1990. Grounding of U.S. Tankship EXXON VALDEZ on Bligh Reef, Prince William Sound Near Valdez, AK March 24, 1989. O'Cinneide, E., 16. January 2009. Death ship drama, Tradewinds. Oltedal, H.A., Engen, O.A., 2010. Tanker versus dry cargo - The use of safety management systems within Norwegian dry cargo shipping. In: Ale, Papazoglou, Zio (Eds.), Reliability, Risk and Safety, Taylor & Francis Group, London, pp. 2118-2125. Oltedal, Helle, Wadsworth, Emma, 2010. Risk perception in the Norwegian shipping industry and identification of influencing factors. Maritime Policy & Management 37 (6), 601-623. Oltedal, H.A., Engen, O.A., 2009. Local management and its impact on safety culture and safety within Norwegian Shipping. In: Artorel, M., Guedes, C., Barnett, J. (Eds.), Safety, Reliability and Risk Analysis. Theory, Methods and Applications, Taylor & Francis Group, London, pp. 1423-1430. Oltedal, H., 2010. The use of safety management systems within the Norwegian tanker industry - do they really improve safety? In: Bris, Guedes Soares, Martorell (Eds.), Reliability, Risk, and Safety: Theory and Applications, Taylor & Francis Group, London. Perrow, C., 2004. A Personal Note on Normal Accidents, Organization & Environment. 17 (1), 9-14. Perrow, C., 1999. Normal accidents: living with high-risk technologies, Princeton University Press, Princeton, N.J. Pett, M.A., Lackey, N.R., Sullivan, J.J., 2003. Making sense of factor analysis: The use of factor analysis for instrument development in health care research, Sage, Thousand Oaks, Calif. Rasmussen, J., 1997. Risk management in a dynamic society: A modelling problem. Safety Science 27(2-3), 183-213. Reason, J., 2001. Managing the risks of organizational accidents, Ashgate, Aldershot. Rothblum, A.R., 2000. Human Error and Marine Safety. Scott, W.R., Davis, G.F., 2007. Organizations and organizing: Rational, natural, and open system perspectives, Pearson Education, Upper Saddle River, N.J.

25

Smith, A., 2007. Adequate Crewing and Seafarers' Fatigue: The International Perspective, Centre for Occupational and Health Psychology, Cardiff University. Snook, S.A., 2000. Friendly fire: the accidental shootdown of U.S. Black Hawks over Northern Iraq, Princeton University Press, Princeton, N.J. Soma, T., 2008. Are the accidental losses increasing in shipping - what can be done?, DNV Maritime Solutions. Stacey, R., 2005. Local and global processes in organizational life. In: Stacey, R.D. (Eds.), Experiencing emergence in organizations: Local interaction and the emergence of global pattern, Routledge, London, pp. 17-47. Stacey, R.D., 2007. Strategic management and organisational dynamics: The challenge of complexity to ways of thinking about organisations. FT Prentice Hall, Harlow. Standard P&I Club 2010. 16 September-last update, Standard Safety, September issue [Homepage of Standard P&I Club], [Online]. Available: http://www.standardclub.com/docs/14468_STANDARD_SEPTEMBER_LR-2.pdf . Studio Apertura, 2004. Sikkerhetskulturer i transportsektoren. Metoder for kartlegging av sikkerhetskultur: Evaluering av noen eksisterende verktøy, Norway. Weick, K.E., 1976. Educational Organizations as Loosely Coupled Systems, Administrative Science Quarterly 21 (1), 1-19. Weick, K.E., Sutcliffe, K.M., 2007. Managing the unexpected. Jossey-Bass, San Francisco, Calif. www.ParisMOU.org, 2010. 21. July-last update [Homepage of Paris MOU], [Online]. Available: www.ParisMOU.org.

26

Appendix 1: Survey questionnaire and letter of introduction

Dear crew member

In this survey I want to assess how safety is handled on Norwegian-controlled bulk cargo and tanker vessels and how the crew onboard perceive safety. This is part of my PhD work, which is aimed at improving our understanding of how various safety-related circumstances work in practice. The results of the survey will help your shipping company make safety-related decisions, enabling your safety to be better ensured. All information that is obtained through the survey is anonymous. It will not be possible to trace any answers to individuals, shipping companies or vessels. My research has been financed by Stord/Haugesund University College and I am therefore acting independently of all shipping companies, public authorities and other organisations and interests. As a PhD student and researcher, my work will be carried out in line with Norwegian guidelines for research ethics, which among other things protect your right to be anonymous. Only I will handle the completed forms or have other access to the data. To optimise the quality of the survey, it is important that as many people as possible complete the form. It will take about 30 minutes to complete the form. It is also important that the questions are answered as frankly as possible. When you have completed the form, place it in the enclosed envelope, seal the envelope and deliver it directly to the vesselâ&#x20AC;&#x2122;s safety delegate or the chosen contact person onboard for this survey. -

As regards the answering of the questions on the form itself, please relate your answers to the circumstances onboard this particular vessel. As regards comments concerning questions and suggestions for improvements, you can relate these to the experience you have of the sector in general.

This type of survey is very common in Norway. Both Norwegian companies and authorities want the safety of employees to be given the highest priority. From similar surveys we have learned that safetyrelated matters often do not function as intended, and that the reason can be traced back to weaknesses linked to the company's management or other organisational factors. Some of the questions may be difficult to answer, but the aim is not to place the blame on individuals. We know for example that procedures can be broken without the blame resting on the individual who breaks the procedure but on other levels within an organisation. In such cases I want to identify the organisational reasons for breakdowns in procedure. Possible reasons here include insufficient involvement in the development of procedures, the existence of too many procedures or the adoption of dangerous procedures. Participation in and completion of the questionnaire is voluntary. If you choose not to take part in the survey, please fill in the enclosed green sheet and return it in the same way as the questionnaire. If you have any questions concerning the survey, you can contact me, Helle Oltedal, on telephone no. (+47) 93 82 61 87 or (+47) 52 70 26 44. Thank you for taking the time to fill in the questionnaire!

Yours sincerely, Helle Oltedal, Stord/Haugesund University College

Safety Culture Survey

Please indicate your answer by crossing off a box for each question like this: _. Mark wrong answers like this: J Where it is not possible to select an answer, please provide an alternative in the space provided for other or fill inn. PLEASE WRITE ALL COMMENTS OR SUGGESTIONS IN ENGLISH

1. Background Information A Gender:

Female

B Job position/title

Captain Electrician

C Vessel class/cargo

Bulk Chemical

D Age

1

Male

1

Mate

5

Catering

2

2

Engineer

6

Apprentice

2

Shuttle tanker

3

AB /Seaman

7

1

Combined

6

Oiltanker

31-40 years

2

41-50 years

3

Under 31 years

E Nationality

3

4

Other (Fill in):_________________ Gas 4 General Cargo

5

Other (Fill in)_______________

7

1

Norwegian

1

Polish

2

Filipino

3

0-2 year(s)

1

3-5 years

2

6-10 years

3

51-60 years 4

Over 60 years 5

Other (Fill in):______________

F How long have you been working within shipping?

G H

11-20 years 4

Over 20 years 5

How long have you been working for this shipping company?Fill in:____________

All in all, how long have you been working at this vessel?

Fill in:____________

I Which kind of employment contract do you have?

Permanent employee

1

2

9 months duration

3

6 months duration

4

3 months duration Other (Fill in):

J

How long is your ordinary work /sailing schedule at the ship? (for contract (Fill in)__________________ workers this will often be the same as the contract duration)

K How long do you usually stay at home or on shore between each work /

(Fill in)__________________

sailing period?

L Do you normally work at the same vessel at every work / sailing period? Very seldom /never Sometime Very often / always

1

2

3

MHow is your ordinary watch system, without overtime?(for example Dayman, or 4-8-4 meaning 4 hours work – 8 hours off – 4 hours work)

1

(Fill in)__________________

English version 65

2. Top Management’s Safety Priorities Only select one answer per question

A The on-shore top management in my company prioritizes safety before economy.

B I experience conflicting requirements from my company and the captain.

C I experience that safety work is more a facade than a real priority area.

D I am familiar with the company’s safety goal.

Strongly disagree

Disagree Not sure

Agree

Strongly agree

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Comments and suggestions:

3. Local Management Please state your evaluation of your closest superior’s attitude toward safety. If you are the captain, relate the questions to the closest on-shore manager. A All in all, for how long have you been working with your closest superior?(Fill inn) ___________________________ Only select one answer per question

B Is your closest superior clear in his engagement to ensuring his co-workers’ safety?

C Does your closest superior follow up to ensure that all work on board is done in a safe manner?

D Is your closest superior a good role model when it comes to attending to his own and others’ safety?

E How often do you participate in meetings with your closest superior where safety is a topic?

Very seldom Seldom Sometime /never

willing to discuss safety-related conditions.

G My closest superior is not afraid of admitting his own mistakes.

H My closest superior has too little confidence in his co-workers.

I

My closest superior is supportive if safety is prioritized in all situations

Very often /always

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Once a week

Twice a month

Once a month

Once every 2nd month

Once every 6 moths

1

Strongly disagree

F My closest superior appreciates that the crew is

Often

2

3

Disagree Not sure

4

5

Agree

Strongly agree

More seldom

6

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Comments and suggestions:

2

English version 65

4 Procedures and Guidelines Only select one answer per question

A Due to the company’s demand for efficiency we sometimes have to violate procedures.

B Due to the captain’s demand for efficiency we sometimes have to violate procedures.

C I have received good training in the company’s procedures.

D I feel that it is difficult to know which procedures are applicable.

E The procedures are helpful in my work. F The procedures are difficult to understand or are poorly written.

G We have the opportunity to influence and form the procedures.

Strongly disagree

Disagree Not sure

Agree

Strongly agree

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

If you don’t follow the procedures in a specific situation, what may be the reason? Please don’t mark more than three options. A The work will be done faster. 1

B The rest of the crew does it. C I feel pressured because I am overloaded with work. D It improves the quality of my work. E I am not familiar with the applicable procedures. F The procedures do not work as intended. G There are too many procedures.

2

3

4

5

6

7

H Others (please specify): Comments and suggestions:

5. Interaction In relation to following questions safety will be any issue or condition that you feel may threaten or cause any injury or damage to yourself, your co-workers or the vessel. Only select one answer per question

A Do you normally work with the same team members

Very seldom Seldom Sometime /never

Often

Very often /always

Don’t know

1

2

3

4

5

6

B Do you discuss safety issues with your co-workers?

1

2

3

4

5

6

C Do you ever feel forced to continue your work even

1

2

3

4

5

6

within your working area / working group?

if safety may be threatened?

3

English version 65

Only select one answer per question

D Does the crew get positive feedback when they raise safety issues?

E Can you tell the captain to ”stop”/”time out” if you feel that safety is threatened?

F Can you say ”stop”/”time out” to the company if you feel that safety is threatened?

Very seldom Seldom Sometime /never

Very often /always

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Strongly disagree

G The working environment on board is characterized

Often

Disagree Not sure

Agree

Strongly agree

Don’t know

1

2

3

4

5

6

H We solve problems and conflicts in a good manner.

1

2

3

4

5

6

I We receive sufficient safety-related information

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

M If I ask for help I will appear incompetent.

1

2

3

4

5

6

N We usually speak up to a co-worker if we notice that

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

by openness and dialog.

when we start a new watch.

J We receive sufficient safety-related information when we sign on / start a new sailing period.

K I am confident that my company always prioritizes the crew’s safety.

L I am confident that the captain always prioritizes the crew’s safety.

he is doing his work in a risky manner.

O We usually speak up to the ship management if we notice that a co-worker is doing his work in a risky manner.

P I stop work if I am not sure that safety is satisfactorily ensured.

Q I feel appreciated by my co-workers. R I feel appreciated by the ship management. S My co-workers do their jobs in a way that makes me feel safe.

T My co-workers can communicate effectively in English.

U Different languages on board may represent a safety risk.

V Different national cultures on board may represent a safety risk.

1

1

2

2

3

3

4

4

5

5

6

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Comments and suggestions:

4

English version 65

6. Work Situation Only select one answer per question

A Do you have the possibility to prioritize safety first in your daily work?

B Do you carry out a “Safe Job Analysis”/“Risk Analysis” before high-risk operations?

C Do you carry out a safety evaluation before new working methods, tools, or routines are introduced?

D Have you experienced situations where you need to expose your self to danger to get the work done?

E Do you take a ”time-out” when unforeseen situations occur?

F Do you use protective equipment in situations when it is mandatory?

G Do you feel sufficiently rested to carry out your tasks in a safe manner on your shift?

H Is the safety documentation you need readily available?

I Is the safety documentation you need up to date? J Is anyone ever intoxicated/drunk on board?

Very seldom Seldom Sometime /never

Very often /always

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

1

Strongly disagree

K The number of crewmembers is not sufficient to

Often

2

2

3

3

Disagree Not sure

4

4

5

5

Agree

Strongly agree

6

6

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

MI miss feedback on the work I do.

1

2

3

4

5

6

N The on-board maintenance is sufficient to ensure

1

2

3

4

5

6

ensure safety on board.

L The management doesn’t care how we do our work as long as the work gets done.

safety.

O I am familiar with the on-board safety goals. P I have to work much overtime to get the work done

1

1

2

2

3

3

4

4

5

5

6

6

Comments and suggestions:

5

English version 65

7. Competence Only select one answer per question

A I have received the training that is necessary in order to work safely.

B I have received the education that is necessary in order to work safely.

C I have received the training that is necessary in order to handle critical or hazardous situations.

D I have received the education that is necessary in order to handle critical or hazardous situations.

E New crew members get a thorough introduction to safety-related issues.

F On our vessel we frequently carry out drills in safety procedures.

G What we learn in courses is not relevant in practice. H Some of my co-workers lack experience.

Strongly disagree

Disagree Not sure

Agree

Strongly agree

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

1

2

2

3

3

4

4

5

5

6

6

Comments and suggestions:

8. Responsibility & Sanctions Only select one answer per question

A In my day-to-day work there is no doubt about who is responsible for the different tasks.

B When an undesirable incident has occurred, people are more preoccupied with placing blame than finding the cause of the incident.

C If I violate the safety regulations, there will be negative consequences for me.

D Vague responsibilities on board contribute toward creating hazardous situations.

E In my opinion, the consequences for violating the company’s safety regulations are fair.

F Responsibility for the safety of others is a motivational factor in the performance of my work.

G I know which tasks I am responsible for if a critical or hazardous situation should occur.

Strongly disagree

Disagree Not sure

Agree

Strongly agree

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Comments and suggestions:

6

English version 65

9. Working Environment Only select one answer per question

A I enjoy my job. B I feel sure that I will not loose my job. C I feel that the work we do on board is too little appreciated by the company. D This company is a good employer compared to others. E I have too little influence on my working situation.

F The working situation is less physically challenging than 2 years ago. G The working situation is less mentally challenging than 2 years ago. H The safety delegates have an important role in ensuring safety at my work site. I I feel certain that I will not be exposed to an injury/ accident at my work site.

Strongly disagree

1 1

1

1

1

1

Disagree Not sure

2 2

2

2

2

2

3 3

3

3

3

3

Agree

Strongly agree

4 4

4

4

4

4

5 5

5

5

5

5

Don’t know

6 6

6

6

6

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Comments and suggestions:

10. Learning from Incidents If accidents or severe incidents happen on board, I believe they happen because…. Do not select more than 3 alternatives. A The crew has a large work load. B The crew does not feel enough responsibility for their tasks. C The crew lacks knowledge and experience in relation to the job they are doing. D There is no tradition for speaking up when someone is working in a hazardous manner. E There are too many interruptions in the work. F Procedures/best practice is not followed. G There are inadequate instructions for using technical equipment. H There are mistakes or deficiencies in the procedures. I There is bad maintenance. J There is defective equipment.

1

2

3

4

5

6

7

8

9

10

K Others (please specify): Comments and suggestions:

7

English version 65

Only select one answer per question

A During the last 2 years, have you been involved in a serious incident/accident?

B During the last 2 years, have any of your co-workers been involved in a serious incident/accident?

Yes

No

Yes

No I don’t know

1

2

1

2

3

If yes, please comment the last incident / accident that happened:

C During the last 2 years, have you been involved in what was almost a serious incident/accident?

D During the last 2 years, have any of your co-workers been involved in what was almost a serious incident/accident?

Yes

No

Yes

No I don’t know

1

2

1

2

3

If yes, please comment the last episode that happened:

Only select one answer per question

E Do minor incidents get reported in writing? F Do close calls get reported in writing? G Are reports of undesirable incidents ever “fixed up” to cover mistakes?

H Do you receive constructive feedback from the company on the conditions you report?

I Do you receive constructive feedback from the captain on the conditions you report?

J Do you get information from incidents/accidents on other vessels?

Very seldom Seldom Sometime /never

1

1

something has gone wrong.

L Reporting is important to prevent the recurrence of accidents or incidents.

M Most of all, I report incidents because I have to. N In my company, they are more preoccupied with the statistics than the human consequences of an incident.

O Reporting in itself take too much time.

2

2

3

3

Very often /always

4

4

5

5

Don’t know

6

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Strongly disagree

K Here it is seldom improvements are made before

Often

Disagree Not sure

Agree

Strongly agree

Don’t know

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

1

2

3

4

5

6

Comments and suggestions:

8

English version 65

Assume that you were involved in an incident. Would some of the following conditions stop you from reporting the incident? Do not mark more than 3 alternatives. A There is no tradition for reporting all incidents that happen. 1

B No improvements ever happen based on the reports. C The incident didn’t have any serious consequences. D I am afraid that the information will be used against me. E I am afraid that the information will be used again my co-

workers.

F This could cause the company to loose contracts. I There could be negative reactions from my co-workers.

J I don’t feel comfortable discussing my actions/mistakes. K We have too much to do and don’t have time to write reports. L Mistakes I make don’t concern anyone but me. M I don’t know how to report an incident.

2

3

4

5

6

7

8

9

10

N Other (please specify):

11

Comments and suggestions:

11. Description of the Organization How would you describe this organization? Please provide your estimate based on the statements below. Select only one box per statement.

The work is characterized by control in detail and overall control.

It is important to do what we are told.

The work is usually performed on an individual basis.

1

2

3

1

1

4

2

2

5

3

3

6

4

4

5

5

9

7

8

6

7

6

7

10

The work is characterized by flexibility and democracy/influence.

9

10

It is important to be creative and original.

9

10

9

8

8

The work is usually performed as a team.

English version 65

12. Risk perception How would you assess the risk involved with your work? Please provide your estimate based on the statements below. Select only one box per statement.. A) All in all, how would you assess the safety in your working situation? Very bad

1

2

3

4

5

6

7

8

9

10

Very good

B) All in all, how do feel that the level of safety has developed over the last two years? It is much poorer

1

2

3

4

5

6

7

8

9

10

It is much better

C) All in all, how likely is it that you will have an accident on the vessel during the next 12 months? Very likely

1

2

3

4

5

6

7

8

9

10

Very unlikely

D) All in all, how likely is it that any of the other crew members will have an accident on the vessel during the next 12 months? Very likely

1

2

3

4

5

6

7

8

9

10

Very unlikely

E) How safe do you feel when you consider the risk involved with your work on board? Very safe

1

2

3

4

5

6

7

8

9

10

Very unsafe

F) How much do you worry when you consider the risk involved with your work on board? Very much

1

2

3

4

5

10

6

7

8

9

10

Very little

English version 65

13. Suggestions for New Safety Actions or Other Comments

11

English version 65