Psi Chi Journal Volume 31.1 | Spring 2026

ISSN: 2325-7342

Published by Psi Chi, The International Honor Society in Psychology ® SPRING 2026 | VOLUME 31 | ISSUE 1

PSI CHI JOURNAL OF PSYCHOLOGICAL RESEARCH SPRING 2026 | VOLUME 31, NUMBER 1

EDITOR

ROBERT R. WRIGHT, PHD

Brigham Young University-Idaho

Email: wrightro@byui.edu

ASSOCIATE EDITORS

TIFANI FLETCHER, PHD West Liberty University

STELLA LOPEZ, PhD University of Texas at San Antonio

TAMMY LOWERY ZACCHILLI, PhD Saint Leo University

ALBEE MENDOZA, PhD

Delaware State University

JULEE POOLE, PhD Purdue University Global

KIMBERLI R. H. TREADWELL, PhD University of Connecticut

EDITOR EMERITUS

DEBI BRANNAN, PhD Western Oregon University

MANAGING EDITOR BRADLEY CANNON

DESIGNER

JANET REISS

EDITORIAL ASSISTANT EMMA SULLIVAN

ADVISORY EDITORIAL BOARD

GLENA ANDREWS, PhD RAF Lakenheath USAF Medical Center

AZENETT A. GARZA CABALLERO, PhD Weber State University

MARTIN DOWNING, PhD Lehman College

HEATHER HAAS, PhD University of Montana Western

ALLEN H. KENISTON, PhD University of Wisconsin–Eau Claire

MARIANNE E. LLOYD, PhD Seton Hall University

DONELLE C. POSEY, PhD Washington State University

LISA ROSEN, PhD Texas Women's University

CHRISTINA SINISI, PhD Charleston Southern University

ABOUT PSI CHI

Psi Chi is the International Honor Society in Psychology, founded in 1929. Its mission: "recognizing and promoting excellence in the science and application of psychology." Membership is open to undergraduates, graduate students, faculty, and alumni making the study of psychology one of their major interests and who meet Psi Chi’s minimum qualifications. Psi Chi is a member of the Association of College Honor Societies (ACHS), and is an affiliate of the American Psychological Association (APA) and the Association for Psychological Science (APS). Psi Chi’s sister honor society is Psi Beta, the national honor society in psychology for community and junior colleges.

Psi Chi functions as a federation of chapters located at senior colleges and universities around the world. The Psi Chi Headquarters is located in Chattanooga, Tennessee. A Board of Directors, composed of psychology faculty who are Psi Chi members and who are elected by the chapters, guides the affairs of the Organization and sets policy with the approval of the chapters.

Psi Chi membership provides two major opportunities. The first of these is academic recognition to all inductees by the mere fact of membership. The second is the opportunity of each of the Society’s local chapters to nourish and stimulate the professional growth of all members through fellowship and activities designed to augment and enhance the regular curriculum. In addition, the Organization provides programs to help achieve these goals including conventions, research awards and grants competitions, and publication opportunities.

JOURNAL PURPOSE STATEMENT

The twofold purpose of the Psi Chi Journal of Psychological Research is to foster and reward the scholarly efforts of Psi Chi members, whether students or faculty, as well as to provide them with a valuable learning experience. The articles published in the Journal represent the work of undergraduates, graduate students, and faculty; the Journal is dedicated to increasing its scope and relevance by accepting and involving diverse people of varied racial, ethnic, gender identity, sexual orientation, religious, and social class backgrounds, among many others. To further support authors and enhance Journal visibility, articles are now available in PsycInfo, EBSCO, Clarivate Emerging Sources Citation Index, Crossref, and Google Scholar databases. In 2016, the Journal also became open access (i.e., free online to all readers and authors) to broaden the dissemination of research across the psychological science community.

JOURNAL INFORMATION

The Psi Chi Journal of Psychological Research (ISSN 2325­7342) is published quarterly in one volume per year by Psi Chi, Inc., The International Honor Society in Psychology. For more information, contact Psi Chi Headquarters, Publication and Subscriptions, 651 East 4th Street, Suite 600, Chattanooga, TN 37403, (423) 756­2044. https://www.psichi.org; psichijournal@psichi.org

Statements of fact or opinion are the responsibility of the authors alone and do not imply an opinion on the part of the officers or members of Psi Chi.

Advertisements that appear in Psi Chi Journal do not represent endorsement by Psi Chi of the advertiser or the product. Psi Chi neither endorses nor is responsible for the content of third­party promotions. Learn about advertising with Psi Chi at https://www.psichi.org/Advertise

COPYRIGHT

Permission must be obtained from Psi Chi to reprint or adapt a table or figure; to reprint quotations exceeding the limits of fair use from one source, and/or to reprint any portion of poetry, prose, or song lyrics. All persons wishing to utilize any of the above materials must write to the publisher to request nonexclusive world rights in all languages to use copyrighted material in the present article and in future print and nonprint editions. All persons wishing to utilize any of the above materials are responsible for obtaining proper permission from copyright owners and are liable for any and all licensing fees required. All persons wishing to utilize any of the above materials must include copies of all permissions and credit lines with the article submission.

Theatre as an Intervention Against Autism Stigma Using Broadway’s How to Dance in Ohio

Melissa S. Garber and Katherine L. Fiori*

Gordon F. Derner School of Psychology, Adelphi University

Resilience and Drinking Behaviors in Emerging Adults With Previous Extracurricular Participation

Ava Avolio

Department of Psychology, Counseling, and Criminology, Carlow University 22 A Comparison of AI and Human-Driven Assistance on Help-Seeking intention: The Mediating Role of Perceived Emotional Support

Lee Yan Ying Department of Psychology, HELP University 36 Hourly Negative Interpersonal Interactions and Momentary Ambulatory Blood Pressure Among African American Emerging Adults

Emilie J. Chai1, Eunji Shin2, Nataria T. Joseph*2, and Laurel M. Peterson*3

1Psychology Department, University of California, San Diego

2Social Science Division, Pepperdine University

3Psychology Department, Bryn Mawr College 47 The Effects of Correcting Misperceived Norms on Sexual Assault Prevention Intentions and Rape Myth Acceptance

Cristal Lopez, Alyssa Martinez, and Heike I. M. Mahler*

Psychology Department, California State University San Marcos 59 Out of Reach, Out of Mind? Mechanisms of Psychological Distance and Emotional Reactions to Tragedy Abroad

Syed M. Wahid1, Christopher T. Dawes2, and Alysson E. Light1

1 Department of Psychology, University of Chicago

2 Wilf Family Department of Politics, New York University 72 The Power of Perspective: How Framing Affects Undergraduate Student Attitudes Toward Graduate School

Hannah V. Hyman, Iris R. Wright, and Ralph G. Hale*

Department of Psychological Science, University of North Georgia

Reviewer Appreciation

Correction to Abrams et al. (2025)

Dedication to Steven V. Rouse, PhD

We dedicate this issue to the memory of Dr. Steven V. Rouse of Pepperdine University, who served as Editor (2021–2025) and Associate Editor (2014–2021) of Psi Chi Journal of Psychological Research

Throughout his tenure, Steve was a steadfast champion of transparency, inclusivity, and methodological rigor. With wisdom, patience, and genuine care, he profoundly shaped the experiences of countless student and faculty authors, as well as the members of our reviewer team and editorial board.

Recently honored with Psi Chi’s highest honor, the Distinguished Member Award, Steve leaves a legacy of integrity, mentorship, and scholarly excellence. His influence will continue to guide and inspire our community for years to come. We are deeply grateful for his extraordinary service and mourn the loss of a remarkable colleague, mentor, and friend.

—The Psi Chi Journal Editorial Board

Theatre as an Intervention Against Autism Stigma Using Broadway’s How to Dance in Ohio

Melissa S. Garber and Katherine L. Fiori*

Gordon F. Derner School of Psychology, Adelphi University

ABSTRACT. This research project examined whether a live theatrical performance of How to Dance in Ohio decreases stigma among audience members. The research was conducted during 2 weeks of Broadway performances in New York City and followed a pretestposttest design, including a survey of audiences right before, 48 hours after, and 2 weeks after the performance. We expected that autism stigma would decrease after the performance and remain lower 2 weeks after the performance compared to the pretest. Results show that participants displayed a significant difference in levels of autism stigma after viewing the performance, F (1,58) = 10.56, Wilks’ Lambda = .85, p = .002, η 2 = .15. There was also a significant interaction with gender, showing that men had higher levels of stigma at pretest and showed a larger decrease after the performance, F(1,58) = 7.51, Wilks’ Lambda = .89, p = .008, η2 = .12. These data support the initial hypothesis that the performance was an effective intervention to lower levels of autism stigma for 2 weeks among audience members of a musical.

Keywords: autism spectrum disorder, stigma, intervention, theater, arts, musical, prejudice

Stigmas and stereotypes towards individuals who differ from the societal norm exist in many different spaces and formats. A stigma is a set of negative beliefs that discredit a person’s claim to a normal identity or discredit a group of people (Hoffman, 2016; Twardzicki, 2008). Stigma towards others can be expressed for various reasons, such as someone’s physical appearance or the group to which they belong. These stigmas can be displayed through a person’s actions or words or expressed in the media. The present study aimed to determine if a theatrical performance can be an effective intervention against stigma, specifically towards the autism community.

Mental Health Stigma

One community that has faced stigma for several decades consists of those who are diagnosed with a mental illness or disorder; this is known specifically as psychiatric stigma (Hawke et al., 2014). Psychiatric stigma has been observed in families and healthcare

providers of people with mental illnesses (Gaebel & Baumann, 2003). This type of stigma can result in those suffering from a mental illness not accepting professional help due to a fear of being labeled (Gaebel & Baumann, 2003). Additionally, psychiatric stigma can cause shame, which can increase psychological pain in individuals (Patterson & Sextou, 2017). Furthermore, the caregivers of people with mental illnesses are also subject to stigma, similar to people who have mental illnesses themselves (Sextou & Patterson, 2014).

Autism as a Mental Illness and a Neurodiverse Condition

Autistic individuals are a part of a largely stigmatized community. Autism spectrum disorder (ASD) is characterized by social and communicative deficits; however, each individual has varying symptomology and severity (White et al., 2016). In the Diagnostic and Statistical Manual of Mental Disorders, ASD is included as a neurodevelopmental disorder (American Psychiatric

Association, 2022). Although this perspective is still currently accepted, there is a movement for ASD to be recognized as a neurodiverse condition rather than as a mental illness. The neurodiversity movement recognizes the neurological differences within the human population as different ways people exist rather than as disabilities (Jaarsma & Welin, 2012).

Theatre as an Intervention to Stigma

Both perspectives about ASD are being explored within the psychological community; however, stigma towards autistic individuals still exists across many facets of society. Previous studies showed that this stigma may result in autistic individuals having fewer friendships and facing higher rates of bullying compared to neurotypical peers (Sterzing et al., 2012; van Roekel et al., 2010). Of course, the difficulties with conversational ability and reduced social skills associated with ASD may also make friendships more difficult, above and beyond the existing stigma. Autistic individuals have described stigma as “destructive,” indicating the extreme effect it can have on them and their families (Turnock et al., 2022). The best strategies to decrease stigma include education, protest, contact, and improving psychoeducation (Gaebel & Baumann, 2003). The strategies to decrease stigma can change and evolve. For example, theatre can be a form of both education and contact for many people. Theatre presents an audience with the opportunity to see issues through a dramatic lens that heightens and clarifies those issues and to observe the interpersonal dynamics associated with mental illnesses (Blignault et al., 2010; Sextou & Patterson, 2014). For example, theatrical performances have been used to promote discussion of illnesses such as schizophrenia and Alzheimer’s Disease (Blignault et al., 2010). Additional studies have shown that theatre can reduce stigma towards bipolar disorder (Michalak et al., 2014).

Existing literature has also shown that many different forms of art can be effective interventions against stigma towards people with mental illnesses or people who are neurodivergent. For example, a study of bipolar disorder stigma used both live performance and DVD performances as an intervention and saw that stigma decreased after both types of intervention. In fact, when used together, these interventions were found to be even more effective (Michalak et al., 2014). Additionally, drama­based teaching methods helped people feel more comfortable discussing sensitive issues and raising questions about these issues, which is a key principle in reducing stigma (Twardzicki, 2008).

Another study showed that viewing a movie portraying an individual’s experience with schizophrenia made audience members empathize more with people with schizophrenia (Gaebel & Baumann, 2003). Interestingly, introducing Julia, a child with autism, to Sesame Street’s programming in 2017

was also shown to increase knowledge and acceptance of autistic children within community settings (Georgetown University & Children’s National, 2023; Sesame Workshop).

How to Dance in Ohio: The Musical How to Dance in Ohio is a musical that premiered in Syracuse, New York, in 2022, before transferring to Broadway in New York City in late 2023. The show is based on the documentary of the same name that follows seven autistic young adults as they prepare to attend a dance being put on by their therapist, Dr. Amigo. The musical incorporates the daily routines of the seven main characters and uses the songs to help the audience understand the inner monologue of each character. Previous literature has demonstrated that other shows depicting autistic young people can act as indirect interventions for autism stigma (Massa et al., 2020). Specifically, a new work entitled Beyond Spectrums, inspired by the play The Curious Incident of the Dog in the Night-Time, was developed by Massa, DeNigris, and Gillespie­Lynch and was designed to provide contact with autistic people. After viewing the Beyond Spectrums performance, many audience members showed a greater willingness to interact with autistic people, supported by decreased social distance scores. Our study aimed to find further support for this type of effect, and to fill the gap in the literature for the use of musical theatre as an intervention against autism stigma.

The Present Study

In the present study, we aimed to identify if a live theatrical performance could be an effective intervention for autism stigma among audience members. As such, we measured levels of autism stigma among audience members before, 48 hours after, and two weeks after viewing a performance of How to Dance in Ohio. We hypothesized (Hypothesis 1) that participants would display a decrease in stigma 48 hours after viewing the performance and that these changes would be sustained two weeks after the performance. We also hypothesized (Hypothesis 2) that men would display higher levels of stigma compared to women before the performance (and therefore a greater decline between Time 1 and Time 2) based on previous literature demonstrating that women tend to have a greater knowledge of autism relative to men (Massa et al., 2020). Furthermore, Massa et al. (2020) also found that emerging adults (ages 18–30 years old) reported higher levels of stigma than adults older than 30 prior to a theatrical intervention. Thus, we also hypothesized (Hypothesis 3) that participants aged 18–30 would display higher levels of stigma than older audience members before the performance (and, therefore, a greater decline between Time 1 and Time 2). Additionally, we hypothesized (Hypothesis 4) that

participants who were more familiar with the show would display lower levels of autism stigma to start (and therefore a smaller decline between Time 1 and Time 2), as these individuals were likely more knowledgeable about autism. Finally, we hypothesized (Hypothesis 5) that familiarity with autism itself may play a role, such that individuals with autism would show the lowest levels of stigma (and smallest declines between Time 1 and Time 2) compared to individuals with a friend/ family member with autism or those with no familiarity.

Method

Participants and Procedure

Our data were collected from a sample of audience members who attended a theatrical performance in New York City in February 2024. Participants were recruited by email the day before they attended the show. Participants were also recruited via the show’s Instagram account with a link that was accessible through the production’s Instagram bio. The participants completed an online survey using Qualtrics before their first time viewing the performance (Time 1), 48 hours after the performance (Time 2), and two weeks after the performance (Time 3). The participants were contacted via email to complete the second and third waves of the survey. If the participants did not respond to the email within a week of its original send date, they were not included in that wave of the survey. Furthermore, some participants who were recruited via social media completed the first wave of the survey after seeing the performance rather than before seeing it. This study focuses on the 64 participants who completed all three waves of the study. In total, 184 participants completed the first wave of the study, 122 participants completed both the first and second wave of the study, and 64 participants completed all three waves of the study. The 120 participants who did not complete all three waves of the study were excluded from the data analyses1.

The 64 participants had a mean age of 32.32 years ( SD = 13.21, range 18–79). Of the 64 participants, 39 were considered emerging adults (between the ages of 18–30), and the remaining 25 were older than 30. Participants were primarily female (68.75%; n = 44), with n = 13 transgender/nonbinary participants and n = 7 men. Of the 64 participants, 53.13% were familiar with the show before seeing the performance. Additionally, whereas 28

1Of the 120 participants who were excluded from the analyses, 62 of the participants were excluded because they viewed the performance before completing Time 1 of the survey, 7 participants were excluded because they did not complete Time 2, 23 participants were excluded because they did not complete Time 3, and 28 participants were excluded because they did not complete Time 2 or Time 3. Attrition analyses constituting a series of Chisquare analyses and t tests indicated no differences in any study variables between the n = 64 individuals who completed all 3 time points and the n = 120 individuals who were excluded.

participants had no familiarity with someone with autism, 9 participants knew a friend or colleague with autism, 12 participants had a family member with autism, and 16 participants self­identified as autistic. Over half of the participants were employed full­time (54.69%), and 75% of the whole sample had earned an associate’s degree or higher.

Measures

Demographics

Demographics requested from the participants included age, gender identity, level of education, occupation, race, employment status, home location, relationship to a person with autism, familiarity with the show, and the date of the performance attended. Familiarity with the show was measured on a scale from 1 (not familiar) to 5 (very familiar). We divided familiarity into two categories for our analyses: those with low familiarity (who scored 1 or 2) and those with high familiarity (who scored 3, 4, or 5).

Stigma Towards Autism

Stigma towards autism was measured using the Social Distance Scale, which was adapted for a focus on autism (ASDS; Gillespie­Lynch et al., 2015). The scale included six items. Participants responded to items on a Likert scale from 1 (definitely willing) to 4 (definitely unwilling) of how likely they would engage in an activity with a person with autism. An example item was: “How willing would you be to move next door to someone with autism?” We created a mean score, and the scale had good reliability (α = .84), with higher scores representing higher levels of stigma.

Results

To test our hypotheses, we used the SPSS General Linear Model (GLM) procedure. Specifically, we included age (18–30, n = 39 vs. 31+, n = 25), gender (male, n = 7 vs. female, n = 44 vs. transgender/binary, n = 13), familiarity with the show (low familiarity, n = 30 vs. high familiarity, n = 34), and familiarity with autism (no familiarity, n = 28, friend n = 9, family n = 12, self n = 16) as covariates in two repeated­measures ANCOVAs predicting changes in autism stigma, first across all 3 time points, and then across just the first two time points (given that we actually expected stability between Times 2 and 3). GLM automatically generates interaction terms between covariates and the within­subjects factors (repeated measures), so we were able to evaluate all of our hypotheses (1–5) with this approach2.

2To supplement these analyses, we also ran two sets of four repeatedmeasures ANOVAs testing for interactions with gender, age, familiarity with the show, and familiarity with autism separately in each ANOVA. In the first set, we predicted changes in stigma across the three time points of the study. Then, we ran a second set of ANOVAs in which we predicted changes in stigma just from Time 1 to Time 2. Results were largely similar with the GLM approach. These supplementary analyses are presented in Appendix.

The first ANCOVA analysis including all three time points showed a significant decrease in levels of stigma, F(2,57) = 5.194, Wilks’ Lambda = .85, p = .008, η2 = .15 (Time 1 M = 1.20 (SD = 0.33), Time 2 M = 1.16 (SD = 0.31), Time 3 M = 1.17 (SD = 0.30); see Tables 1–3), consistent with Hypothesis 1. We also observed a significant interaction between level of autism stigma and gender, consistent with Hypothesis 2, such that women showed higher levels of stigma across all three time points compared to participants who identified as nonbinary or transgender, F(2,57) = 3.89, Wilks’ Lambda = .88, p = .03, η2 = .07 (see Figure 1), and women showed higher levels of stigma at Times 2 and 3 compared to men. As can be seen in Figure 1, only men showed a decline in autism stigma from Time 1 to Time 2, as confirmed with a paired­samples t test (see Appendix). No other interactions were significant, contrary to Hypotheses 3–5.

To supplement our three timepoint repeatedmeasures ANCOVA, we conducted a second ANCOVA predicting changes only from Time 1 to Time 2, given our hypothesis in stability in stigma from Time 2 to Time 3. We again observed a significant change in levels of autism stigma such that stigma levels were significantly lower at Time 2 compared to Time 1, F(1,58) = 10.56, Wilks’ Lambda = .85, p = .002, η2 = .15 (Time 1 M = 1.20, SD = 0.33; Time 2 M = 1.16, SD = 0.31). We saw a significant interaction between gender and autism stigma, F(1,58) = 7.51, Wilks’ Lambda = .89, p = .008, η2 = .12. This analysis also displayed a significant interaction between familiarity with the show and autism stigma, such that only participants who had lower familiarity with the show displayed a decrease in stigma after viewing the performance, F(1,58) = 4.43, Wilks’ Lambda = .93, p = .04, η2 = .07 (see Figure 2). These results are consistent with Hypothesis 4.

As a follow ­ up to these results, we conducted a Chi­square analysis and found a significant association between gender and familiarity with autism, χ2(6) = 32.24, p = < .001. That is, of the 13 participants who identified as nonbinary or transgender, 10 of those participants self­identified as having autism, compared to male participants (n = 1) and female participants (n = 5). This implies that the interaction between gender and autism stigma may be due at least in part to the higher concentration of autistic individuals who are also nonbinary or transgender. However, we did not see a significant association between show familiarity and autism familiarity, χ2(6) = 5.79, p = .45. Indeed, a similar number of participants who had high familiarity with the show (n = 13) had no familiarity with someone with autism compared to those with low familiarity with the show (n = 14). This implies that the interaction effect of familiarity with the show was not solely due to knowing someone with autism.

Discussion

In this study, we aimed to determine if a theatrical performance could serve as an intervention to reduce autism stigma. We compared the levels of stigma immediately before a Broadway musical performance of How to Dance in Ohio, 48 hours after the performance, and one month after the performance in a sample of 64 audience members. We also looked for interactions with gender, age, familiarity with autism, and familiarity with the show. Overall, consistent with our first hypothesis, we saw a significant change in stigma across the three waves of the study, with a decrease between Time 1 and Time 2, and stable levels from Time 2 to Time 3. Although the difference in stigma we observed was quite small, we still believe that this finding has

Note. Stigma mean was measured on a scale from 1 to 4, with a score closer to 1 representing lower levels of stigma.

Autism Stigma From Pre-Performance to 48 Hours Post to 2 Weeks Post

Stigma From

3

practical importance given that even small decreases in stigma can potentially lead to important changes in how the autistic community is perceived.

Consistent with our second hypothesis, we also found a significant interaction between stigma and gender, such that women showed higher levels of stigma across all three time points compared to participants who identified as nonbinary or transgender, and women showed higher levels of stigma at Times 2 and 3 compared to male participants. It is possible that women still hold a bias towards the autistic community despite previous research showing they tend to be more knowledgeable about autism (Massa et al., 2020). Alternatively, the participants who identify as nonbinary or transgender showed the lowest levels of stigma possible across all time points, M = 1.00. These gender identities are part of the LGBTQIA+ community, which is also stigmatized in many settings. Therefore, members of this community may hold less stigma because of their own experiences within a stigmatized community. Additionally, our follow­up chi­square analysis showed that there was a significant overlap between participants who identified as transgender or nonbinary and who had autism. Thus, it is possible that these individuals have lower levels of stigma because they are also part of the neurodiverse community. Future research using a larger sample of transgender/nonbinary individuals could explore these possibilities.

There was also a significant interaction between stigma and familiarity with the show (at least in the analyses examining changes from Time 1 to Time 2 only), which partially supports our fourth hypothesis. Participants who had no or low familiarity with the musical displayed higher levels of autism stigma across all time points compared to participants with high levels of familiarity with the show, but also showed a decrease in stigma between Time 1 and Time 2. This is an important finding because it showcases the show’s effectiveness, increasing familiarity and knowledge of the autistic community for those who previously were not familiar. This sample of our participants echoes one of the key tenets of the musical and documentary, as they were created to help increase awareness and knowledge of autistic individuals and their experiences.

Our data did not support our third hypothesis that younger adults (aged 18–30) would show higher levels of stigma compared to adults aged 31 and older. This may be due in part to the high levels of education across our sample (75% of the sample had earned an associate’s degree or higher; Gaebel & Baumann, 2003), potentially washing out any age differences. Furthermore, the older age group in our sample (31+) varied widely, ranging up to 79. Combined with the small sample, this wide

variability may have limited age/cohort differences; perhaps with a more homogenous older adults age group (e.g., 65+), we might have seen greater differences with a younger sample.

Additionally, our fifth hypothesis, that participants who were not familiar with an autistic person would show higher levels of stigma compared to people who had a friend, colleague, or family member with autism or had autism themselves (and therefore a greater decline over time), was not supported. This may be due to the stigma that families, friends, or autistic individuals are subject to, as has been seen with other medical and psychiatric disorders (Sextou & Patterson, 2014). By experiencing this stigma, these populations may be more likely to internalize these negative beliefs, leading to higher stigma levels overall.

In sum, our data show that this production of How to Dance in Ohio may be a successful intervention for small decreases in autism stigma among audience members, particularly among men and those with less familiarity with the show. Our findings are consistent with previous literature showing that theatre can be an intervention against stigma toward other mental illnesses or neurodivergent conditions (Blignault et al., 2010; Michalak et al., 2014).

Limitations and Future Research

Our study was limited by a relatively small sample size, especially compared to the number of audience members who see a Broadway performance over the course of a show’s run. Due to the resulting low power of our analyses, we must interpret our results with caution. It is possible that additional interactions between stigma and covariates may exist that our data could not identify. Future research would benefit from a longer period of recruiting participants to increase the sample size. Furthermore, our study only showed that the decrease in stigma was maintained over a short period of time, 2 weeks, and we cannot make inferences about whether this effect would be longer lasting. Another limitation exists within the musical itself. The seven central characters have relatively low support needs, which does not represent the full spectrum of autistic people. It may be possible that the show’s representation of autistic people changed the schema held by our participants compared to before the show. If this schema changed, then it is possible that the changes in stigma levels we observed were because the schema participants had of autistic individuals changed and impacted their responses to the questions. To better assess this, future research should include more specific information in the measures used to ensure that all participants answer questions with the same reference point, or find shows that showcase the wide range of support needs that autistic individuals can have.

Additionally, the participants were recruited based on convenience, which may have influenced who took part in the study. For example, individuals who choose to attend a Broadway show about an autistic group of adolescents may have lower levels of stigma than the average population. Also, audience members at Broadway shows typically come from a more affluent background, which may have impacted their levels of stigma prior to the performance. We did not ask participants if this was their first experience at a Broadway show; however, this may be an impactful difference among some participants and should be included in future studies. Furthermore, personality characteristics not assessed in the current study, such as openness, may also account for differences in changes in levels of stigma.

Furthermore, not every participant saw the same performance, which may impact the consistency of a theatre intervention (compared to a movie, which is a

permanent art form). Moreover, we cannot draw conclusions about causality because of this study’s correlational design. Future research could randomly assign participants to see either a show designed to combat autism stigma, or a show unrelated to autism, and look at stigma outcomes. Finally, the changes in stigma we observed may have occurred because of practice effects, because the participants were asked about their stigma levels multiple times. By asking the participants about autism stigma before seeing the performance, they were more likely to be aware of their levels of stigma during and after the performance. Future research may benefit from measuring levels of stigma farther in advance of the performance to minimize this effect.

Conclusions

The purpose of this study was to determine if theatre could be used to decrease autism stigma among audience members at a performance of a Broadway show in New York City. In spite of its limitations and correlational nature, our study supports the existing literature that theatre can be an effective intervention. Additionally, we saw that gender may play a role in how stigma changes after seeing such a performance. This research helps to expand on this existing literature by including autism, which can be considered both a mental illness and neurodiversity, as a target for stigma reduction in theatre.

References

American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed., text rev.). https://doi.org/10.1176/appi.books.9780890425787

Blignault, I., Smith, S., Woodland, L., Ponzio, V., Ristevski, D. & Kirov, S. (2010). Fear and shame: Using theatre to destigmatise mental illness in an Australian Macadonian community. Health Promotion Journal of Australia, 21(2), 120–126. Gaebel, W., & Baumann, A. E. (2003). Interventions to reduce the stigma associated with severe mental illness: Experiences from the Open Doors Program in Germany. The Canadian Journal of Psychiatry, 48(10), 657–662. Georgetown University & Children’s National (2023). See amazing in all children: Evaluation report executive summary https://sesameworkshop.org/wpcontent/uploads/2023/03/See-Amazing-Executive-Summary.pdf

Gillespie-Lynch, K., Brooks, P., J., Someki, F., Obeid, R., Shane-Simpson, C., Kapp, S., K., Daou, N., & Smith, D., S. (2015). Changing college students’ conceptions of autism: An online training to increase knowledge and decrease stigma. Journal of Autism and Developmental Disorders, 45, 2553–2566. https://doi.org/10.1007/s10803-015-2422-9

Hawke, L. D., Michalak, E. E., Maxwell, V., & Parikh, S. V. (2014). Reducing stigma toward people with bipolar disorder: Impact of a filmed theatrical intervention based on a personal narrative. International Journal of Social Psychiatry, 60(8), 741–750. https://doi.org/10.1177/0020764013513443

Hoffman, G. (2016). Mental illness through the lens of theatre. [Honors program project, University of Tennessee, Knoxville]. Tennessee Research and Creative Exchange.

Jaarsma, P., & Welin, S. (2012). Autism as a natural variation: Reflections on the claims of the neurodiversity movement. Health Care Analysis, 20, 20–30. https://doi.org/10.1007/s10728-011-0169-9

Massa, A., DeNigris, D., & Gillespie-Lynch, K. (2020). Theatre as a tool to reduce autism stigma? Evaluating ‘Beyond Spectrums’. Research in Drama Education, 25(4), 613–630. https://doi.org/10.1080/13569783.2020.1732810

Michalak, E. E., Livingston, J. D., Maxwell, V., Hole, R., Hawke, L. D., & Parikh, S., V. (2014). Using theatre to address mental illness stigma: A knowledge translation study in bipolar disorder. International Journal of Bipolar Disorders, 2(1). https://dx.doi.org/10.1186/2194-7511-2-1

Patterson, P., & Sextou, P. (2017). ‘Trapped in the labyrinth’: Exploring mental

illness through devised theatrical performance. Medical Humanities, 43, 86–91. https://doi.org/10.1136/medhum-2016-011094

Sesame Workshop. Autism initiative: Celebrating and supporting kids with autism and every child.

https://sesameworkshop.org/our-work/what-we-do/autism-initiative/

Sextou, P., & Patterson, P. (2014). Theatre, society, and stigma: Mental illness on stage. The International Journal of Social, Political and Community Agendas in the Arts, 9(1), 1–10. https://doi.org/10.18848/2326-9960/CGP/v09i01/1-10

Sterzing, P. R., Shattuck, P. T., Narendorf, S. C., Wagner, M., & Cooper, B. P. (2012). Bullying involvement and autism spectrum disorders: Prevalence and correlates of bullying involvement among adolescents with an autism spectrum disorder. Archives of Pediatrics & Adolescent Medicine, 166(11), 1058–1064. https://doi.org/10.1001/archpediatrics.2012.790

Turnock, A., Langley, K. & Jones, C. R. G. (2022). Understanding stigma in autism: A narrative review and theoretical model. Autism in Adulthood, 4(1), 76–91. https://doi.org/10.1089/aut.2021.0005

Twardzicki, M. (2008). Challenging stigma around mental illness and promoting social inclusion using the performing arts. The Journal of the Royal Society for the Promotion of Health, 128(2), 68–72. https://doi.org/10.1177/1466424007087804

van Roekel, E., Scholte, R. H. J., & Didden, R. (2010). Bullying among adolescents with autism spectrum disorders: Prevalence and perception. Journal of Autism and Development Disorders, 40, 63–73. https://doi.org/10.1007/s10803-009-0832-2

White, D., Hiller, A., Frye, A., & Makrz, E. (2016). College students’ knowledge and attitudes towards students on the autism spectrum. Journal of Autism and Developmental Disorders, 49(7), 2699–2705. https://doi.org/10.1007/s10803-016-2818-1

Author Note.

Melissa S. Garber https://orcid.org/0009­0001­4065­1698

Katherine L. Fiori https://orcid.org/0000­0003­3386­5898

This study received approval from the Adelphi University Institutional Review Board as Protocol #20240004. We have no known interests to disclose. Special thanks to Sammy Lopez and the entire P3 Productions and How to Dance in Ohio teams for making this collaboration possible, as well as the members of the ReACH Lab at Adelphi University for their collaboration and insight during lab meetings.

The participants of this study did not give written consent for their data to be shared publicly, so due to the sensitive nature of the research, supporting data is not available.

Melissa S. Garber played a lead role in conceptualization, data collection, and original writing. Katherine L. Fiori played a lead role in editorial assistance and a supporting role in minor original writing. Both authors contributed to research design and data analysis and interpretation.

Melissa and Katherine identify as heterosexual, cisgender White women. All authors are nondisabled and neurotypical and acknowledge that their perspectives are influenced by their positions within all of these dimensions of identity.

Correspondence concerning this article should be addressed to Melissa S. Garber. Email: melissagarber@mail.adelphi.edu

APPENDIX

The first repeated-measures ANOVA examining changes in levels of stigma across the three waves did not show a significant difference in autism stigma between the time points, F(2, 62) = 1.01, Wilks’ Lambda = .97, p = .37, η2 = .03; Time 1 M = 1.20 (SD = 0.33), Time 2 M = 1.16 (SD = 0.31), Time 3 M = 1.17 (SD = 0.30; Table 1 and 2). Consistent with the GLM approach, a repeated-measures ANOVA with gender as a factor showed a significant interaction between stigma and gender across all three time points, F(2,60) = 4.01, Wilks’ Lambda = .88, p = .02, η2 = .12. We conducted follow-up paired-samples t tests to compare differences by gender. This analysis showed a significant decrease in stigma level from Time 1 to Time 2 for men only, Time 1 M = 1.40 (SD = 0.43), Time 2 M = 1.17 (SD = 0.25), t(6) = 2.34, p = .029 (one-sided). The paired-samples t tests were not significant for women or transgender/nonbinary individuals. Additionally, and again consistent with the GLM approach, a repeated-measures ANOVA with show familiarity as a factor predicting changes in stigma across two waves showed a significant interaction between stigma and show familiarity from Time 1 to Time 2, F(1,62) = 3.99, Wilks’ Lambda = .94, p = .05, η2 = .03. There were no other significant interactions in the repeated-measures ANOVA analyses, consistent with the GLM approach.

TABLE 1

Autism Stigma From Pre-Performance to 48 Hours Post to 2 Weeks Post

TABLE 2

Autism Stigma From Pre-Performance to 48 Hours Post Variables

Familiarity

Resilience and Drinking Behaviors in Emerging Adults With Previous Extracurricular Participation

Ava Avolio  Department of Psychology, Counseling, and Criminology, Carlow University1

ABSTRACT. Incorporating a Positive Youth Development framework, this study sought to demonstrate how adolescent extracurricular participation facilitated resilience development and promoted long ­ term behavioral and educational benefits into emerging adulthood. Resilience was previously identified as a protective factor and promoter of advantageous life outcomes despite childhood adversity. Data were collected from 281 participants aged 18 to 25 across 50 American universities and trade schools. The resulting correlational analysis suggested that extracurricular participation had a significant positive correlation with resilience ( r = .23, p < .001). Moreover, this resilience was negatively associated with binge drinking (r = ­.23, p < .001) and positively associated with higher educational attainment (r = .24, p < .001). A path analysis was performed based on a PYD­informed model: χ²(6) = 29.84, p < .001, Comparative Fit Index (CFI) = .96, Tucker­Lewis Index (TLI) = .91, Root Mean Square Error of Approximation (RMSEA) = .11 (90% CI [.08, .16]), and Standardized Root Mean Square Residual (SRMR) = .09. The results suggested a moderate fit. The analysis confirmed that extracurricular participation is linked to greater resilience, which in turn supports positive life outcomes in emerging adulthood, regardless of childhood adversity. These findings encourage further research incorporating intentional resilience­building into established extracurricular activities as a method of extending developmental benefits to underserved communities.

Keywords: resilience, extracurricular participation, emerging adulthood, Positive Youth Development

Few contemporary high school archetypes have entered the cultural consciousness quite like the “model student.” As prestigious universities faced thousands more qualified applicants than available seats, admission committees used extracurricular involvement to predict future success. In response, student extracurricular participation grew at record rates (Mayol ­ García, 2022). However, data analysis of over six million college applications revealed that participants from historically privileged demographic

1Jennifer

Roth, PhD is the faculty mentor.

Open Materials badge earned for transparent research practices. Materials are available at https://osf.io/yngw5/

categories (e.g., White, high socioeconomic status, and private school educated) reported significantly more extracurricular participation, leadership positions, and awards (Park et al., 2023).

Nevertheless, viewing extracurricular participation solely to impress college admissions would be reductive. The admissions shift toward valuing these activities emerged due to their long­documented association with positive developmental outcomes (Agans et al., 2014; Busseri et al., 2006; Forneris et al., 2015). It has been well ­ established that

extracurricular participation is an impactful part of American students’ development as it has repeatedly been linked to school success and decreased behavioral issues (Martin et al., 2013; Weitzman & Chen, 2005). Notably, extracurricular participation has been demonstrated as particularly developmentally beneficial for youths exposed to adverse events (Mahoney & Cairns, 1997). Although highly influential during adolescence, the long­term impact of extracurricular participation has yet to be thoroughly explored. Unanswered questions remain regarding the durability of these effects and whether such participation contributes to enduring behavioral patterns into emerging adulthood.

Emerging Adulthood

Emerging adulthood, ages 18–25, was conceptualized to describe shifting timelines for major life milestones across generations (Arnett, 2014). Though legally adults, individuals in this age range are typically in the early stages of establishing their adult independence. In the past, it was common to marry and have children during the mid­twenties; however, today the average marriage age is 30 for U.S. men and 28 for U.S. women (U.S. Census Bureau, 2023). Despite the shift, typical U.S. adults still leave their parents’ homes between the ages of 18–19 and do not enter the workforce until their late twenties (Arnett, 2014). This delayed transition has created an extended developmental period characterized by increased autonomy, identity exploration, and self­focus. Across 300 interviews, Dr. Jeffery Arnett conceptualized emerging adulthood by classifying five typical developmental experiences: identity exploration, instability, self­focus, feeling “in­between,” and optimism (Arnett, 2014). Although an exciting time, this period is also associated with increased substance use. A 45­year­long longitudinal study found that substance use peaked during this time across generations (Schulenberg et al., 2020).

Binge Drinking

Binge drinking has remained a common occurrence among emerging adults as supported by a nationally representative study of more than 50,000 college students. About two in five students reported drinking at binge levels at least once every two weeks (Wechsler & Kuo, 2000). Intoxication has also been frequently found among young adults as 48% of college­student drinkers reported that reaching drunkenness was their main motivation for drinking, 23% drank more than ten times a month, and 29% became intoxicated more than 3 times a month (Wechsler & Nelson, 2008). Moreover, a clear association between binge drinking and frequent drinking has been established in this age range. In a

survey of American college students, 64% of drinkers met criteria for both binge drinking and weekly drinking. Only 19% of drinkers qualified as binge drinkers alone and 17% were weekly drinkers only (DeMartini & Carey, 2012). Frequent binge drinking in emerging adulthood has been linked to neural structural changes, academic decline, increased criminal behavior, and future substance use disorder (Pérez­García et al., 2022).

Adverse Childhood Experiences

Felitti et al. (1998) sought to determine if events in childhood correlated with later health complications. After analyzing health data from nearly 10,000 adults, the research team identified ten traumatic childhood experiences that had a significant dose­response relationship with later serious health complications. Over 50% of respondents reported one or more Adverse Childhood Experiences (ACEs), and around 25% reported two or more ACEs. People who reported four or more ACEs were 4–12 times more at risk for substance use disorder, depression, and suicide; 2–4 times more at risk for smoking, sexually transmitted diseases, and poor health; and 1.4–1.6 times more at risk for obesity (Felitti et al., 1998). This study marked the first empirical linkage between childhood experience and later health outcomes.

Challenges

Linked to ACEs

Since the original publication, studies across healthcare, education, and social science disciplines have used Adverse Childhood Experiences to quantify childhood adversity. ACEs have consistently correlated with negative physical, mental, and interpersonal consequences if left without intervention (Howell et al., 2021). An analysis of over 1000 children in the Fragile Families and Child Wellbeing Study found that 55% of the children were exposed to at least one ACE and 12% experienced more than three (Jimenez et al., 2016). Generational ACE impact literature from Howell et al. (2021) suggested that ACEs limit skills acquisition opportunities and lead to self­blame or externalizing behaviors. ACE exposure was found to correlate with lower social and emotional learning capabilities, low­affection parent­child relationships, and more frequent behavioral and emotional problems during childhood (Howell et al., 2021). A 2024 meta­analysis found that ACE­exposed young adults reported significantly more psychosocial problems than their peers (Silva et al., 2024). These internal struggles have been shown to externalize as risk taking behaviors (Felitti et al., 1998). Adults with higher ACE exposure displayed significantly decreased emotional regulation and self­regulation abilities which have been linked to poor decision making (Silva et al., 2024).

Positive Youth Development

After the ACE study solidified that health consequences can originate from disrupted childhood development, researchers sought ways to protect the process. Educators soon recognized the significant amount of time children spend in school, and research began emphasizing childhood development enhancements through schoolbased programming. This initiated the Positive Youth Development (PYD) movement which holds the position that all children hold innate strengths that can be intentionally developed (Martin et al., 2013). The goal of PYD is not to eliminate challenges but instead introduce age­appropriate challenges that allow youths to practice positive personal agency (Romer & Hansen, 2021). The leading model of PYD, known as the five characteristics or “5 Cs,” describes five core traits needed to facilitate holistic youth development. These Cs are connection, competence, character, caring, and confidence (Lerner, 2004). Collectively, these traits have helped youths foster relationships, hone cognitive and motor skills, solidify moral judgment, display empathy, and build self­esteem. One longitudinal study demonstrated that these five traits helped youths successfully navigate challenges which in turn promoted well­rounded development (Bowers et al., 2010). Furthermore, research found that individual positive development most effectively occurred when expressing these characteristics through community interactions, making prosocial approaches crucial for development (Romer & Hansen, 2021).

Extracurricular Participation

For school aged children, extracurricular activities are any regularly occurring, planned activities that occur outside of required academics. Although not required, extracurricular participation has become a staple of the American education system. As of 2015, over 80% of 7 th–12 th graders participated in at least one organized extracurricular activity (Grover et al., 2015). The most popular include athletics, arts, special interest clubs, and service­based organizations (Grover et al., 2015).

Researchers have identified seven factors of extracurricular activities that contribute to beneficial adolescent development. These factors include physical and psychological safety, supportive interpersonal relationships, regular structure, opportunities for social belongingness, skill­building challenges, self­efficacy discovery, and collaboration between school, family, and community (Gardner et al., 2012). Given the theoretical similarities between these factors and PYD’s 5 Cs, increasing extracurricular participation has become a core Positive Youth Development intervention.

Benefits Linked to Extracurricular Participation

Research has sought to validate PYD’s approach by quantifying participation benefits. One longitudinal study collected 5 Cs, behavioral, and extracurricular data from 927 students during middle and high school. Students who consistently participated in extracurricular activities scored significantly higher on all five characteristics, particularly competence and connection. However, the findings on participation and behavioral decisions were mixed. More involved students did not report significantly less substance use but did refrain from risky behaviors more often than less involved peers (Agans et al., 2014).

Academically, activity outside of the classroom has been positively correlated with performance inside the classroom. In a study of 643 students spanning from elementary school through high school, extracurricular participation correlated positively with class participation rates, adaptive motivation, meaningful class engagement, and academic satisfaction after controlling for socioeconomic status (Martin et al., 2013). Furthermore, extracurricular participation has also been found to encourage school completion. In another longitudinal study following 392 students from 7th grade until 12th grade, at­risk students who participated in extracurricular activities dropped out of school significantly less than at­risk peers who did not participate (Mahoney & Cairns, 1997). A more recent study of 545 students had similar results. Students who regularly participated in extracurriculars had significantly less risk of dropout, but those who attended activities infrequently had similar dropout rates to students who did not participate at all (Thouin et al., 2022). Crucially the threshold for this protective factor was attainable for most students as students only had to participate in one activity for at least two hours weekly to significantly decrease dropout risk (Thouin et al., 2022).

Importantly, environments that included adult mentors who engaged with the participants were found to facilitate social­emotional learning more effectively (Almeida et al., 2023). Mentors provide leadership to extracurriculars, often being the person to define an activity’s culture. A mentor’s positive or negative dynamic with participants has been shown to be one of the determining factors in youths’ decision to continue participation (Dworkin, 2007). Even within participating samples, youths who had strong mentorship bonds with teachers had significantly higher academic achievement and lower disciplinary action compared to youths without bonds (Crosnoe et al., 2004). Likewise, Choi et al. (2015) found that the most successful mentors built authentic rapport with their participants which engendered mutual trust. With this bond, mentors integrated

prosocial lessons such as good sportsmanship and conscientiousness into their coaching. Participants and their parents endorsed increased participant confidence, competence, and life skills that translated on and off the field (Choi et al., 2015).

From a PYD perspective, it is hypothesized that these behavioral benefits are facilitated through the self­regulation skills developed during extracurricular participation. Self­regulation is one’s ability to mediate internal states despite external stimuli. This homeostatic responsibility extends across psychobiological processes including controlling circadian rhythms, behavioral reactivity, and goal ­ directed action (Gestsdottir & Lerner, 2007). Proponents of PYD suggest that extracurricular activities allow self­regulation skills to develop through providing a safe environment for youth to experiment with self­directed behavior and appropriate adaptive coping to activity ­ based challenges. It is believed that this process is behaviorally reinforcing, as positive adaptive self­regulation is rewarded through successful goal completion (Gestsdottir & Lerner, 2007). Participants echoed this sentiment reporting more ease building relationships, collaborating with people of diverse backgrounds, and resisting peer pressure (Forneris et al., 2015). A Gestsdottir and Lerner (2007) study of over 1,000 extracurricular­participating youths found self­regulation scores positively correlated with the 5 Cs of PYD and negatively correlated with known risk behaviors. A follow up analysis of 2,357 high schoolers indicated that self­regulation ability once again correlated with PYD signifiers and negatively predicted substance use, depression symptoms, and problem behaviors in youth (Gestsdottir et al., 2010).

Challenges From Extracurricular Participation

Although research has shown extracurricular activities offer adolescents benefits across academics and personality development, participation may also have social consequences. Bullying within peer groups and unsupportive adult leaders were the leading causes of dissatisfaction. Youths reported quitting extracurriculars as a result and thus lost access to the possible developmental opportunities (Dworkin, 2007). Additionally, some extracurriculars may expose participants to performance pressures and associated mental health risks. U.S. youths who played team sports displayed significantly lower conduct problems, anxiety symptoms, depression symptoms, peer conflict, and concentration issues than youths who did not participate. However, youths who played individual sports showed significantly greater anxiety and depression scores, social withdrawal, and attention issues (Hoffmann et al., 2022). Another study found sports participation as a whole was not associated with internalizing symptoms, however

increased frequency and perceived incompetency were positively associated with internalizing symptoms amongst athletes (Carter et al., 2023). Finally, extracurricular participation may reinforce socioeconomic advantages. In a study of music and athletic extracurriculars, it was found that youths from families with higher education levels experienced greater cognitive benefits than peers who came from less educated families (Bering & Schulz, 2024). Although research has shown that extracurricular participation particularly enhances development for participants with higher ACE exposure, socioeconomic status may prevent children from fully committing to activity participation (Gardner et al., 2012).

Resilience

Resilience is a nonfixed trait that allows a person to adapt, overcome, and thrive after facing adversity (Connor & Davidson, 2003). Self­efficacy often precedes resilience as self­efficacy’s goal­directed decision­making abilities are essential for resilience (Hamill, 2003).

Beyond its definition as a personality trait, resilience has also been hypothesized to be a measure of an individual’s ability to successfully cope in high stress situations. Unmanaged chronic stress can lead to negative health implications and lower quality of life (Felitti et al., 1998; Felix et al., 2019). Resilience may be crucial to overcoming adversity without gaining increased risk for poor life outcomes (Campbell­Sills et al., 2006). A representative undergraduate sample had their coping styles, personality traits, and resilience assessed. Coping styles were split between task­oriented and emotion­oriented coping. Although both types of coping correlated with resilience, task­oriented coping correlated more significantly (Campbell ­ Sills et al., 2006). Moreover, through the creation of the Connor Davidson Resilience scale (CD RISC), it was found that resilience was strongly negatively associated with stress (Connor & Davidson, 2003). These findings further endorse resilience’s active role in coping.

Resilience in the Context of ACE-Exposed Individuals

As ACE exposure positions people in high stress environments, it is necessary for individuals to effectively cope. However, research has shown that resilience was less likely to naturally develop in individuals with ACE exposure (Howell et al., 2021). Longitudinal studies assessing ACE exposure and life outcomes found that ACE exposure was significantly negatively correlated with both resilience and posttraumatic growth in adulthood. Moreover, these deficits were greater in individuals who were exposed to ACEs at younger ages (Howell et al., 2021).

Evidence has suggested resilience is a key factor against the negative developmental consequences of ACE exposure. In studies particularly addressing ACEexposed youth, resilience significantly correlated with decreased stress symptoms and predicted age­appropriate development. These were meaningful findings as that same study found that ACE­exposed youth reported significantly greater trauma symptoms and qualified for a PTSD diagnosis more frequently than non­exposed youths (Bethell et al., 2014). This protective factor has been observed in emerging adulthood. Schaefer et al. (2018) found resilience correlated with prosocial attitudes in ACE­exposed people as they transitioned to college. These emerging adults with greater resilience also reported higher optimism, posttraumatic growth, and improved familial bonds.

Drinking Behaviors Related to Resilience

Trauma exposure studies have explored resilience as a protective factor against substance use. For three years, 1,810 adolescents were surveyed annually about substance use, family risk factors, and adverse events. Regardless of the number of adverse effects faced, those who reported higher resilience also reported significantly less substance use (Wills et al., 2001). Likewise, a buffering effect was found between resilience and alcohol use in the aftermath of trauma. Over 6,000 students were surveyed throughout their college years on alcohol use, alcohol use disorder symptoms, resilience, and newly occurring traumatic experiences. Those who reported a newly occurring traumatic experience also reported significantly more alcohol use and alcohol dependence. However, for those who reported higher amounts of resilience, increased alcohol use reported after new trauma was significantly reduced (Cusack et al., 2023). These findings suggest that resilience may decrease the stress induced motivations that increase drinking behaviors.

Current Study

Although PYD research has frequently studied current adolescent extracurricular participants, there is limited research into the duration of developmental impacts. Particularly, there is interest in whether these attitudes and behaviors developed during high school last through the highly transformative period of emerging adulthood. Self­efficacy and self­regulation have been identified as general contributors to lifelong well­being and their development has been directly linked to extracurricular participation (Agans et al., 2014; Forneris et al., 2015; Gestsdottir & Lerner, 2007). Across psychology and education research, self­efficacy and resilience have been repeatedly found to have a strong positive correlation in

high school students and emerging adults, with some results suggesting these traits may strengthen each other (Hamill, 2003; Konaszewski et al., 2021; Qamar & Akhter, 2020).

The Positive Youth Development theoretical framework assumes that personality and behavioral patterns developed during adolescence have lasting impact into adulthood. Although the current study aims to add to the literature in this regard, other studies have previously provided findings to support this trajectory. Haider and von Stumm (2022) found emerging personality traits like conscientiousness recorded during adolescence correlated with greater academic attainment and performance in emerging adulthood. When compared to intelligence and socioeconomic status, personality traits during adolescence were found to be the strongest and most consistent predictors of adult life outcomes. Furthermore, a 12­year longitudinal study recorded adolescents’ personality trait development and educational careers path yearly. It was found that those who developed greater amounts of emotional stability, conscientiousness, and extraversion reported significantly greater income, job, and career satisfaction during adulthood (Hoff et al., 2021).

Only a single study, by Gardner et al. (2008), has examined how previous extracurricular activity in high school affected life outcomes in emerging adults. This study limited its focus to educational and occupational attainment. The project evaluated data from the National Education Longitudinal Study, which consisted of a nationally representative sample of 24,599 8th graders surveyed again in 10th grade, 12th grade, and eight years after high school graduation. It was found that students who participated in school sponsored extracurricular activities during high school reported significantly higher grades and college attendance even after accounting for previous grades and confounding socioeconomic variables. Intensity of participation mattered as well. For every year of extracurricular participation in high school, the odds of being employed full time after graduation increased by 8% and those who participated for at least two years during high school had a 54% greater chance of attending college. Overall, students who participated more frequently in extracurricular activities attained greater educational and occupational outcomes (Gardner et al., 2008).

The current study aimed to build upon the previous literature by examining both the attitudes and behaviors of those who participated in previous extracurricular activity as they enter adulthood. Notably, this study investigated the duration and specific behavioral manifestations of the developmental benefits that have been previously associated with involvement, both of which

are unexplored within Positive Youth Development literature. Additional examination of extracurricular participation and positive youth development impact was evaluated in context of childhood adversity. We hypothesized that resilience, educational attainment, and less frequent binge drinking in emerging adulthood would positively correlate with the amount of previous extracurricular participation. Moreover, we expected that ACEs would positively correlate with more frequent binge drinking and lower educational attainment, if not paired with extracurricular participation and increased resilience. As such, resilience was hypothesized to be a direct outcome of adolescent extracurricular participation and a mediator between extracurricular participation and positive outcomes in emerging adulthood.

Methods

Sample

The sample contained emerging adults ages 18–25 ( N = 281). Of the respondents, 60.5% identified as women, 38.4% identified as men, and 1.1% identified as nonbinary. As for ethnic identity, 64.1% of respondents were White, 12.5% were Asian American, 10.8% were Black, 9.3% were Latino, 1.5% were Pacific Islander, and 1.8% identified as other. All participants entered their age to verify that only emerging adults were participating (M = 21.85, SD = 2.17).

The collected sample was made of particularly high achieving emerging adults, possibly due to self­selection bias or internet recruitment sampling bias. To quantify achievement, educational attainment was defined as the highest level of education received, and college GPA data were collected. Of the respondents, 0.4% reported a partial high school education, 25.6% reported a completed high school education, 32.4% reported a partial bachelor’s degree, 24.6% reported a completed bachelor’s degree, 17.1% reported a partial graduate degree. This study was open to a wide spectrum of emerging adults to gain a greater perspective of life outcomes across U.S. regions and backgrounds. Participants were able to voluntarily submit the name of the last educational institution they attended, including universities or vocational centers. Participants originated from over 50 institutions including multiple Ivy League and R1 universities. Of the three­fourths of the sample pursuing higher education, the average college GPA (M = 3.61, SD= 0.42) exceeded the typical population. The 50th percentile was 3.70, and the mode was 4.0 on a 4­point scale. This expansive sample was made possible by online survey recruitment. The researcher used the online platforms LinkedIn, CampusGroups, Reddit, and Instagram to spread the survey to social circles of emerging adults. These adults were then encouraged to spread the survey to their networks of peers ages 18–25.

Materials

The survey included 35 questions divided into 4 subsections.The first survey section gathered demographic information from the participants including gender, age, ethnicity, educational attainment, college GPA (if applicable), number of high school extracurriculars, number of years participated, type of athletic/nonathletic activity, fraternity or sorority affiliation, and current group activity participation. Questions pertaining to the type of activity and intensity of participation were asked to remedy gaps in the literature specifically mentioned in previous studies (Kort­Butler & Martin, 2015). As described by Fischer et al. (2020), there have been multiple approaches to quantifying extracurricular participation. Given that each activity differs, and students have their own attendance records, successful measurements of participation reflect temporal and categorical involvement. As such, the current study collected data on the number of years participated in at least one extracurricular and how many different extracurriculars each respondent was involved in. The extracurricular participation questionnaire can be viewed at https://osf.io/spr52.

The second section included the ConnorDavidson Resilience Scale 10 (CD­RISC 10) (Connor & Davidson, 2003). This unidimensional scale has 10 items on which respondents choose their level of agreement on a 5­point scale. The options included 0 (not true at all), 1 (rarely true), 2 (sometimes true), 3 (often true), and 4 (true nearly all of the time). The sum of these items was calculated to create an overall resilience score. Psychometrically, it was validated at three undergraduate institutions. Other studies validated that the CD­RISC 10 has good internal consistency with a Cronbach’s alpha of .86 (Kuiper et al., 2019). Third party review supports its use for resilience quantification (Campbell­Sills & Stein, 2007).

The third section consisted of the Adverse Childhood Experiences (ACE) scale (Felitti et al., 1998). This scale has 10 items listing possible adverse events. Respondents answer yes or no (coded as 1 or 0, respectively) if that event happened to them during their first 18 years of life. It has been previously found that the ACE scale has acceptable internal consistency with a Cronbach’s alpha of .70 (Oláh et al., 2023). Research has shown that scores of 4 or more correspond significantly with negative health outcomes (Felitti et al., 1998). It is imperative to understand that this score is meant to assess trends and that no one score should be used to quantify a specific individual’s risk (Narayan et al., 2021).

The final section held the Alcohol Use Disorders Identification Test­Concise (AUDIT­C) (Bush et al., 1998). The AUDIT­C was modified from the full version to be able to assess risky drinking behaviors effectively

but quickly. This is a three­item scale that asks how frequently people drink, the typical number of drinks consumed on days they do drink, and how often people drink 6 or more drinks in a single occasion (Bush et al., 1998). The AUDIT­C has good internal consistency with Cronbach’s alpha of 0.81 (Reinert & Allen, 2007). Not only was this scale psychometrically validated for use, but it also performed more effectively than the full version for screening college­aged samples (Bush et al., 1998; DeMartini & Carey, 2012).

Procedure

Prior to data collection, this study received institutional review board approval. Participants were first directed to an informed consent page. After giving voluntary consent, participants completed a demographics questionnaire, extracurricular participation questions, the ConnorDavdison Resilience Scale 10 (CD­RISC 10), the Adverse Childhood Experiences Scale (ACE), and the Alcohol Use Disorders Identification Test­Concise (AUDIT­C). For the current study, participants’ CD­RISC 10 scores were used to quantify their overall resilience, ACE scores were used to quantify adverse childhood experience exposure, and the sum of typical number of drinks consumed per night drinking and binge drinking frequency were used to quantify regular heavy drinking. Previous extracurricular participation (PEP) was determined by the number of high school extracurricular activities reported by participants. Once data collection concluded, the researcher used statistical analysis software including SPSS and the lavaan package for R to assess trends. Procedures included correlation calculations and path analyses. A path analysis was performed to fit a model which depicted a possible way PEP, resilience, and ACEs can influence life outcomes and drinking behaviors in emerging adults.

TABLE 1

Results

General Life Outcome Correlations

We performed correlational analysis to determine if there were associations between previous extracurricular activity, resilience, and behaviors in emerging adulthood (see Table 1). Overall, 96.4% (n = 271) of respondents participated in at least one extracurricular activity during high school. The number of previous extracurricular activities was positively correlated with resilience (r = .23, p < .001) and current extracurricular participation (r = .66, p < .001). The number of previous extracurriculars had a weak negative correlation with ACEs (r = ­.19, p = .001). The number of years spent participating in extracurriculars was found to positively correlate with college GPA (r = .22, p = .001). Correlations found through this study’s naturalistic approach suggest that relationships between previous extracurricular participation and life outcomes are present across widely varying activities.

Resilience Correlations

Resilience was found to positively correlate with educational attainment (r = .24, p < .001) and negatively correlate with binge drinking (r = ­.23, p < .001), regular heavy drinking (r = ­0.21, p < .001), and ACEs (r = ­.20, p < .001). These findings suggest that those with greater resilience tend to participate in extracurriculars more in high school, attain higher education, and participate in less risky drinking behaviors than people with less resilience. Moreover, when assessing the correlations of drinking behaviors, heavy drinking was found to correlate significantly with resilience at (r = ­.21, p < .001), but drinking frequency did not significantly correlate with resilience. These findings suggest those

Correlations Between Previous Extracurricular Participation and Major Life Outcomes

Variables 1 2 3 4 5 6 7 8 9

1. Number of Previous Extracurriculars -

2. Resilience .277** [.11, .33] -

3. Adverse Childhood Experiences -.193** [-.30, -.08] -.199** [.31, -.08] -

4. Current Extracurriculare Participation .660** [.59, .72] .150* [.03, .26] -.101 [-.22, .02] -

5. Drinking Frequency .045 [-.07, .16] -.100 [-.21, .02] .151* [.03, .26] .001 [-.12, .12] -

6. Binge Drinking Episodes -.089 [-.20, .03] -.226 [-.33, -.11] .318 [.21, .42] .003 [-.11, .12] .546 [.46, .62] -

7. Regular Heavy Drinking -.118* [-.23, .00] .328** [.22, .43] .328** [.22, .43] .008 [-.11, .12] .504** [.41, .59] .918* [.90, .94] -

8. Educational Attainment .065 [-.05, .18 .243** [.13, .35] -.165** [.05,-.28]-.026 [-.14, .09] -.009 [-.13, .11] -.227* [-.34, -.11] -.229** [-.34, -.12] -

9. College GPA -.078 [-.19, .04] .004 [-.11, .12] .002 [-.11, .12] .007 [-.11, .12] -.071 [-.19, .05] -.007 [-.12, .11] .013 [-.10, .13] .218** [.10, .33] -

Note. All values display 2-tailed significance. * p < .05. ** p < .01.

with high and low resilience tend to drink at a similar frequency, but those with higher resilience tend to drink in moderation when they do drink compared to those with less resilience. Age was not found to significantly correlate with resilience or drinking behaviors.

ACEs Correlations

In addition to previously mentioned associations, ACEs were found to negatively correlate with years spent participating in extracurriculars (r = ­.13, p = .03) and educational attainment (r = ­.17, p = .006). ACEs positively correlated with binge drinking (r = .32, p < .001) and regular heavy drinking (r = .33, p < .001). The findings suggest that those with more ACEs tend to participate in extracurriculars less frequently, attain higher education less frequently, be less resilient, and participate in riskier drinking behaviors more often than those with less ACE exposure.

Path Analysis

A path analysis was conducted to examine the relationships between PEP, ACEs, and resilience on high achieving emerging adults’ life outcomes and risky drinking. The proposed path model expected resilience to mediate participation, ACE exposure, and life outcomes. Participation and ACE exposure initiated indirect pathways in which resilience influenced later drinking behaviors and educational attainment (see Figure 1). The Comparative Fit Index (CFI) was .96 and Tucker­Lewis Index (TLI) was .91. Both values suggest the model is a good fit. However, the Chi­Square value for the model was significant (6) = 29.84, p < .001, the Root Mean Square Error Approximation (RMSEA) value was 0.11, and the Standardized Root Mean Square Residual (SRMR) was 0.09, dampening the strong CFI and TLI. In all, the model suggests a moderate fit and does support the validity of previous extracurricular participation and ACE exposure affecting life outcomes in emerging adulthood through resilience.

Discussion

The current study provides novel insight into how adolescent extracurricular activity may promote prosocial development with lasting protective effects on behavior and educational outcomes throughout emerging adulthood. Consistent with study hypotheses, significant positive correlations between PEP and resilience, educational attainment, and current participation in adulthood were found. However, resilience did not significantly correlate with college GPA. In the PEPresilience analysis, resilience served as an outcome variable which established a direct effect between the two. PEP’s significant positive correlation with resilience is particularly significant as it further supports existing literature’s

claims that youth programming can foster prosocial development (Caldwell & Witt, 2011; Gestsdottir et al., 2010; Gestsdottir & Lerner, 2007).

Exploring drinking behaviors, the current study hypothesized that high resilience would correlate with less risky drinking behaviors. This hypothesis was informed by previous studies that showed correlation between resilience, effective coping skills, and less frequent alcohol usage (Antelo et al., 2021; Wills et al., 2001). It was expected that greater coping ability would decrease the need for substance assisted emotional regulation and thus drive down heavy drinking behaviors. Resilience correlated significantly with less binge drinking episodes in the current dataset. However, there was not a significant correlation between resilience and drinking frequency which contrasts with previous research (Wills et al., 2001). This pattern suggests that those with higher resilience may be motivated to drink for social benefit as opposed to coping needs. These findings align with a previous study which revealed that among those who had recent trauma exposure, those with higher trait resilience drank significantly less in the aftermath (Cusack et al., 2023).

As hypothesized, ACEs significantly correlated with increased risky drinking and less frequent extracurricular participation. These findings agreed with previous studies that showed high ACE individuals suffering from increased substance use, substance use disorder, and multimorbidity at higher rates than those not exposed to ACEs (Felitti et al., 1998; Senaratne et al., 2024). This pattern of increased risk­taking behavior and substance use were specifically identified in adolescents (Silva et al., 2024). The current study depicts this identified population continuing their adolescent behavioral patterns into adulthood.

Path Analysis of Resilience’s Role in Life Outcomes

Postextracurricular Involvement

The resulting path analysis sought to validate patterns established through previous youth development research. The analysis did produce a moderate fit to the proposed model informed by Positive Youth Development theory (see Figure 1). Previously, high self­efficacy was found to positively correlate with other prosocial attitudes and behaviors after successfully navigating challenges (Bandura et al., 2003). Proponents of the Positive Youth Development Theory expanded this idea and suggested that intentional programming could assist kids in developing prosocial traits by providing them a safe environment to build skills through overcoming group challenges (Romer & Hansen, 2021). Longitudinal studies of kids participating in group extracurricular activities did show children developing increased prosocial attitudes and successful development (Busseri et al., 2006; Forneris et al., 2015; Gestsdottir & Lerner, 2007). The current model sought to validate this premise by using resilience as a mediator of PEP, ACEs, and life outcomes. As a mediator, resilience could reflect pro­social development from PEP while also reflecting possible pro­social deficits from adversity exposure. Resilience was used as a proxy measure for self­efficacy given resilience has been found to be an advanced manifestation of self­efficacy and its protective effect for ACE­exposed populations (Hamill, 2003; Howell et al., 2021; Schaefer et al., 2018). The current study’s CFI and TLI suggested a good fit but the RMSEA and SRMR values were mixed.

Generally, the CFI compares if data fits a proposed model better than a null model (Kline, 2011). The current study’s .96 CFI suggests an excellent fit and implies meaningful relationships between variables. TLI is a similar fit index to CFI, but it also reflects the proposed model complexity (Kline, 2011). The current .91 TLI indicates a good fit suggesting the current paths explain the dataset to an acceptable degree. These findings hold significant weight as they support PYD’s fundamental beliefs that activities can promote prosocial development and this involvement and prosocial trait development lead to better life outcomes (Lerner, 2004; Romer & Hansen, 2021).

The RMSEA estimates how well the model would fit the population when accounting for covariances (Kline, 2011). It is hypothesized this value was affected by its application to a particularly high achieving sample whereas the model was meant to reflect the general emerging adult population. Thus, these participants achieved favorable educational outcomes above the degree the model expected resulting in an increased error approximation (Silva et al., 2024). Moreover, the strength of the model may have been weakened by neglecting prosocial traits besides resilience. Because SRMR reflects the average standardized difference

between the observed and model­predicted outcomes, the slightly high (.09) value suggests that a contributing path may be missing (Kline, 2011). Similarly, the model’s chi­square value was significant, which means the model has opportunity for enhancement. This supports the claim that resilience is only one part of the larger developmental process. Resilience is not the only prosocial trait linked to PYD and ACE buffering effects. The 5 Cs are the leading model of PYD and are traits that have been shown to lessen ACE exposure effects. Connection has been shown to have a particularly strong buffering effect on anxiety, depression, and social disconnection in ACE­exposed samples (Crouch et al., 2019). Overall, the consensus of the various fit statistics suggests that the current model is a moderate fit. Although there may be room for further expansion, the present fit endorses previous research claiming extracurricular activities facilitate prosocial development that has lasting behavioral effects and buffers consequences from ACE exposure (Lerner, 2004; Romer & Hansen, 2021; Schaefer et al., 2018).

Notably, the current study contributes to the scarce literature examining the long­term impact of extracurricular involvement. Gardner et al. (2008) performed the first life outcome study which found adolescent extracurricular participation did significantly correlate to educational attainment, career, and civic engagement. However, Gardner did not explore trait outcomes. Building upon Gardner et al. (2008), the current study’s findings suggest that extracurricular involvement may indirectly encourage favorable life outcomes like higher educational attainment via directly developing resilience and prosocial traits that last into emerging adulthood.

Limitations

Data collection occurred through internet ­ assisted snowball sampling and a single free­recall format due to location and testing duration limits. This sampling method created a large sample which improved statistical strength, but it may have limited generalizability. Similarly, this method was effective in reaching a geographically diverse sample, but possible selection bias from recruitment within internet social networks created a sample with particularly high college GPAs. There also may have been self ­ selection bias where possible participants only responded if they were confident with their academic performance in emerging adulthood. Given the sample of high­achieving emerging adults, the collected data most likely had restricted ranges for educational outcomes. For example, other studies had more drastic negative academic impacts for ACE­exposed individuals, yet the current study did not show a significant correlation between ACEs and GPA (Mahoney & Cairns, 1997; Silva et al., 2024). Because this

study collected college GPA, those who may have had larger academic impacts may have self­selected out of attending college and thus this impact is not reflected in current data. Where GPA calculations may fall short, collected educational attainment data provides insight into participants who may have self ­ selected out of higher education. Moreover, the single recall format may have impacted data accuracy, especially for young adults retrospectively reporting adolescent extracurricular participation. Although the study assumes accurate recall, improper recollection could have affected subsequent data analysis. The sample was restricted to U.S. emerging adults, and results should only be generalized to that population. Likewise, the current study did not account for socioeconomics which has the potential to influence extracurricular participation and education (Gardner et al., 2012). Previous data controlling for socioeconomic status have supported the PYD framework, but status was not included in the current path analysis (Gardner et al., 2008; Martin et al., 2013).

Additionally, many correlations were found to be significant, but their strengths were not strong. Despite weak correlational strengths, these relationships were found to be naturally occurring across all extracurricular activities. Given the wide spectrum of accepted data and naturalistic approach, any significant effect provides meaningful theoretical support for these activities’ potential.

Finally, the current study only collected data on resilience for prosocial traits. Although literature links resilience to PYD and ACE exposure buffering, it is not the only prosocial trait to contribute to these outcomes (Hamill, 2003; Howell et al., 2021; Schaefer et al., 2018). The path analysis could have reached greater strength if a wider range of advantageous prosocial traits were included in the model.

Directions for Future Research

Future research to be performed analyzing the longevity of extracurricular participation’s benefits into emerging adulthood and how these activities can be intentionally utilized in schools to promote long­term success. The current study had a single data collection period for a retrospective, self­report study. Longitudinal studies with multiple data collection points and more focused extracurricular inclusion parameters may result in more robust data. Data collection through tangible means such as employment and academic records is suggested to complement self­report data. Likewise, exploration of other life outcomes such as employment, finances, and belief systems and expansion into other prosocial attitudes or socioeconomic status may provide insightful context. Elaboration on self­efficacy measures may also provide insight into the development of these attitudes

and behaviors. Specifically possible prosocial development to buffer binge drinking behaviors may encourage more responsible social drinking.

Furthermore, experimental research into adding intentional prosocial interventions into the framework of pre­existing extracurriculars could provide a means of accessible and readily engaged with prosocial development within schools. Socioeconomic status can restrict youths from participation and receiving developmental benefits (Gardner et al., 2012). Despite varying levels of resources, most schools do provide funding for athletics and arts programming. Providing psychoeducation and basic, easy to implement resiliency skills to these established coaches and directors could provide an avenue to reach students in schools that are typically underserved. This research would be a proactive approach to providing schools with nonclinical, prosocial development tools. These resiliency skills could be taught to improve performance in the extracurricular activity. As they are employed during the activity, they are then adopted into the youths’ repertoire and can be called upon when needed to handle the stressors of daily life.

Preliminary versions of this concept are already being tested within high school athletics (Bryant, 2016). Through the 2010s into early 2020s, some youth sport coaches started to adopt and use “mental skills” for their athletes. This involved quick lessons during practices on sport psychology tools such as visualization, goal setting, and mindful breathing for gametime use. This movement gained momentum within coaching circles and some coaches began to develop their own curriculums (Bryant, 2016). This pursuit does hold promise as systematic review of short, tool ­ focused sports psychology interventions did find these lessons provided meaningful enhancement for high performing athletes (Reyes­Bossio et al., 2022).

Conclusion

In all, the current study aimed to examine the possible impact of extracurricular activity and resulting protective factors’ longevity into emerging adulthood. Specifically, resilience was explored as a possible protective factor developed within these extracurriculars that could be employed during emerging adulthood to achieve greater educational outcomes and resist risky drinking behaviors. Adverse Childhood Experiences were also examined within this dynamic as their occurrence is common throughout the school population and has been historically recognized as significant risk factors. Previous research provides many examples of resilience development through extracurricular participation and correlation with academic and behavioral benefits. These benefits extended to those who were exposed to multiple Adverse Childhood Experiences.

However, there is a current lack of research into the longevity of these benefits. The current study hopes to minimize this literature gap. The results suggest that extracurricular participation and resilience do significantly correlate. Moreover, resilience had significant positive correlations with educational attainment and safe drinking behaviors even after adolescent participation ceased. The data was meaningfully fit to a model via path analysis that suggests that resilience developed through extracurricular participation and impacted by ACEs has a significant positive effect on educational outcomes and risk­adverse behavior into emerging adulthood. The findings may be valuable in encouraging more meaningful participation and the possible implementation of intentional prosocial skill­building lessons within established extracurriculars to enhance positive youth development in an accessible manner.

References

Agans, J. P., Champine, R. B., DeSouza, L. M., Mueller, M. K., Johnson, S. K., & Lerner, R. M. (2014). Activity involvement as an ecological asset: Profiles of participation and youth outcomes. Journal of Youth and Adolescence, 43(6), 919–932. https://doi.org/10.1007/s10964-014-0091-1

Almeida, L., Dias, T., Corte-Real, N., Menezes, I., & Fonseca, A. (2023). Positive youth development through sport and physical education: A systematic review of empirical research conducted with grade 5 to 12 children and youth. Physical Education and Sport Pedagogy, 0(0), 1–27. https://doi.org/10.1080/17408989.2023.2230208

Antelo, E., Saldaña, O., & Carballeira, A. (2021). The impact of group psychological abuse on distress: The mediating role of social functioning and resilience. European Journal of Psychotraumatology, 12, 1954776. https://doi.org/10.1080/20008198.2021.1954776

Arnett, J. (2014). Emerging adulthood: The winding road from the late teens through the twenties (2nd edition) https://doi.org/10.1093/acprof:oso/9780199929382.001.0001

Bandura, A., Caprara, G. V., Barbaranelli, C., Gerbino, M., & Pastorelli, C. (2003). Role of affective self-regulatory efficacy in diverse spheres of psychosocial functioning. Child Development, 74(3), Article 3. https://doi.org/10.1111/1467-8624.00567

Bethell, C. D., Newacheck, P., Hawes, E., & Halfon, N. (2014). Adverse childhood experiences: Assessing the impact on health and school engagement and the mitigating role of resilience. Health Affairs (Project Hope), 33(12), 2106–2115. https://doi.org/10.1377/hlthaff.2014.0914

Bering, H., & Schulz, W. (2024). Can sports and music level the playing field? Adolescents’ extracurricular activities and the reproduction of social inequalities in cognitive skills. Research in Social Stratification and Mobility, 90, 100895. https://doi.org/10.1016/j.rssm.2024.100895

Bowers, E. P., Li, Y., Kiely, M. K., Brittian, A., Lerner, J. V., & Lerner, R. M. (2010). The Five Cs model of positive youth development: A longitudinal analysis of confirmatory factor structure and measurement invariance. Journal of Youth and Adolescence, 39(7), 720–735. https://doi.org/10.1007/s10964-010-9530-9 Bryant. (2016). Mental skills training: high school athletes need it. Coaches Toolbox. https://www.coachestoolbox.net/mental-toughness/mental-skillstraining-high-school-athletes-need-it Bush, K., Kivlahan, D. R., McDonell, M. B., Fihn, S. D., & Bradley, K. A. (1998). The AUDIT alcohol consumption questions (AUDIT-C): An effective brief screening test for problem drinking. Ambulatory Care Quality Improvement Project (ACQUIP). Alcohol Use Disorders Identification Test. Archives of Internal Medicine, 158(16), 1789–1795. https://doi.org/10.1001/archinte.158.16.1789

Busseri, M., Rose-Krasnor, L., Willoughby, T., & Chalmers, H. (2006). A longitudinal examination of breadth and intensity of youth activity involvement and successful development. Developmental Psychology 42, 1313–1326. https://doi.org/10.1037/0012-1649.42.6.1313

Caldwell, L., & Witt, P. (2011). Leisure, recreation, and play from a developmental context. New Directions for Youth Development, 2011, 13–27. https://doi.org/10.1002/yd.394

Campbell-Sills, L., Cohan, S. L., & Stein, M. B. (2006). Relationship of resilience to personality, coping, and psychiatric symptoms in young adults. Behaviour Research and Therapy, 44(4), 585–599. https://doi.org/10.1016/j.brat.2005.05.001

Campbell-Sills, L., & Stein, M. B. (2007). Psychometric analysis and refinement of the Connor-Davidson Resilience Scale (CD-RISC): Validation of a 10-item measure of resilience. Journal of Traumatic Stress, 20(6), 1019–1028. https://doi.org/10.1002/jts.20271

Carter, J. S., McNair, G., Bushnell, A., Saldana, L., & Grant, K. E. (2023). Sports participation, frequency, and competence differentially impact youth depressive, anxious, and somatic symptoms: gender, neighborhood, and sports type effects. Mental Health and Physical Activity, 25 https://doi.org/10.1016/j.mhpa.2023.100562

Choi, E., Lee, O., Park, J.-J., & Lee, O. (2015). The influence of a sports mentoring program on children’s life skills development. Journal of Physical Education and Sport, 15(2), 2264–2271. https://doi.org/10.7752/jpes.2015.02040

Connor, K. M., & Davidson, J. R. T. (2003). Development of a new resilience scale: The Connor-Davidson Resilience Scale (CD-RISC). Depression and Anxiety, 18(2), 76–82. https://doi.org/10.1002/da.10113

Crosnoe, R., Johnson, M. K., & ElderJr., G. H. (2004). Intergenerational bonding in school: The behavioral and contextual correlates of student–teacher relationships. Sociology of Education, 77(1), 60–81. https://doi.org/10.1177/003804070407700103

Crouch, E., Radcliff, E., Strompolis, M., & Srivastav, A. (2019). Safe, stable, and nurtured: Protective factors against poor physical and mental health outcomes following exposure to adverse childhood experiences (ACEs). Journal of Child & Adolescent Trauma, 12(2), 165–173. https://doi.org/10.1007/s40653-018-0217-9

Cusack, S. E., Bountress, K. E., Sheerin, C. M., Spit for Science Work Group, Dick, D. M., & Amstadter, A. B. (2023). The longitudinal buffering effects of resilience on alcohol use outcomes. Psychological Trauma: Theory, Research, Practice, and Policy, 15(6), 1000–1011. https://doi.org/10.1037/tra0001156

DeMartini, K. S., & Carey, K. B. (2012). Optimizing the use of the AUDIT for alcohol screening in college students. Psychological Assessment, 24(4), 954–963. https://doi.org/10.1037/a0028519

Dworkin, J. (2007). Adolescents’ negative experiences in organized youth activities. Journal of Youth Development, 1(3), 44–62. https://doi.org/10.5195/jyd.2007.373

Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., Koss, M. P., & Marks, J. S. (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. American Journal of Preventive Medicine, 14(4), 245–258. https://doi.org/10.1016/S0749-3797(98)00017-8

Felix, A. S., Lehman, A., Nolan, T. S., Sealy-Jefferson, S., Breathett, K., Hood, D. B., Addison, D., Anderson, C. M., Cené, C. W., Warren, B. J., Jackson, R., & Williams, K. P. (2019). Stress, resilience and cardiovascular disease risk among African-American women: Results from the Women’s Health Initiative. Circulation. Cardiovascular Quality and Outcomes, 12(4), e005284. https://doi.org/10.1161/CIRCOUTCOMES.118.005284

Fischer, N., Steiner, C., & Theis, D. (2020). How many, how long, how diverse? Measuring and analysing participation in extracurricular activities. In International developments in research on extended education: Perspectives on extracurricular activities, after-school programmes, and all-day schools. (pp. 87–113). Verlag Barbara Budrich. Forneris, T., Camiré, M., & Williamson, R. (2015). Extracurricular activity participation and the acquisition of developmental assets: Differences between involved and noninvolved Canadian high school students. Applied Developmental Science, 19(1), 47–55. https://doi.org/10.1080/10888691.2014.980580

Gardner, M., Browning, C., & Brooks-Gunn, J. (2012). Can organized youth activities protect against internalizing problems among adolescents living in violent homes. Journal of Research on Adolescence, 22(4), 662–677. https://doi.org/10.1111/J.1532-7795.2012.00811.X

Gardner, M., Roth, J., & Brooks-Gunn, J. (2008). Adolescents’ participation in organized activities and developmental success 2 and 8 years after high school: Do sponsorship, duration, and intensity matter? Developmental Psychology, 44(3), 814–830. https://doi.org/10.1037/0012-1649.44.3.814

Gestsdottir, S., Bowers, E., & von Eye, A. (2010). Intentional self regulation in middle adolescence: The emerging role of loss-based selection in positive youth development. Journal of Youth and Adolescence, 39(7), 764–782. https://doi.org/10.1007/s10964-010-9537-2

Gestsdottir, S., & Lerner, R. M. (2007). Intentional self-regulation and positive youth development in early adolescence: Findings from the 4-h study of positive youth development. Developmental Psychology

https://www.academia.edu/52780386/Intentional_self_regulation_and_ positive_youth_development_in_early_adolescence_Findings_from_ the_4_h_study_of_positive_youth_development

Grover, H. M., Limber, S. P., & Boberiene, L. V. (2015). Does it matter if students experience school as a place of community? American Journal of Orthopsychiatry, 85(6, Suppl), S79–S85. https://doi.org/10.1037/ort0000131

Haider, Z. F., & von Stumm, S. (2022). Predicting educational and socialemotional outcomes in emerging adulthood from intelligence, personality, and socioeconomic status. Journal of Personality and Social Psychology, 123(6), 1386–1406. https://doi.org/10.1037/pspp0000421

Hamill, S. (2003). Resilience and self-efficacy: The importance of efficacy beliefs and coping mechanisms in resilient adolescents. Colgate University Journal of the Sciences, 35. Hoff, K. A., Einarsdóttir, S., Chu, C., Briley, D. A., & Rounds, J. (2021). Personality changes predict early career outcomes: Discovery and Replication in 12year longitudinal studies. Psychological Science, 32(1), 64–79. https://doi.org/10.1177/0956797620957998

Hoffmann, M. D., Barnes, J. D., Tremblay, M. S., & Guerrero, M. D. (2022). Associations between organized sport participation and mental health difficulties: Data from over 11,000 US children and adolescents. PLoS ONE, 17(6), e0268583. https://doi.org/10.1371/journal.pone.0268583

Howell, K. H., Miller-Graff, L. E., Martinez-Torteya, C., Napier, T. R., & Carney, J. R. (2021). Charting a course towards resilience following adverse childhood experiences: Addressing intergenerational trauma via strengths-based intervention. Children, 8(10), 844. https://doi.org/10.3390/children8100844

Jimenez, M. E., Wade, R., Lin, Y., Morrow, L. M., & Reichman, N. E. (2016). Adverse experiences in early childhood and kindergarten outcomes. Pediatrics, 137(2), e20151839. https://doi.org/10.1542/peds.2015-1839

Kline, R. B. (2011). Principles and practice of structural equation modeling, 3rd ed (pp. xvi, 427). Guilford Press.

Konaszewski, K., Kolemba, M., & Niesiobędzka, M. (2021). Resilience, sense of coherence and self-efficacy as predictors of stress coping style among university students. Current Psychology, 40(8), 4052–4062. https://doi.org/10.1007/s12144-019-00363-1

Kort-Butler, L., & Martin, D. (2015). The influence of high school activity portfolios on risky behaviors in emerging adulthood. Justice Quarterly, 32(3), 381–409. https://doi.org/10.1080/07418825.2013.770547

Kuiper, H., van Leeuwen, C. C. M., Stolwijk-Swüste, J. M., & Post, M. W. M. (2019). Measuring resilience with the Connor-Davidson Resilience Scale (CDRISC): Which version to choose? Spinal Cord, 57(5), 360–366. https://doi.org/10.1038/s41393-019-0240-1

Lerner, R. M. (2004). Liberty: Thriving and civic engagement among America’s youth (pp. xxi, 203). Sage Publications, Inc.

Mahoney, J. L., & Cairns, R. B. (1997). Do extracurricular activities protect against early school dropout? Developmental Psychology, 33(2), 241–253. https://doi.org/10.1037/0012-1649.33.2.241

Martin, A. J., Mansour, M., Anderson, M., Liem, G. A. D., & Sudmalis, D. (2013). The role of arts participation in students’ academic and nonacademic outcomes: A longitudinal study of school, home, and community factors. Journal of Educational Psychology, 105(3), 709–727. https://doi.org/10.1037/A0032795

Mayol-García, Y. (2022). Girls take lessons, join clubs more often than boys but boys play more sports Census.Gov. https://www.census.gov/library/stories/2022/07/children-continue-to-beinvolved-in-extracurricular-activities.html

Narayan, A. J., Lieberman, A. F., & Masten, A. S. (2021). Intergenerational transmission and prevention of adverse childhood experiences (ACEs). Clinical Psychology Review, 85, 101997. https://doi.org/10.1016/j.cpr.2021.101997

Oláh, B., Fekete, Z., Kuritárné Szabó, I., & Kovács-Tóth, B. (2023). Validity and reliability of the 10-Item Adverse Childhood Experiences Questionnaire (ACE-10) among adolescents in the child welfare system. Frontiers in Public Health, 11. https://doi.org/10.3389/fpubh.2023.1258798

Park, J. J., Zheng, J., & Kim, B. H. (2023). Unveiling the inequalities of extracurricular activities in admissions. Change: The Magazine of Higher Learning, 55(5), 45–54. https://doi.org/10.1080/00091383.2023.2235253

Pérez-García, J. M., Suárez-Suárez, S., Doallo, S., & Cadaveira, F. (2022). Effects of

binge drinking during adolescence and emerging adulthood on the brain: A systematic review of neuroimaging studies. Neuroscience & Biobehavioral Reviews, 137, 104637. https://doi.org/10.1016/j.neubiorev.2022.104637

Qamar, S., & Akhter, M. (2020). Relationship between Students’ self-efficacy and resilience at secondary school level. Bulletin of Education and Research, 42(3), 215–224.

Reinert, D. F., & Allen, J. P. (2007). The Alcohol Use Disorders Identification Test: An update of research findings. Alcoholism: Clinical and Experimental Research, 31(2), 185–199. https://doi.org/10.1111/j.1530-0277.2006.00295.x

Reyes-Bossio, M., Corcuera-Bustamante, S., Veliz-Salinas, G., Junior, M. V. B., Delgado-Campusano, M., Brocca-Alvarado, P., Caycho-Rodríguez, T., CasasApayco, L., Tutte-Vallarino, V., Carbajal-León, C., & Brandão, R. (2022). Effects of psychological interventions on high sports performance: A systematic review. Frontiers in Psychology, 13, 1068376. https://doi.org/10.3389/fpsyg.2022.1068376

Romer, D., & Hansen, D. (2021). Positive youth development in education (pp. 75–108). Palgrave Macmillan, Cham. https://doi.org/10.1007/978-3-030-64537-3_4

Schaefer, L., Howell, K., Schwartz, L., Bottomley, J., & Crossnine, C. (2018). A concurrent examination of protective factors associated with resilience and posttraumatic growth following childhood victimization. Child Abuse & Neglect, 85 https://doi.org/10.1016/j.chiabu.2018.08.019

Schulenberg, J. E., Johnston, L. D., O’Malley, P. M., Bachman, J. G., Miech, R. A., & Patrick, M. E. (2020). Monitoring the future national survey results on drug use, 1975-2019. Volume II, college students & adults ages 19-60. Institute for Social Research. https://eric.ed.gov/?id=ED608266

Senaratne, D. N. S., Thakkar, B., Smith, B. H., Hales, T. G., Marryat, L., & Colvin, L. A. (2024). The impact of adverse childhood experiences on multimorbidity: A systematic review and meta-analysis. BMC Medicine, 22(1), 1–14. https://doi.org/10.1186/s12916-024-03505-w

Silva, C., Moreira, P., Moreira, D. S., Rafael, F., Rodrigues, A., Leite, Â., Lopes, S., & Moreira, D. (2024). Impact of adverse childhood experiences in young adults and adults: A systematic literature review. Pediatric Reports, 16(2), 461–481. https://doi.org/10.3390/pediatric16020040

Thouin, É., Dupéré, V., Dion, E., McCabe, J., Denault, A.-S., Archambault, I., Brière, F. N., Leventhal, T., & Crosnoe, R. (2022). School-based extracurricular activity involvement and high school dropout among at-risk students: Consistency matters. Applied Developmental Science 26(2), 303–316. https://doi.org/10.1080/10888691.2020.1796665

U.S. Census Bureau, D. C. (2023). Median age at first marriage: 1890 to present. 1890 to 1940, and Current Population Survey, Annual Social and Economic Supplements, 1947 to 2023 census.gov

Wechsler, H., & Kuo, M. (2000). College students define binge drinking and estimate its prevalence: Results of a national survey. Journal of American College Health, 49(2), 57–64. https://doi.org/10.1080/07448480009596285

Wechsler, H., & Nelson, T. (2008). What we have learned from the Harvard School of Public Health College Alcohol Study: Focusing attention on college student alcohol consumption and the environmental conditions that promote it. https://www.academia.edu/15875009

Weitzman, E., & Chen, Y. (2005). Risk modifying effect of social capital on measures of heavy alcohol consumption, alcohol abuse, harms, and secondhand effects: National survey findings. Journal of Epidemiology and Community Health, 59(4), 303–309. https://doi.org/10.1136/jech.2004.024711

Wills, T. A., Sandy, J. M., Yaeger, A., & Shinar, O. (2001). Family risk factors and adolescent substance use: Moderation effects for temperament dimensions. Developmental Psychology, 37(3), 283–297. https://doi.org/10.1037/0012-1649.37.3.283

Author Note.

Ava J. Avolio https://orcid.org/0009­0002­5445­9549 Materials for this study can be accessed at https://osf.io/yngw5/. There are no known conflicts of interest to disclose. Special thanks to Jennifer Roth for supervising this project. Correspondence concerning this article should be addressed to Ava J. Avolio, Department of Psychology, Counseling, and Criminology, Carlow University, 3333 Fifth Ave, Pittsburgh, PA 15213, United States. Email: ajavolio@live.carlow.edu

A Comparison of AI and Human-Driven Assistance on Help-Seeking Intention: The Mediating Role of Perceived Emotional Support

Lee Yan Ying Department of Psychology, HELP University

ABSTRACT. This between­subjects lab experimental design examined whether perceived emotional support mediated the relationship between types of emotional support providers and help­seeking intention. Sixty undergraduates, aged 19 to 24, reflected on and discussed a recent issue with either AI chatbots or humans. They completed the Perceived Emotional Support Scale and the General Help­Seeking Questionnaire. The first hypothesis that individuals who received emotional support from humans perceived greater emotional support than those receiving emotional support from AI chatbots, was not supported (r = ­.22, 95% CI [­12.64, 1.11], p = .099). As hypothesized, perceived emotional support predicted help­seeking intention (r = .34, 95% CI [0.06, 0.37], p = .009). Perceived emotional support did not mediate the relationship between types of emotional support providers and help­seeking intention ( BCa CI [ ­ 3.51, 0.11] ); the third hypothesis was not supported. These findings may contribute to the integration of AI chatbots into support services.

Keywords: AI, chatbot, perceived emotional support, help­seeking intention, human–AI communicationhabitant

摘要：

本研究采用受试者间实验室实验设计，探讨感知到的情感支持是否 在情感支持提供者类型与求助意图之间起中介作用。六十名年龄介 于19至24岁的大学生参与研究，他们需回顾并与人工智能聊天机 器人或真人讨论近期的一项烦恼。参与者随后填写了《感知情感支 持量表》和《一般求助意向问卷》。第一项假设认为，相较于接受人工 智能聊天机器人支持的个体，接受真人情感支持的个体会感知到更 高程度的情感支持，然而该假设未获支持。与假设一致，感知到的情 感支持能够预测个体的求助意图。然而，感知到的情感支持并未在 情感支持提供者类型与求助意图之间起中介作用，第三项假设亦未 获得支持。本研究结果或可为人工智能聊天机器人在支持服务中的 整合提供理论依据。

关键词：人工智能、聊天机器人、感知情感支持、求助意图、人机沟通

The COVID ­ 19 pandemic led to an increase in reported mental health problems, highlighting the growing need for emotional support (Robinson et al., 2022). Since the onset of the COVID­19 pandemic, the global prevalence of mental health issues was 50% for psychological distress, 36.5% for stress, 28% for depression, and 26.9% for anxiety, all of which have been linked to rising suicide rates (Lew et al., 2022; Nochaiwong et al., 2021). Despite this, help­seeking rates remain low, and mental health services continue to receive little public attention due to barriers, namely perceived stigma, feelings of shame, and limited accessibility (Duong et al., 2021; Kim & Lee, 2022). Help­seeking intention is an individual’s conscious decision and willingness to seek support for personal, emotional, or mental health difficulties (Rickwood et al., 2005). Low help­seeking intention negatively impacts an individual’s mental health and well­being, leading to poorer psychological outcomes, increased distress, and reduced social connectedness (Ratnayake & Hyde, 2019).

Another barrier to help­seeking intention is the lack of social networks, which can lead to feelings of isolation (Kim & Lee, 2022). Individuals facing difficulties often seek social support from others in hopes of receiving encouragement, empathy, or resources. However, traditional social support channels have limitations, including time constraints, privacy concerns, and stigma, which discourage individuals from seeking help (Aguirre Velasco et al., 2020; Pretorius et al., 2019). The rapid advancement of artificial intelligence (AI) technology has introduced an alternative means of obtaining social support. AI is a field of science dedicated to studying and engineering the development of intelligent machines (Luxton, 2016). With its ability to process large amounts of data, analyse human emotions, and generate human­like language, AI has been considered a potential alternative for providing online social support (D’Alfonso, 2020). Therefore, AI­based applications may serve as an additional method for delivering social support through technology.

Social Support

Social support refers to any resources provided by an individual’s social network that enhance their ability to cope with difficulties (Cohen & Wills, 1985). It can be categorized into instrumental support and emotional support. Instrumental support consists of tangible assistance aimed at problem ­ solving (Adams et al., 1996), whereas emotional support involves expressing empathy, understanding, concern, or encouragement for another person (Cohen & Wills, 1985). These two forms of social support influence an individual’s wellbeing differently. Compared to instrumental support,

emotional support has been found to be a stronger predictor of greater well­being, higher marital satisfaction, and reduced work­family conflict (Sheikh et al., 2016; Uysal Irak et al., 2020; Yedirir & Hamarta, 2015). This may be because emotional support fulfils individuals’ fundamental needs for reassurance, validation, and a sense of being cared for.

Emotional support can be provided through open, compassionate, and accepting communication to convey messages that help regulate the emotional state of recipients (Burleson, 2008). It is a vital relational resource, as it facilitates relationship closeness with listeners, enhances subjective well­being, and provides emotional relief (Pauw et al., 2018; Pauw et al., 2022; Roksa & Kinsley, 2019). The positive impact of emotional support may depend on perceived emotional support, which refers to an individual’s subjective perception of the quality of emotional support received from others (Cogan et al., 2022; Haber et al., 2007). When individuals perceive emotional support as genuine, they are more likely to express their concerns openly, experience psychological safety, and feel heard (Yan, 2020). The perception of emotional support may play a crucial role in promoting favourable outcomes. Individuals who perceived emotional support as effective reported better mental well­being (Umucu & Lee, 2020). Additionally, perceived emotional support has been associated with reduced psychological distress, including lower levels of depression, anxiety (García‐Torres et al., 2020), and stress (Woodhead et al., 2016). Beyond mitigating distress, perceived emotional support also strengthened protective factors, for example, effective coping strategies (Namkoong et al., 2013), resilience (Hu et al., 2022), and life satisfaction (Peng et al., 2019). However, despite the continued emphasis on the importance of perceived emotional support, there remains a lack of evidence regarding its impact on prospective recipients' help­seeking intention.

The Role of AI Chatbots

AI technologies have become an integral part of modern daily life, appearing in various applications including automobiles, smart technology devices, and navigation systems. Their implementation enables faster information processing and more accurate predictions. Recently, AI has advanced to the point where it can process, analyse, and generate human­like language, a capability known as communicative AI. This type of AI is a learning machine that mimics human intelligence to process information and perform communication tasks (Frankish & Ramsey, 2014). An AI chatbot is a form of communicative AI that serves as an interpersonal communicator, engaging with individuals

through text­based interactions (Meng & Dai, 2021). AI chatbots have been widely implemented in customer service, daily text­based communication, and mental health services (Brandtzaeg et al., 2022; De Andrade & Tumelero, 2022). Notable examples, such as Replika and Xiaoice, have demonstrated significant progress in social interaction with humans (Skjuve et al., 2021; Zhou et al., 2020). Their human­like intelligence made them suitable as interlocutors or virtual companions, as they possessed social and empathetic conversational skills that enhanced user interactions (Skjuve et al., 2021; Zhou et al., 2020).

In recent years, an increasing number of individuals have sought emotional support from AI chatbots, particularly since the onset of the COVID ­ 19 pandemic, when mental health issues surged and access to professional support became more challenging (Kiron & Unruh, 2019). Although web­based mental health services have improved the delivery of mental health care and hold important potential to adapt to societal needs, challenges including long wait times continue to hinder timely access to telehealth (Ratheesh & AlvarezJimenez, 2022). Given these challenges, AI chatbots such as Replika are increasingly being used as alternative sources of emotional support due to their accessibility, anonymity, and availability outside regular therapy hours (Brandtzaeg et al., 2022; Malik et al., 2022). Additionally, individuals preferred AI chatbots when they feared stigma, lacked access to affordable care, or had no available human peer support (Sullivan et al., 2023). They often turned to AI chatbots during moments of emotional distress to vent negative emotions, seek emotional validation and support, or obtain coping suggestions (Andrade­Arenas et al., 2024).

Recent studies have highlighted the benefits of AI chatbots as companions in text­based conversations. In one study, interviewees described their reciprocal relationships with Replika as trusting, supportive, and secure after interacting with it for three months (Brandtzaeg et al., 2022). They noted that Replika’s high availability made it easier to nurture a sense of friendship with it. Some participants also reported that Replika exhibited empathetic and caring behaviours that fulfilled their emotional needs and reduced feelings of loneliness. Similarly, Sullivan et al. (2023) found that individuals who interacted with AI chatbots experienced decreased loneliness and anxiety, as well as increased emotional support and mental well­being. Participants also reported experiencing substantial emotional benefits (e.g., feeling better). These findings suggest that individuals who engage with AI chatbots may perceive emotional support similar to that received from human interactions.

The emotional support provided by AI chatbots

may play a role in encouraging individuals’ help­seeking intention. Planey et al. (2019) found that social networks motivate individuals to seek help. Similarly, human­AI friendship has been suggested as a new form of intimate relationship comparable to human­human friendship (Brandtzaeg et al., 2022). When interacting with users online, AI chatbots recognize emotions by detecting keywords and identifying problems raised, then respond with expressions of understanding and empathy that validate users’ experiences (Sullivan et al., 2023). According to the Computer Are Social Actors (CASA) framework, individuals perceive and react to emotional support from AI chatbots similarly to how they do with human interactions (Reeves & Nass, 1996). Furthermore, AI chatbots may be perceived as free from the biases that humans might hold, making individuals more willing to express their concerns (Luxton, 2016). This suggests that the emotional support provided by AI chatbots may be regarded by users as unbiased and genuine, helping users cope with negative emotions by reassuring them that they are not alone and that someone is willing to understand their situation (Arias et al., 1997).

Meanwhile, perceiving genuine emotional support may motivate individuals to seek help. Encouragement and empathy, as key components of emotional support, have been found to enhance help­seeking intention by making individuals feel that their struggles are taken seriously (Gulliver et al., 2012; Thompson et al., 2004; Vogel et al., 2007). AI chatbots may encourage individuals to confront difficulties positively and seek help by providing mental health­related information or available therapy contacts (Song, 2011). In a study by Shah et al. (2022), participants who interacted with an AI chatbot named “Alex,” which delivered empathetic and supportive responses, reported feeling understood, validated, and emotionally supported. They also indicated that the chatbot’s encouragement to seek help, along with its supportive interactions, prompted them to reflect on their mental health and consider professional support. These forms of emotional support may foster an environment in which individuals feel accepted and cared for (Cobb, 1976). These findings suggest that perceiving genuine emotional support from AI chatbots may inspire individuals to seek help.

AI Chatbot Versus Human as an Emotional Support Provider

Although the CASA framework suggests that individuals interact with AI chatbots as though they are social entities, it does not claim that individuals perceive or engage with AI chatbots in the same way they would with real humans across all contexts. Different types of emotional support providers, whether that is an AI chatbot or a

human, can influence individuals’ expectations and subsequently shape how they perceive and respond to the emotional support provided. These differing expectations may lead to unique psychological and emotional impacts that would not arise if individuals assumed the provider was always human.

Previous research has indicated that emotional support is perceived differently depending on whether it is provided by an AI chatbot or a human, highlighting the importance of understanding how these variations affect help­seeking intention. Smith and Masthoff (2018) found that participants preferred receiving emotional support from humans rather than virtual agents, despite the emotional support rendered by virtual agents being considered valuable. Other studies have also suggested that the type of emotional support provider influenced the positive impacts of perceived emotional support. Meng and Dai (2021) found that emotional support provided by a human led to greater perceived supportiveness, which in turn resulted in lower stress and worry compared to emotional support from a chatbot. Similarly, Medeiros et al. (2021) reported that participants who received support from a human perceived the highest level of emotional support, which was associated with reduced self­perceived stress, followed by those in the chatbot condition and those in the control group.

In contrast, Gelbrich et al. (2021) found that participants who received emotional support from an AI chatbot perceived greater warmth and demonstrated more persistence than those who received emotional support from a human. The authors explained that in a digital environment, the mere presence of a human assistant created warm impressions, thereby reduced the necessity for the human to provide additional emotional support. However, AI chatbots were generally perceived as aloof; therefore, when they provided emotional support, it effectively compensated for this perceived lack of warmth. Conversely, Ho et al. (2018) found that participants experienced similar and enhanced emotional, psychological, and rational benefits regardless of the identity of their support provider. These findings suggest that the perception of emotional support from different types of providers may influence psychological outcomes differently. Thus, different types of emotional support providers may shape how individuals perceive emotional support, affecting their help­seeking intention.

Theoretical Framework

According to the model of perceived understanding (Reis et al., 2017), the psychological benefits of emotional support depend on whether individuals perceive the emotional support they receive as genuine. When individuals feel that the provider truly understands them

and accepts their authentic selves and experiences, the emotional support is more likely to be effective. In this context, receiving emotional support from a human partner may be more advantageous, as individuals are more likely to view humans as genuine emotional support providers, capable of empathy and emotional support. Awareness of receiving emotional support from a human may further enhance perceived emotional support, leading individuals to experience a deep sense of emotional connection and the feelings that the provider genuinely empathizes with, cares for, encourages, and understands them on a fundamental level (Stein & Ohler, 2017). The perception of authentic emotional support fosters social inclusion and acceptance, as it reinforces an individual’s sense of belonging and emotional security (Reis et al., 2017). It also activates brain regions linked to connection and reward, ultimately strengthening motivation to pursue personal goals or adopt healthy coping strategies (e.g., enhancing their intention to seek help; Reis et al., 2017).

Conversely, communicating with AI chatbots may elicit a different reaction. This can be explained by the machine heuristic, a cognitive bias in which individuals make assumptions about interactions based on machinerelated attributes (Sundar & Kim, 2019). Specifically, when individuals recognize that the emotional support provider is a machine (i.e., an AI chatbot), they may hold preconceived notions that it is mechanical, impersonal, and emotionless (Sundar & Kim, 2019). They may also apply stereotypes suggesting that AI chatbots are incapable of emotional tasks—for instance, experiencing emotions or providing genuine emotional support—and that their responses are scripted (Madhavan et al., 2006; Sundar & Kim, 2019). This perception may, in turn, affect interactions, leading individuals to perceive emotional support from AI chatbots as inauthentic and ineffective. As a consequence of perceiving a lack of emotional support, individuals may feel unmotivated to seek help in the future (Ali et al., 2017).

Research Gaps

Past studies have focused on individual and contextual factors, specifically past positive help­seeking experiences, recognition of a problem, and positive attitudes toward help ­ seeking, as predictors of help ­ seeking intention (Aguirre Velasco et al., 2020; Ali et al., 2017; Randles & Finnegan, 2022). However, the enabling factor (i.e., perceived emotional support) has not been investigated as a predictor of help­seeking intention. Additionally, other enabling factors closely related to perceived emotional support, namely social support and encouragement, have been found to facilitate help­seeking intention (Aguirre Velasco et al., 2020;

Ali et al., 2017; Randles & Finnegan, 2022). Yet, most of these findings are based on correlational evidence, and the specific effect of perceived emotional support on help­seeking intention remains unclear (Magaard et al., 2017). Addressing this gap is critical, as persistently low help­seeking intention remains a major barrier to early intervention and treatment, despite increasing rates of mental health issues and suicide (Duong et al., 2021; Lew et al., 2022; Yasuhiro et al., 2021). Without understanding the factors that promote help­seeking intention, individuals in distress may continue to avoid seeking necessary support, leading to worsened mental health outcomes.

Past research has consistently indicated that a person’s social network is one of the primary factors influencing their help ­ seeking intention (Planey et al., 2019). However, these studies have predominantly focused on human­to­human interactions, leaving a gap in understanding how emerging forms of human­AI relationships may impact this dynamic (Planey et al., 2019). With the increasing prevalence of AI chatbots as emotional support providers and the growing acceptance of human­AI friendships, it is essential to explore whether perceived emotional support from AI chatbots influences individuals’ intention to seek help and how this may differ from emotional support provided by humans (Brandtzaeg et al., 2022). This is relevant as AI chatbots have become integrated into daily life, reshaping traditional notions of social network, emotional support and help­seeking intention. Understanding this relationship could provide valuable insights into the evolving role of AI in mental health and emotional well­being.

There is currently a lack of research comparing the differences and effects of perceived emotional support between AI chatbots and humans, with existing studies reporting inconsistent findings. Meng and Dai (2021) found that individuals who received emotional support from a human perceived greater supportiveness, leading to a reduction in certain psychological effects compared to those who received emotional support from a chatbot. In contrast, Gelbrich et al. (2021) found the opposite relationship, but Ho et al. (2018) suggested there was no significant difference. However, theoretical frameworks and past findings generally agree that perceiving emotional support from different providers may lead to distinct psychological outcomes (Gelbrich et al., 2021; Meng & Dai, 2021; Smith & Masthoff, 2018). The type of emotional support provider appears to influence perceived emotional support, which in turn affects helpseeking intention. Therefore, further empirical research is needed to clarify this relationship. Additionally, although existing studies suggest a difference between perceived emotional support from AI chatbots and

humans, little is known about how this difference influences help­seeking intention. Understanding the role of different types of emotional support providers as conversational partners in emotionally supportive communication may be a key factor in enhancing helpseeking intention.

Current Study

This study aimed to compare the perceived emotional support from an AI chatbot and a human partner regarding their effect on help­seeking intention. The research question was: Will perceived emotional support be a mediator of the relationship between types of emotional support providers and help­seeking intention? There were three hypotheses. The first hypothesis stated that individuals who received emotional support from a human partner would perceive greater emotional support than those who received emotional support from an AI chatbot. Second, it was hypothesised that there would be a positive relationship between perceived emotional support and help­seeking intention. Third, it was anticipated that perceived emotional support would mediate the relationship between types of emotional support providers and help­seeking intention (See Figure 1).

Design

A between ­ subjects lab experimental design was utilized, incorporating one independent variable (IV), one dependent variable (DV), and one mediator. The IV was the type of emotional support provider, with two levels: AI chatbot and human partner. The DV was help­seeking intention, and the mediator was perceived emotional support.

Participants

A G*Power calculation, based on a conventional medium effect size of 0.30, an alpha level of .05, and a power of .80, suggested a required sample size of 36

Statistical Diagram of the Relationship Between Types of Emotional Support Providers and Help-Seeking Intention as Mediated by Perceived Emotional Support

individuals (https://osf.io/9r4eh). However, to enhance statistical power and the robustness of the study’s findings, approximately 50% more participants ( n = 24) were added, resulting in a total sample size of 60.

Sixty psychology students (53 women and seven men), aged between 19 and 24 (M = 21.53, SD = 1.05), were recruited from a university in West Malaysia for the actual experiment. All participants were Malaysian, with 53 identifying as Chinese (88.3%), four as Indian (6.7%), one as mixed Chinese and Indian (1.7%), one as Iban (1.7%), and one as Eurasian (1.7%). A convenience sampling method was used, with the experiment advertised on the university’s experiment portal. Participants received 0.50% extra credit for their participation. For both the pilot and actual experiments, only individuals who met the following criteria were recruited: aged between 18 and 30, fluent in English, and having prior experience interacting with either an AI chatbot or a mental health practitioner to discuss personal or emotional problems. These criteria aimed to ensure that participants were genuinely willing to share their experiences and actively engage in the experiment. However, individuals who participated in the pilot test were not permitted to participate in the actual experiment.

Materials

Demographic Questionnaire

A demographic questionnaire was prepared for participants to provide their gender, age, ethnicity, nationality, and participant code. Additionally, participants were asked whether they had prior experience interacting with either an AI chatbot or a mental health practitioner to discuss personal problems and whether they had participated in the pilot study (https://osf.io/9r4eh).

Perceived Emotional Support

Perceived emotional support was measured using the Perceived Emotional Support Scale (PESS), which combined the Recipient Support Perception Scale (RSPS) and the Robot’s Perceived Empathy Scale (RoPE), resulting in a total of 20 items. This approach was chosen due to the absence of an existing scale that aligned with the context of the current study and the conceptualization of perceived emotional support. The RSPS was used to assess the extent to which individuals perceived their conversational partner’s ability to provide emotional support ( https://osf.io/9r4eh). Nine items from the nurturant support subscale of the RSPS were included, two of which were negatively worded. Sample items included “made me feel cared for” and “shared my concerns” (Trees, 2000). In Trees’s (2000) study, this subscale demonstrated good internal consistency, with Cronbach’s α = .87.

The RoPE scale consisted of two subcategories: empathic understanding (EU) and empathic response (ER), each comprising eight items (https://osf.io/9r4eh; Charrier et al., 2019). This scale was developed based on human empathy metrics. All items from the EU subscale and three items from the ER subscale were included. Sample items included “The robot encourages me” and “The robot knows me and my needs” (Charrier et al., 2019). The RoPE scale demonstrated excellent internal consistency, with Cronbach’s α = .91 (Reghunath, 2021).

In the Perceived Emotional Support Scale (PESS), all items were transformed into past tense to improve participant intuition, as the scale was completed post ­ interaction ( https://osf.io/9r4eh ). Responses were recorded on a 5­point Likert scale ranging from 1 ( strongly disagree ) to 5 ( strongly agree ). The scale was scored by calculating the total score, with specific items reverse­scored according to scoring instructions. Perceived emotional support was operationally defined as the total score on the PESS, where a higher score indicated greater perceived emotional support. The internal consistency of the scale in the current study was excellent, with Cronbach’s α = .92.

Help-Seeking Intention

The General Help­Seeking Questionnaire (GHSQ) was used to assess participants’ help­seeking intention from various social support sources (Wilson et al., 2005). The questionnaire consisted of 10 items, rated on a 7­point Likert scale ranging from 1 (extremely unlikely) to 7 (extremely likely; https://osf.io/9r4eh). To align with the context of the study, item 8 “teacher” was replaced with “lecturer,” and item 9 “someone else” was replaced with “AI chatbot.” Participants were asked to indicate their likelihood of seeking help from nine different sources of support (e.g., partners and friends) within the next four weeks (Wilson et al., 2005). The questionnaire was scored by aggregating all item responses, with specific items reverse­scored according to the scoring instructions. A higher total score indicated a greater intention to seek help (Wilson et al., 2005). Ibrahim et al. (2019) reported acceptable internal consistency for the scale, with Cronbach’s α = .75. However, in the present study, the overall internal consistency was lower, with Cronbach’s α = .64.

Emotional Support Providers

An AI chatbot, Chatfuel (www.chatfuel.com), integrated with ChatGPT, was used to provide emotional support by identifying keywords in participant inputs and generating responses accordingly. The chatbot was activated in Facebook Messenger to initiate conversations. Meanwhile, five trainee clinical psychologists or counsellors were recruited as part of the research team to serve as human

partners. All human partners were blinded to the study’s aim and were informed that their role was solely to provide emotional support during recruitment.

In line with Meng and Dai’s (2021) study, both emotional support providers underwent training before the experiment and followed the same predefined script to ask questions and respond to participants by identifying relevant keywords. The predefined script was taken directly from Meng and Dai’s (2021) study and implemented in the current research. The script included one greeting message, six questions, and six corresponding emotional support responses relevant to each question ( https://osf.io/9r4eh ). Emotional support providers first asked a question that encouraged participants to disclose their problems. They then responded by offering emotional support. This exchange constituted one turn in the conversation, with a total of six turns per interaction.

In the current study, human partners were granted some flexibility to modify the predefined script when responding to participants. This adjustment was made because participants’ responses might not always align perfectly with the scripted responses, necessitating some degree of personalization and adaptability. This approach allowed for natural variation while maintaining consistency across conditions.

Procedure

Pilot Test

Ethical approval from the review board was obtained before data collection. A pilot test was conducted to assess the applicability of the AI chatbot. Nine female participants aged between 20 and 22 (M = 21.22, SD = 0.94) were recruited. Eight participants were Malaysian (88.9%), and one was Indonesian (11.1%), with all identifying as Chinese. During the pilot test, participants signed an informed consent form and completed a demographic questionnaire. They were then asked to reflect on and describe a recent issue that had caused them to experience negative emotions, with a minimum requirement of 100 words, within eight minutes (https://osf.io/9r4eh). After completing the reflection, participants interacted with the AI chatbot via Facebook Messenger by clicking the “Get Started” entry point to initiate the conversation. Finally, they completed the scale and questionnaire. The entire study lasted approximately 55 minutes.

Following the pilot study, two issues were identified and addressed. The first issue was that participants did not follow the structured flow of the conversation. Some participants immediately shared their issues without first responding to the chatbot’s greeting message, and others provided multiple responses to a single question, disrupting the planned conversation flow. To address this, detailed instructions were provided in the main

experiment to participants through a slide presentation, which included a short example of a proper conversation. The second issue was a technical problem in which the AI chatbot failed to launch on certain smartphones when participants clicked the “Get Started” entry point. To resolve this, all conversations in the main experiment were conducted using laptops.

Main Experiment

After receiving ethical approval from the review board, an advertisement was posted on the university’s experiment portal, inviting undergraduate students to voluntarily participate in the study. Interested individuals who met the selection criteria and signed up on the portal received an email containing their participant code and experiment details, including their assigned group. Participants in Group A were assigned to the AI chatbot condition, and those in Group B were assigned to the human condition. They then selected and registered for a session based on their availability. Participants were randomly assigned to one of the two conditions, with 30 placed in the AI chatbot condition and 30 in the human condition. The procedure during the session was identical to that of the pilot test, except that after the reflection task, participants were instructed to engage in a conversation about their issue either with an AI chatbot or a human partner.

Results

Participants in both groups reported moderate levels of perceived emotional support and help­seeking intention (see Table 1). However, participants in the human condition perceived greater emotional support and exhibited a higher intention to seek help compared to those in the AI chatbot condition. Pearson’s correlation analyses indicated that perceived emotional support was positively associated with help­seeking intention.

TABLE 1

Mean Scores, Standard Deviations, and Correlations for the Types of Emotional Support Providers, Perceived Emotional Support, and Help-Seeking Intention

Independent Samples T Test

The Shapiro­Wilk test revealed that perceived emotional support scores for participants in the AI chatbot condition, Shapiro ­ Wilk (30) = .94, p = .089, and the human condition, Shapiro ­ Wilk (30) = .98, p = .778, were normally distributed. The assumption of normality was met. Levene’s test showed that the assumption of homogeneity of variances was met, F = 0.05, p = .825. The independent samples t test showed no statistically significant difference in perceived emotional support between participants in AI chatbot and human groups, with t(58) = ­1.68, p = .099, 95% CI [­12.64, 1.11]. Thus, the first hypothesis that individuals who received emotional support from a human partner would perceive greater emotional support than those who received emotional support from an AI chatbot was not supported.

Exploratory Factor Analysis (EFA)

Due to the combination of items from two scales, an exploratory factor analysis (EFA) was conducted to identify dimensions within the Perceived Emotional Support Scale. The pre­test process concluded with the Kaiser­Meyer­Olkin index value of .78, and Bartlett’s Test of Sphericity was significant (p < .001). Thus, the sample size of 60 was robust and could confidently continue with the results of the exploratory factor analysis. Five factors were selected, following the eigenvalues greater than 1.0 rule. The cumulative variance of the five factors had eigenvalues of 69.15%, accounting for more than 60% of the total variance indicators (see Table 2). The scree plot results showed five plots before the point of inflexion, which denoted the existence of the five factors.

Given that some items were highly loaded on multiple factors, an oblique factor solution was utilized via the direct oblimin with the Kaiser normalization method, as the factors were assumed to be theoretically correlated. After that, item 15 was removed. This cross­loading might be due to participants having read or understood the item in a way that was different from how the researcher intended to. Then, the EFA was rerun and the results showed that no items needed to be excluded, indicating a simple structure.

The items with values significant to their corresponding factors were found. For the specific item classification, please refer to https://osf.io/9r4eh. Five factors were involved: understanding and support, empathy, care/ concern, recognition, and encouragement.

Assumptions Testing

The assumptions of normally distributed errors (residuals), homoscedasticity, linearity, absence of multicollinearity, and outliers were tested using multiple linear regression.

The P­P plot and bell­shaped histogram revealed that the assumption of normal distributed errors was met. The scatterplot showed that the assumptions of homoscedasticity and linearity were met for the mediating role of perceived emotional support in the relationship between types of emotional support providers and help­seeking intention. The assumption of absence of multicollinearity was met, as Pearson’s r coefficient was ­.22 (p = .049), the variance inflation factor was 1.049 (less than 5), and the tolerance was .954 (greater than .2), meaning that there was no significant relationship between types of emotional support providers and perceived emotional support. Cook’s distance showed that there was no outlier, with all values smaller than 1.

Hypothesis Tests

The IV (types of emotional support providers; categorical and binary), was included in the mediation model. Before including the IV in the model, it was converted into a dummy variable, coded as 1 for the AI chatbot and 0 for the human partner.

Mediation Analysis for Perceived Emotional Support as a Mediator in the Relationship Between Types of Emotional Support Providers and Help-Seeking Intention

Note CI = confidence interval; B = unstandardized Beta coefficient; LL = lower limit; UL = upper limit; TESP = Types of Emotional Support Providers; PES = Perceived Emotional Support; HSI = Help-seeking Intention.

A regression­based simple mediation analysis showed that the total effect model of the types of emotional support provider accounted for 0.4% of the variance in help­seeking intention and did not significantly predict the frequency of help ­ seeking intention, R 2 = .004, F(1, 58) = 0.22, p = .639. Types of emotional support provider was not a significant predictor of help­seeking intention, with B c = ­1.03, 95% CI [­5.42, 3.35], t(58) = ­0.47, p = .639 (see Table 3).

The type of emotional support providers was entered into the regression against perceived emotional support. The model was non­significant, with F(1, 58) = 2.82, p = .099, explaining 4.6% of the variance in perceived emotional support (R2 = .046). Types of emotional support providers did not significantly predict perceived emotional support, with B a = ­5.77, 95% CI [­12.64, 1.11], t(58) = ­1.68, p = .099.

Types of emotional support providers and perceived emotional support were entered into the regression against help­seeking intention. The model was significant, with F(2, 57) = 3.76, p = .029, explaining 11.7% of the variance in help­seeking intention (R2 = .117). Path b of the mediation model was established as perceived emotional support was found to be a significant and positive predictor of help ­ seeking intention while controlling for types of emotional support providers, Bb = 0.22, 95% CI [0.06, 0.37], t(57) = 2.70, p = .009. Hence, the second hypothesis that there was a positive relationship between perceived emotional support and help­seeking intention was supported. After controlling for perceived emotional support, the types of emotional support providers was not a significant predictor of help­seeking intention, Bc’ = 0.20, 95% CI [­4.06, 4.47], t(57) = 0.10, p = .924.

The path ab indirect effect of types of emotional support providers on help­seeking intention through perceived emotional support was not significant (Bab = 0.94), with bias­corrected bootstrapped confidence interval containing zero, BCa CI [­3.51, 0.11], indicating that mediation did not occur. Hence, the third hypothesis that perceived emotional support mediated the relationship between types of emotional support providers and help­seeking intention was not supported.

Discussion

This study explored whether perceived emotional support mediated the relationship between types of emotional support providers and help­seeking intention. The results showed no significant difference in perceived emotional support between the AI chatbot and human conditions. This finding aligns with the studies of Ho et al. (2018) and Meng et al. (2023) but contradicts the findings of Gelbrich et al. (2021), Medeiros et al. (2021),

Meng and Dai (2021), and Smith and Masthoff (2018). One possible explanation for this discrepancy is that participants in the present study, as well as those in Ho et al. (2018) and Meng et al. (2023), were young adults who may be more familiar with digital interactions and AI chatbots compared to participants in other studies that recruited middle­aged adults. Young adults may feel more comfortable receiving emotional support from AI chatbots, perceiving it as similar to human­provided emotional support, whereas middle ­ aged adults may perceive AI chatbots as less effective in offering emotional support.

This finding provides additional evidence supporting the Computers Are Social Actors (CASA) framework and previous CASA studies that found no significant differences in perceptions of emotional support between AI chatbots and humans (Ho et al., 2018; Meng et al., 2023). This suggests that individuals apply the same social scripts and expectations when communicating with AI chatbots, without critically analysing the authenticity of each message or its source, thereby responding to AI chatbots as if they were human (Ho et al., 2018; Reeves & Nass, 1996). As a result, the interaction process and outcomes (i.e., perceived emotional support) remain similar regardless of the emotional support provider. This may indicate that individuals perceive emotional support from AI chatbots to be as effective as that provided by humans.

Perceived emotional support was found to be a significant predictor of help­seeking intention. This study is among the first to provide significant insights supporting this relationship, aligning with previous research that found perceived emotional support to be associated with various beneficial outcomes (García‐Torres et al., 2020; Umucu & Lee, 2020; Woodhead et al., 2016). Consistent with other enabling factors (e.g., encouragement and social support), this study demonstrated that perceived emotional support predicts help­seeking intention (Aguirre Velasco et al., 2020; Ali et al., 2017; Randles & Finnegan, 2022). Additionally, this study complements previous research on the predictive role of individual and contextual characteristics in helpseeking intention by examining perceived emotional support as an enabling factor (Aguirre Velasco et al., 2020; Ali et al., 2017; Randles & Finnegan, 2022). It offers a holistic perspective on the predictors of helpseeking intention, suggesting that perceived emotional support may enhance individuals’ willingness to seek help and providing causal evidence for this relationship. Furthermore, the study expands upon prior research that has primarily focused on other factors influencing helpseeking intention or has reported positive findings when the emotional support provider was a human rather than

an AI chatbot (Aguirre et al., 2020; Pretorius et al., 2019).

Perceived emotional support did not mediate the relationship between types of emotional support providers and help ­ seeking intention. This finding contradicts the results of Medeiros et al. (2021), Meng and Dai (2021), and Gelbrich et al. (2021). The absence of a mediating relationship may be attributed to participants’ past experiences with AI chatbots or mental health professionals, which could have influenced how they perceived the emotional support they received. Individuals who previously had negative experiences with emotional support providers may have been more critical and developed biased perceptions of the emotional support received in the current study, whereas those with positive past experiences may have been more receptive (Aguirre Velasco et al., 2020). Alternatively, participants’ perceived need for help may have confounded the mediating relationship. In their reflections, several participants in the present study indicated an unwillingness to seek help, believing they could resolve their problems independently. Others expressed concerns about the severity of their issues and the necessity of external support. It is possible that individuals who recognize their difficulties and believe that receiving assistance could improve their circumstances are more inclined to consider seeking help. Thus, past experiences with emotional support providers and perceived need for help may have weakened the mediating role of perceived emotional support. Future research could incorporate a pre­test to examine these factors.

Moreover, the current finding did not support the machine heuristic. According to the Expectancy Violations theory, individuals form expectations about how others will behave in a given context, and violation from these expectations—can influence their perceptions (Burgoon & Jones, 1976). In this study, participants initially had pre­existing expectations about AI chatbots, shaped by the machine heuristic, which suggests that machines are incapable of emotional tasks. However, when the AI chatbot demonstrated unexpected levels of emotional support during the interaction, this would constitute a positive violation of participants’ expectations. Such a violation could lead participants to re­evaluate their initial beliefs, resulting in perceptions of emotional support that are comparable to those provided by a human. In this case, the AI chatbot’s ability to exceed expectations by providing effective emotional support may have overridden participants’ reliance on the machine heuristic, leading to no significant differences in perceived emotional support and help­seeking intention.

Interestingly, the current findings revealed a positive and significant relationship between perceived

emotional support and help­seeking intention but did not support the mediating role of perceived emotional support in the relationship between types of emotional support providers and help­seeking intention. This suggests that individuals who perceive genuine emotional support may feel comfortable seeking help, regardless of the specific type of emotional support provider. This is in line with the model of perceived understanding (Reis et al., 2017). It is plausible that individuals in need of emotional support prioritize the ability to address their problems and needs over the provider’s identity in motivating help­seeking intention. When individuals feel that someone is willing to spend time listening to them, offering emotional support, and taking their concerns seriously, they may be more inclined to seek help (Medeiros et al., 2021; Yan, 2020). The role of the emotional support provider may become less significant when the perception of emotional support is strong. This implies that as long as emotional support is perceived as effective and genuine in addressing individuals’ needs, they will demonstrate a greater intention to seek help.

Limitations and Suggestions for Future Research

Several limitations should be considered. The first limitation is the duration of the experiment, as it was conducted as a one ­ time study. Establishing a close relationship with an emotional support provider requires time, which may have restricted the study’s ability to capture changes in participants’ perceived levels of emotional support over time and their subsequent intention to seek help. Specifically, previous research has shown that after repeated interactions with an AI chatbot, individuals reported perceiving it as lacking conscientiousness and credibility due to its occasional irrelevant questions or inappropriate responses, which diminished their willingness to engage with it (Sullivan et al., 2023). In such cases, individuals who initially perceived high levels of emotional support from AI chatbots may eventually experience a decline in perceived emotional support. Future research may benefit from conducting a longitudinal study to replicate the present experiment, allowing for a better understanding of how participants’ perceptions evolve over time and providing deeper insights into the causal relationship. Secondly, this study is limited by the use of a scripted conversation, in which emotional support messages were static and did not fully reflect real­world interactions. Although both the AI chatbot and the human partner delivered the similar scripted messages, the way emotional support is conveyed in real ­ life conversations may differ due to variations in sentence structure, tone of voice, and word choice. Future research may consider incorporating these linguistic

features and adopting a semi ­ scripted conversation approach, allowing conversation partners to adapt their responses based on the recipient’s input and context. This approach would provide greater flexibility to accommodate the dynamic and unpredictable nature of human communication.

Thirdly, the overall internal consistency of the GHSQ in the present study was questionable (Cronbach’s α = .64). It is important to consider this limitation when interpreting the results. A possible explanation is that participants had varying preferences regarding whom they sought help from, leading them to select different items. For example, some participants preferred seeking support from intimate others (e.g., parents and partners), but others favoured professional sources (e.g., mental health professionals). This variation may have introduced inconsistency in the data, making it challenging to identify meaningful patterns.

Another limitation is that the sample in this study consisted of psychology students, who were young adults primarily experiencing academic­related stressors and were digital natives. Psychology students may be more familiar with research expectations and more open to seeking help (Foot & Sanford, 2004). This could result in a potential participant effect, in which participants modify their responses in alignment with the study’s aim or hypotheses. Consequently, the findings may not be generalizable to other populations, particularly individuals with lower educational backgrounds or those outside the field of psychology. Additionally, older adults may be less familiar with and more resistant to AI technologies for emotional support (Xiang et al., 2023). Future research should replicate these hypotheses and findings in different samples and contexts to enhance generalizability.

Future research should develop a scale specifically designed to measure perceived emotional support in post­interaction contexts. Currently, there is a lack of a standardized scale that assesses this construct and its key dimensions. Existing emotional support scales primarily evaluate the quality or frequency of perceived emotional support in everyday life or identify sources of emotional support rather than focusing on conversational interactions. For example, the Communication­Based Emotional Support Scale (Weber & Patterson, 1996) measures emotional support in general communication but does not capture its nuances in direct conversations. Given that the present study identified five key factors of perceived emotional support—understanding and support, empathy, care/concern, recognition, and encouragement—these findings may serve as a foundation for scale development. Establishing such a scale would contribute to the theoretical understanding of perceived

emotional support and provide a more precise tool for assessing its impact in various interpersonal and digital interactions.

Implications

The current study tested and did not support the machine heuristic. As AI chatbots become more common in daily life, people are increasingly exposed to AI­driven interactions, which may further reduce reliance on the machine heuristic (Brandtzaeg et al., 2022). Individuals may become more accustomed to receiving emotional support from AI. Over time, familiarity and trust in AI­driven interactions may lead to a shift in perception, where AI chatbots are seen as legitimate emotional support providers rather than impersonal machines. Thus, the machine heuristic may become less influential in shaping individuals’ perceptions of emotional support. Future research should explore the conditions under which people override the machine heuristic, for example, repeated AI interactions or accuracy of responses.

This study contributes to existing research, which has focused on traditional social networks that emphasize the role of humans in shaping emotional support and help­seeking intention (Planey et al., 2019). It expands the scope of social network theories by proposing that AI chatbots may function as a new form of support network, providing emotional support and enhancing individuals’ help­seeking intentioan. This perspective necessitates reconsidering how social network theories account for AI­driven interactions. Furthermore, the recognition of AI chatbots as legitimate sources of support challenges the long­standing assumption that emotional support can only be meaningfully provided through human connection. The evolving nature of AI­human relationships also prompts a re­evaluation of how emotional bonds are formed, maintained, and perceived within the context of digital interactions.

The nonsignificant mediating role of perceived emotional support in the relationship between types of emotional support providers and help­seeking intention was identified. This finding contributes to the theoretical understanding of perceived emotional support, suggesting that although it is essential, it may not serve as a strong mediator in this relationship. Future research may explore alternative mediating factors that play a more significant role while also considering the complexity of human­human and human­chatbot interactions, as well as cultural influences. Particularly, perceived trustworthiness and closeness of the relationship with the provider have been found to predict help­seeking intention (Kim & Lee, 2022). Additionally, the present study highlighted potential confounding variables, specifically past experiences with emotional support providers and perceived need for

help, which may weaken the mediating relationship. To further investigate this issue, future research could conduct exploratory studies incorporating qualitative methods to provide deeper insights into individuals’ subjective experiences with perceived emotional support from different providers and their help­seeking intention.

The present study contributes to the understanding of perceived emotional support and help­seeking intention in the context of mental health. Given the low rates of help­seeking intention (Kim & Lee, 2022), mental health professionals may develop tailored support programs and psychological interventions that prioritize emotional support for the public. Resources and strategies can be implemented to foster supportive connections and enhance emotional support within social networks. For online mental health support services, professionals should ensure that they provide sufficient and genuine emotional support to encourage individuals to seek help and maintain long­term engagement in therapy. Additionally, mental health services may consider integrating AI chatbots into their support systems, particularly for individuals who are hesitant to seek human support due to perceived stigma or limited social networks (Luxton, 2016). Emotional support from AI chatbots could offer a more accessible alternative for these individuals, bridging the gap in mental health care.

References

Adams, G. A., King, L. A., & King, D. W. (1996). Relationships of job and family involvement, family social support, and work–family conflict with job and life satisfaction. Journal of Applied Psychology, 81, 411–420. https://dx.doi.org/10.1037/0021-9010.81.4.411

Aguirre Velasco, A., Cruz, I. S. S., Billings, J., Jimenez, M., & Rowe, S. (2020). What are the barriers, facilitators and interventions targeting help-seeking behaviours for common mental health problems in adolescents? A systematic review. BMC Psychiatry, 20(1), 1–22. https://doi.org/10.1186/s12888-020-02659-0

Ali, K., Farrer, L., Fassnacht, D. B., Gulliver, A., Bauer, S., & Griffiths, K. M. (2017). Perceived barriers and facilitators towards help‐seeking for eating disorders: A systematic review. International Journal of Eating Disorders, 50(1), 9–21. https://doi.org/10.1002/eat.22598

Andrade-Arenas, L., Yactayo-Arias, C., & Pucuhuayla-Revatta, F. (2024). Therapy and Emotional Support through a Chatbot. International Journal of Online & Biomedical Engineering, 20(2), 114–130. https://dx.doi.org/10.3991/ijoe.v20i02.45377

Arias, I., Lyons, C. M., & Street, A. E. (1997). Individual and marital consequences of victimization: Moderating effects of relationship efficacy and spouse support. Journal of Family Violence, 12, 193–210. https://doi.org/10.1023/A:1022888728475

Brandtzaeg, P. B., Skjuve, M., & Følstad, A. (2022). My AI friend: How users of a social chatbot understand their human–AI friendship. Human Communication Research, 48(3), 404–429. https://doi.org/10.1093/hcr/hqac008

Burgoon, J. K., & Jones, S. B. (1976). Toward a theory of personal space expectations and their violations. Human Communication Research, 2(2), 131–146. https://doi. org/10.1111/j.1468-2958.1976.tb00706.x

Burleson, B. R. (2008). What counts as effective emotional support? Explorations of situational and individual differences. In M. T. Motley (Ed.), Studies in applied interpersonal communication (pp. 207–227). Sage. https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi= 05d6 c2816c7c90e83d675926a6b7cc2c49acd666

Charrier, L., Rieger, A., Galdeano, A., Cordier, A., Lefort, M., & Hassas, S. (2019, March 11). The rope scale: a measure of how empathic a robot is perceived

[Paper presentation]. 2019 14th ACM/IEEE International Conference on Human-Robot Interaction (HRI), Daegu, Korea (South). https://doi.org/10.1109/HRI.2019.8673082

Cobb, S. (1976). Social support as a moderator of life stress. Psychosomatic Medicine, 38(5), 300–314. https://doi.org/10.1097/00006842-197609000-00003

Cogan, A., Pret, T., & Cardon, M. S. (2022). Everyday social support processes: Household members’ instrumental and emotional support of entrepreneurs. International Small Business Journal, 40(5), 537–563. https://doi.org/10.1177/02662426211054098

Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98(2), 310–357. https://doi.org/10.1037/0033-2909.98.2.310

D’Alfonso, S. (2020). AI in mental health. Current Opinion in Psychology, 36, 112–117. https://doi.org/10.1016/j.copsyc.2020.04.005

De Andrade, I. M., & Tumelero, C. (2022). Increasing customer service efficiency through artificial intelligence chatbot. Revista De Gestão, 29(3), 238–251. https://doi.org/10.1108/REGE-07-2021-0120

Duong, M. T., Bruns, E. J., Lee, K., Cox, S., Coifman, J., Mayworm, A., & Lyon, A. R. (2021). Rates of mental health service utilization by children and adolescents in schools and other common service settings: A systematic review and meta-analysis. Administration and Policy in Mental Health and Mental Health Services Research, 48, 420–439. https://doi.org/10.1007/s10488-020-01080-9

Foot, H., & Sanford, A. (2004). The use and abuse of student participants. The Psychologist, 17(5), 256–259. https://www.abdn.ac.uk/psychology/documents/0504foot.pdf

Frankish, K., & Ramsey, W. M. (2014). The Cambridge handbook of artificial intelligence. Cambridge University Press. https://doi.org/10.1017/CBO9781139046855

García‐Torres, F., Jacek Jabłoński, M., Gómez Solís, Á., Moriana, J. A., Jaén‐Moreno, M. J., Moreno‐Díaz, M. J., & Aranda, E. (2020). Social support as predictor of anxiety and depression in cancer caregivers six months after cancer diagnosis: A longitudinal study. Journal of Clinical Nursing, 29(5-6), 996–1002. https://doi.org/10.1111/jocn.15123

Gelbrich, K., Hagel, J., & Orsingher, C. (2021). Emotional support from a digital assistant in technology-mediated services: Effects on customer satisfaction and behavioral persistence. International Journal of Research in Marketing, 38(1), 176–193. https://doi.org/10.1016/j.ijresmar.2020.06.004

Gulliver, A., Griffiths, K. M., & Christensen, H. (2012). Barriers and facilitators to mental health help-seeking for young elite athletes: A qualitative study. BMC Psychiatry, 12, 1–14. https://doi.org/10.1186/1471-244X-12-157

Haber, M. G., Cohen, J. L., Lucas, T., & Baltes, B. B. (2007). The relationship between self-reported received and perceived social support: A metaanalytic review. American Journal of Community Psychology, 39, 133–144. https://doi.org/10.1007/s10464-007-9100-9

Ho, A., Hancock, J., & Miner, A. S. (2018). Psychological, relational, and emotional effects of self-disclosure after conversations with a chatbot. Journal of Communication, 68(4), 712–733. https://doi.org/10.1093/joc/jqy026

Hu, X., Song, Y., Zhu, R., He, S., Zhou, B., Li, X., Bao, H., Shen, S., & Liu, B. (2022). Understanding the impact of emotional support on mental health resilience of the community in the social media in Covid-19 pandemic. Journal of Affective Disorders, 308, 360–368. https://doi.org/10.1016/j.jad.2022.04.105

Ibrahim, N., Amit, N., Shahar, S., Wee, L. H., Ismail, R., Khairuddin, R., Siau, C. S., & Safien, A. M. (2019). Do depression literacy, mental illness beliefs and stigma influence mental health help-seeking attitude? A cross-sectional study of secondary school and university students from B40 households in Malaysia. BMC Public Health, 19, 1–8. https://doi.org/10.1186/s12889-019-6862-6

Kim, S. B., & Lee, Y. J. (2022). Factors associated with mental health help-seeking among Asian Americans: A systematic review. Journal of Racial and Ethnic Health Disparities, 9(4), 1276–1297. https://doi.org/10.1007/s40615-021-01068-7

Kiron, D., & Unruh, G. (2019). Even if AI can cure loneliness-should it? MIT Sloan Management Review, 60(2), 1–4. https://www.proquest.com/scholarlyjournals/even-if-ai-can-cure-loneliness-should/docview/2168846958/se-2

Lew, B., Kõlves, K., Lester, D., Ibrahim, N., & Siau, C. S. (2022). Looking into recent suicide rates and trends in Malaysia: a comparative analysis. Frontiers in Psychiatry, 12, 1–11. https://doi.org/10.3389/fpsyt.2021.770252

Luxton, D. D. (2016). Artificial intelligence in behavioral and mental health care. Academic Press. https://doi.org/10.1016/B978-0-12-420248-1.00001-5

Madhavan, P., Wiegmann, D. A., & Lacson, F. C. (2006). Automation failures on tasks easily performed by operators undermine trust in automated aids. Human Factors, 48(2), 241–256.

https://doi.org/10.1518/001872006777724408

Magaard, J. L., Seeralan, T., Schulz, H., & Brütt, A. L. (2017). Factors associated with help-seeking behaviour among individuals with major depression: A systematic review. PloS One, 12(5), 1–17. https://doi.org/10.1371/journal.pone.0176730

Malik, T., Ambrose, A. J., & Sinha, C. (2022). Evaluating user feedback for an artificial intelligence–enabled, cognitive behavioral therapy–based mental health app (Wysa): Qualitative thematic analysis. JMIR Human Factors, 9(2), 1–10. https://doi.org/10.2196/35668

Medeiros, L., Bosse, T., & Gerritsen, C. (2021). Can a chatbot comfort humans? Studying the impact of a supportive chatbot on users’ self-perceived stress. IEEE Transactions on Human-Machine Systems, 52(3), 343–353. https://doi.org/10.1109/THMS.2021.3113643

Meng, J., & Dai, Y. (2021). Emotional support from AI chatbots: Should a supportive partner self-disclose or not? Journal of Computer-Mediated Communication, 26(4), 207–222. https://doi.org/10.1093/jcmc/zmab005

Meng, J., Rheu, M., Zhang, Y., Dai, Y., & Peng, W. (2023). Mediated social support for distress reduction: AI Chatbots vs. Human. Proceedings of the ACM on Human-Computer Interaction, 7(CSCW1), 1–25. https://doi.org/10.1145/3579505

Namkoong, K., McLaughlin, B., Yoo, W., Hull, S. J., Shah, D. V., Kim, S. C., Moon, T. J., Johnson, C. N., Hawkins, R. P., McTavish, F. M., & Gustafson, D. H. (2013). The effects of expression: how providing emotional support online improves cancer patients’ coping strategies. Journal of the National Cancer Institute Monographs, 2013(47), 169–174. https://doi.org/10.1093/jncimonographs/lgt033

Nochaiwong, S., Ruengorn, C., Thavorn, K., Hutton, B., Awiphan, R., Phosuya, C., Ruanta, Y., Wongpakaran, N., & Wongpakaran, T. (2021). Global prevalence of mental health issues among the general population during the coronavirus disease-2019 pandemic: A systematic review and meta-analysis. Scientific Reports, 11(1), 1–18. https://doi.org/10.1038/s41598-021-89700-8

Pauw, L. S., Sauter, D. A., Van Kleef, G. A., & Fischer, A. H. (2018). Sense or sensibility? Social sharers’ evaluations of socio-affective vs. cognitive support in response to negative emotions. Cognition and Emotion, 32(6), 1247–1264. https://doi.org/10.1080/02699931.2017.1400949

Pauw, L. S., Sauter, D. A., Van Kleef, G. A., Lucas, G. M., Gratch, J., & Fischer, A. H. (2022). The avatar will see you now: Support from a virtual human provides socio-emotional benefits. Computers in Human Behavior, 136, 1–10. https://doi.org/10.1016/j.chb.2022.107368

Peng, C., Kwok, C. L., Law, Y. W., Yip, P. S., & Cheng, Q. (2019). Intergenerational support, satisfaction with parent–child relationship and elderly parents’ life satisfaction in Hong Kong. Aging & Mental Health, 23(4), 428–438. https://doi.org/10.1080/13607863.2017.1423035

Planey, A. M., Smith, S. M., Moore, S., & Walker, T. D. (2019). Barriers and facilitators to mental health help-seeking among African American youth and their families: A systematic review study. Children and Youth Services Review, 101, 190–200. https://doi.org/10.1016/j.childyouth.2019.04.001

Pretorius, C., Chambers, D., & Coyle, D. (2019). Young people’s online helpseeking and mental health difficulties: Systematic narrative review. Journal of Medical Internet Research, 21(11), 1–17. https://doi.org/10.2196/13873

Randles, R., & Finnegan, A. (2022). Veteran help-seeking behaviour for mental health issues: A systematic review. BMJ Mil Health, 168(1), 99–104. https://dx.doi.org/10.1136/bmjmilitary-2021-001903

Ratheesh, A., & Alvarez-Jimenez, M. (2022). The future of digital mental health in the post-pandemic world: Evidence-based, blended, responsive and implementable. Australian & New Zealand Journal of Psychiatry, 56(2), 107–109. https://doi.org/10.1177/00048674211070984

Ratnayake, P., & Hyde, C. (2019). Mental health literacy, help-seeking behaviour and wellbeing in young people: Implications for practice. The Educational and Developmental Psychologist, 36(1), 16–21. https://doi.org/10.1017/edp.2019.1

Reeves, B., & Nass, C. (1996). The media equation: How people treat computers, television, and new media like real people. Cambridge, 10(10), 19–36. https://www.researchgate.net/publication/37705092

Reghunath, A. (2021). Expression of empathy in social virtual bots used for genetic counselling [Master’s thesis, Uppsala University]. DiVA Portal. https://www.diva-portal.org/smash/get/diva2:1571903/FULLTEXT01.pdf

Reis, H. T., Lemay, E. P., & Finkenauer, C. (2017). Toward understanding understanding: The importance of feeling understood in relationships. Social and Personality Psychology Compass, 11(3), 1–22. https://doi.org/10.1111/spc3.12308

Rickwood, D., Deane, F. P., Wilson, C. J., & Ciarrochi, J. (2005). Young people’s help-seeking for mental health problems. Australian E-journal for the Advancement of Mental Health, 4(3), 218–251.

https://doi.org/10.5172/jamh.4.3.218

Robinson, E., Sutin, A. R., Daly, M., & Jones, A. (2022). A systematic review and meta-analysis of longitudinal cohort studies comparing mental health before versus during the COVID-19 pandemic in 2020. Journal of Affective Disorders, 296, 567–576. https://doi.org/10.1016/j.jad.2021.09.098

Roksa, J., & Kinsley, P. (2019). The role of family support in facilitating academic success of low-income students. Research in Higher Education, 60, 415–436. https://doi.org/10.1007/s11162-018-9517-z

Shah, J., DePietro, B., D’Adamo, L., Firebaugh, M. L., Laing, O., Fowler, L. A., Smolar, L., Sadeh‐Sharvit, S., Taylor, C. B., Wilfley, D. E., & Fitzsimmons‐Craft, E. E. (2022). Development and usability testing of a chatbot to promote mental health services use among individuals with eating disorders following screening. International Journal of Eating Disorders, 55(9), 1229–1244. https://doi.org/10.1002/eat.23798

Sheikh, M. A., Abelsen, B., & Olsen, J. A. (2016). Clarifying associations between childhood adversity, social support, behavioral factors, and mental health, health, and well-being in adulthood: A population-based study. Frontiers in Psychology, 7, 1–24. https://doi.org/10.3389/fpsyg.2016.00727

Skjuve, M., Følstad, A., Fostervold, K. I., & Brandtzaeg, P. B. (2021). My chatbot companion—A study of human–chatbot relationships. International Journal of Human-Computer Studies, 149, 1–14. https://doi.org/10.1016/j.ijhcs.2021.102601

Smith, K. A., & Masthoff, J. (2018). Can a virtual agent provide good emotional support? [Paper presentation]. Proceedings of the 32nd International BCS Human Computer Interaction Conference, United Kingdom. https://dx.doi.org/10.14236/ewic/HCI2018.13

Song, L. (2011). Social capital and psychological distress. Journal of Health and Social Behavior, 52(4), 478–492. https://doi.org/10.1177/0022146511411921

Stein, J. P., & Ohler, P. (2017). Venturing into the uncanny valley of mind—The influence of mind attribution on the acceptance of human-like characters in a virtual reality setting. Cognition, 160 43–50. https://doi.org/10.1016/j.cognition.2016.12.010

Sullivan, Y., Nyawa, S., & Fosso Wamba, S. (2023). Combating loneliness with artificial intelligence: An AI-based emotional support model [Paper presentation]. Proceedings of the 56th Hawaii International Conference on System Sciences, United States. https://hdl.handle.net/10125/103173

Sundar, S. S., & Kim, J. (2019). Machine heuristic: When we trust computers more than humans with our personal information. In S. Brewster & G. Fitzpatrick (Eds.), Proceedings of the 2019 CHI Conference on human factors in computing systems (pp. 1–9). Association for Computing Machinery. https://doi.org/10.1145/3290605.3300768

Thompson, A., Hunt, C., & Issakidis, C. (2004). Why wait? Reasons for delay and prompts to seek help for mental health problems in an Australian clinical sample. Social Psychiatry and Psychiatric Epidemiology, 39, 810–817. https://doi.org/10.1007/s00127-004-0816-7

Trees, A. R. (2000). Nonverbal communication and the support process: Interactional sensitivity in interactions between mothers and young adult children. Communications Monographs, 67(3), 239–261. https://doi.org/10.1080/03637750009376509

Umucu, E., & Lee, B. (2020). Examining the impact of COVID-19 on stress and coping strategies in individuals with disabilities and chronic conditions. Rehabilitation Psychology, 65(3), 193–198. https://doi.org/10.1037/rep0000328

Uysal Irak, D., Kalkışım, K., & Yıldırım, M. (2020). Emotional support makes the difference: Work–family conflict and employment related guilt among employed mothers. Sex Roles, 82, 53–65. https://doi.org/10.1007/s11199-019-01035-x

Vogel, D. L., Wade, N. G., Wester, S. R., Larson, L., & Hackler, A. H. (2007). Seeking help from a mental health professional: The influence of one’s social network. Journal of Clinical Psychology, 63(3), 233–245. https://doi.org/10.1002/jclp.20345

Weber, K. D., & Patterson, B. R. (1996). Construction and validation of a communication based emotional support scale. Communication Research Reports, 13(1), 68–76. https://doi.org/10.1080/08824099609362072

Wilson, C. J., Deane, F. P., & Ciarocchi, J. (2005). Measuring help seeking intention: Properties of the General Help Seeking Questionnaire. Canadian Journal of Counselling, 39(1), 15–28. https://ro.uow.edu.au/cgi/viewcontent.cgi?referer=&http sredir=1&article=2580&context=hbspapers

Woodhead, E. L., Northrop, L., & Edelstein, B. (2016). Stress, social support, and burnout among long-term care nursing staff. Journal of Applied Gerontology, 35(1), 84–105. https://doi.org/10.1177/0733464814542465

Xiang, H., Zhou, J., & Xie, B. (2023). AI tools for debunking online spam reviews? Trust of younger and older adults in AI detection criteria. Behaviour & Information Technology, 42(5), 478–497. https://doi.org/10.1080/0144929X.2021.2024252

Yan, L. (2020). The kindness of commenters: An empirical study of the effectiveness of perceived and received support for weight‐loss outcomes. Production and Operations Management, 29(6), 1448–1466. https://doi.org/10.1111/poms.13171

Yasuhiro, K., Su-Hie, T., & Neary, S. (2021). Mental health of Malaysian university students: UK comparison, and relationship between negative mental health attitudes, self-compassion, and resilience. Higher Education, 81(2), 403–419. https://doi.org/10.1007/s10734-020-00547-w

Yedirir, S., & Hamarta, E. (2015). Emotional expression and spousal support as predictors of marital satisfaction: The case of Turkey. Educational Sciences:

Theory and Practice, 15(6), 1549–1558. https://doi.org/10.12738/estp.2015.6.2822

Zhou, L., Gao, J., Li, D., & Shum, H. Y. (2020). The design and implementation of xiaoice, an empathetic social chatbot. Computational Linguistics, 46(1), 53–93. https:// doi.org/10.1162/coli_a_00368

Author Note.

Lee Yan Ying https://orcid.org/0009­0006­6202­1263 All materials used for this study can be found at https://osf.io/9r4eh I have no known conflict of interest to disclose. Special thanks to Dr. Eugene Tee, Mr. James Yeow, and Dr. Andi for insightful comments during the thesis process.

Correspondence concerning this article may be addressed to Lee Yan Ying. Email: yylee0513@gmail.com

NHourly Negative Interpersonal Interactions and Momentary Ambulatory Blood Pressure Among African American Emerging Adults

Emilie J. Chai1*, Eunji Shin2*, Nataria T. Joseph2**, and Laurel M. Peterson3**

1Psychology Department, University of California, San Diego

2Social Science Division, Pepperdine University

3Psychology Department, Bryn Mawr College

ABSTRACT. Prior research established an association between negative interpersonal interactions (NII) and increases in ambulatory blood pressure (ABP). However, the role of moderators in this relationship is underexplored. This study aimed to examine the association between momentary NII and ABP with perceived social support as a moderator. A sample of 54 healthy African American emerging adults completed a baseline survey, 2 days of ABP monitoring, and 2 days of hourly ecological momentary assessment (EMA) in which they reported on various aspects of their social interactions. Mixed modeling controlling for age, sex, relationship status, income, BMI, and ABP momentary covariates revealed that the positive association between NII and ABP was moderated by baseline perceived social support (systolic ABP F(1, 407) = 9.17, p = .003, ƒ2 = .02 and diastolic ABP F(1, 392) = 8.17, p = .004, ƒ2 = .03). NII increased the diastolic ABP of those who reported low perceived social support (b = 0.90, p = .03, ƒ2 = .02). NII significantly decreased the systolic ABP of those who reported high perceived social support ( b = ­ 1.47, p = .03, ƒ 2 = .02). Unexpectedly, we found NII to be associated with lower systolic BP among those with high perceived social support. Plausible explanations for this finding are outlined. Future research should explore replication of this finding as well as other psychosocial moderators of the relationship between NII and ABP.

Keywords: negative interpersonal interaction, ambulatory blood pressure, perceived social support, ecological momentary assessment

egative interpersonal interaction (NII) refers to mistreatment between individuals, which can range from subtle incivilities to overt aggression (Andersson & Pearson, 1999). It often manifests through the withholding of positive behaviors or the engagement in negative ones (Andersson & Pearson, 1999; Zawadzki et al., 2023). Those who experience such interactions typically encounter adverse health consequences. For instance, NII contributes to negative mood and feelings of stress (Brondolo et al., 2003; Shapiro et al., 1997; Sneed & Cohen, 2014), which are linked to increases in blood pressure (BP; Schoenthaler et al., 2010; Shapiro et al., 1997; Sloan et al., 2001). BP, a well­established indicator of cardiovascular system function, is integral in assessing overall health and is a predictive measure for a range of health conditions (Perkovic et al., 2007). These health conditions include hypertension, which is a key risk

factor for stroke and organ damage in later life and has broad implications for morbidity and mortality rates (Holt­Lunstad et al., 2003). Ambulatory blood pressure (ABP) provides a comprehensive profile of momentary BP fluctuations in day­to­day life in individuals’ natural environments and is thus considered a more accurate indicator of health compared to clinical BP (Angeli et al., 2014). Accordingly, ABP has critical importance in understanding the relationship between psychological stressors and physical health.

To our knowledge, only one article has examined the relationship between NII and ABP using EMA (Zawadzki et al., 2023). In a sample of 565 Black and Hispanic urban adults (ages 23–65, M = 39.06, SD = 9.35), negative emotion was assessed in moments across the day and found to be associated with both the intensity of NII and systolic and diastolic ABP levels. Zawadzki et al. reported a main effect of NII on ABP, where more intense NIIs were associated with higher systolic and diastolic ABP levels. They also determined that neither race nor discrimination moderated the relation between NII and ABP and suggested that future research continue to investigate moderating factors. Although the researchers kept the EMA questionnaires brief to reduce participant burden, the frequency of ABP assessment (every 20 minutes over 24 hours) was noted as a potential limitation of the study as it could have interrupted daily social interactions, possibly leading to an underestimation of the true effect of NII on ABP. We sought to extend Zawadzki et al.’s work in illustrating the role of negative emotion in the NII and ABP relation by conducting an EMA study exploring alternative moderators between NII and ABP and reducing the frequency of ABP assessments.

One factor that may influence the relationship between NII and ABP is social support, which is strongly implicated as a protective mechanism against stress—a phenomenon known as the stress buffering effect of social support (Berkman et al., 2000; Cohen & Wills, 1985; Thoits, 2011). In particular, perceived social support—defined as the amount of support a person believes is available to them (Uchino et al., 2022)—has a greater protective effect on mental health than received social support (i.e., resources and acts of social support received from one’s social network; Bolger et al., 2000; Brown, 1978; Wethington & Kessler, 1986). Additionally, studies have linked higher levels of perceived social support to an increased sense of belonging, self­esteem, and a heightened perception of control over one’s life (Berkman et al., 2000; Chun & Lee, 2017; Thoits, 2011). Perceived social support is also associated with lower risk of coronary heart disease, and low levels of perceived social support have been linked to higher risk of cardiovascular

disease (CVD; Datta et al., 2023; Kuper & Marmot, 2003).

Although these associations intuitively suggest that social support has a direct influence on ABP, a recent meta­analysis disputes social support’s direct association to ABP (Uchino et al., 2022). Nonetheless, this finding does not contradict the stress buffering effect hypothesis, because social support may still play an indirect role in decreasing BP by acting as a psychological barrier in stressful situations. In other words, it appears that social support’s influence on ABP may vary by individual and depend on that person’s stress levels, including the stress that emanates from NII. Combined, these findings give us reason to believe that perceived social support may moderate the relationship between NII and ABP by protecting a person against the stress and perceived threat that accompany negative interactions.

Although previous research on BP and social relations have typically focused on older populations, a study conducted with older adults found that age did not moderate the association between NII and hypertension risk for those above age 65; however, a comparatively younger population ranging from ages 51 to 64 exhibited a somewhat stronger moderating effect between those variables (Sneed & Cohen, 2014). Furthermore, research contends that younger populations are less likely to have conditions like hypertension that predominantly affect older adults. Thus, studying younger populations would allow researchers to better understand the psychosocial avenues that impact health indicators such as BP with fewer preexisting and confounding health issues (Goosby et al., 2015).

In particular, it may be important to investigate the NII and ABP relationship among emerging adults (about ages 18–30), the population transitioning from adolescence to adulthood (Arnett et al., 2014). Emerging adulthood is characterized by heightened instability, unique social challenges, and significant life changes (e.g., moving away from family or getting married) that may heighten vulnerability to the effects of NII and subsequent health issues (e.g., increased BP, stress, suicidality, and depressed mood; Arnett et al., 2014; Matud et al., 2020). Although definitions have traditionally capped emerging adulthood at 25 years, an expanded age range has been increasingly utilized in recent years (with some going up to 34 years) to reflect findings indicating continued brain maturation throughout this period and to compensate for the growing delay in commitment to traditional roles associated with adulthood, such as child­rearing or moving away from home (Hochberg & Konner, 2020; Huang et al., 2021). We opted to define emerging adulthood between ages 18–30 instead of 25 because perception of emerging adulthood has been found to vary across demographics,

with factors such as racial profiling and increasingly limited employment opportunities postponing successful transition to adulthood (Jones et al., 2023). During this turbulent period, perceived social support may act as a critical buffer against the psychological and physiological impacts of stressors such as NII (Matud et al., 2020). Additionally, research emphasizes the importance of identifying biomarkers of cardiovascular risk in emerging adults, as they are more predictive of health conditions such as coronary calcification and CVD in younger populations than in middle age (Behbodikhah et al., 2021). Identifying elevated ABP is particularly important because it can serve as a precursor to future CVD, which often goes undetected in early stages by traditional clinical methods. Thus, understanding the relationship between NII and ABP during this formative period could enable early interventions that may prevent the progression of heart conditions.

Population demographics notwithstanding, there remains a notable gap in our understanding of how the effects of social interactions on cardiovascular health manifest in real time. Ecological momentary assessment (EMA) generates repeated samples of an individual’s interactions over time, allowing for a more ecologically valid and comprehensive examination of the transient and cumulative effects of daily social interactions on ABP. EMA’s real ­ time, in ­ situ approach has various advantages. For example, it reduces the inaccuracies or recall errors associated with periodic interviews or in­person surveys, which involve retrospective assessments of emotions and other experiences (Moskowitz & Young, 2006; Shiffman et al., 2008). The context­specific nature of EMA data is further advantageous in studying within­person variables sensitive to change over short periods of time. EMA­assessed psychosocial variables may more strongly predict cardiovascular outcomes than traditional global self­reports due to their ecological and temporal sensitivity. Relevant to the current study, EMA allows for more accurate assessment of social interactions shortly after they occur. The combination of EMA and ABP provides a more nuanced perspective on how NII influences ABP in real­world settings by capturing both subjective experiences and objective physiological responses. Historically, studies that examine cardiovascular function and social interaction occur in laboratory settings (Kamarck, 1992; Sneed & Cohen, 2014) and ecological assessment with ABP typically occurs through a single day (e.g., Brondolo et al., 2003; Shapiro et al., 1997; Smith et al., 2012; Zawadzki et al., 2023). Although these methodological approaches have laid a strong foundation for understanding negative social interactions and blood pressure dynamics, they are limited in capturing the variability of day­to­day social interactions, which

typically include both negative and positive interactions with different people across diverse social contexts. Thus, more EMA studies—including EMA protocols that last longer than one day—are necessary.

In sum, previous research has established a baseline understanding of the impact of negative social interactions on BP. However, because BP is a fundamental and context­dependent health indicator, it is vital to assess this relationship in real­life settings that account for the variability of day­to­day interactions using EMA. Moreover, specific moderators of the relationship between NII and BP are underexplored, especially among emerging adults. Given the positive implications of perceived social support on cardiovascular health, the goal of the present study is to examine the association between momentary NIIs and ABP and the extent to which social support may influence this relationship. We hypothesized that the relationship between momentary NII and ABP would be moderated by perceived social support, with NII being more strongly associated with higher ABP in those perceiving low levels of social support.

Method

Participants

All participants provided written informed consent before participating in the study. The sample consisted of 54 African American emerging adults (ages 18–30, M = 23.09, SD = 3.02) with no history of CVD and a mean body mass index (BMI) of 27.50 ( SD = 6.43). Across all participants (74.1% female and 25.9% male), 53.7% were enrolled part­time or full­time in school, and 77.7% of participants worked part­time or fulltime. Participants’ annual household incomes varied widely, and participants had achieved many different levels of educational attainment. See Table 1 for a full demographic description of the sample.

Materials

ABP

ABP was measured using the SunTech Medical Oscar Ambulatory Blood Pressure monitor (Suntech Inc.), conveniently worn on a waist belt or arm strap. At the baseline visit, the ABP cuff was programmed to inflate hourly throughout the day and overnight. The Oscar utilizes the oscillometric method for ABP measurement, which detects arterial pressure waves as the cuff inflates; this method provides systolic and diastolic BP readings along with heart rate. The Oscar has demonstrated validity and reliability (Jones et al., 2004).

In line with previous research, we excluded potentially artifactual readings marked with error messages due to weak Korotkoff sounds, air leaks, or other errors. As is standard, we also excluded outlier readings that

included a systolic ABP reading above 190 or below 85 or a diastolic ABP reading above 135 or below 40.

EMA and NII

EMA surveys assessed factors influencing BP, emotional state, and social interactions. In line with Zawadzki et al. (2023), each survey was kept brief to avoid influencing changes in participants’ interactions. Participants accessed the EMA on electronic devices provided to them (Google Nexus S, Gingerbread Operating System). Procedures, wording, and response scaling of interpersonal interaction questions were based on the Diary of Ambulatory Behavioral States, a widely applied and well­validated assessment for capturing social interactions in daily life (Kamarck et al., 1998). At each EMA assessment, participants were windowed into reporting about social interactions by, first, reporting whether a social interaction had occurred in the last hour. If yes, participants were prompted to report when (e.g., 15 minutes or less/ongoing) and with whom (e.g., number of people in the social interaction, type of relation such as coworker, family, or stranger, and perceived racial/ ethnic identity of members in the social interaction). NII in regards to participants’ most recent hourly

TABLE 1

Descriptive Statistics of Participant Demographics

interactions was assessed using the following questions: (a) “Did someone treat you badly?” and (b) “Was someone in conflict with you?” Responses were assessed using a six ­ item Likert Scale ranging from strongly disagree to strongly agree, articulated as follows: NO!, No, no, yes, Yes, YES! Responses to the scale exhibited a moderate internal consistency (α = .74), acceptable for within­persons designs (Bonito et al., 2012; Nezlek, 2017). The intraclass correlation coefficient for the NII variable was .44, suggesting a substantial degree of within­persons variability in NII and supporting the rationale for examining the associations between momentary NII and ABP using EMA. This aligns with expectations that, for each participant, negative social interaction scores should vary (i.e., we expected people to have different types of social interactions throughout the day, some of which are more negative than others). Global measures of negative social interactions inquiring regarding typical or general exposure to negative social interactions would not capture this rich moment to moment variability of negativity within social interactions throughout individuals’ days.

Perceived Social Support

Perceived social support was assessed at the baseline visit using the Social Provisions Scale (Cutrona & Russell, 1987). This 24­item measure has demonstrated past reliability (α = .92) and validity for assessing perceived social support across diverse groups (e.g., students, teachers, nurses; see Cutrona & Russell, 1987 for a comprehensive overview). As others have done, we utilized the 20­item version of the measure that excludes the Opportunities for Nurturance subscale. This subscale (a) assesses perceptions of whether others need the person’s support rather than perceptions of whether others are available to provide social support to the person, (b) is not as highly correlated with the other subscales, and (c) has a lower factor loading, which support the notion that it might be tapping into a different dimension of social support than the other subscales (Cutrona & Russell, 1987). Response options from the scale range from 1 (strongly disagree ) to 4 (strongly agree ), with the minimum and maximum possible scores on the 20­item version being 20 and 80. Internal reliability of the 20­item version of the scale in the present sample was acceptable (α = .93).

Covariates

Posture (e.g., standing, sitting), physical activity, and personal temperature were controlled for as withinsubject covariates for momentary ABP readings, with prior research consistently implicating these factors as variables that influence BP (Barone­Gibbs et al., 2023;

Fan et al., 2023; Takahiro et al., 2022). Participant­level covariates were also accounted for, including age, sex (Wang et al., 2006), relationship status, yearly household income, and BMI (Obeid et al., 2016), which are commonly associated with BP and psychosocial variables. Relationship status was measured on a scale from 0 (no relationship) to 7 (married), which numbers in between these anchors representing different levels of commitment to a partner. BMI was calculated based on height and weight measured by the researchers.

Procedure

Participants were recruited through word of mouth, flyers, bus ads, university participant research registry, and Craigslist. Participants eligible for screening were called. During this call, participants were given an overview of the study (i.e., study purpose, procedure, and compensation) and asked for verbal consent to proceed with the screening. Participants were eligible if they were African American, between the ages of 18 and 30, native English speakers, technologically proficient, had no past or present cardiovascular health or serious mental health conditions and were not taking medication affecting BP, pregnant or attempting to conceive, or working overnight shifts. Further, due to other aims not relevant to the present report, individuals who had abstained completely from any substance use (alcohol, tobacco, or marijuana) within the past year were not eligible.

At the first in­person laboratory visit, participants were given a more in ­ depth overview of the study, reviewed and signed a written consent form, completed a baseline questionnaire, and were taught how to use the equipment. During the monitoring phase, which lasted two days, participants wore BP cuffs and completed EMAs. Participants self­initiated the beginning­of­day questionnaire as soon as they woke up, were prompted to complete the hourly EMA questionnaires, and selfinitiated an end­of­day questionnaire before they went to sleep. Participants were called by research assistants at the end of the first day of monitoring.

At the final in ­ person laboratory visit, participants returned the equipment and completed a final survey. Participants were compensated up to $135 for completing the full study protocol, which included both laboratory visits, at least 80% of the hourly EMA questionnaires, and at least 80% of the blood pressure readings. Participants who met these criteria were also entered into a raffle for a chance to win an additional $55 bonus. Those who did not meet full completion thresholds were provided with partial compensation based on the extent of their participation: $25 for completing Visit 1, $65 for completing the monitoring period with 50–79% compliance, $85 for 80% or greater

compliance, and $25 for completing Visit 2. We reimbursed participants for parking or public transportation expenses for all in­person visits.

Analysis

SPSS Version 27 was used to conduct all statistical analyses. Mixed multilevel modeling accounted for the hierarchical (i.e., observations within days, days within people) nature of our data. Models used an autoregressive variance structure. Statistical significance was determined at the p < .05 level. The primary independent variables were NII, social support, and the interaction term combining NII and social support. The dependent variable in the analyses was ABP (systolic ABP and diastolic ABP were the dependent variable in separate analyses). Models controlled for the between­subjects factors of age, sex, relationship status, income, and BMI as well as the within­subjects factors of posture, physical activity, and temperature. Significant interactions were followed up with simple effects testing at low and high levels of social support as indicated by scores above and below the median social support score.

Results

Descriptive Statistics

Nearly half of the EMA data points (43%) involved a newly reported social interaction. Of these new social interactions, approximately 31% occurred less than 10 minutes prior to the associated hourly ABP assessment or were ongoing during the hourly ABP assessment, and 34% occurred between 10 and 45 minutes prior to the associated hourly ABP assessment. The majority of interactions (63%) involved only one other person, and interactions involving a large group of four or more people were less common (11.6%). Approximately 31% of the interactions involved a friend, 21% involved a coworker, 20% involved a spouse or partner, 21% involved a family member other than a spouse, and 14% involved a stranger. Approximately 65% of interactions involved a Black interaction partner, 32% involved a White interaction partner, and 22% involved interaction partners that were either Asian, Latino, Native American, or multiracial.

Mean systolic ABP was 132.06 (SD = 16.96), and mean diastolic ABP was 78.64 (SD = 13.27). Mean resting systolic blood pressure was 117.26 (SD = 11.74) and mean resting diastolic blood pressure was 75.24 (SD = 9.06), reflecting the fact that we excluded individuals with diagnosed hypertension or cardiovascular issues. Approximately 89% of participants exhibited resting systolic blood pressures within ranges that would not be considered high blood pressure, and approximately 74% exhibited resting diastolic blood pressures within ranges that would not be considered high blood pressure.

In terms of interaction quality, 61.6% of new social interactions involved some negativity according to at least one of the negative interaction items, with 21% involving at least moderate negativity according to at least one of the negative interaction items (i.e., at least a 4 on the 6­point response scale). Mean global social support scores were 66.74 (SD = 8.88), with a range of 45 to 78.

NII and ABP

Mixed modeling controlling for age, sex, relationship status, BMI, income, and ABP momentary covariates (posture, temperature comfort, movement) found that the association between NII and ABP was moderated by perceived baseline social support (systolic ABP F(1,407) = 9.17, p = .003, ƒ2 = .021 and diastolic ABP F(1,392) = 8.17, p = .004, ƒ2 = .03).

See Table 2 for results of the systolic and diastolic multilevel models. Although main effects are not of focus in this study and will not be interpreted further in light of the significant interactions, each analysis also demonstrated main effects of NII (systolic ABP F(1,407) = 7.98, p = .005, ƒ2 = .01 and diastolic ABP F(1,393) = 8.91, p = .003, ƒ2 = .02) and social support (systolic ABP F(1,249) = 14.90, p < .001, ƒ2 = .03 and diastolic ABP F(1,279) = 12.56, p = .001, ƒ2 = .03). As can be seen in more detail in Table 2, significant covariates were age (only systolic significant), sex (with male sex associated with higher ABP), relationship status, income, posture, and movement (only systolic significant). BMI and temperature comfort were not significant, ps > .27.

Social Support Moderation Follow-Up

Follow­up analyses demonstrated that, for systolic ABP, NII did not influence the ABP of those who reported having low social support (p = .43) but significantly decreased the ABP of those who reported having high social support (b = ­1.47, p = .025, ƒ2 = .02). For diastolic ABP, NII increased the ABP of those who reported having low social support (b = 0.90, p = .031, ƒ2 = .02) and did not influence the ABP of those who reported having high social support (p = .55).

Discussion

The current study results demonstrated, for the first time, that social support moderates the association between momentary NII and ABP. Prior research has demonstrated the impact of social interactions on cardiovascular health, but the real­time effects of daily social interactions—and particularly NII—on ABP were understudied and social support had not been explored

1Please note that there is currently no widespread consensus as to calculating effect sizes for multilevel models. We utilized the local Cohen’s ƒ2 effect size as recommended by many (e.g., Lorah, 2018; Selya et al., 2012) for multilevel regression models.

as a moderator. The present study addressed these gaps by examining the moderating effects of social support on the relationship between NII and ABP among African American emerging adults. Due to a small sample size and specificity of our sample, we advise against generalizing the results of this paper across different populations. Nevertheless, we believe this paper contributes to extant literature for several reasons. Firstly, the EMA method allowed us to collect multiple observations of the participants, which increased the overall number of data points. Repeated ABP readings exhibit high reproducibility, which strengthens the reliability of our findings and partially compensates for the small sample size (Barnes et al., 2002). Secondly, EMA data points possess greater predictive capacity for cardiovascular measures than the same number of readings conducted in an office setting, which makes our data points more reliable and ecologically valid (Tanner et al., 2025).

Moderating Effects of Perceived Social Support

Although the effects of NII on ABP were relatively small, effect sizes across the literature focused on predicting ABP from psychosocial factors tend to be small as well. For example, a meta ­ analysis has shown that work­related stressors such as job strain are associated with modest increases in systolic ABP—4 to 5 mm Hg systolic and 3 mm Hg diastolic on average across the day (Landsbergis et al., 2013). Zawadzki et al. (2023) found that momentary increases in NIIs corresponded

TABLE 2

Multilevel Model of Momentary Ambulatory Blood Pressure

Note. The following abbreviations reflect their respective variables: Temp Comf = temperature comfort; NII = negative interpersonal interaction; SS = social support.

* p < .05. ** p < .01. *** p < .001. ƒ2 = localized effect size for multilevel regressions. - = effect size less than .01, i.e., trivial or nonexistent effect.

to small but significant elevations in ABP (systolic b = 0.06 mm Hg; diastolic b = 0.05 mm Hg). Although smaller in magnitude compared to the Landsbergis et al. (2013) study, effects in EMA studies like the current study and that of Zawadzki were captured at the momentary level and may accumulate over time. Thus, the smaller effect sizes observed in our study seems consistent with a broader pattern in ABP research. Among individuals with low perceived social support, NII had no significant effect on systolic ABP but was significantly associated with increases in diastolic ABP. In contrast, among individuals with high perceived social support, NII was associated with lower systolic ABP but not diastolic ABP. Our findings both converge with and diverge from previous findings.

The significant decrease in systolic ABP during moments of higher NII for individuals with high perceived social support, though unexpected, aligns with research highlighting the protective effects of strong social networks on cardiovascular health (Boen & Yang, 2018; Bowen et al., 2014; Cornwell & Waite, 2012). This finding also supports the buffering hypothesis, which suggests that social support can mitigate stress and its adverse effects on BP (Bowen et al., 2014). It is possible that the positive influence of perceived social support on ABP may be activated during times of high stress—high levels of NII in this case. Although social support was assessed globally in the present study, experimental paradigms assessing social support provide potential insight into this idea of conjuring one’s social support to confront challenges. A meta­analysis exploring whether experimentally manipulated social support (e.g., activating ideas of social support or presence of a supportive other) buffers cardiovascular reactivity when exposed to lab­based stress shows that participants who have or imagine social support have reduced cardiovascular reactivity compared to those who do not (Teoh & Hilmert, 2018). In daily NII, a person might cope by recalling that there are people who love and care for them, which could lead to comfort, relief, and positive emotions that then lower ABP. Further study is needed to test whether ecologically assessed presence or conjuring of social support among individuals with high global social support is the mechanism for reducing systolic blood pressure in times of NII. Other studies have found similar patterns of lower ABP in those with more stress but more social support. Specifically, Tobe and colleagues (2007) found that high spousal support was associated with lower systolic ABP in those with high work stress. An alternative explanation could be variability across negative social interactions; greater variability in intensity of the negative interaction has been linked to better stress coping and lower reactivity in systolic blood pressure (Don

et al., 2023). Likewise, oxytocin is a hormone consistently tied to lower blood pressure; it has been found to decrease anxiety and cortisol levels primarily under conditions of stress combined with high social support (Riem et al., 2020). An additional study found that active displays of social support reduced cardiovascular reactivity—a phenomenon Gerin et al. attributed to the presence of social support, which they posited may have altered participants’ cognitive appraisals of the situation to perceive the interaction as less threatening, consequently dampening the body’s stress response and lowering blood pressure (1992). These theories offer a potential rationalization of the observed results and would benefit from further research on the matter.

Although the simple effects follow ­ up on the moderation effect for diastolic ABP did not yield significance, the interaction and the strong simple effects trend help complete the picture and align with our original hypothesis: that NII may be more strongly associated with higher ABP in those with low perceived social support. Again, although the interaction was significant, the simple effects were not, so this interaction should be interpreted with caution. Further, previous research suggests that diastolic ABP may, in general, be less reactive to psychosocial buffers such as perceived social support compared to systolic ABP. In support of this notion, some studies found that social support is significantly linked to systolic but not diastolic ABP (e.g., Bowen et al., 2014; Ituarte et al., 1999; Tobe et al., 2007). For example, Tobe and colleagues (2007) found moderated effects of social support and stress on systolic but not diastolic ABP.

Limitations

This study is not without limitations. Prior research suggests that ABP varies by race, ethnicity, sex, age, and other demographic characteristics (e.g., Beatty Moody et al., 2016; Busse et al., 2017; Mencía­Ripley et al., 2015; Wang et al., 2006). For example, it is widely recognized that African Americans and other minority groups generally have higher BP compared to White people, highlighting the need to study factors that may lead to racialized health inequities (Lackland, 2014; Mensah et al., 2005; Munter et al., 2015). Other research indicates that younger and older adults exhibit differing patterns in ABP, such that BP tends to increase with age (Beatty Moody et al., 2017; Ishikawa et al., 2011). These variations emphasize the need for future research to explore how NII and social support affect ABP across diverse populations, as these factors may interact with demographic characteristics to influence health outcomes uniquely. Other demographic factors may have influenced the outcomes of this study as well. For instance, relationship

quality was not assessed, but many studies support the finding that ambivalent relationships—including those with platonic friends, family members, and spouses—are associated with higher diastolic and systolic ABP than purely negative relationships (Birmingham et al., 2019; Holt­Lunstad & Clark, 2014; Rook et al., 2012). Age is a salient variable in this association, as older age has been tied to fewer ambivalent relationships (Rook et al., 2012). Because this study focused on emerging adults, who are more likely to encounter and maintain ambivalent ties, they may not have experienced the protective effects of high social support in the same way.

Another limitation involves EMA. There is the risk of participant fatigue due to frequent and continuous data collection, which can be more time­consuming than other methods such as periodic interviews (Moskowitz & Young, 2006). Moreover, the need to complete hourly surveys may additionally alter participants’ natural behavior. To combat these concerns, the hourly survey was kept brief, but we were unable to assess NII and other constructs through more thorough measures. Despite this, the NII scale was reliable and showed substantial within­person variability.

Because BP readings were initiated and recorded once every hour, this allowed for the possibility of a substantial lapse in time between the occurrence of a NII and the recorded ABP reading. During this time, other factors or events such as a positive interpersonal interaction may have transpired, which has been linked to lower average BP readings (Cornelius et al., 2018). Nevertheless, our hourly readings were conducted over the course of two days, resulting in a more comprehensive representation of day­to­day interactions.

Avenues for Future Research

Despite limitations, this study had many strengths, including using the combination of EMA and ABP to identify rich fluctuations in daily life social interactions and blood pressure physiology and, thus, to explore the relationship between the two in reliable, sensitive, and dynamic ways. Future work is needed to understand when and under which contexts social support is most effective as a protective factor in the dynamic relationship between NII and ABP. In this regard, it may be particularly beneficial to assess whether associations found vary within the context of different modalities of social interaction, as negative social interactions can occur across a variety of platforms (e.g., over text, online, in­person, video call, audio call).

Additionally, as moderation of the relationship between NII and ABP is underexplored, it would be useful to consider other moderating factors related to perceived social support such as religiosity and age,

two variables closely involved in fostering a sense of community and self ­ esteem, which may also act as protective buffers against NII (Fayçal & Nouredinne, 2015; von Soest et al., 2016; Wantiyah et al., 2020). For instance, religiosity may provide a coping mechanism that mitigates stress­related BP increases (Hovey et al., 2014), and age can affect BP variability through factors such as social isolation or increased socioemotional resilience that become increasingly common with older age (Charles & Carstensen, 2010; Masini & Barrett, 2008).

We also recommend future studies use intersectionality as a theoretical framework to analyze the relationship between NII, perceived social support, and ABP. Rooted in Black feminist theory, intersectionality provides a lens to observe how power relations within and between social groups intertwine to influence individual experiences (Hudson et al., 2024; Kapilashrami & Hankivsky, 2018). This framework is crucial for understanding the complex ways in which different dimensions of identity–such as race, sex, sexual orientation, socioeconomic status, and disability–intersect to affect relationships between variables. Furthermore, intersectionality acknowledges that overlapping identities can contribute to differences in coping mechanisms or the experience of interpersonal stressors, which other analyses may overlook. Through this framework, researchers can develop a more nuanced understanding of how systemic inequalities and social determinants of health collectively impact ABP. Though our study focused on African American emerging adults, applying an intersectional framework to a broader sample can reveal unique patterns that exist within demographic groups and develop understanding of these relationships across diverse populations.

Conclusion

Our sample consisted of African American emerging adults and examined the moderating role of social support in the association between NII and ABP. NII was significantly associated with decreases in systolic ABP for those reporting high social support. This finding is important, as systolic BP has been more strongly linked to cardiovascular risk than diastolic BP (Benetos et al., 2003; Conen & Bamberg, 2008; Uchino et al., 2022).

The findings of this study also contribute to a growing body of research emphasizing the importance of strong social networks as a protective measure against increasing BP, which has been robustly linked to detrimental conditions such as CVD (Birmingham et al., 2023; Byrd et al., 2022; Lei et al., 2019; Suchy­Dicey et al., 2022). The clinical implications of this study therefore include the promotion of social support in instances of heightened stress as well as the avoidance

of negative social interaction. In addition, healthcare providers can encourage the creation and maintenance of strong supportive relationships as a holistic method of managing risk factors associated with increasing BP.

References

Andersson, L. M., & Pearson, C. M. (1999). Tit for tat? The spiraling effect of incivility in the workplace. Academy of Management Review, 24(3), 452–471. https://doi.org/10.5465/amr.1999.2202131

Angeli, F., Gianpaolo Reboldi, Poltronieri, C., Bartolini, C., Cinzia D’Ambrosio, Valentina de Filippo, & Verdecchia, P. (2014). Clinical utility of ambulatory blood pressure monitoring in the management of hypertension. Expert Review of Cardiovascular Therapy, 12(5), 623–634. https://doi.org/10.1586/14779072.2014.903155

Arnett, J. J., Žukauskienė, R., & Sugimura, K. (2014). The new life stage of emerging adulthood at ages 18–29 years: Implications for mental health. The Lancet Psychiatry, 1(7), 569–576. https://doi.org/10.1016/s2215-0366(14)00080-7

Barnes, V. A., Johnson, M. H., Dekkers, J. C., & Treiber, F. A. (2002). Reproducibility of ambulatory blood pressure measures in African-American adolescents. Ethnicity & Disease, 12(4), 101–106. https://www.jstor.org/stable/45410870

Barone-Gibbs, B., Muldoon, M. F., Conroy, M. B., Paley, J. L., Shimbo, D., & Perera, S. (2023). Influence of recent standing, moving, or sitting on daytime ambulatory blood pressure. Journal of the American Heart Association, 12(17), e029999. https://doi.org/10.1161/JAHA.123.029999

Beatty Moody, D. L., Waldstein, S. R., Tobin, J. N., Cassells, A., Schwartz, J. C., & Brondolo, E. (2016). Lifetime racial/ethnic discrimination and ambulatory blood pressure: The moderating effect of age. Health Psychology, 35(4), 333–342. https://doi.org/10.1037/hea0000270

Behbodikhah, J., Ahmed, S., Elyasi, A., Kasselman, L. J., De Leon, J., Glass, A. D., & Reiss, A. B. (2021). Apolipoprotein B and cardiovascular disease: Biomarker and potential therapeutic target. Metabolites, 11(10), 690. https://doi.org/10.3390/metabo11100690

Benetos, A., Thomas, F., Bean, K., Gautier, S., Smulyan, H., & Guize, L. (2003). Prognostic value of systolic and diastolic blood pressure in treated hypertensive men. Archives of Internal Medicine, 162(5), 577–581. https://doi.org/10.1001/archinte.162.5.577

Berkman, L. F., Glass, T., Brissette, I., & Seeman, T. E. (2000). From social integration to health: Durkheim in the new millennium. Social Science & Medicine, 51(6), 843–857. https://doi.org/10.1016/S0277-9536(00)00065-4

Birmingham, W. C., Wadsworth, L. L., Hung, M., Li, W., & Herr, R. M. (2019). Ambivalence in the early years of marriage: Impact on ambulatory blood pressure and relationship processes. Annals of Behavioral Medicine, 53(12), 1069–1080. https://doi.org/10.1093/abm/kaz017

Birmingham, W. C., Jorgensen, A., Hancock, S., Wadsworth, L. L., & Hung, M. (2023). Social support: The effect on nocturnal blood pressure dipping. International Journal of Environmental Research and Public Health, 20(5), 4579. https://doi.org/10.3390/ijerph20054579

Boen, C., & Yang, Y. C. (2018). Social relationships and cardiovascular health: Underlying mechanisms, life course processes, and future directions. Elsevier EBooks, 291–296. https://doi.org/10.1016/b978-0-12-811279-3.00025-2

Bolger, N., Zuckerman, A., & Kessler, RC. (2000). Invisible support and adjustment to stress. Journal of Personality and Social Psychology, 79(6), 953–961. https://doi.org/10.1037/0022-3514.79.6.953

Bonito, J. A., Ruppel, E. K., & Keyton, J. (2012). Reliability estimates for multilevel designs in group research. Small Group Research, 43(4), 443–467. https://doi.org/10.1177/1046496412437614

Bowen, K. S., Uchino, B. N., Birmingham, W., Carlisle, M., Smith, T. W., & Light, K. C. (2014). The stress-buffering effects of functional social support on ambulatory blood pressure. Health Psychology, 33(11), 1440–1443. https://doi.org/10.1037/hea0000005

Brondolo, E., Rieppi, R., Erickson, S. A., Bagiella, E., Shapiro, P. A., McKinley, P., & Sloan, R. P. (2003). Hostility, interpersonal interactions, and ambulatory blood pressure. Psychosomatic Medicine, 65(6), 1003–1011. https://doi.org/10.1097/01.psy.0000097329.53585.a1

Brown, B. B. (1978). Social and psychological correlates of help-seeking behavior among urban adults. American Journal of Community Psychology, 6(5), 425–39. https://doi.org/10.1007/BF00941419

Busse, D., Yim, I. S., & Campos, B. (2017). Social context matters: Ethnicity, discrimination and stress reactivity. Psychoneuroendocrinology, 83, 187–193. https://doi.org/10.1016/j.psyneuen.2017.05.025

Byrd, D. R., Jiang, Y., Zilioli, S., Thorpe, R. J., Lichtenberg, P. A., Whitfield, K. E., & Newman, A. B. (2022). The interactive effects of education and social support on blood pressure in African Americans. The Journals of Gerontology, 77(2), e98–e106. https://doi.org/10.1093/gerona/glab289

Charles, S., & Carstensen, L. L. (2010). Social and emotional aging. Annual Review of Psychology, 61(1), 383–409. https://doi.org/10.1146/annurev.psych.093008.100448

Chun, J. W., & Lee, M. J. (2017). When does individuals’ willingness to speak out increase on social media? Perceived social support and perceived power/ control. Computers in Human Behavior, 74, 120–129. https://doi.org/10.1016/j.chb.2017.04.010

Conen, D., & Bamberg, F. (2008). Noninvasive 24-h ambulatory blood pressure and cardiovascular disease: A systematic review and meta-analysis. Journal of Hypertension, 26(7), 1290–1299. https://doi.org/10.1097/hjh.0b013e3282f97854

Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin, 98(2), 310–357. https://doi.org/10.1037/0033-2909.98.2.310

Cornelius, T., Birk, J. L., Edmondson, D., & Schwartz, J. E. (2018). The joint influence of emotional reactivity and social interaction quality on cardiovascular responses to daily social interactions in working adults. Journal of Psychosomatic Research, 108, 70–77. https://doi.org/10.1016/j.jpsychores.2018.02.010

Cornwell, E. Y., & Waite, L. J. (2012). Social network resources and management of hypertension. Journal of Health and Social Behavior, 53(2), 215–231. https://doi.org/10.1177/0022146512446832

Cutrona, C. E., & Russell, D. W. (1987). The provisions of social relationships and adaptation to stress. In W. H. Jones & D. Perlman (Eds.), Advances in Personal Relationships (Vol. 1, pp. 37–67). JAI Press. https://www.researchgate.net/publication/271507385_The_Provisions_ of_Social_Relationships_and_Adaptation_to_Stress

Datta, B. K., Coughlin, S. S., Gummadi, A., Mehrabian, D., & Ansa, B. E. (2023). Perceived social support and cardiovascular risk among nonelderly adults in the United States. The American Journal of Cardiology, 209, 146–153. https://doi.org/10.1016/j.amjcard.2023.09.106

Don, B. P., Gordon, A. M., & Berry Mendes, W. (2023). The good, the bad, and the variable: Examining stress and blood pressure responses to close relationships. Social Psychological & Personality Science, 15(3), 288–298. https://doi.org/10.1177/19485506231156018

Fan, P., Xue, X., Hu, J., Qiao, Q., Yin, T., Yang, X., Chen, X., Hou, Y., & Chen, R. (2023). Ambient temperature and ambulatory blood pressure: An hourly-level, longitudinal panel study. Science of the Total Environment, 864, 160854. https://doi.org/10.1016/j.scitotenv.2022.160854

Fayçal, K., & Nouredinne, J. (2015). Relationship between religiosity and selfefficacy among patients with cardiovascular disorders: Field study at the Center CHU Hospital Saadna Abdel Nour, Setif Algeria. Mağallaẗ Al-Dirāsāt al-Tarbawīyyaẗ Wa-al-Nafsīyyaẗ, 9(1), 109. https://doi.org/10.24200/jeps.vol9iss1pp109-123

Gerin, W., Pieper, C., Levy, R., & Pickering, T. G. (1992). Social support in social interaction: A moderator of cardiovascular reactivity. Psychosomatic Medicine, 54(3), 324–336. https://doi.org/10.1097/00006842-199205000-00008

Goosby, B. J., Malone, S., Richardson, E., Cheadle, J. E., & Williams, D. (2015). Perceived discrimination and markers of cardiovascular risk among low-income African American youth. American Journal of Human Biology, 27(4), 546–552. https://doi.org/10.1002/ajhb.22683

Hochberg, Z., & Konner, M. (2020). Emerging adulthood, a pre-adult life-history stage. Frontiers in Endocrinology (Lausanne), 10, 918–918. https://doi.org/10.3389/fendo.2019.00918

Holt-Lunstad, J., Uchino, B. N., Smith, T. W., Olson-Cerny, C., & Nealey-Moore, J. B. (2003). Social relationships and ambulatory blood pressure: Structural and qualitative predictors of cardiovascular function during everyday social interactions. Health Psychology, 22(4), 388–397. https://doi.org/10.1037/0278-6133.22.4.388

Holt-Lunstad, J., & Clark, B. D. (2014). Social stressors and cardiovascular response: Influence of ambivalent relationships and behavioral ambivalence. International Journal of Psychophysiology, 93(3), 381–389. https://doi.org/10.1016/j.ijpsycho.2014.05.014

Hovey, J. D., Hurtado, G., Morales, L. R. A., & Seligman, L. D. (2014). Religion-based emotional social support mediates the relationship between intrinsic religiosity and mental health. Archives of Suicide Research, 18(4), 376–391. https://doi.org/10.1080/13811118.2013.833149

Huang, Y., Heflin, C. M., & Validova, A. (2021). Material hardship, perceived stress, and health in early adulthood. Annals of Epidemiology, 53, 69–75.e3. https://doi.org/10.1016/j.annepidem.2020.08.017

Hudson, S. T. J., Myer, A., & Berney, E. C. (2024). Stereotyping, prejudice, and discrimination at the intersection of race and sex: An intersectional theory primer. Social and Personality Psychology Compass, 18(2). https://doi.org/10.1111/spc3.12939

Ishikawa, J., Ishikawa, Y., Edmondson, D., Pickering, T. G., & Schwartz, J. E. (2011). Age and the difference between awake ambulatory blood pressure and office blood pressure. Blood Pressure Monitoring, 16(4), 159–167. https://doi.org/10.1097/mbp.0b013e328346d603

Ituarte, P. H. G., Kamarck, T. W., Thompson, H. S., & Bacanu, S. (1999). Psychosocial mediators of racial differences in nighttime blood pressure dipping among normotensive adults. Health Psychology, 18(4), 393–402. https://doi.org/10.1037/0278-6133.18.4.393

Jones, D. M., Masyn, K. E., & Spears, C. A. (2023). Associations among discrimination, psychological functioning, and substance use among US Black adults aged 18–28: Moderation by racial attribution and sex. Journal of Substance Use and Addiction Treatment, 153, Article 209080. https://doi.org/10.1016/j.josat.2023.209080

Jones, S. C., Bilous, M., Winship, S., Finn, P., & Goodwin, J. (2004). Validation of the OSCAR 2 oscillometric 24-hour ambulatory blood pressure monitor according to the International Protocol for the validation of blood pressure measuring devices. Blood Pressure Monitoring, 9(4), 219–223. https://doi.org/10.1097/00126097-200408000-00007

Kamarck, T. W. (1992). Recent developments in the study of cardiovascular reactivity: Contributions from psychometric theory and social psychology. Psychophysiology, 29(5), 491–503. https://doi.org/10.1111/j.1469-8986.1992.tb02023.x

Kamarck, T. W., Shiffman, S. M., Smithline, L., Goodie, J. L., Thompson, H. S., Ituarte, P. H. G., . . Perz, W. (1998). The diary of ambulatory behavioral states: A new approach to the assessment of psychosocial influences on ambulatory cardiovascular activity. In D. S. Krantz, & A. Baum (Eds.), Technology and methods in behavioral medicine (pp. 163–193, Chapter xviii, 305 Pages). Lawrence Erlbaum Associates Publishers.

Kapilashrami, A., & Hankivsky, O. (2018). Intersectionality and why it matters to global health. The Lancet, 391(10140), 2589–2591. https://doi.org/10.1016/s0140-6736(18)31431-4

Kuper, H., & Marmot, M. (2003). Job strain, job demands, decision latitude, and risk of coronary heart disease within the Whitehall II Study. Journal of Epidemiology & Community Health, 57(2), 147–153. https://doi.org/10.1136/jech.57.2.147

Lackland, D. T. (2014). Racial differences in hypertension: Implications for high blood pressure management. The American Journal of the Medical Sciences, 348(2), 135–138. https://doi.org/10.1097/MAJ.0000000000000308

Landsbergis, P. A., Dobson, M., Koutsouras, G., & Schnall, P. (2013). Job strain and ambulatory blood pressure: A meta-analysis and systematic review. American Journal of Public Health, 103(3), e61–71. https://doi.org/10.2105/ajph.2012.301153

Lei, P., Wu, Y., Ye, R., Sunil, T. S., & Zhou, H. (2019). Social support and blood pressure: A cross-sectional study. The Lancet, 394(S51). https://doi.org/10.1016/S0140-6736(19)32387-6

Lorah, J. (2018). Effect size measures for multilevel models: Definition, interpretation, and TIMSS example. Large-Scale Assessments in Education, 6(1). https://doi.org/10.1186/s40536-018-0061-2

Masini, B. E., & Barrett, H. A. (2008). Social support as a predictor of psychological and physical well-being and lifestyle in lesbian, gay, and bisexual adults aged 50 and over. Journal of Gay & Lesbian Social Services, 20(1–2), 91–110. https://doi.org/10.1080/10538720802179013

Matud, M. P., Díaz, A., Bethencourt, J. M., & Ibáñez, I. (2020). Stress and psychological distress in emerging adulthood: A sex analysis. Journal of Clinical Medicine, 9(9), 2859. https://doi.org/10.3390/jcm9092859

Mencía-Ripley, A., Schwartz, J., & Brondolo, E. (2015). Sex identity, interpersonal interactions, and ambulatory blood pressure. Revista Interamericana de Psicología, 49(2), 261–271. https://doaj.org/article/3f17526b8281460c98a799b40a70bcf4

Mensah, G. A., Mokdad, A. H., Ford, E. S., Greenlund, K. J., & Croft, J. B. (2005). State of disparities in cardiovascular health in the United States. Circulation, 111(10), 1233–1241. https://doi.org/10.1161/01.CIR.0000158136.76824.04

Moskowitz, D. S., & Young, S. N. (2006). Ecological momentary assessment: What it is and why it is a method of the future in clinical

psychopharmacology. Journal of Psychiatry and Neuroscience, 31(1), 13–20. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1325062/

Muntner, P., Lewis, C. E., Diaz, K. M., Carson, A. P., Kim, Y., Calhoun, D., Yano, Y., Viera, A. J., & Shimbo, D. (2015). Racial differences in abnormal ambulatory blood pressure monitoring measures: Results from the Coronary Artery Risk Development in Young Adults (CARDIA) study. American Journal of Hypertension, 28(5), 640–648. https://doi.org/10.1093/ajh/hpu193

Nezlek, J. B. (2017). A practical guide to understanding reliability in studies of within-person variability. Journal of Research in Personality, 69, 149–155. https://doi.org/10.1016/j.jrp.2016.06.020

Obeid, T., Megill, R., Schmidt, E., Wootan, M., Greco, D.J., Alsulaim, H., Schweitzer, M., Magnuson, T., & Steele, K. (2016). Factors that influence blood pressure following bariatric surgery. Surgery for Obesity and Related Diseases, 12(7), 83–84. https://doi.org/10.1016/j.soard.2016.08.129

Perkovic, V., Huxley, R., Wu, Y., Prabhakaran, D., & MacMahon, S. (2007). The burden of blood pressure-related disease. Hypertension, 50(6), 991–997. https://doi.org/10.1161/hypertensionaha.107.095497

Riem, M. M. E., Kunst, L. E., Bekker, M. H. J., Fallon, M., & Kupper, N. (2020). Intranasal oxytocin enhances stress-protective effects of social support in women with negative childhood experiences during a virtual Trier Social Stress Test. Psychoneuroendocrinology, 111, Article 104482. https://doi.org/10.1016/j.psyneuen.2019.104482

Rook, K. S., Luong, G., Sorkin, D. H., Newsom, J. T., & Krause, N. (2012). Ambivalent versus problematic social ties: Implications for psychological health, functional health, and interpersonal coping. Psychology and Aging, 27(4), 912–923. https://doi.org/10.1037/a0029246

Schoenthaler, A. M., Schwartz, J., Cassells, A., Tobin, J. N., & Brondolo, E. (2010). Daily interpersonal conflict predicts masked hypertension in an urban sample. American Journal of Hypertension, 23(10), 1082–1088. https://doi.org/10.1038/ajh.2010.141

Selya, A. S., Rose, J. S., Dierker, L. C., Hedeker, D., & Mermelstein, R. J. (2012). A practical guide to calculating Cohen’s f2, a measure of local effect size, from PROC MIXED. Frontiers in Psychology, 3 https://doi.org/10.3389/fpsyg.2012.00111

Shapiro, D., Jamner, L. D., & Goldstein, I. B. (1997). Daily mood states and ambulatory blood pressure. Psychophysiology, 34(4), 399–405. https://doi.org/10.1111/j.1469-8986.1997.tb02383.x

Shiffman, S., Stone, A. A., & Hufford, M. R. (2008). Ecological momentary assessment. Annual Review of Clinical Psychology, 4(1), 1–32. https://doi.org/10.1146/annurev.clinpsy.3.022806.091415

Sloan, R. P., Schwarz, E., McKinley, P. S., Weinstein, M., Love, G., Ryff, C., Mroczek, D., Choo, T.-H., Lee, S., & Seeman, T. (2017). Vagally-mediated heart rate variability and indices of well-being: Results of a nationally representative study. Health Psychology, 36(1), 73–81. https://doi.org/10.1037/hea0000397

Smith, T. W., Birmingham, W., & Uchino, B. N. (2012). Evaluative threat and ambulatory blood pressure: Cardiovascular effects of social stress in daily experience. Health Psychology, 31(6), 763–766. https://doi.org/10.1037/a0026947

Sneed, R. S., & Cohen, S. (2014). Negative social interactions and incident hypertension among older adults. Health Psychology, 33(6), 554–565. https://doi.org/10.1037/hea0000057

Suchy-Dicey, A., Eyituoyo, H., O’Leary, M., Cole, S.A., Traore, A., Verney, S., Howard, B., Manson, S., Buchwald, D., & Whitney, P. (2022). Psychological and social support associations with mortality and cardiovascular disease in middle-aged American Indians: The strong heart study. Social Psychiatry and Psychiatric Epidemiology, 57(7), 1421–1433. https://doi.org/10.1007/ s00127-022-02237-7

Takahiro, K., Hoshide, S., Kanazawa, H., Taki, M., Suzuki, N., Kotruchin, P., & Kario K. (2022). The effect of mountaineering on the association between blood pressure and physical activity: A new multi-sensor ambulatory bloodpressure monitoring device. The Mount Fuji Study. The Journal of Clinical Hypertension, 24(9), 1236–1241. https://doi.org/10.1111/jch.14525

Tanner, R. M., Jaeger, B. C., Bradley, C. K., Thomas, S. J., Min, Y.-I., Hardy, S. T., Irvin, M. R., Shimbo, D., Schwartz, J. E., & Muntner, P. (2025). Blood pressure on ambulatory monitoring and risk for cardiovascular disease and all-cause mortality: Ecological validity or measurement reliability? American Journal of Hypertension, 38(2), 111–119. https://doi.org/10.1093/ajh/hpae133

Teoh, A. N., & Hilmert, C. (2018). Social support as a comfort or an encouragement: A systematic review on the contrasting effects of social support on cardiovascular reactivity. British Journal of Health Psychology, 23(4), 1040–1065. https://doi.org/10.1111/bjhp.12337

Thoits, P. A. (2011). Mechanisms linking social ties and support to physical

and mental health. Journal of Health and Social Behavior, 52(2), 145–-161. https://doi.org/10.1177/0022146510395592

Tobe, S. W., Kiss, A., Sainsbury, S., Jesin, M., Geerts, R., & Baker, B. (2007). The impact of job strain and marital cohesion on ambulatory blood pressure during 1 year: The double exposure study. American Journal of Hypertension, 20(2), 148–153. https://doi.org/10.1016/j.amjhyper.2006.07.011

Uchino, B. N., Baucom, B. R. W., Landvatter, J., de Grey, R. G. K., Tacana, T., Flores, M., & Ruiz, J. M. (2022). Perceived social support and ambulatory blood pressure during daily life: A meta-analysis. Journal of Behavioral Medicine, 45(4), 509–517. https://doi.org/10.1007/s10865-021-00273-3

von Soest, T., Wichstrøm, L., & Kvalem, I. L. (2016). The development of global and domain-specific self-esteem from age 13 to 31. Journal of Personality and Social Psychology, 110(4), 592–608. https://doi.org/10.1037/pspp0000060 Wang, X., Poole, J. C., Treiber, F. A., Harshfield, G. A., Hanevold, C. D., & Snieder, H. (2006). Ethnic and sex differences in ambulatory blood pressure trajectories: Results from a 15-year longitudinal study in youth and young adults. Circulation, 114(25), 2780–2787. https://doi.org/10.1161/CIRCULATIONAHA.106.643940

Wantiyah, W., Putri Rivani, F. R., & Hakam, M. (2020). The correlation between religiosity and self-efficacy in patients with coronary artery disease. Belitung Nursing Journal, 6(5), 172–176. https://doi.org/10.33546/bnj.1134

Wethington, E., & Kessler, R. C. (1986). Perceived support, received support, and adjustment to stressful life events. Journal of Health and Social Behavior, 27(1), 78. https://doi.org/10.2307/2136504

Zawadzki, M. J., Graham-Engeland, J. E., Robles, P. L., Hussain, M., Fair, E. V., Tobin,

J. N., Cassells, A., & Brondolo, E. (2023). Acute experiences of negative interpersonal interactions: Examining the dynamics of negative mood and ambulatory blood pressure responses among Black and Hispanic urban adults. Annals of Behavioral Medicine, 57(8), 630–639. https://doi.org/10.1093/abm/kaad033

Author Note.

This research was financially supported in part by the Center for Race and Social Problems, School of Social Work, University of Pittsburgh pilot grant program (co­PIs: Drs. Nataria Tennille Joseph and Laurel M. Peterson) and the Clinical and Translational Science Institute at the University of Pittsburgh via National Institutes of Health (Grant Number UL1TR000005). The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Emilie J. Chai and Eunji Shin contributed equally to this work and share first authorship. Emilie J. Chai and Eunji Shin took the lead role in hypothesis development, data interpretation, and manuscript writing and revisions. Nataria T. Joseph and Laurel M. Peterson had lead roles in study design, data analysis, table creation, and draft supervision.

Correspondence concerning this article should be addressed to Eunji Shin, Social Science Division, Pepperdine University, Malibu CA, 90263­4372, United States. Email: eunji.shin@pepperdine.edu

The Effects of Correcting Misperceived Norms on Sexual Assault Prevention Intentions and Rape Myth Acceptance

Cristal Lopez, Alyssa Martinez, and Heike I. M. Mahler*

Psychology Department, California State University San Marcos

ABSTRACT. Incidences of sexual assaults on college campuses remain high despite education and prevention efforts. Previous research suggests that correcting misperceptions about behaviors others approve of (injunctive norms) or engage in (descriptive norms) may improve preventative programming efficacy. Providing individuals with feedback that highlights discrepancies between perceived and actual norms may be an effective method of correcting misperceived norms. College students (n = 86) were randomly assigned to receive sexual assault information (educational video condition), feedback to correct misperceptions of injunctive and descriptive norms regarding sexual assault (normative feedback condition), or both (combined condition). Combined condition participants reported significantly higher perceived norms for sexual aggression prevention relative to those who only watched the video, t(82) = 2.39, p = .019, d = 0.53, and those who only received normative feedback, t (82) = 2.68, p = .009, d = 0.59. The combined condition also resulted in greater intentions to engage in sexual assault prevention relative to the normative feedback condition, t(82) = 1.97, p = .052, d = 0.44, but not relative to the educational video condition, t(82) = 1.10, p = .276, d = 0.24. A 2 ­ week follow ­ up revealed that all groups exhibited reduced endorsement of rape myths compared to baseline levels, F (1, 29) = 7.27, p = .012, f = 0.50, and the reduction did not vary reliably by condition. The findings suggest that including feedback to correct misperceived norms holds promise for enhancing the efficacy of college campus initiatives aimed at preventing sexual assault.

Keywords: social norms, sexual assault prevention, college students, rape myths

RESUMEN. Las agresiones sexuales en los universitarios siguen siendo frecuentes a pesar de los esfuerzos de educación y prevención. Estudios previos indican que corregir las ideas erróneas sobre las conductas que otras personas aprueban (normas injuntivas) o realizan (normas descriptivas) puede mejorar la eficacia de los programas preventivos. Ofrecer retroalimentación que muestre las diferencias entre las normas percibidas y las reales puede ser una forma efectiva de corregir esas percepciones equivocadas. En este estudio, estudiantes universitarios ( n = 86) fueron asignados al azar para recibir información sobre agresión sexual (condición de video educativo), retroalimentación para corregir percepciones erróneas sobre normas injuntivas y descriptivas (condición de retroalimentación normativa), o ambas (condición combinada). Participantes en la condición combinada reportaron normas percibidas significativamente más altas sobre la prevención de la agresión sexual en comparación con quienes solo vieron el video, t(82) = 2.39, p = .019, d = 0.53, y con quienes solo recibieron retroalimentación normativa, t(82) = 2.68, p = .009, d = 0.59. La condición combinada también mostró mayores intenciones de participar en la prevención de agresiones sexuales en comparación con la condición de retroalimentación normativa, t(82) = 1.97, p = .052, d = 0.44, aunque no

en comparación con la condición de video educativo, t(82) = 1.10, p = .276, d = 0.24. Dos semanas después, todos los grupos mostraron una menor aceptación de los mitos sobre la violación en comparación con el inicio del estudio, F(1, 29) = 7.27, p = .012, f = 0.50, y esta reducción no varió significativamente entre las condiciones. Los resultados sugieren que incluir retroalimentación para corregir normas erróneamente percibidas puede mejorar la eficacia de las iniciativas universitarias dirigidas a prevenir las agresiones sexuales.

Palabras clave: normas sociales, prevención de agresiones sexuales, estudiantes universitarios, mitos sobre la violación

The persistent and alarming rates of sexual assault on U.S. college campuses demand urgent action. According to the 2019 Campus Climate Survey on Sexual Assault and Misconduct, nearly one in four undergraduate women at 33 of the nation’s major universities has faced sexual assault or misconduct (Cantor et al., 2020). The numbers are unsettling, particularly when considering estimates that 65% of sexual assaults go unreported to law enforcement (Langton et al., 2012) and are therefore not reflected in official prevalence rates. Furthermore, this issue is not exclusive to women as 6.8% of men experience sexual assault during their college years (Cantor et al., 2020). Overall, sexual assault accounts for the largest percentage (43%) of all types of crimes occurring on college campuses in the United States (National Center for Education Statistics, 2022).

It is particularly disconcerting that sexual assault rates remain high, despite the 2013 Campus Sexual Violence Elimination Act requiring colleges to implement training programs aimed at reducing sexual assault (Coker et al., 2016), and evidence that most colleges have complied.

For example, in a survey of over 180,000 students conducted by the Association of American Universities, 80% reported that they had received training and information regarding sexual assault and other types of sexual misconduct (Cantor et al., 2020). The same survey showed that approximately one­third of students believed they had a “very” or “extremely” high level of understanding regarding the definition of sexual assault (37.1%), where to seek assistance if needed (37.1%), and the process of reporting sexual assault (31.5%), an increase from 2015 (Cantor et al., 2020). Despite this increase in knowledge, the survey also found that the rate of sexual assault actually increased by 3% between 2015 and 2019. More recently, alarming reports of increased incidents at a number of universities (Vara, 2023; White, 2024) include a “135% increase in rapes” at Texas State University (Leschber, 2023), five sexual assaults within just one month at a California campus

(Williams, 2023), and the University of Minnesota reporting its highest rate of residence hall rapes in years (Roessler, 2024). Whether these data reflect an increase in sexual assaults or an increase in awareness and reporting of sexual assaults, there clearly is room to improve (or supplement) the preventive training currently provided.

In general, training programs at most colleges are delivered online and focus on educating students about the definitions and prevalence of sexual assault, precautions to take to keep oneself and friends safe, campus policies, reporting procedures, and resources (Banyard et al., 2020). These programs typically prioritize the modification of students’ beliefs in rape myths, which typically serve as an outcome measure in research evaluating the effectiveness of preventative interventions. Rape myths are widely and persistently held misconceptions about sexual assault shaped by sexism and biases individuals may hold that serve to deny and justify sexual aggression (McMahon & Farmer, 2011; O’Connor, 2023). Commonly referenced rape myths include the belief that a woman’s attire or behavior implies consent. Research has established that the acceptance of rape myths not only signifies problematic attitudes and beliefs but also serves as a predictive factor in the commission of sexual assault or a propensity for rape (McMahon & Farmer, 2011). Although increasing awareness, knowledge of preventive behaviors, and the rejection of rape myths are important, these training programs have not sufficiently reduced the incidence of sexual assaults on campuses. A recent systematic review of 31 research studies (conducted between 2001 and 2017) with educational programming concluded with a call for interventions to highlight consequences for perpetrators and to incorporate new social and environmental approaches to sexual assault prevention (Mahoney et al., 2020).

It has been suggested that the efficacy of sexual assault prevention programs may be enhanced by including information to correct mis­perceived social

norms (Collins et al., 2014; Miller & Prentice, 2016).

Social norms are informal guidelines that define acceptable behaviors within a community (e.g., not talking in a library). There are two types of norms: injunctive norms, which refer to what a person believes they should do based upon what society approves or disapproves of, and descriptive norms, which are beliefs regarding what others actually do (Jacobson et al., 2011). An example of an injunctive norm might be the belief that one should stop when someone rejects their sexual advances, whereas a descriptive norm might be the belief that most college students stop the first time someone rejects their sexual advances. Both types of norms can be misperceived. That is, one can be mistaken about the extent to which society approves of a particular behavior (injunctive norm) and the frequency with which others actually engage in the behavior (descriptive norm).

Misperceptions about norms often occur in situations where there is pervasive inconsistency between private and public attitudes and behaviors. This can lead to pluralistic ignorance—a phenomenon where people assume that others’ public behavior accurately reflects their private attitudes even though their own public behaviors are not a true reflection of their private attitude (Miller & Prentice, 2016). Thus, in such circumstances, relying on public attitudes or behaviors can lead to misperception of the true norms (Miller & Prentice, 2016).

Interventions to correct misperceived norms have demonstrated efficacy in shaping beliefs and behaviors across various settings, including fruit and vegetable consumption (Crocker et al., 2009), risky driving (Simons­Morton et al., 2014), and energy conservation (Schultz et al., 2007), among others. Typically, such interventions involve informing participants that most of their peers approve of (injunctive norm) or engage in (descriptive norm) a particular behavior (e.g., conserving energy), thereby making the behavior more normative. In the context of sexual violence, there have been some limited efforts to examine the impact of interventions to correct misperceived social norms (Berkowitz, 2010). For example, Zapp et al. (2018) examined the effects of a sexual assault prevention course which included information that most peers would intervene, and respect others who intervened, to prevent a sexual assault. Although the study found increases in intentions to intervene to prevent sexual violence, the results must be interpreted with caution because providing accurate social norms was only a small part of the multi­component educational program used. Thus, it is not possible to determine whether the change was due to the social normative information or some other aspect of the program (e.g., information about the benefits of building

healthy relationships, training in the best way to obtain consent, etc.). Nevertheless, the evidence that college students misperceive the prevalence and acceptability of sexual violence (Paul & Gray, 2011; Zapp et al., 2018) and the extensive literature affirming the effectiveness of social normative interventions in other contexts, suggests that such interventions should also be efficacious in the context of sexual assault prevention.

To increase the likelihood that students more fully attend to and process corrective normative information, researchers have used personalized feedback that explicitly contrasts students’ beliefs about injunctive and descriptive norms with the actual peer norms. For example, multiple studies have shown that providing participants with personalized feedback showing that their peers actually consume less alcohol (descriptive norm) and disapprove of excessive consumption to a greater extent (injunctive norm) than participants had assumed, can reduce alcohol consumption among college students (Lewis & Neighbors, 2006; Wolter et al., 2021). Although the efficacy of personalized normative feedback has most extensively been examined in the context of college drinking, other work has found this approach effective for reducing misperceived social norms and inducing behavior change in contexts as varied as health promotion and precious resource conservation (e.g., Balvig & Holmberg, 2011; Neighbors et al., 2015; Reid & Aiken, 2013; Rosas et al., 2017; Schultz et al., 2014).1

Although not as extensive as the literature addressing alcohol usage, some previous work has examined the efficacy of personalized corrective normative feedback to address sexual assault among college students. One such study examined the effects of targeting both sexual assault risk and alcohol misuse among college women at

1Personalized normative feedback may face unique challenges in sexual assault prevention given the role of gender norms, power, consent, and bystander intervention which are less central to the behavioral contexts in which the correction of misperceived norms has frequently been implemented. Moreover, unlike behaviors such as energy conservation, sunscreen use, and maintaining a healthy diet, the proscriptions against sexual assault are more than merely informal guidelines that define acceptable behaviors (i.e., norms), but rather there are laws against sexual violence. Therefore, there is reason to suspect that social norms­based interventions developed in other contexts may not transfer directly to the sexual assault context. On the other hand, it has been argued that sexual assault in the college student population meets most of the criteria used to determine whether a particular behavior is likely to be amenable to a social norms­based intervention (Berkowitz, 2003; e.g., a number of misperceptions exist about sexual assault, students exert influence on each other’s sexual­assault related behavior, misperceptions about sexual assault influence behavior, it is possible for the correction of misperceptions to change behavior). Also, the success of correcting misperceived norms across many different behaviors, as well as some initial promising efforts in the sexual assault domain (e.g., Gilmore et al., 2015; 2022; Testa et al., 2020; Zapp et al., 2018), justifies further exploration of the prevention enhancing efficacy of correcting misperceived sexual assault norms.

high risk for sexual assault (i.e., those who reported high rates of heavy episodic drinking; Gilmore et al., 2015). Participants were provided personalized normative corrective feedback for sexual assault, alcohol misuse, or both. Lower incidence and severity of sexual assault at 3 ­ month follow ­ up was found in the combined condition, relative to a control condition, but only among those with more severe sexual assault histories. Another study included personalized normative feedback in a multi ­ component intervention to address alcohol misuse, sexual assault victimization and perpetration, and bystander intervention (Gilmore et al., 2022). The results demonstrated increased intentions to engage in sexual assault preventative behaviors, decreased belief in rape myths, and decreased intention to have sex with someone who was incapacitated (i.e., drunk). Although these findings support the notion that correcting misperceived norms may enhance interventions aimed at reducing sexual assault among college students, it should be noted that both studies (Gilmore et al., 2015; 2022) combined the personalized normative feedback (e.g., a message indicating how participants’ estimates of the percentage of their peers who had been sexually assaulted while drinking differed from the actual rate on campus) with sexual assault education (i.e., the definition of sexual assault, risk factors for sexual assault, methods of reducing and resisting sexual assault). Confounding the personalized feedback with sexual assault education makes it impossible to determine the extent to which the corrective feedback was responsible for the results. In another study, Testa et al. (2020) found that correcting misperceived descriptive norms about sexual “hook ups” reduced female college students’ odds of sexual assault compared to a control group who received no intervention at all. Although this study did not confound personalized normative feedback with sexual assault education, because there was not a comparison condition that received only sexual assault education, it is unclear whether personalized feedback is more effective than sexual assault education.

The current study had two primary objectives. First, we sought to add to the existing literature by assessing the effectiveness of personalized corrective feedback in increasing intentions to engage in sexual assault preventative behaviors, correcting perceived norms for sexual aggression prevention, and reducing rape supportive beliefs. In addition, the study was designed to enable comparison of the relative efficacy of normative corrective feedback versus sexual assault education, as well as their combination. College students were randomly assigned to either receive only information about sexual assault (education video condition), only customized feedback about how their perceptions

of injunctive and descriptive norms regarding sexual assault compared to the actual norms (normative feedback condition), or both of these interventions (combined condition). Drawing upon the extensive literature that highlights the effectiveness of interventions rooted in social norms, particularly those enhanced by correcting misperceived norms, we hypothesized that participants in the combined condition would report (A) greater intentions to engage in preventive measures against sexual assault, (B) more positive perceived norms for sexual aggression prevention, and (C) lower acceptance of rape myths relative to those in either intervention alone.

Method

Participants

Participants were 86 undergraduates (63% women, 37% men, 0% nonbinary), ages 18–26 ( M = 19.62, SD = 1.70) who received course credit for their participation. With sample size N = 86 the study obtained 80% power to detect effects sizes of d > 0.77 for between subjects t tests, and f > 0.34 for analyses of variance (Cohen, 1969). The ethnic background of the sample was 39.5% Hispanic/Latino, 31.4% White, 10.5% Asian/Pacific Islander, 7.0% African American, 5.8% identified as both Hispanic/Latino and White, 1.2% checked “Other,” and 1.2% identified as each of the following multiracial categories: (a) Asian/Pacific Islander and African American and Hispanic/Latino; (b) Asian/Pacific Islander and Hispanic/Latino and White; (c) Asian/ Pacific Islander and White and other; (d) African American and Hispanic/Latino and White. The sample was 46.5% first year students, 25.6% sophomores, 22.1% juniors, and 4.7% seniors. This study was reviewed and approved by the California State University San Marcos (CSUSM) Institutional Review Board (Protocol #892901).

Design and Conditions

We randomly assigned participants to one of three conditions. Twenty ­ nine were assigned to receive only sexual assault information (educational video condition), 29 were assigned to receive only feedback to correct misperceived norms (normative feedback condition), and 29 were assigned to receive both sexual assault information and feedback to correct misperceived norms (combined condition). One participant who had been randomly assigned to the educational video condition declined to continue after learning that the video might include a survivor story, leaving a total sample of 86. It should be noted that all students enrolled at CSUSM are required to complete a commercially available online sexual assault prevention program called “Not Anymore” (https://cultureofrespect.org/program/ not­anymore/). The required training is completed in

the fall semester and the current study was conducted during the spring semester.

Materials

Baseline Rape Myths and Perceived Injunctive and Descriptive Norms

Participants completed measures of demographics, their baseline acceptance of rape myths, and their perceptions of the injunctive and descriptive norms regarding sexual assault. We measured participants’ acceptance of rape myths using the Updated Illinois Rape Myth Acceptance scale (u­IRMA; McMahon & Farmer, 2011), designed to measure acceptance of rape myths across different contexts and situations. The u­IRMA contains 22 statements (e.g., “If a girl initiates kissing or hooking up, she should not be surprised if a guy assumes she wants to have sex;” “If a girl doesn’t say ‘no’ she can’t claim rape”). Level of agreement with each item is rated on a 5­point Likert­type scale from 1 (strongly agree) to 5 (strongly disagree). The u­IRMA has demonstrated good reliability (internal consistency and test­retest) and convergent and discriminant validity in previous work (Ewa Lys et al., 2023; Fansher & Zedaker, 2022; McMahon & Farmer, 2011; Skov et al., 2022). McMahon and Farmer (2011) reported a Cronbach’s alpha of .87. Alpha for the 22 items in the current study was also high (α = .88) and thus they were reverse scored and summed to produce a total baseline rape myth acceptance score (a higher score indicates higher acceptance of rape myths).

We assessed participants’ baseline perceptions of injunctive norms using one item in which they were asked to report how the typical student attending their school would feel about stopping the first time someone declines their sexual advances; from 1 (extremely bad) to 7 (extremely good). Two additional items were used to measure perceived descriptive norms. The first item asked participants to estimate the percentage of CSUSM students who would stop the first time someone rejects their sexual advances. The second item asked participants to estimate the percentage of CSUSM students who have tried to get someone drunk in order to make them more willing to engage in sexual activity.

Educational Video

Information regarding sexual assault was provided via an approximately 13­min. video that contained basic information regarding sexual assault, such as the definition, prevalence rates (e.g., the percentage of assaults that occur within the survivor’s residence, the percentage of assaults that are perpetrated by someone known to the victim, etc.), as well as information about how sexual assaults can be prevented. In addition, the video discussed the psychological and health consequences that survivors of sexual assault often experience. This

was bolstered by including a composite story that we created using two actual survivors’ stories; both stories were posted on the website for Pandora’s Project (pandys.org). In the composite story, the survivor describes the emotional turmoil and life consequences that she experienced for several years after she was date raped by her high school boyfriend. As some literature has found that incorporating the consequences faced by perpetrators reduces attraction to sexual aggression and acceptance of rape myths (Tharp et al., 2013), two actual perpetrators’ stories were also included in the video. Both stories describe the remorse and life consequences experienced by the perpetrators. The first was obtained from an online discussion forum (accessed on Reddit) and the other was shared at a Narcotics Anonymous meeting (permission was given for inclusion in the study); both stories were only minimally edited. All three stories were written in the first­person perspective and narrated by actors in the video.

Normative Feedback

To correct misperceptions of social norms regarding sexual assault, we provided feedback with a personally customized feedback sheet that showed the difference between the participant’s own perception of each norm (assessed at baseline) and the actual norm (assessed using a survey of over 200 undergraduate students drawn from the same participant pool a few months prior to the current study). The personally customized feedback sheet contained feedback for one injunctive and two descriptive norm items. Participants’ baseline perception of each norm was written onto the feedback sheet in direct comparison to the actual norm. Specifically, one of the descriptive norm items stated “You thought that ____ % of CSUSM students would stop the first time someone says no to their sexual advances. On average, actually 88% of CSUSM students would stop the first time someone said no to their sexual advances.” The second descriptive norm item stated “You thought that _____% of CSUSM students have tried to get someone else drunk to make them more willing to engage in sexual activity. On average, actually less than 5% of CSUSM students have tried to get someone else drunk for this purpose.” The injunctive norm item displayed the same 7­point scale (1 = extremely bad; 7 = extremely good ) participants had completed at baseline, and above the scale it stated: “You thought that the typical CSUSM student believes that stopping the first time someone says no to their sexual advances is:” The response they had given on the baseline item was circled in ink on the scale and there was an arrow pointing to the “7 extremely good” label on the scale and inside the arrow it stated: “On average, CSUSM students believe stopping the first time is extremely good.” This normative feedback sheet was based on one used by

Reid and Aiken (2013). Participants were instructed to carefully read this feedback sheet and compare their own perceptions of the norms to the actual norms.

Intentions

We developed an 8­item scale to measure intentions to prevent sexually aggressive behavior (e.g., “I plan to express my disapproval if a friend is making unwanted sexual advances toward someone;” “I plan to make sure that my partner is okay with any sexual contact before I initiate anything;” “I plan to drink less alcohol around others if I find that it makes me sexually aggressive”). Level of agreement with each item was assessed using a 5 ­ point Likert ­ type scale (1 = strongly disagree ; 5 = strongly agree ). We developed this scale for this study and thus its reliability and validity have not been established previously. The item that concerned drinking less alcohol if it produces sexual aggressiveness had a weaker item­total correlation than any of the others (r = .28 versus r = .37 ­ .75). Moreover, it was the only item that resulted in a higher internal consistency score when deleted. Thus, we omitted this item from the scale. The seven remaining items demonstrated good internal consistency (α = .83) and were thus summed to create an overall intentions index.

Perceived Norms for Sexual Aggression Prevention

We used 4 items to assess participants’ perception of the extent to which their friends and family are supportive of sexual aggression prevention behaviors. Specifically, participants indicated their level of agreement with the statements “Most people who are important to me think that I should speak out when someone tries to minimize and/or justify sexually aggressive behavior” and “Most of my friends think that I should speak out when someone tries to minimize and/or justify sexually aggressive behavior” (1 = completely disagree; 5 = completely agree). Participants then indicated the degree to which those who are important to them and, separately, their friends would approve if they were to take steps to minimize sexually aggressive behavior (1 = strongly disapprove ; 5 = strongly approve ). The validity of this scale has not been previously established. The internal consistency for the four items was good (α = .84) and they were summed to create an index of perceived norms for sexual aggression prevention.

Manipulation Checks

We used four items to assess participants’ recall of the information provided in the educational video. Specifically, participants (a) listed the potential short and long­term psychological and/or physical effects that sexual assault can have on a survivor (open­ended written response); (b) indicated the percentage of sexual assaults that are

perpetrated by someone who is known to the survivor; (c) listed some ways that a person could potentially prevent an assault from occurring (open­ended written response); and (d) reported the percentage of sexual assaults that occur in the residence of the survivor, a friend, or a neighbor. Participants’ recall of the injunctive and descriptive norms information provided was assessed by having them complete the same three items that assessed their baseline perceived injunctive and descriptive norms.

Follow-up Rape Myth Acceptance

The u­IRMA was used to assess rape myth acceptance at the two­week follow­up. Again, the 22 items demonstrated strong internal consistency at follow­up (α = .90) and were summed into an overall follow­up rape myth acceptance score (with higher values indicating higher rape myth acceptance).

Procedure

Participants signed up for the study via the campus Human Participant Pool (HPP) and data collection sessions were run individually and in­person. Participants were not aware that the study concerned sexual assault when they signed up. Upon arrival at the lab, each participant reviewed and signed a consent form and then completed the demographic and baseline measures. Thereafter, the procedure depended on the condition to which the participant was randomly assigned. Participants assigned to the educational video condition viewed the video. Participants assigned to the normative feedback condition reviewed the customized feedback sheet. Those assigned to the combined condition first watched the video and then received the customized feedback sheet. Next, all participants completed the outcome measures and manipulation checks. Finally, all participants were probed for suspicion (none was detected), debriefed, and then thanked for their participation.

During the debrief, participants were informed about the opportunity to complete a follow­up portion of the study in order to earn additional course credit. All participants were sent an e­mail with the rape myth measure approximately two weeks following their visit to the lab. Only 38% of the participants returned the u­IRMA at follow­up (9 of 27 in the educational video condition, 12 of 29 in the normative feedback condition, and 11 of 29 in the combined condition). Those who completed the measure received another e ­ mail in which they were fully debriefed and thanked for their participation in the follow­up portion of the study.

Results

Preliminary Analyses

Group Equivalence

To determine the initial equivalence of the conditions,

separate one­way analyses of variance (ANOVAs) were performed on the demographic, baseline u ­ IRMA, and baseline norms variables. The results indicated no significant differences in age, gender, ethnicity, year in college, baseline u­IRMA scores, or baseline injunctive and descriptive norms measures as a function of condition (all ps > .162, fs < 0.20). Thus, it appears that participants were effectively randomized to condition.

Manipulation Checks

We used four items as manipulation checks for the educational video and three for the normative feedback. All participants, regardless of condition, completed all seven items. Assuming that participants attended to the information provided in each condition and that the information was new to most, we would expect that those who viewed (versus did not view) the educational video would be more likely to answer the corresponding manipulation checks correctly, and that those who received (versus did not receive) the normative feedback would provide more correct responses on the three normative manipulation checks. We created two manipulation checks to simplify presentation of the findings and to minimize the number of statistical tests performed. Specifically, the four educational video manipulation checks were standardized (z­scored) and combined to create one composite and, separately, the three normative feedback manipulation checks were reverse scored, as necessary, and then standardized and combined to create the second composite.

The one­way ANOVA performed on the composite educational video manipulation check demonstrated a significant overall condition effect, F(2, 78) = 30.79, p < .001, f = 1.01. As one would expect, post hoc analyses showed that, relative to those who only received the normative feedback, participants in either of the conditions who viewed the educational video were significantly more likely to answer the manipulation checks correctly, t(78) = 7.19, p < .001, d = 1.63, for combined versus normative feedback and, t(78) = 6.16, p < .001, d = 1.40, for educational video versus normative feedback. The two conditions that viewed the video did not differ significantly, t(78) = 0.87, p = .387, d = 0.20.

A significant overall condition effect was also obtained for the one­way ANOVA performed on the normative feedback composite manipulation check, F(2, 81) = 13.68, p < .001, f = 0.58). The post hoc analyses demonstrated, as one would expect, that the manipulation check responses of those in either of the conditions that had received the normative feedback were reliably more correct than the responses of those who had not received the feedback (t(81) = 4.48, p < .001, d = 1.00

for combined vs educational video, and t(81) = ­4.62, p < .001, d = 1.03 for normative feedback versus educational video). The two conditions that received the normative feedback did not differ significantly, t(81) = ­0.18, p = .860, d = 0.04. It should be noted that the four significant post hoc comparisons described above remain significant when applying a Bonferroni correction to control for multiple comparisons (corrected alpha is p < .0083 for six comparisons).

Thus, in general it appears that participants paid attention and processed the information provided in the respective interventions, and that the information provided is not merely general knowledge. Means and standard deviations for each of the seven manipulation checks can be found in Table 1.

Primary Analyses

As mentioned previously, in designing this experiment we were specifically interested in determining whether the addition of normative feedback would enhance the effects of sexual assault information. Thus, given our specific a priori hypotheses, we analyzed both the intentions and the perceived norms for sexual aggression prevention measures using planned orthogonal comparisons (Keppel, 1973). Specifically, to test the prediction that participants who received the normative feedback in addition to viewing the educational video

Feedback Manipulation Checks:

would report higher sexual assault prevention intentions and perceived norms for sexual aggression prevention than those who received either intervention alone, the educational video condition and the normative feedback condition were each separately contrasted against the condition that included both interventions (i.e., the combined condition). The rape myths data were analyzed with a repeated measures analysis of variance. Means and standard deviations for each outcome as a function of condition are in Table 2.

Intentions

The planned orthogonal contrasts conducted on the intentions index showed, as predicted, that those who received both the educational video and the normative feedback (i.e., the combined condition) reported greater intentions to engage in sexual assault prevention behaviors relative to those who had only received the normative feedback, t(82) = 1.97, p = .052, d = 0.44 (see Table 2 for condition means and standard deviations). However, although the means are in the predicted direction, the combined condition did not result in reliably greater intentions to engage in sexual assault prevention behaviors relative to the educational video condition, t(82) = 1.10, p = .276, d = 0.24.

Perceived Norms for Sexual Aggression Prevention

The planned orthogonal contrasts conducted on the perceived norms for sexual aggression prevention index demonstrated that those participants who received both interventions (combined condition) reported significantly higher perceived sexual aggression prevention norms (i.e., that their friends and important others would want them to speak out when someone tries to minimize or justify sexually aggressive behavior) than did those who only watched the educational video, t(82) = 2.39, p = .019, d = 0.53, or only received the normative feedback, t(82) = 2.68, p = .009, d = 0.59 (see Table 2).

Rape Myth Acceptance

The 38% who did, versus those who did not, complete the follow­up were more likely to be women (26 women vs 6 men), t(77.35) = ­3.00, p = .004, d = 0.62, and had reported somewhat higher intentions to engage in sexual assault prevention behaviors, t(82.66) = ­1.94, p = .056, d = 0.37, and higher perceived norms for sexual aggression prevention, t(79.15) = ­3.20, p = .002, d = 0.60, immediately after the intervention. There were no differences between those who did, versus did not, complete the follow­up in age, year in college, ethnicity, or baseline u­IRMA scores (all ps > .337, all ds < 0.21). Those who did complete the follow­up also did not differ from non­completers on any of the three

baseline injunctive and descriptive norms perceptions (all ps > .246, all ds < 0.25). Thus, those who chose to complete the follow­up generally had similar demographic and baseline characteristics as those who did not. Also, among those who chose to participate in the follow­up, there were no significant differences across condition in demographic characteristics (age, gender, ethnicity, year in college) or baseline u­IRMA scores (all ps > .299, all ds < 0.28).

A 3 (Condition: Educational Video vs Normative Feedback vs Combined) x 2 (Time: baseline vs follow­up)

ANOVA, with repeated measures on the last factor was performed on the rape myth acceptance scores of those who completed the follow­up. There was not a main effect of condition, F(2, 29) = 0.60, p = .555, f = 0.20. However, the analysis revealed a significant effect of time period, such that participants expressed lower acceptance of rape myths at follow­up than they did prior to the interventions, F(1, 29) = 7.27, p = .012, f = 0.50. The Condition x Time effect was not significant, F(2, 29) = 0.26, p = .777, f = 0.13, indicating rape myth acceptance decreased similarly across all conditions from baseline to follow­up.2

Discussion

Sexual assault remains a significant problem on college TABLE 2

Means (and Standard Deviations) of Outcomes as a Function of Condition

Note. Rape myth acceptance was assessed using the Updated Illinois Rape Myth Acceptance (u-IRMA) scale (McMahan & Farmer, 2011). Higher scores on the scale indicate higher acceptance of rape myths.

2Given that the educational video and normative feedback conditions differed not only in content but also method of delivery (i.e., video versus written), it could be argued that any outcome differences between these conditions might be due to differences in delivery method rather than content. However, in this experiment there were no significant differences between the condition that only received the educational video and the one that only received the written normative feedback on any of the dependent variables (p = .406 for intentions; p = .810 for perceived norms for sexual aggression prevention; p = .706 for follow­up rape myth acceptance).

campuses despite the mandated implementation of training programs designed to reduce its prevalence. It is likely that no single intervention will be optimally effective in all contexts. Thus, it is important to continue to develop effective interventions to assemble an array of tools for combating sexual assault. The results of this study suggest some promise for the combination of interventions utilized. That is, receiving a personalized message to correct misperceived norms and viewing an educational video resulted in significantly higher perceived norms for sexual aggression prevention than did either intervention alone. The combination of interventions also produced greater intentions to engage in preventive action than did the normative feedback alone. In addition, each intervention and their combination showed decreased rape myth endorsement at follow­up relative to baseline. The effect sizes obtained were generally in the moderate range, offering some promise that the greater power afforded by a large­scale randomized controlled trial (RCT) may demonstrate more definitive and enduring results.

Several prominent theories argue that initiating and maintaining behavior change is a complicated, multifaceted process (Fishbein & Ajzen, 2010) and extensive literature has found information to be necessary, but often not sufficient, to produce behavior change (Corace & Garber, 2014; Fishbein & Ajzen, 2010). The generally beneficial results in this study of combining normative feedback with education are consistent with this previous literature. That is, to avoid or prevent sexual assault, one must know what it is, but this knowledge may not be sufficient to overcome the myriad and complex cognitive, emotional, and social factors (including perceived social norms) involved in behavior change. Combining information with feedback to correct misperceived norms may weaken one of the multiple processes maintaining a particular behavior, perhaps in part by enhancing the processing and retention of the educational content (Lewis & Neighbors, 2006). The findings of this study are also generally consistent with the limited previous literature that has examined whether the inclusion of corrective normative feedback can affect sexual assault­related cognitions and behaviors (Gilmore et al., 2015; 2022; Testa et al., 2020; Zapp et al., 2018). However, the previous work has often either confounded corrective feedback with sexual assault education (e.g., Gilmore et al., 2015; 2022; Zapp et al., 2018) or failed to include a comparison condition that only received sexual assault education (Testa et al., 2020). This study contributes to the literature by explicitly comparing the efficacy of a personalized corrective normative feedback intervention, sexual assault education, and their

combination. It is costly for universities to implement sexual assault prevention programs and time consuming for students to participate. Thus, determining which components of such programs actually produce beneficial effects has important practical implications.

Strengths

This study featured several noteworthy methodological strengths. First, participants were randomized into conditions and analyses of demographic and baseline variables suggest that random assignment was effective. Also, rape myth beliefs were assessed two weeks after the interventions, extending beyond the typical assessment of immediate beliefs and intentions. The u­IRMA has demonstrated good reliability and validity in previous work, and the intentions and perceived norms for sexual aggression prevention measures were based on measures used in previous studies (e.g., Rosas et al., 2017) and showed strong internal consistency in this study. In addition, real­life stories from perpetrators detailing the negative consequences of their actions were included to enhance the impact and efficacy of the educational video. Finally, manipulation checks confirmed that participants processed both the normative feedback and the sexual assault information, strengthening confidence that observed changes were due to the interventions.

Limitations

Given the extensive previous literature demonstrating the efficacy of sexual assault education (Hudspith et al., 2023), we did not include a group that received no intervention in this study. Inclusion of a control (no intervention) condition in a future large­scale RCT would help to determine whether the interventions were truly responsible for the lower rape myth acceptance at follow­up, or whether the reduction was due to a pretest sensitization or demand effect. A future large ­ scale RCT might also include an assessment of all outcome measures at baseline to enable examination of change in outcomes as a function of intervention. However, the evidence of effective random assignment to conditions in this study makes it unlikely that participants’ baseline intentions and perceived norms for sexual aggression prevention differed reliably across conditions. Another notable limitation of this study is a poor two ­ week follow­up rate. The 38% of participants who returned the follow­up u­IRMA were more likely to identify as women and appeared to have been more immediately impacted by the interventions, as evidenced by somewhat higher sexual assault prevention intentions and perceived norms for sexual aggression prevention than those who did not return the follow­up. Thus, there is some question about whether the lower rape myth acceptance

two weeks after the interventions would generalize to the 62% who chose not to complete the follow­up. Although one must be especially cautious when the research topic concerns an issue as sensitive as sexual assault, it may be possible to increase follow­up rates in the future by offering incentives without being coercive. Due to the nature of the intervention (specifically because the experimenters started the video and provided the normative feedback sheet), experimenters in this study were unable to be blind to condition. However, the experimenters were not in the room as participants completed the outcome measures, thus limiting the potential to influence responses. Additionally, the scales used to assess intentions and perceived norms for sexual aggression prevention were developed specifically for this study and their validity has not been established. Future work is needed to formally confirm that they are measuring the intended constructs.

The sample for this study was racially diverse, but largely women, students in their first or second year in college, and exclusively between the ages of 18 and 26. Given that both sexual assault ­ related attitudes (Aronowitz et al., 2012) and the impact of social norm interventions have been shown to vary by gender (Banyard et al., 2020; Lewis & Neighbors, 2006), future studies with larger samples may consider gender­specific normative correction. Future research should also examine the impact that age and year in college may have on outcomes as younger women in the first few weeks of college tend to be more vulnerable to sexual assault (Cranney, 2015; Kimball et al., 2008). It should also be noted that we did not ask our participants how much previous sexual violence prevention training they had received or whether they had first­hand experience with sexual violence (either as perpetrator or survivor). Individuals with more frequent sexual violence prevention training or previous sexual violence involvement could react differently to the interventions.

Lastly, there are a number of questions related to intervention delivery that could be explored in future work. For example, it would be beneficial to determine the optimal order of intervention delivery. That is, if normative feedback enhances the processing and encoding of sexual assault education then it would be beneficial to provide normative feedback first. However, it may be that social norms information is most effective when one understands what constitutes sexual assault. Additionally, both sexual assault education and personalized normative feedback have often been delivered online, and such online interventions have been shown to be cost effective (Dotson et al., 2015). Although there are obvious limitations to online delivery (e.g., divided attention), conducting an online large­scale RCT to further examine

the efficacy of the interventions developed for this study would more readily enable recruitment of participants from a broader age and gender range and from multiple colleges, thereby both enhancing the generalizability of the results and allowing for the exploration of many of the issues discussed above.

Conclusion

Sexual assault on U.S. college campuses remains a critical issue despite widespread prevention efforts. The need for continued innovation in intervention strategies is crucial. The present study provides encouraging evidence that correcting misperceived injunctive and descriptive norms regarding sexual assault holds promise when paired with sexual assault education. These promising results warrant further investigation through larger­scale randomized controlled trials, and we suggest a number of additional avenues for further work to improve the impact and efficacy of these interventions. This research offers insights for universities aiming to implement cost­effective and impactful sexual assault prevention programs, advancing the ultimate goal of creating safer campuses.

References

Aronowitz, T., Lambert, C. A., & Davidoff, S. (2012). The role of rape myth acceptance in the social norms regarding sexual behavior among college students. Journal of Community Health Nursing, 29(3), 173–182. https://doi.org/10.1080/07370016.2012.697852

Balvig, F., & Holmberg, L. (2011). The ripple effect: A randomized trial of a social norms intervention in a Danish middle school setting. Journal of Scandinavian Studies in Criminology and Crime Prevention, 12(1), 3–19. https://doi.org/10.1080/14043858.2011.561619

Banyard, V. L., Edwards, K. M., & Moynihan, M. M. (2020). Exploring the potential campus level impact of online universal sexual violence prevention education for college students. Journal of American College Health, 68(4), 389–397. https://doi.org/10.1080/07448481.2019.1613401PMC

Berkowitz, A. D. (2003). Applications of social norms theory to other health and social justice issues. In H. W. Perkins (Ed.), The social norms approach to preventing school and college age substance abuse: A handbook for educators, counselors, clinicians (Chapter 16). Jossey-Bass.

Berkowitz, A. D. (2010). Fostering healthy norms to prevent violence and abuse: The social norms approach. In K. Kaufman (Ed.), The prevention of sexual violence: A practitioner’s sourcebook (Chapter 8). NEARI Press.

Cantor, D., Fisher, B., Chibnall, S., Harps, S., Townsend, R., Thomas, G., Lee, H., Kranz, V., Herbison, R., & Madden, K. (2020). Report on the AAU Climate Survey on Sexual Assault and Sexual Misconduct. The Association of American Universities. https://www.aau.edu/sites/default/files/AAU-Files/ Key-Issues/Campus-Safety/Revised%20Aggregate%20report%20%20 and%20appendices%201-7_(01-16-2020_FINAL).pdf

Cohen, J. (1969). Statistical power analysis for the behavioral sciences. Academic Press.

Coker, A. L., Bush, H. M., Fisher, B. S., Swan, S. C., Williams, C. M., Clear, E. R., & DeGue, S. (2016). Multi-college bystander intervention evaluation for violence prevention. American Journal of Preventive Medicine, 50(3), 295–302. https://doi.org/10.1016/j.amepre.2015.08.034

Collins, S. E., Kirouac, M., Lewis, M. A., Witkiewitz, K., & Carey, K. B. (2014). Randomized controlled trial of web-based decisional balance feedback and personalized normative feedback for college drinkers. Journal of

Studies on Alcohol and Drugs, 75(6), 982–992. https://doi.org/10.15288/jsad.2014.75.982

Corace, K., & Garber, G. (2014). When knowledge is not enough: Changing behavior to change vaccination results. Human Vaccines & Immunotherapeutics, 10(9), 2623–2624. https://dx.doi.org/10.4161/21645515.2014.970076

Cranney, S. (2015). The relationship between sexual victimization and year in school in U.S. colleges: Investigating the parameters of the “red zone.” Journal of Interpersonal Violence, 30(17), 3133–3145. https://doi.org/10.1177/0886260514554425

Crocker, H., Whitaker, K. L., Cooke, L., & Wardle, J. (2009). Do social norms affect intended food choice? Preventive Medicine, 49(2–3), 190–193. https://doi.org/10.1016/j.ypmed.2009.07.006

Dotson, K. L., Kearney, M. D., & Roper, M. C. (2015). Stand-alone personalized normative feedback for college student drinkers: A meta-analytic review, 2004 to 2014. PLOS ONE, 10(8), 1–17. https://doi.org/10.1371/journal.pone.0139518

Ewa Lys, A., Bargiel-Matusiewicz, K., Krasuski, T., & Studzinska, A. (2023). Psychometric properties of the Polish updated Illinois rape myth acceptance scale. Current Psychology, 42, 445–459. https://doi.org/10.1007/s12144-020-01249-3

Fansher, A. K., & Zedaker, S. B. (2022). The relationship between rape myth acceptance and sexual behaviors. Journal of Interpersonal Violence, 37(1–2), NP903–NP924. https://doi.org/10.1177/0886260520916831

Fishbein, M., & Ajzen, I. (2010). Predicting and changing behavior: The reasoned action approach. Taylor & Francis.

Gilmore, A. K., Leone, R. M., Oesterle, D. W., Davis, K. C., Orchowski, L. M., Ramakrishnan, V., & Kaysen, D. (2022). Web-based alcohol and sexual assault prevention program with tailored content based on gender and sexual orientation: Preliminary outcomes and usability study of positive change (+Change). JMIR Formative Research, 6(7), 1–13 https://doi.org/10.2196/23823

Gilmore, A. K., Lewis, M. A., & George, W. H. (2015). A randomized controlled trial targeting alcohol use and sexual assault risk among college women at high risk for victimization. Behaviour Research and Therapy, 74, 38–49. https://doi.org/10.1016/j.brat.2015.08.007

Hudspith, L. F., Wager, N., Willmott, D., & Gallagher, B. (2023). Forty years of rape myth acceptance interventions: A systematic review of what works in naturalistic institutional settings and how this can be applied to educational guidance for jurors. Trauma, Violence, & Abuse, 24(2), 981–1000. https://doi.org/10.1177/15248380211050575

Jacobson, R. P., Mortenson, C. R., & Cialdini, R. B. (2011). Bodies obliged and unbound: Differentiated response tendencies for injunctive and descriptive social norms. Journal of Personality and Social Psychology, 100(3), 433–448. https://doi.org/10.1037/a0021470

Keppel G. (1973). Design and analysis: A researcher’s handbook. Prentice-Hall. Kimball, M., Neacsiu, A. D., Flack, W. F., Jr., & Horner, J. (2008). Risk of unwanted sex for college women: Evidence for a red zone. Journal of American College Health, 57(3), 331–338. https://doi.org/10.3200/JACH.57.3.331-338

Langton, L., Berzofsky, M., Krebs, C. P., & Smiley-McDonald, H. (2012). Victimizations not reported to the police, 2006–2010. U.S. Department of Justice, Bureau of Justice Statistics. https://bjs.ojp.gov/content/pub/pdf/vnrp0610.pdf

Leschber, B. (2023, September 27). Texas state reports 135% increase in rapes, 200% increase in hate crimes in 2023 annual security fire and safety report. The University Star. https://universitystar.com/24080/news/texas-statereports-135-increase-in-rapes-200-increase-in-hate-crimes-additionalincreases-in-assaults-and-burglaries-in-2023-annual-security-fire-andsafety-report/

Lewis, M. A., & Neighbors, C. (2006). Social norms approaches using descriptive drinking norms education: A review of the research on personalized normative feedback. Journal of American College Health, 54(4), 213–218. https://doi.org/10.3200/JACH.54.4.213-218

Mahoney, P., Gielen, A. C., Bailey, M. M., & Gabel, C. (2020). Applying the Haddon Matrix to evaluate sexual assault interventions on college campuses. Journal of American College Health, 68(6), 579–586. https://doi.org/10.1080/07448481.2019.1583658

McMahon, S., & Farmer, G. L. (2011). An updated measure for assessing subtle

rape myths. Social Work Research, 35(2), 71–81. https://doi.org/10.1093/swr/35.2.71

Miller, D. T., & Prentice, D. A. (2016). Changing norms to change behavior. Annual Review of Psychology, 67, 339–361. https://doi.org/10.1146/annurev-psych-010814-015013

National Center for Education Statistics (NCES). (2022). Criminal incidents at postsecondary institutions. Condition of Education. U.S. Department of Education, Institute of Education Sciences. https://nces.ed.gov/programs/coe/indicator/a21

Neighbors, C., Rodriguez, L. M., Rinker, D. V., Gonzales, R. G., Agana, M., Tackett, J. L., & Foster, D. W. (2015). Efficacy of personalized normative feedback as a brief intervention for college student gambling: A randomized controlled trial. Journal of Consulting and Clinical Psychology, 83(3), 500–511. https://doi.org/10.1037/a0039125

O’Connor, J. (2023). Profiles of men’s rape myth beliefs and the association with rape proclivity. Journal of Interpersonal Violence, 38(9–10), 6366–6388. https://doi.org/10.1177/08862605221130395

Paul, L. A., & Gray, M. J. (2011). Sexual assault programming on college campuses: Using social psychological belief and behavior change principles to improve outcomes. Trauma, Violence & Abuse, 12(2), 99–109. https://doi.org/10.1177/1524838010390710

Reid, A. E., & Aiken, L. S. (2013). Correcting injunctive norm misperceptions motivates behavior change: A randomized controlled sun protection intervention. Health Psychology, 32(5), 551–560. https://doi.org/10.1037/a0028140

Roessler, A. (2024, April 3). Residence hall rapes at the highest percentage in recent years. The Minnesota Daily. https://mndaily.com/283049/top-story/ residence-hall-rapes-at-highest-percentage-in-recent-years/

Rosas, C. E., Gregorio-Pascual, P., Driver, R., Martinez, A., Price, S. L., Lopez, C. & Mahler, H. I. M. (2017). Effects of social norms information and selfaffirmation on sugar-sweetened beverage consumption intentions and behaviors. Basic and Applied Social Psychology, 39(2), 112–126. https://dx.doi.org/10.1080/01973533.2017.1283503

Schultz, P. W., Nolan, J. M., Cialdini, R. B., Goldstein, N. J., & Griskevicius, V. (2007). The constructive, destructive, and reconstructive power of social norms. Psychological Science, 18(5), 429–434. https://doi.org/10.1111/j.1467-9280.2007.01917.x

Schultz, P. W., Messina, A., & Estrada, M. (2014). Personalized normative feedback and the moderating role of personal norms: A field experiment to reduce residential water consumption. Environment and Behavior, 48(5), 686–710. https://doi.org/10.1177/0013916514553835

Simons-Morton, B. G., Bingham, C. R., Falk, E. B., Li, K., Pradhan, A. K., Ouimet, M. C., Almani, F., & Shope, J. T. (2014). Experimental effects of injunctive norms on simulated risky driving among teenage males. Health Psychology, 33(7), 616–627. https://doi.org/10.1037/a0034837

Skov, M., van Mastrigt, S., & Jensen, A. V. (2022). Comparing rape myth acceptance among police trainees and medical students: A preliminary Danish validation of the updated Illinois rape myth acceptance scale. Violence Against Women, 28(11), 2649–2676. https://doi.org/10.1177/10778012211038967

Tharp, A. T., DeGue, S., Valle, L. A., Brookmeyer, K. A., & Matjasko, J. L. (2013). A systematic qualitative review of risk and protective factors for sexual violence perpetration. Trauma, Violence, & Abuse, 14(2), 133–167. https://doi.org/10.1177/1524838012470031

Testa, M., Livingston, J. A., Wang, W., & Lewis, M. A. (2020). Preventing college sexual victimization by reducing hookups: A randomized controlled trial of a personalized normative feedback intervention. Prevention Science, 21(3), 388–397. https://doi.org/10.1007/s11121-020-01098-3

Vara, J. (2023, September 27). Four sexual assaults reported at Cal State San Marcos during Fall Semester FOX 5 San Diego. https://fox5sandiego.com/ news/local-news/four-sexual-assaults-reported-at-cal-state-san-marcosduring-fall-semester/

White, E. (2024, October 24). UM sees a rise in rape cases for the second year in a row. Montana Kaimin. https://www.montanakaimin.com/news/um-seesa-rise-in-rape-cases-for-the-second-year-in-a-row/article_9a2da28a9188-11ef-bc49-0b7b3e41f0ab.html

Williams, D. (2023, September 29). Five sexual assaults reported at Cal State University San Marcos since semester started. NBC 7 San Diego.

https://www.nbcsandiego.com/news/local/five-sexual-assaults-reportedat-cal-state-university-san-marcos-since-semester-started/3316207/

Wolter, C., Lesener, T., Thomas, T. A., Hentschel, A. C., & Gusy, B. (2021). Finding the right balance: A social norms intervention to reduce heavy drinking in university students. Frontiers in Public Health, 9, 653435. https://doi.org/10.3389/fpubh.2021.653435

Zapp, D., Buelow, R., Soutiea, L., Berkowitz, A., & DeJong, W. (2018). Exploring the potential campus-level impact of online universal sexual assault prevention education. Journal of Interpersonal Violence, 36(5-6), NP2324–NP2345. https://doi.org/10.1177/08862605187624

Author Note.

Education in the Sciences (OTRES) at California State University San Marcos. Materials used in this study are available upon request of the authors.

The authors have no known conflict of interest to disclose. The authors wish to thank Adrian L. Price, Esmeralda Curiel, Spencer Michalek, and A. Martinez for their help in carrying out this project.

Cristal Lopez https://orcid.org/0000­0002­1900­4365

Cristal Lopez is now Clinical Faculty at Rady Children’s Health. This research was supported by the MARC GM­08807 grant received from the Center for Training, Research, and Educational Excellence (CTREE) formally the Office for Training, Research and

Cristal Lopez played a lead role in conceptualization, data collection, and substantive original writing, and a supporting role in research design, data analysis and interpretation. Alyssa Martinez played a lead role in conceptualization, data collection, and substantive original writing, and a supporting role in research design, data analysis and interpretation. Heike I. M. Mahler played a lead role in research design, data analysis and interpretation, and substantive original writing, and a supporting role in conceptualization.

Correspondence concerning this article should be addressed to Heike I. M. Mahler, Psychology Department, California State University San Marcos. Email: hmahler@csusm.edu

Out of Reach, Out of Mind? Mechanisms of Psychological Distance and Emotional Reactions to Tragedy Abroad

Syed M. Wahid1, Christopher T. Dawes*2, and Alysson E. Light*1

1 Department of Psychology, University of Chicago

2 Wilf Family Department of Politics, New York University

ABSTRACT. Increased psychological distance can dampen emotions and decrease self­efficacy for prosocial behavior like helping. But can distance and its downstream effects impact emotional responses to news reports of tragedies, and by what mechanisms? We hypothesized that common theories in which distance increases abstraction are insufficient for explaining affective consequences of distance. By contrast, we proposed that distance reduces affective responses to tragedies by reducing helping self­efficacy, which motivates a “collapse of compassion.” In 2 studies (Ntotal = 804), we manipulated news of a bombing to occur in many countries (Experiment 1) and in the same country with differing primed distances to the attack (Experiment 2). Across studies, greater perceived geographic distance was associated with less emotional response (r = .32, p < .001), mediated by lower compassion (B = .03, SE = .04, bootstrapped 90% CI: [.002, .135]), consistent with a compassion collapse mechanism.

Keywords: emotion, compassion, reaction, psychological distance

Diversity badge earned for conducting research focusing on aspects of diversity. Preregistration, Open Data, and Open Materials badges earned for transparent research practices. Preregistration can be viewed at https://osf.io/ nbtm5/. Materials and data can be accessed at https://osf.io/6m32r/

It seems that almost every day, people in the United States hear news of tragedies that occur abroad. In an ever­connected world and with many ongoing conflicts, this information is readily available. Often viewed as the world’s policeman, the U.S. public and its government’s reaction are of critical importance. But would a U.S. resident feel equally shocked or saddened to hear of a suicide bombing that occurred in Spain as opposed to Iran? One example of a stark disparity in public reaction to tragedy is that of the 2015 Ankara bombings and the Paris attacks that took place

shortly after. Both tragedies were carried out by the same actor and with similar casualties, but the former received relative indifference while the latter received an outpouring of public outrage and support (Healy & Fausset, 2015; Yeginsu & Arango, 2015). Politically, following the Ankara bombings, President Obama expressed his condolences in a private phone call to Turkish Prime Minister Erdogan. In contrast, the Paris attacks elicited powerful public remarks from both the President and Vice President (Garunay, 2015). Perhaps this is because the United States is geographically closer

to France than it is to Turkey. Nonetheless, time and time again, one can observe differences in reactions to tragedies that occur in various places abroad that cannot be explained based on loss of life or the nature of tragedy. Such inequity in reactions would conflict with the value of universalism, to protect the welfare of others equally, which transcends across cultures (Schwartz, 2012). The dissonance between universalist values and behavior is evident in disparate reactions to tragedy. This raises the question: What psychological mechanisms are at play when reacting to tragedies abroad?

Theories of Psychological Distance

A foundation for investigating reactions to tragedy lies in the theory of psychological distance. Psychological distance refers to the perceived separation between an individual and a distal object, person(s), or event (Trope et al., 2007). This separation lies along four unique dimensions, including hypothetical, temporal, spatial, and social distance (Liberman & Trope, 2014; Trope et al., 2007). Psychological distance along its many dimensions has a significant influence over behavior and other cognitive functions, with even early studies finding relations to motivation, conflict, and emotion (Miller, 1944). Social distance refers to a sense of personal familiarity, which can vary based on factors such as social bonds, perceived similarity, or cultural affiliation. For example, two individuals sharing the same cultural heritage would express less social distance than those of vastly different cultures. Spatial distance is simply the perceived physical space between one individual and another. This distance can vary greatly by scale, from between two individuals in a room to those in different countries. In studying reactions to tragedy, social and spatial distance would be most applicable, given that news covers real events and is generally released shortly after they occur. In the context of U.S. reactions to tragedies abroad, psychological distance would apply to victims who are strangers in other countries, vast distances away. Thus, social distance would primarily vary according to culture, and spatial distance would vary in terms of physical geography.

The relevant dimensions of social and spatial distance both suggest impacts on emotional reactions. Wong and Bagozzi (2005) found that, when manipulating social distance in reading a story of immoral behavior, distance had a negative relationship with emotional intensity, which was consistent across cultural groups. Additionally, there is a clear relationship between spatial distance and emotional response for positive and negative stimuli, with stimuli moving away decreasing both positive and negative affect, respectively (Davis et al., 2011; Mühlberger et al., 2008). This suggests that both spatial

and cultural distances from a tragedy abroad contribute to attenuating emotional reactions.

An overarching theory that posits the general effect of psychological distance on thinking is construal­level theory (CLT). CLT asserts that greater psychological distance away from an object yields a higher level construal or more abstract perception of the object (Liberman & Trope, 2014). Inversely, objects of closer psychological distance are construed as more detailed. Much of the present literature regarding psychological distance uses CLT as a framework, and there is evidence of CLT applying to various dimensions such as spatial and temporal distance (Henderson et al., 2011; Liberman & Trope, 2003; Norman et al., 2016). This presents a straightforward route of understanding varied reactions to tragedy abroad: tragedies that affect socially and spatially closer victims would have lower level construals, hence more detailed perceptions that would evoke more intense emotional reactions. By contrast, victims in distant countries with very different cultures would appear more abstract, inviting less emotion.

Although much of the literature assumes the relationship between CLT and psychological distance in applied contexts, there is a growing body of evidence that they are distinct and can be studied separately (Li et al., 2019; Maglio, 2019; Mrkva et al., 2018; Van Boven & Caruso, 2015; Wong & Wyer, 2016). In fact, for emotion and emotional regulation specifically, psychological distance and CLT appear particularly separable with evidence of diverging effects and unique mechanisms of impact (Abraham et al., 2023; Williams et al., 2014). Research examining the impact of psychological distance and CLT on emotion has yielded findings that conflict with CLT (Williams & Bargh, 2008) or reveal impacts of psychological distance independent of CLT (Van Boven et al., 2010), specifically, a negative relationship between distance and emotional reaction without any significant effects of construal level. Moran and Eyal (2022) solidified this distinction by conducting and comparing two meta­analyses: one of studies testing the impact of abstraction on emotion and the other on psychological distance on emotion. In assessing 98 manipulations of abstraction on emotional experience, the authors found mixed impacts of emotional intensity by emotion type. Meanwhile, their meta­analysis of 230 studies on psychological distance found a consistent negative relationship between distance and emotional intensity. Thus, although CLT argues that physical distance impacts emotional reactions by increasing the abstraction with which events are perceived, this mechanism cannot account for the impact of physical distance, given that physical distance and abstraction have different patterns of association with emotional

intensity. Taken together, these findings indicate that psychological distance is more likely to be a determinant of emotional reaction to news of tragedy than level­ofabstraction as explained by CLT. However, given the theoretical separation of psychological distance from CLT in its effect on emotional reaction, other mediating mechanisms must explain why distance has this effect.

Mechanisms of Compassion

In terms of the impact of distance on emotional reactions to tragedy, any account of reactions to tragedy must include consideration of the fundamental human response of compassion. Compassion is a distinct affective state characterized by response to another’s suffering, the experience of negative affect, and a motivation to help (Goetz et al., 2010). Individuals tend to express more compassion and subsequent helping behavior towards members of one’s self­categorized ingroups (vs. outgroups)—effectively close vs. far strata of social distance (Levine et al., 2002; Marshall & Wilks, 2024). Additionally, emotion is a key driving factor of helping behavior (Batson et al., 1991; Xiao et al., 2021). In examining disparities of emotional reactions to tragedies abroad, potential explanations may lie in the absence of compassion or helping behavior. The collapse of compassion refers to the phenomenon in which individuals are less emotionally sensitive to the suffering of particular groups of people (Slovic, 2007). This phenomenon is particularly salient in the case of the identified victim effect in which the suffering of individuals or smaller groups elicits stronger reactions and consequently greater helping behavior compared to larger groups (Kogut & Ritov, 2005). A prevalent explanation is that individuals inherently attend to or process the plight of individuals differently from groups. A second explanation for the identified victim effect, motivated emotional regulation, may better apply to explain disparities in reactions to tragedy. This explanation is grounded in the notion that witnessing another’s suffering induces negative physiological arousal, and that compassion is costly, creating a balance between steps to reduce the other’s suffering and in­turn one’s own negative arousal alongside the costs of compassion (Dovidio et al., 1991) . Researchers have shown that individuals only engage in prosocial behavior when there are limited financial and cognitive costs (Penner et al., 2005; Shaw et al., 1994). Individuals often take proactive steps to reduce emotional sensitivity when expecting intense emotional arousal (Lazarus & Folkman, 1984). With this, Cameron and Payne (2011) supported that, when facing the suffering of large groups, individuals are motivated to downregulate their emotional responses out of an expected overwhelming intensity of emotion and inability to help others.

Crucially, just as mass suffering can feel more challenging to alleviate than the suffering of an individual victim, tragedies that occur in more distant places may seem more difficult to address than those that occur closer to home. Indeed, Touré­Tillery and Fishbach (2017) found that, when manipulating spatial distance to those in suffering, individuals perceived lesser helping self­efficacy and thus helping behavior. Taken together, reactions to tragedies abroad may vary due to an individual’s spatial distance from victims in which a greater distance decreases their perceived ability to help, subsequently leading to a motivated and proactive downregulation of compassion and emotional response. This would indicate a serial mechanism of Distance → Helping Self­Efficacy → Compassion → Reaction.

Perceived Threat to Self

Another mechanism of reactions to tragedy could be more directly out of self­interest—a perceived threat to oneself. It is well­documented that psychological distance can predict threats to oneself in a variety of applied contexts. Psychological distance from environmental threats is negatively associated with the perceived severity of these threats (Carmi & Kimhi, 2015). In the context of disease, all four dimensions of psychological distance (i.e., social, spatial, temporal and hypothetical) are negatively correlated with perceived threat (White et al., 2014). In the context of crime, psychological distance predicts fear of victimization (Mellberg et al., 2022). In manipulating distance, increasing spatial distance from an object can decrease perceived threat (Stamps, 2011). There is further evidence of bidirectional effects with people tending to perceive threatening objects as closer (Fini et al., 2018). With clear connections between distance and threat perception, we must examine their downstream effects on emotional response.

Researchers have found that even spatial distance primes on a Cartesian plane, unrelated to subsequently consumed violent media, have been shown to affect emotional response with greater distance yielding lesser distress (Williams & Bargh, 2008). This effect of distance and threat perception on emotion has been shown to apply to a number of contexts of varying scales. Several studies have demonstrated that immediate and local proximity to terrorist attacks is associated with intensity of emotional response and PTSD (North et al., 1994; Sprang, 1999; Vázquez et al., 2006). On a national scale, following the September 11 attacks, Americans reported stress responses and perceptions of personal threat throughout the country (Schuster et al., 2001). On an international scale but varied circumstance, social media analysis in the Netherlands found that psychological distance and perceived threat corresponded with the

increased spread and proximity of Ebola with greater responses of fear (van Lent et al., 2017). In the context of reactions to tragedy, we reason that increased spatial distance, operationalized as geographic distance, may reduce the perceived threat of the tragedy afflicting oneself, reducing emotional response. This would indicate a simple mechanism of Distance → Perceived Threat → Reaction.

The Present Research

Across two experiments, we pursued answers to the question: How does psychological distance affect reactions to tragedy abroad? We predicted to replicate the findings of past studies that greater psychological distance reduces emotional response, applied to the novel context of news stories from abroad. Crucially, we sought to address the open question regarding mechanism: If CLT­related abstraction cannot explain distance’s dampening effect on emotions in this context, what does? Using mediation methods, we explored two potential mechanisms: motivated emotional regulation (via helping self­efficacy) and perceived threat to oneself (see Figure 1). To achieve this, in Experiment 1 we manipulated the perceived location of a tragedy abroad, and in Experiment 2 we manipulated the perceived distance of the same location. To our knowledge, this is the first study to investigate the effect of psychological distance on reactions to tragedies abroad and its potential mechanisms. In doing so, we broadened the applied literature for theories of psychological distance, motivated emotional regulation and threat perception.

Experiment 1

Method

In this first study, we tested for differences in emotional reactions to a fabricated tragedy based on its perceived location. Specifically, we varied the country in which it allegedly had taken place, including Australia, Canada, Indonesia, Mexico, and Venezuela. With a U.S.­based participant pool, these countries vary along dimensions of geographic distance, cultural similarity and general stability relative to the United States. We collected measures of participants’ psychological distance as well as fear for oneself to identify evidence of prospective mechanisms for varied reactions. We additionally gathered measures of desensitization to violence from the country on a peripheral basis, independent of our primary mechanisms of distance. We predicted that emotional reactions would be greater in response to countries that were geographically close (vs. distant), culturally similar (vs. different), and perceived as stable (vs. unstable). Furthermore, we predicted emotional reactions to have a negative association with perceived geographic distance, cultural distance and desensitization.

Participants

Two hundred forty­eight participants were recruited online using Amazon’s Mechanical Turk (mTurk) and were compensated $1.00 for participation in the study. A sensitivity analysis using G*Power indicated that this sample size was sufficient to detect an effect size of F = .22 for the main effect of country condition and r = .18 for correlations with Power = .80. The sample pool was restricted to U.S.­based participants above the age of 18. Self­reported measures of race (78.9% White/ European American, 9.3% Black/African American, 6.5% Asian American/Pacific Islander, and 5.2% Hispanic/Latino) and income (29% less than $30,000, 45.2% $30,000–$60,000, 20.2% $60,000–$100,000 & 5.6% More than $100,000) were collected. The sample size was determined based on available funding at the time of data collection.

Study Materials

A template of an article was written to depict a report on a recent terrorist attack. This vignette included a balance of both objective reporting and descriptive rhetoric. The reporting was straightforward and informative, including plainly stated facts, (i.e., “The next explosion occurred at 8:56 a.m. near (building name), where the death toll was 21, the police said.”), and circumstances (i.e., “As of now, no organization has taken credit for inspiring or executing the attacks.”). The descriptive rhetoric was aimed to elicit emotions using charged language (i.e., “commuters on the city’s buses were

plunged into the nightmare of an apocalyptic bloodbath”) and quotes (i.e., “ ‘They were shaken, bleeding,’ Mr. (name) said. “One woman was badly injured, burned on her face. She had it covered. People were just shocked.’ ”). The fabricated event largely mirrors the 2005 London Bombings with a few notable changes. First, in this event bombs were set off exclusively on buses as opposed to the three trains and one bus in 2005. This was done to ensure realism as some of the locations used in this study do not possess subway infrastructure. Second, it was determined that the 2005 London Bombings were executed at the hands of Islamic extremists. In the article, the motivations of the attackers were omitted to avoid any potential political bias in participants’ reactions and to increase the plausibility of the article across all country conditions. The article was presented as being published by the news organization Reuters, an internationally focused news agency that is largely perceived as a neutral source (Elejalde et al., 2018). The template was then applied for the following countries: Canada, Australia, Indonesia, Mexico, and Venezuela. To do this, only proper nouns such as names of locations, individuals, and buildings involved within the article were changed based on real locations, common names, and entities within the country. This was done to keep the magnitude of the tragedy constant while manipulating the perceived location of the tragedy to the participant. The countries used in the manipulation were selected to create a range for geographic distance and cultural similarity. We determined geographic distance in terms of miles from central locations within each country and cultural distance using the Hofstede Index. The Hofstede Index, a tool developed by Hofstede Insights, effectively compares countries by cultural similarity. This index quantifies cultural similarity based on six factors of dimensions: power distance, individualism, masculinity, uncertainty avoidance, long term orientation, and indulgence (Hofstede Insights, 2024). For the purposes of this study, we generalized cultural similarity by totaling the differences between the United States and selected countries in each of the six factors. Based on these criteria, four countries were selected: Canada (geographically close/culturally similar), Mexico (geographically close/culturally different), Australia (geographically distant/culturally similar), and Indonesia (geographically distant/culturally different). A fifth country, Venezuela, was included to support a peripheral analysis of how desensitization to violence influences responses to tragedy. Venezuela is one of the most unstable and fragile states in the world according to the Fragility States Index (2019) and additionally contributed as a country that is both geographically distant and culturally different from the United States according to the Hofstede Index.

Procedure

Using a between­subjects design, participants recruited on Mechanical Turk were directed to a survey and randomly assigned to one of the five country groups: Australia (n = 49), Canada (n = 46), Indonesia (n = 53), Mexico (n = 50), and Venezuela (n = 50). Following completion of a consent form, participants were instructed to first read a provided vignette and then continue with answering a questionnaire on a separate page. Participants were not informed that the event was fabricated until debriefing after the study to obtain more realistic responses to the tragedy. A response to every question was required to reach the conclusion of the survey. Participants were led to a debrief page and compensated $1.00 following survey completion.

Emotional Reaction

Participants responded to nine items adapted from the Discrete Emotions Questionnaire (DEQ; Harmon­Jones et al., 2016). The DEQ is a scale that collates responses to measure eight distinct emotions. For this study, we assessed fear, anxiety, and sadness as those deemed as the most relevant component emotions. All nine items were averaged to create an index of emotional reaction. Additionally, the corresponding items (three each) for fear, anxiety, and sadness were averaged to create indices of component emotions. All emotion indices exhibited high internal reliability (αoverall = .95, αfear = .93, αanxiety = .91, αsadness = .84). All items used the same 5­point fully labeled scale from 1 (not at all) to 5 (very much). These items shared the same structure of “To what extent has this article made you experience (insert measure)?”

Attention Checks

Two attention­check items were included, recounting details from the article in naming the city of the attack and the total number of casualties. These items were multiple­choice questions with four choices each.

Single-Item Measures

Participants answered single­item measures of perceived prosperity and stability, expectation of tragedy, general relatedness, perceived cultural distance, perceived geographic distance, and perceptions of the extent to which tragedies from the region in the article generally affected their day­to­day life and mood. These items shared the same structure of “To what extent do you (insert measure),” and participants responded to the same 5­point scale from 1 (not at all) to 5 (very much). In example for General Relatedness, participants answered the question “To what extent do you associate yourself or relate to the people of (Country) or the victims?” Additionally, a similarly worded item measuring fear for oneself was included with its five­point scale ranging from 1(strongly disbelieve) to 5 (strongly believe).

Mechanisms of Reaction to Tragedy Abroad | Wahid, Dawes, and Light

Analytic Plan

We first screened data quality by calculating the percentage of participants who correctly answered each attention ­ check item. We did not exclude any participants from data analysis. To test for differences in emotional reactions across country conditions, we conducted a one­way between­subjects ANOVA on overall emotional reaction. We then examined bivariate correlations between emotional reaction and predictors (relatedness, perceived geographic distance, perceived cultural distance, perceived stability, and expectation). To test the independent effects of geographic and cultural distance on component emotions (fear, anxiety, sadness), we ran multiple regression analyses with both predictors entered simultaneously. Finally, mediation analyses using bootstrapped generalized linear models (1,000 resamples) tested whether predictors mediated the relationship between perceived geographic distance and fear. All statistical tests were conducted using an alpha level of .05 (two­tailed). Analyses were conducted using Jamovi Version 2.5.7.0.

Results

Primary Analyses

We found that 91.9% of participants correctly named the city of the attack and 79.0% correctly chose the number of casualties. A one­way between­subjects ANOVA was conducted to compare the effect of country condition on overall emotional reaction, which found no significant differences between any two means, F(4, 243) = 1.55, p = .187. We then assessed correlations (α = .05) between predictors: general relatedness, perceived geographic distance, perceived cultural distance, expectation and perceived prosperity/stability and the dependent variable, overall emotional reaction. As shown in Table 1, both relatedness and perceived prosperity were significantly positively correlated with emotional reaction. Both perceived cultural distance, and in line with our theory, geographic distance were significantly negatively correlated with emotional reaction.

Exploratory Analyses

Although there was no significant main effect of country, we did find a significant correlation between perceived geographic distance and emotional reactions. We further explored this relationship and whether perceived geographic distance predicted emotions independently of perceived cultural distance by running multiple regressions predicting each component emotion (fear, anxiety, and sadness) with both perceived geographic and cultural distance as predictors (see Figure 2). Notably, we found that, when controlling for perceived cultural distance, perceived geographic distance was significantly negatively correlated with fear, B = ­0.16, SE = .08, t(247),

p = .035, but nonsignificantly correlated with sadness, B = ­ 0.02, SE =.08 , t(247), p = .747, and anxiety, B = ­0.12, SE = .08, t(237), p = .127. Thus, in line with our theory, perceived geographic distance had a unique relationship to fear reactions over and above perceived cultural distance.

In examining potential mediators, we found that adding relatedness to the equation eliminated the relationship between perceived geographic distance and fear, B = ­0.08, SE = .07, t(247), p = .288. Mediation analysis using a GLM mediation model with 1,000 bootstrap resamples confirmed that relatedness was a significant indirect pathway for the effects of perceived geographic distance on fear (B = ­.19, SE = .04, bootstrapped 95% CI: [­.26, ­.11]; see Figure 3). Geographic distance significantly predicted relatedness (B = −0.49, SE = 0.05, bootstrapped 95% CI: [−0.61, −0.37], p < .001), and relatedness significantly predicted fear (B = 0.37, SE = 0.06, bootstrapped 95% CI: [0.23, 0.49],

TABLE 1

Correlation Matrix for Experiment 1 1 2 3 4 5

Geographic Distance

Cultural distance

Relatedness

Prosperity

Expectiation

Emotional Reaction

Note * p < .05. ** p < .01. *** p < .001.

FIGURE 2

Average Participants' Ratings of Perceived Geographic Distance, Cultural Distance, and Prosperity of Assigned Country

Note. Error bars represent 95% confidence intervals.

Mechanisms

p < .001). When participants perceived that the tragedy had occurred at a greater geographic distance, they identified less with the victims of the tragedy, and in turn felt less fear in response.

Although Experiment 1 found results in line with the hypothesis that perceived geographic distance results in less emotional response to tragedies, only correlational results supported this hypothesis, and no significant differences were found between country conditions. This may be due to pre­existing assumptions and associations that participants had with the specific countries. Although we attempted to control for some such associations by varying the degree of cultural difference, there are likely other ways that perceptions of these countries differ that were not captured in our operationalization (e.g., the perception of Australia as an exciting tourist destination, etc.). In Experiment 2, we adopted a new manipulation of perceived distance that allowed us to hold the country in which the attack occurred constant.

Experiment 2 Methods

In this second study, we tested for differences in emotional reactions to a fabricated tragedy at the same location while manipulating participants’ perceived geographic distance to it. The aim of the study was to isolate the role of geographic distance and further investigate two potential mechanisms for its effect on reactions. Given that Experiment 1 identified perceived relatedness to victims as a mechanism linking perceived geographic distance, one possible explanation for the relationship between distance and emotional reaction is that nearer tragedies are perceived as more personally threatening, in that such events may thus be more likely to occur in one’s own geographic area. Simultaneously, and in line with our reasoning, distance may also reduce emotional reactions through mechanisms other than fear for personal safety. Research on the identified victim effect finds that a greater number of victims increases the perceived difficulty of helping the victims, leading to emotional disengagement as a means of emotion regulation (Cameron & Payne, 2011). More distant tragedies may evoke a collapse of compassion because distance reduces helping self­efficacy (Touré­Tillery & Fishbach, 2017), leading to down­regulation of compassion to reduce negative emotions that cannot be acted upon.

We predicted that manipulating the perceived distance of the tragedy’s location as closer, would yield greater emotional reactions from participants than farther away. We further predicted that this would be mediated by feelings of compassion and helping

self­efficacy with greater feelings of compassion and helping self­efficacy in the close condition in a serial model (Distance → Helping Self­Efficacy → Compassion → Reaction). Additionally, we predicted that this would be mediated by threat perception with greater levels of perceived threat in the close condition in a simple model (Distance → Perceived Threat → Reaction). This experiment was preregistered prior to data collection using the Open Science Framework (OSF) preregistration template, which can be viewed at https://osf.io/nbtm5/ . This preregistration further includes data files and study materials. We tested further mediational models on an exploratory basis looking at individual mediators such as compassion or self­efficacy.

Participants

Five hundred fifty­six participants were recruited online using CloudResearch Connect and were compensated $1.00 for participation in the study. The sample pool was restricted to U.S.­based participants above the age of 18. Four participants were excluded (nclose=3, ndistant=1) from analyses due to answering two of two attentioncheck questions incorrectly1. Self­reported measures of age (M = 38.1, SD = 12.6), gender (58.1% male, 48.2% female), race (65.4% White/European American, 16.8% Black/African American, 7.8% Asian American/Pacific Islander, 6% two or more races, 3.8% Hispanic/Latino, and 0.2% Middle­Eastern), place of birth (94.3% in the U.S., 5.7% abroad), first language (97.6% English, 2.4% non­English), student status (88.5% nonstudents, 11.5% students), and education ­ level (0.4% less than high school, 9.2% high school or equivalent, 19.6% some college, 8.5% two­year degree, 46.4% four­year degree, 14.7% professional degree, and 1.3% doctorate) were collected. To determine sample size, we performed a power analysis using G*Power. We specified an a priori power analysis for an independent­samples t test with a one­tailed test, an even allocation between conditions, α = .05, Power = .95, and an estimated effect size of

3

Concept Diagram of Exploratory Simple Mediation Model for Experiment 1

1 Inclusion of these four participants did not change either the statistical significance or pattern of any reported results.

d = 0.28, based on Touré­Tillery & Fishbach (2017). This analysis suggested a sample size of n = 554.

Study Materials

We applied the same article template used in Experiment 1 to Mexico but rather than the tragedy occurring in Mexico City, had it placed in the city of Merida, located on the curled, southern tip of Mexico. Its unique geographic position enables the manipulation of its perceived distance as with the reference point of the U.S. city, Houston, it is over a 1500­mile drive away or less than half this distance by flight. Its relative obscurity compared to the country’s capital, Mexico City, prevents more biased responses from participants due to knowledge of its location.

To manipulate perceived distance, we followed the method employed by Touré­Tillery and Fishbach (2017). We prepared two analogous descriptions of Merida, using close or distant descriptors of its location relative to the United States. The descriptions (close vs. distant) read as follows:

On the following page, you will read an article about an event that occurred in the (nearby vs distant) city of Merida, located (in the neighboring country of Mexico vs. on the southern tip of Mexico). Merida is a vibrant city, rich in history and culture with many colonial­era churches and local dishes. It is a well­visited place by American tourists (just a two­hour flight from Houston and vs. requiring a two­day car drive from Houston but) well worth it. It is further known for its beautiful beaches and nature reserves.

Additionally, the first question of the main questionnaire varied in wording to further prime close or distant perceptions. In the close condition, participants were asked how nearby Merida is with answer choices repeating the word “close.” In the distant condition, participants were asked how far away Merida is with answer choices repeating the word “far.” These questions solely functioned as a part of the manipulation given their biased wording, and responses were not analyzed. This was followed by a general question measuring perceived distance, which acted as a manipulation check.

Procedure

Using a between­subjects design, participants recruited on CloudResearch Connect were directed to a survey and randomly assigned to the close or distant condition. Following completion of a consent page, participants were presented with their respective close or distant description of Merida. Then participants were instructed

to read the same article depicting the tragedy occurring in Merida. Participants were not informed that the event was fabricated until debriefing after the study. After the article, participants answered two attention­check items recounting details from the article (city and number of casualties). Four participants answered both questions incorrectly and were excluded from analyses (nclose = 275, ndistant = 277). Participants were then led to the main questionnaire and were required to answer all questions to proceed with the study. Following this, participants answered a separate set of demographic questions and were lastly debriefed. Participants were compensated $1.00 through their CloudResearch Connect account following survey completion.

Emotional Reaction

As measured in Experiment 1, participants responded to the same nine items adapted from the DEQ (HarmonJones et al., 2016). This included emotions of fear, anxiety, and sadness with an index calculated with averages for overall emotional reaction. This 9­item index exhibited high internal consistency (αoverall = .95).

Compassion

Participants responded to a 9­item scale measuring compassion for the victims and families affected by the tragedy. This scale was adapted from Cameron and Payne (2011) with adjustments in wording to apply to the context of the article and the original 7­point scale was converted to a 5­point scale from 1 (not at all) to 5 (very much) for consistency with other measures. All items were averaged to create an index for compassion. This 9­item index exhibited high internal consistency (αcompassion = .93).

Helping Self-Efficacy

Participants responded to two items measuring helping self­efficacy, adapted from Cameron and Payne (2011): “Imagine that you donated $10 to help the families of the victims. How effective do you think that $10 would be in helping the victims and their families in Merida?” and “How much of a difference do you think that $10 would make in helping the victims and their families in Merida?” Participants responded to each item on a 5­point scale from not at all to very much. Responses for both items were averaged into an index of helping self­efficacy. This 2­item index exhibited high internal consistency (αefficacy = .94).

Perceived Threat

Participants responded to two items measuring perceived threat, adapted from Lee & Lemyre (2009): “How likely do you think it is that explosive terrorism like this will occur near where you live?” and “To what

Concept Diagram of Exploratory Parallel Mediation Model for Experiment 2 Mechanisms

extent do you currently worry about terrorism happening near where you live?” Participants responded to each item on a 5­point scale from 1 (not at all) to 5 (very much). Responses for both items were averaged into an index of perceived threat. This 2­item index exhibited high internal consistency (αthreat = .83).

Analytic Plan

We first screened data quality by calculating the percentage of participants who correctly answered each attention­check item and identifying participants who answered both incorrectly. We excluded participants who answered both items incorrectly according to our preregistered exclusion criteria. To confirm the distance manipulation, we conducted an independent­samples t test on perceived distance ratings between groups. We conducted preregistered one­tailed t tests to compare close and distant conditions on emotional reaction, compassion, helping self­efficacy, and perceived threat. To test our main hypothesis, we ran preregistered mediation analyses using bootstrapped generalized linear models (1,000 resamples) to test whether (a) helping self­efficacy followed by compassion and (b) perceived threat mediated the effect of distance on emotional reaction. Exploratory models further tested helping self­efficacy and compassion as parallel mediators to assess their independent indirect effects. All statistical tests were conducted using an alpha level of .05. Analyses were conducted using Jamovi Version 2.5.7.0.

Results

Preregistered Analyses

Following the removal of four participants for incorrectly answering both attention­check questions, 98.7% of participants correctly named the city of the attack and 80.6% correctly chose the number of casualties. In comparing perceived geographic distance between groups, a manipulation check, participants in the distant group perceived significantly greater distance than the close group, t(550) = 3.99, p < .001, d = 0.34.

We conducted additional independent ­ sample one­tailed t tests with directionality as predicted in our preregistered hypotheses to compare across groups for differences in emotional reaction, compassion, helping self­efficacy, and perceived threat. We did not find a significant main effect of distance condition on emotional reaction, t(550) = 0.53, p = .702, d = ­0.05. Furthermore, we did not find a significant difference between groups in helping self­efficacy, t(550) = ­1.32, p = .094, d = 0.11. We did not find a significant difference between groups in perceived threat, t (550) = ­ 0.41, p = .342, d = 0.03. However, we did find a significant effect of distance condition on compassion; participants

in the close condition (M = 4.02, SD = 0.93) had significantly greater compassion for the victims of the tragedy than the distant condition ( M = 3.88, SD = 0.90), t(550) = ­1.85, p = .032, d = 0.16.

Mediation analysis using a GLM mediation model with 1,000 bootstrap resamples found that perceived threat was not a significant indirect pathway for the effects of distance on emotional reaction (B = .006, SE = .031, bootstrapped 90% CI: [­.04, .07]). A second GLM mediation model with 1,000 bootstrap resamples found that helping self­efficacy followed by compassion was not a significant indirect pathway for the effects of distance on emotional reaction (B = .006, SE = .009, bootstrapped 90% CI: [­.002, .028]).

Exploratory Analyses

Mediation analysis using a GLM mediation model with 1,000 bootstrap resamples found that, with helping self­efficacy and compassion as parallel mediators (see Figure 4), compassion acted as a significant indirect pathway for the effects of distance on emotional reaction (B = .03, SE = .04, bootstrapped 90% CI: [.002, .135]).

Discussion

Across both studies we found a connection between spatial distance and emotional reaction. In Experiment 1, perceived geographic distance was significantly negatively correlated with emotional reaction. We further found that perceived geographic distance had a unique relationship with the component emotion of fear when controlling for cultural distance, but not with sadness or anxiety. In Experiment 2, manipulating perceived distance directly resulted in lower compassion for the victims, albeit with small effect sizes, which mediated indirect effects on emotional response. In combination, these studies support that previously observed effects of distance on emotional response (Moran & Eyal, 2022) may extend to reactions to foreign tragedies and suggest

particular mechanisms of the effect.

Specifically, results in Experiment 2 implicated diminished compassion when the event was framed as geographically distant as the explanatory mechanism for lower emotional response. This is in line with the prediction that a collapse of compassion occurs in the case of increased perceived distance from tragedy abroad, reducing the intensity of emotional reaction to it. However, we had predicted that a collapse of compassion would be mediated by helping self­efficacy—that in line with previous work, participants who perceived the tragedy to be more geographically distant would feel that their help was less impactful (Touré­Tillery & Fishbach, 2017), and would then be motivated to reduce compassion for suffering they felt incapable of alleviating. Although we did find that compassion was correlated with helping self­efficacy, we did not find that self­efficacy mediated the effect of distance on compassion. It is possible that the measure of helping self­efficacy we used failed to capture the construct. We focused on the impact of donations to victims, but perhaps participants considered other means of helping to be more important in response to a terrorist attack. Additionally, the statistical reliability of the two­item scale might have limited our power to assess relationships with other variables, especially relative to the lengthier compassion index.

Alternatively, participants in both the distant and close conditions might have felt unable to alleviate the victim’s suffering, and something besides helping selfefficacy might have motivated participants to engage in compassion when the tragedy was depicted as nearby. Prior research (Scheffer et al., 2022) has highlighted that, although empathy and compassion are experienced as effortful, people can be motivated to engage in them (e.g., by rewarding the decision to empathize; Ferguson et al., 2020). It seems likely that many factors in daily life motivate people to engage in empathy and compassion in some situations, but not others (Cameron, 2018). In this case, for example, perceiving victims as being geographically close may affect how victims are categorized in terms of group membership (Glasford & Caraballo, 2016), in which case deploying compassion might be perceived as benefiting one’s own welfare (Stürmer et al., 2006). In other words, a broader consideration of relevant factors that motivate (vs. deter) compassion and empathy might help to elucidate how geographic distance undermines compassion.

This study serves as a valuable starting point in evaluating psychological mechanisms of reactions to tragedies abroad, which is still a fairly open question. It is important to identify these mechanisms to better understand ourselves and progress towards consciously creating more equitable reactions to human suffering.

It is a well ­ studied phenomenon that people across cultures value universalism and intend to treat others equitably (Schwartz, 2012). In an ever more globally connected world, people are not restricted in response to only nearby tragedy. This work can help enable greater empathy and help where it is needed the most, not guided by spatial differences and their downstream impacts on emotional response.

Limitations and Future Directions

This study has a number of observable limitations. Initially, this study was prompted by disparities in public reactions to real tragedies abroad. Although our example comparing the 2015 Ankara bombings and Paris attacks holds key similarities with a difference in geographical distance, notably the United States has stronger historical and political ties to France, which might have contributed to a greater reaction. This reflects the inherent limitation of comparing real­world tragedies abroad with complex, varying contexts. Across both experiments, although we intended to simulate a representative reaction of the U.S. public to tragedy, demographic factors such as race and income were not proportionally reflected. In Experiment 1, we predicted to find differences in emotional response between country conditions corresponding to varied metrics of geographic and cultural distance. Contrary to expectations, we did not find significant differences. Notably, Experiment 1 employed several single­item measures including perceived geographic distance, which may limit its capture of the variable. We suspect that operationalizing distance by varying the country confounded geographic distance with other associations participants might have had with the target country, obscuring the effects of distance alone. Both Mexico and Canada, countries included in Experiment 1, border the U.S. and have great significance in terms of trade, policy, and immigration which may skew participants’ reactions. This is further obscured by the varying extent of geographic knowledge between individual participants across all countries. By consequence, the results linking perceived geographic distance to emotional response in Experiment 1 are correlational, leaving open the possibility of third variables (e.g., participants’ dispositional tendency to empathize across national borders) or directionality (i.e., having a stronger emotional response to an event increasing the perception of geographic closeness to the event). We were able to address this issue by manipulating perceived distance in Experiment 2 and holding actual location constant, which resulted in generally consistent findings regarding compassionate and emotional responses. Although we notably did not find a main effect of this manipulation on emotional response,

participants reported significantly greater compassion for victims in the close (vs. distant) condition, which mediated a significant indirect effect on emotional response. It may be that reading about a tragedy primarily activates more proximal emotions and motivations like compassion, which in turn have downstream consequences for emotions, but it is also plausible that the measure of compassion was more sensitive than the measure of emotion. Given social desirability concerns with self­report measures, and the general difficulty of measuring emotion via self­report (Larson, 2018; Marcus et al., 2006), it is plausible that participants might have felt more comfortable reporting compassion in this context relative to more self­focused or context­free emotions like sadness and fear. We acknowledge that, although this finding was significant, its effect size was small, which may reflect a weakness of our manipulation or the general sensitivity of such measures.

In addition, although we predicted a main effect of condition on emotional reactions in Experiment 2, we observed only an indirect effect, mediated by compassion. Notably we found that the manipulation of distance itself had a rather small effect on perceptions of geographic distance. Given that the relationship between the condition and the dependent variables is limited by the strength of the manipulation, it is very plausible that these weaker effects on the dependent variables were due to the relatively small effect of the manipulation on the independent variable.

In addition, the experience of reading a written news article in the context of a study might have had relatively weak impacts on participants’ emotions in general. Although news media is certainly capable of evoking strong emotional reactions, the effects of an isolated article in the artificial context of an online study might be markedly less impactful than naturalistic experiences of exposure to reporting on world events. This relatively weak emotional impact might have limited our ability to observe and explain differences in emotional reactions depending on perceived geographic distance.

In addition, in Experiment 1, the smaller participant pool of 248 respondents with roughly 50 participants per country condition presents a low statistical power, which might have concealed true differences in emotional reaction between countries.

Finally, this study used mediation methods to assess mechanism. Although statistical mediation is one of the most common methods of providing evidence for mechanism, it is inherently limited in its ability to distinguish between causal models and biased by the reliability of underlying measures (Pek & Hoyle, 2016). Future research more definitely addressing the collapse of compassion in response to distant tragedies may benefit

by using other methods of establishing mechanism, such as directly manipulating the proposed mediators to see if the original main effect is enhanced or eliminated (Bullock & Green, 2021). In this case, we implicate emotion regulation processes, which suggest that participants are strategically up­regulating compassion in response to geographically close events, and down­regulating compassion in response to distant events. Studies that either assess individual capacity for emotion regulation (e.g., via trait self­control) or that manipulate emotion regulation goals independently from geographic distance would provide stronger evidence regarding this mechanism.

This study poses a number of future research directions. We argue that the results in these studies support a collapse of compassion in response to tragedies in distant countries (relative to close countries), as indicated by compassion mediating indirect effects of distance on emotional reactions more generally. Current perspectives on compassion collapse describe it as a function of emotion regulation. Future research should focus on the role of emotion regulation in explaining emotional reactions to events in distant places—for example, we would expect that individual differences in emotion regulation abilities (e.g., trait self ­ control) and conditions of anticipated effort or cognitive load would moderate the observed effects. By contrast, assessing or manipulating emotion regulation goals could result in participants up ­ regulating their compassionate and emotional responses in some situations. This would help address an important limitation of our study, namely that we did not examine regulation in a targeted manner. Although our findings did not support helping self­efficacy as a precursor, we observed a collapse of compassion that must have occurred through a different mechanism. We predict that, in the context of reactions to tragedy abroad, spatial distance affects a different factor for motivated emotional regulation. Further research can strengthen the evidence supporting this phenomenon and elucidate its mechanism(s). On a broader scale, this may encourage applied psychological research on emotional reactions to events.

Conclusion

The present research extended the literature of the effect of psychological distance on emotion, finding its applicability to reactions to tragedy abroad. We found that participants tend to react with less intensity to tragedies when perceived as further geographically with a particular significance for fear. This research provides initial evidence of compassion as a mediator between distance and emotional reaction. We suggest that future work further explore the mechanisms involved with compassion to deepen understanding of psychological distance in this applied context.

Mechanisms of Reaction to Tragedy Abroad | Wahid, Dawes, and Light

References

Abraham, D., Powers, J. P., & McRae, K. (2023). Emotional appraisal, psychological distance and construal level: Implications for cognitive reappraisal. Emotion Review, 15(4), 313–331. https://doi.org/10.1177/17540739231196577

Batson, C. D., Batson, J. G., Slingsby, J. K., Harrell, K. L., Peekna, H. M., & Todd, R. M. (1991). Empathic joy and the empathy-altruism hypothesis. Journal of Personality and Social Psychology, 61(3), 413–426. https://doi.org/10.1037//0022-3514.61.3.413

Cameron, C. D. (2018). Motivating empathy: Three methodological recommendations for mapping empathy. Social and Personality Psychology Compass, 12(11). https://doi.org/10.1111/spc3.12418

Cameron, C. D., & Payne, B. K. (2011). Escaping affect: How motivated emotion regulation creates insensitivity to mass suffering. Journal of Personality and Social Psychology, 100(1), 1–15. https://doi.org/10.1037/a0021643

Carmi, N., & Kimhi, S. (2015). Further than the eye can see: Psychological distance and perception of environmental threats. Human and Ecological Risk Assessment: An International Journal, 21(8), 2239–2257. https://doi.org/10.1080/10807039.2015.1046419

Davis, J. I., Gross, J. J., & Ochsner, K. N. (2011). Psychological distance and emotional experience: What you see is what you get. Emotion, 11(2), 438–444. https://doi.org/10.1037/a0021783

Dovidio, J. F., Piliavin, J. A., Gaertner, S. L., Schroeder, D. A., & Clark, R. D. (1991). The arousal: Cost-reward model and the process of intervention: A review of the evidence. In M. S. Clark (Ed.), Prosocial behavior (pp. 86–118). Sage Publications, Inc.

Elejalde, E., Ferres, L., & Herder, E. (2018). On the nature of real and perceived bias in the mainstream media. PloS one, 13(3), e0193765. https://doi.org/10.1371/journal.pone.0193765

Ferguson, A. M., Cameron, C. D., & Inzlicht, M. (2020). Motivational effects on empathic choices. Journal of Experimental Social Psychology, 90, 104010. https://doi.org/10.1016/j.jesp.2020.104010

Fini, C., Verbeke, P., Sieber, S., Moors, A., Brass, M., & Genschow, O. (2018). The influence of threat on perceived spatial distance to out-group members. Psychological Research, 84(3), 757–764. https://doi.org/10.1007/s00426-018-1091-7

Garunay, M. (2015, November 13). President Obama offers a statement on the attacks in Paris. Obama White House Archives https://obamawhitehouse.archives.gov/ blog/2015/11/13/watch-president-obamas-statement-attacks-paris

Glasford, D. E., & Caraballo, K. L. (2015). Collective action from a distance: Distance shapes how people view victims of injustice and decreases willingness to engage in collective action. Group Processes & Intergroup Relations, 19(1), 27–42. https://doi.org/10.1177/1368430215570505

Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: An evolutionary analysis and empirical review. Psychological Bulletin, 136(3), 351–374. https://doi.org/10.1037/a0018807

Harmon-Jones, C., Bastian, B., & Harmon-Jones, E. (2016). The Discrete Emotions Questionnaire: A new tool for measuring state self-reported emotions. PLOS One, 11(8). https://doi.org/10.1371/journal.pone.0159915

Healy, J., & Fausset, R. (2015, November 20). Americans view Paris attacks with empathy, fear and resolve. The New York Times. https://www.nytimes.com/2015/11/21/us/ americans-view-paris-attacks-with-empathy-fear-and-resolve.html

Henderson, M. D., Wakslak, C. J., Fujita, K., & Rohrbach, J. (2011). Construal level theory and spatial distance. Social Psychology, 42(3), 165–173. https://doi.org/10.1027/1864-9335/a000060

Hofstede Insights. (2024). Country comparison tool. The Culture Factor Group Retrieved from https://www.theculturefactor.com/country-comparison-tool

Kogut, T., & Ritov, I. (2005). The “identified victim” effect: An identified group, or just a single individual? Journal of Behavioral Decision Making, 18(3), 157–167. https://doi.org/10.1002/bdm.492

Larson, R. B. (2018). Controlling social desirability bias. International Journal of Market Research, 61(5), 534–547. https://doi.org/10.1177/1470785318805305

Lazarus, R., & Folkman, S. (1984). Stress, appraisal, and coping. Springer. Lee, J. E., & Lemyre, L. (2009). A social‐cognitive perspective of terrorism risk perception and individual response in Canada. Risk Analysis, 29(9), 1265–1280. https://doi.org/10.1111/j.1539-6924.2009.01264.x

Levine, M., Cassidy, C., Brazier, G., & Reicher, S. (2002). Self‐categorization and bystander non‐intervention: Two experimental studies1. Journal of Applied Social Psychology, 32(7), 1452–1463. https://doi.org/10.1111/j.1559-1816.2002.tb01446.x

Li, K., Lv, Y., Dong, Y., Wang, T., Wu, J., Zhang, Z., Li, X., Han, R., & Lin, F. (2019). Orderliness/disorderliness is mentally associated with construal level and

psychological distance. Frontiers in Psychology, 10 https://doi.org/10.3389/fpsyg.2019.02521

Liberman, N., & Trope, Y. (2003). Construal level theory of intertemporal judgment and decision. Time and decision: Economic and psychological perspectives on intertemporal choice (pp. 245–276). Russell Sage Foundation.

Liberman, N., & Trope, Y. (2014). Traversing psychological distance. Trends in Cognitive Sciences, 18(7), 364–369. https://doi.org/10.1016/j.tics.2014.03.001

Maglio, S. J. (2019). Psychological distance in consumer psychology: Consequences and antecedents. Consumer Psychology Review, 3(1), 108–125. https://doi.org/10.1002/arcp.1057

Marcus, G. E., Mackuen, M., Wolack, J., & Keel, L. (2006). The measure and mismeasure of emotion. In D. P. Redlawsk (Ed), Feeling politics: Emotion in political information processing (1st ed., pp. 31–45). Palgrave Macmillan. Marshall, J., & Wilks, M. (2024). Does distance matter? How physical and social distance shape our perceived obligations to others. Open Mind, 8, 511–534. https://doi.org/10.1162/opmi_a_00138

Mellberg, J., Chataway, M. L., Ball, M. J., & Miles-Johnson, T. (2022). Psychological distance and fear of crime: Towards a new understanding of risk perception formation. Journal of Criminology, 55(3), 377–399. https://doi.org/10.1177/26338076221105899

Miller, N. E. (1944). Experimental studies of conflict. In J. M. Hunt (Ed), Personality and the behavior disorders: A handbook based on experimental and clinical research (pp. 431–464). The Ronald Press Company.

Moran, T., & Eyal, T. (2022). Emotion regulation by psychological distance and level of abstraction: Two meta-analyses. Personality and Social Psychology Review, 26(2), 112–159. https://doi.org/10.1177/10888683211069025

Mrkva, K., Travers, M., & Van Boven, L. (2018). Simulational fluency reduces feelings of psychological distance. Journal of Experimental Psychology: General, 147(3), 354–376. https://doi.org/10.1037/xge0000408

Mühlberger, A., Neumann, R., Wieser, M. J., & Pauli, P. (2008). The impact of changes in spatial distance on emotional responses. Emotion, 8(2), 192–198. https://doi.org/10.1037/1528-3542.8.2.192

Norman, E., Tjomsland, H. E., & Huegel, D. (2016). The distance between us: Using construal level theory to understand interpersonal distance in a digital age. Frontiers in Digital Humanities, 3. https://doi.org/10.3389/fdigh.2016.00005

North, C. S., Spitznagel, E. L., & Smith, E. M. (1994). Posttraumatic stress disorder in survivors of a mass shooting. American Journal of Psychiatry, 151(1), 82–88. https://doi.org/10.1176/ajp.151.1.82

Penner., L. A., Dovidio., J. F., Piliavin., J. A., & Schroeder., D. A. (2005). Prosocial behavior: Multilevel perspectives. Annual Review of Psychology, 56(1), 365–392. https://doi.org/10.1146/annurev.psych.56.091103.070141

Scheffer, J. A., Cameron, C. D., & Inzlicht, M. (2022). Caring is costly: People avoid the cognitive work of compassion. Journal of Experimental Psychology: General, 151(1), 172–196. https://doi.org/10.1037/xge0001073

Schuster, M. A., Stein, B. D., Jaycox, L. H., Collins, R. L., Marshall, G. N., Elliott, M. N., Zhou, A. J., Kanouse, D. E., Morrison, J. L., & Berry, S. H. (2001). A national survey of stress reactions after the September 11, 2001, terrorist attacks. New England Journal of Medicine, 345(20), 1507–1512. https://doi.org/10.1056/nejm200111153452024

Schwartz, S. H. (2012). An overview of the Schwartz theory of basic values. Online Readings in Psychology and Culture, 2(1). https://doi.org/10.9707/2307-0919.1116

Shaw, L. L., Batson, C. D., & Todd, R. M. (1994). Empathy avoidance: Forestalling feeling for another in order to escape the motivational consequences. Journal of Personality and Social Psychology, 67(5), 879–887. https://doi.org/10.1037/0022-3514.67.5.879

Slovic, P. (2007). “If I look at the mass I will never act”: Psychic numbing and genocide. Judgment and Decision Making, 2(2), 79–95. https://doi.org/10.1017/s1930297500000061

Sprang, G. (1999). Post-disaster stress following the Oklahoma City bombing. Journal of Interpersonal Violence, 14(2), 169–183. https://doi.org/10.1177/088626099014002005

Stamps, A. E. (2011). Distance mitigates perceived threat. Perceptual and Motor Skills, 113(3), 751–763. https://doi.org/10.2466/07.21.pms.113.6.751-763

Stürmer, S., Snyder, M., Kropp, A., & Siem, B. (2006). Empathy-motivated helping: The moderating role of group membership. Personality and Social Psychology Bulletin, 32(7), 943–956. https://doi.org/10.1177/0146167206287363

Touré-Tillery, M., & Fishbach, A. (2017). Too far to help: The effect of perceived distance on the expected impact and likelihood of charitable action. Journal of Personality and Social Psychology, 112(6), 860–876. https://doi.org/10.1037/pspi0000089

Mechanisms of Reaction to Tragedy Abroad | Wahid, Dawes, and Light

Trope, Y., Liberman, N., & Wakslak, C. (2007). Construal levels and psychological distance: Effects on representation, prediction, evaluation, and behavior. Journal of Consumer Psychology, 17(2), 83–95. https://doi.org/10.1016/s1057-7408(07)70013-x

Van Boven, L., & Caruso, E. M. (2015). The tripartite foundations of temporal psychological distance: Metaphors, ecology, and teleology. Social and Personality Psychology Compass, 9(11), 593–605. https://doi.org/10.1111/spc3.12207

Van Boven, L., Kane, J., McGraw, A. P., & Dale, J. (2010). Feeling close: Emotional intensity reduces perceived psychological distance. Journal of Personality and Social Psychology, 98(6), 872–885. https://doi.org/10.1037/a0019262

van Lent, L. G., Sungur, H., Kunneman, F. A., van de Velde, B., & Das, E. (2017). Too far to care? Measuring public attention and fear for ebola using Twitter. Journal of Medical Internet Research, 19(6). https://doi.org/10.2196/jmir.7219

Vázquez, C., Pérez-Sales, P., & Matt, G. (2006). Post-traumatic stress reactions following the March 11, 2004 terrorist attacks in a Madrid community sample: A cautionary note about the measurement of psychological trauma. The Spanish Journal of Psychology, 9(1), 61–74. https://doi.org/10.1017/s1138741600005989

White, A. E., Johnson, K. A., & Kwan, V. S. (2014). Four ways to infect me: Spatial, temporal, social, and probability distance influence evaluations of disease threat. Social Cognition, 32(3), 239–255. https://doi.org/10.1521/soco.2014.32.3.239

Williams, L. E., & Bargh, J. A. (2008). Keeping one’s distance: the influence of spatial distance cues on affect and evaluation. Psychological Science, 19(3), 302–308. https://doi.org/10.1111/j.1467-9280.2008.02084.x

Williams, L. E., Stein, R., & Galguera, L. (2014). The distinct affective consequences of psychological distance and Construal level. Journal of Consumer Research, 40(6), 1123–1138. https://doi.org/10.1086/674212

Wong, N. Y., & Bagozzi, R. P. (2005). Emotional intensity as a function of psychological distance and cultural orientation. Journal of Business Research, 58(4), 533–542. https://doi.org/10.1016/s0148-2963(03)00144-9

Wong, V. C., & Wyer, R. S. (2016). Mental traveling along psychological distances:

The effects of cultural syndromes, perspective flexibility, and construal level. Journal of Personality and Social Psychology, 111(1), 17–33. https://doi.org/10.1037/pspa0000048

Xiao, W., Lin, X., Li, X., Xu, X., Guo, H., Sun, B., & Jiang, H. (2021). The influence of emotion and empathy on decisions to help others. Sage Open, 11(2). https://doi.org/10.1177/21582440211014513

Yeginsu, C., & Arango, T. (2015, October 10). Explosions during peace rally in Ankara, Turkey’s capital, kill scores. New York Times. https://www.nytimes. com/2015/10/11/world/europe/ankara-turkey-explosion-deaths.html

Author Note.

This study was reviewed and approved by the University Committee on Activities Involving Human Subjects (UCAIHS) at New York University (Experiment 1) and Automating UniversityWide Research Administration IRB at the University of Chicago (Experiment 2). The authors declare no conflicts of interest. The authors received no specific funding for this work.

Experiment 2 was preregistered prior to data collection using the Open Science Framework (OSF) preregistration template, which can be viewed at: https://osf.io/nbtm5/ Materials and data can be viewed at https://osf.io/6m32r/

Syed M. Wahid played a lead role in data collection, data analysis and interpretation, and substantive original writing and a supporting role in conceptualization, research design, and editorial assistance. Christopher T. Dawes played a lead role in editorial assistance and a supporting role in conceptualization, research design, and substantive original writing. Alysson E. Light played a lead role in conceptualization, research design, and supervision and a supporting role in data collection, data analysis and interpretation, and substantive original writing.

Correspondence concerning this article should be directed to Syed M. Wahid at 863 Olmstead Avenue, Bronx, NY 10473. Phone: (347) 691­8724. Fax: 773­702­0886. Email: syedmw02@gmail.com

The Power of Perspective: How Framing Affects Undergraduate Student Attitudes Toward Graduate School

Hannah V. Hyman, Iris R. Wright, and Ralph G. Hale* Department of Psychological Science, University of North Georgia

ABSTRACT. This study examined how framing influences undergraduate students’ attitudes toward graduate school. Using a between­subjects design, 92 students from the University of North Georgia were randomly assigned to 1 of 3 framing conditions: positive, neutral, or negative. Participants read 10 statements about the graduate school experience that reflected their assigned framing condition (i.e., positive, neutral, or negative). After reviewing these statements, they rated their overall attitude toward graduate school on a 6­point scale. A 1­way ANOVA revealed a significant effect of framing on attitudes toward graduate school, F(2, 89) = 5.53, p = .005, ηp² = .12. Specifically, students exposed to negatively framed information reported significantly less favorable attitudes toward graduate school compared to those in the positive and neutral conditions (ps < .03), whereas no significant difference emerged between the positive and neutral groups (p = .13). These results suggest that undergraduates may hold generally favorable views of graduate school—but that these attitudes are vulnerable to decline when exposed to negatively framed messaging. Findings underscore the importance of how information about graduate education is communicated, with implications for advising, institutional outreach, and efforts to support informed academic decision­making. Future research should examine whether framing effects differ across demographic groups and persist over time.

Keywords: framing effect, attitudes, graduate school

RESUMEN. Este estudio examino como el encuadre influye las actitudes de los estudiantes universitarios hacia los estudios de posgrado. Empleando un diseño entre sujeto, 92 estudiantes de la Universidad fueron asignados al azar a 1 de 3 condiciones de encuadre: positiva, neutral, o negativa. Los participantes leyeron 10 frases sobre la experiencia de los estudios de posgrado alineados con la condición de encuadre asignada (positiva, neutral o negativa). Después de leer estas frases, valoraron su actitud general hacia los estudios de posgrado utilizando una escala de 6 puntos. El análisis de varianza de 1 vía mostro un efecto significativo del encuadre sobre las actitudes hacia los estudios de posgrado, F (2, 89) = 5.53, p = .005, η p ² = .12. En específico, el encuadre negativo se asoció con actitudes significativamente menos

favorables hacia los estudios de posgrado en comparación con las condiciones positiva y neutral (ps <.03), mientras que no se observo una diferencia significativa entre los grupos positivo y neutral (p = .13). Estos resultados apuntan que los estudiantes universitarios suelen tener percepciones generalmente favorables sobre los estudios de posgrado, aunque estas actitudes son frágiles y tienden disminuir ante la exposición a mensajes con encuadre negativo. Los hallazgos destacan la importancia de la forma en que se comunica la información sobre la educación de posgrado, con implicaciones para la orientación académica, la divulgación institucional, y los esfuerzos dirigidos a apoyar la toma de decisiones académicas informadas. Futuras líneas de investigación deberían examinar si los efectos del encuadre presentan diferencias entre distintos grupos demográficos y si persisten con el tiempo.

Palabras clave: efecto de encuadre, actitudes, estudios de posgrado

Graduate education serves as a critical steppingstone for individuals seeking to deepen their expertise, advance their careers, or transition into specialized fields. A graduate degree provides advanced training, fosters research skills, and opens pathways to higher paying and intellectually rewarding opportunities (Carnevale et al., 2011). Yet, the decision to pursue graduate education is not based solely on academic interest. It is shaped by a range of social, financial, and psychological factors. Rising tuition costs, unstable federal and state funding, concerns about student loan debt, and uncertainty about the return on investment all contribute to the complexity of deciding whether to continue into graduate school (Perna, 2004). Considering these challenges, understanding how undergraduate students form attitudes toward graduate education is vital for educators, policymakers, and institutions committed to supporting informed and strategic decision­making.

Despite its importance, research has indicated that both undergraduate students and the public often hold inaccurate, incomplete, or distorted perceptions of what graduate school entails. Some students view it merely as an extension of undergraduate education, whereas others see it as an elite, exclusive pathway reserved for those pursuing academic careers (Mullen et al., 2003). Misconceptions about affordability, workload, and the necessity of graduate school for career advancement further shape students’ willingness or reluctance to apply (Hearn & Rosinger, 2014). Students’ perceptions are influenced by multiple sources, including direct exposure to mentors and faculty, interactions with peers, institutional marketing, and broader societal

narratives (Posselt & Grodsky, 2017). These influences play a significant role in shaping decision­making, often leaving students with preconceived expectations before they even begin exploring concrete options. Because attitudes are shaped by the way information is presented, the concept of framing is particularly relevant. The framing effect, introduced by Kahneman and Tversky (1981), refers to the cognitive bias in which individuals respond differently to the same information depending on its presentation. Positive framing, which emphasizes benefits, opportunities, successes, and gains, tends to elicit favorable responses, whereas negative framing, which highlights risks, failures, obstacles, and losses, often produces reluctance or aversion (Tversky & Kahneman, 1986). This bias has been observed across many domains, including health communication, risk perception, financial decision­making, and consumer behavior. For instance, patients are more likely to choose treatment when survival rates are emphasized rather than mortality rates (Levin et al., 1998), and consumers are more inclined to buy a product when it is marketed as having a “90% success rate” rather than a “10% failure rate” (Shah et al., 2016). Framing effects can influence real­world choices even when the manipulation is minimal, underscoring the importance of applying caution (or using framing strategically) when presenting information about higher education (Espinosa & Gardeazabal, 2023).

In educational settings, framing has been shown to shape student motivation, risk assessment, and confidence in decision­making (Yeager & Dweck, 2012). Recent research has extended this relevance further. For example, Suroyo and Putra (2022) found that both

high­ and low­performing students in Indonesia were influenced by framing biases when selecting a university major, emphasizing the importance of balanced communication in academic decision­making. Similarly, research has demonstrated that the way educators and institutions frame learning experiences, career prospects, and challenges can influence students’ willingness to take risks, apply to competitive programs, and persist through adversity (Blackwell et al., 2007). Baker and colleagues (2017) reported that students engaged more deeply when small­group discussions were structured with autonomy support, clear expectations, and positive classroom norms. More recently, Santana­Monagas and colleagues (2024) showed that framing interacts with teaching style: positively framed messages were most effective when delivered in motivating, autonomysupportive contexts, enhancing students’ academic self­efficacy, emotional responses, and performance. Together, these findings highlight that both the content of the message and the messenger play a critical role in influencing student engagement and confidence.

Framing also affects students’ broader academic behaviors. Guo and colleagues (2024), in a large­scale study of over 2,500 college students, found that positively framed behavioral nudges increased willingness to participate in online learning, with effects spreading socially as influenced students encouraged peers to engage as well. Such findings illustrate how framing not only shapes individual decision­making but can also diffuse through peer networks. In the context of graduate education, this suggests that consistent exposure to messages emphasizing rigorous workload, financial burdens, and competitiveness could lead to more negative attitudes, even for students who might benefit from pursuing advanced degrees. Conversely, consistent exposure to success stories and long­term professional advantages could foster overly optimistic perceptions that downplay the challenges involved. Therefore, understanding framing effects is essential for ensuring that students make well­informed and realistic decisions about their academic and professional futures.

The present study examined how different framing conditions (i.e., positive, neutral, and negative) affect undergraduate students’ perceptions of graduate school. We hypothesized that students exposed to positively framed information would report stronger intentions to pursue graduate education, whereas those exposed to negative framing would show greater hesitancy or aversion. The neutral framing condition served as a baseline to assess students’ inherent attitudes. By exploring how framing influences attitudes, we aimed to contribute to a deeper understanding of how students process educational and career information. Ultimately,

the findings may inform best practices in advising, mentorship, and institutional outreach, offering valuable insights for faculty, career advisors, and graduate programs seeking to communicate expectations in ways that are both motivating and realistic.

Method

Participants

For this study, we implemented a between ­ subjects design and recruited 92 college students from the University of North Georgia via the university’s research participant pool. We used G*Power version 3.1.9.4 (Faul et al., 2007) to confirm that the sample size was adequate for detecting the expected effect. The power analysis results indicated that this sample size was necessary to achieve a minimum of 80% power for detecting a medium effect with α = .05. Participants were at least 18 years old at the time of participation. No other demographic data was collected (e.g., gender, race, ethnicity). Informed consent was obtained prior to participation. Participants were compensated with partial course credit following study completion. This study adhered to all ethical guidelines for human subject’s research and was approved by the University’s Institutional Review Board prior to study onset.

Measures

This study was conducted via Qualtrics, in which participants were randomly assigned to one of three conditions: “Positive” (n = 30), “Neutral” (n = 32), or “Negative” (n = 30). All groups were exposed to a series of 10 framed statements about the graduate school experience, aligned with their assigned framework. These statements were centered on specialization, research, class sizes, thesis or dissertations, networking, assistantships and tuition, time commitment, competitive admissions and autonomy.1 True/false questions were also created to ensure comprehension and retention of these statements.2 A neutrally framed final question was also developed to assess participants’ opinions on graduate school after reading the statements. Specifically, participants were asked, “To what extent do you agree that graduate school provides value for the time and effort invested?” Responses were recorded using a 6­point scale from 1(strongly disagree) 6 (strongly agree). To minimize bias and ensure consistency across conditions, all statements and the final question were created with assistance from ChatGPT (OpenAI, 2024).

1A complete appendix of positive, neutral, and negative framed items can be accessed at OSF: https://osf.io/q723d

2Performance on the true/false questions indicated that all participants responded accurately, confirming adequate attention throughout the task.

Procedures

Participants were recruited through Sona/NERD, the university’s research participation system, and completed the study online asynchronously after providing informed consent via Qualtrics and confirming they were at least 18 years of age. Following consent, they were randomly assigned to one of three framing conditions: Positive, Neutral, or Negative. Then participants were instructed to read a series of framed statements about the graduate school experience, which were presented one at a time. They had unlimited time to read each item before clicking a button to advance, and after every three statements, they answered true/false questions to ensure attention and comprehension. Once all 10 statements were presented, participants rated their overall attitude toward graduate school on a 6­point scale. The study required less than 15 minutes for all participants to complete. Upon finishing, participants were debriefed with a summary of the study’s purpose, the hypotheses being tested, and the anticipated outcomes. They were thanked for their participation, informed of the study’s contribution to understanding attitudes toward graduate school, and provided with researcher contact information should they have any questions.

Results

We hypothesized that participant attitudes would be impacted by framing, specifically with positive and negative framing making attitudes more or less favorable, respectively. A one ­ way ANOVA was used as it provides an appropriate test of mean differences across multiple groups while controlling Type I error. The analysis revealed a significant effect of framing, F(2, 89) = 5.53 p = .005, ηp² = 0.11. Given this overall effect, pairwise comparisons were conducted to clarify which groups differed. These follow­up tests found significant differences between positive framing (M = 4.63, SD = 1.10) and negative framing (M = 4.03, SD = 1.25), as well as between negative framing and neutral framing (M = 4.91, SD = 0.77; ps < .03). However, no difference was found between positive framing and neutral framing (p = .13; see Figure 1).

Discussion

This study examined how framing influences undergraduate students’ attitudes toward graduate school. Consistent with our hypothesis, negative framing significantly reduced positive perceptions of graduate education, whereas positive framing did not significantly enhance attitudes compared to a neutral condition. Importantly, however, even under negative framing, student attitudes remained somewhat favorable. The mean score of 4.03, while lower than the neutral or

positive conditions, is still above the midpoint on a 6­point scale and does not reflect negative attitudes. Rather, it suggests that strong positive attitudes observed in the neutral and positive conditions were dampened somewhat in the negative condition.

These findings indicate that students may inherently hold favorable perceptions of graduate education, such that additional positive reinforcement may be unnecessary, whereas susceptibility to negative messaging could reduce just how favorable graduate school seems. Indeed, the neutral condition produced near­ceiling attitudes, suggesting that undergraduates already believe graduate education is valuable. These perceptions are likely rooted in widely held associations between graduate education and career advancement, intellectual development, and increased earning potential (Perna, 2004; Posselt & Grodsky, 2017).

Although negatively framed information decreased favorability, it did not deter students from recognizing the benefits of graduate school. For example, students exposed to messages emphasizing financial burden, competitive admissions, or time commitments rated the value of graduate education lower than their peers; yet they still maintained somewhat positive attitudes. Other individual difference factors (e.g., self­efficacy, locus of control, or mental health) may also influence susceptibility to framing and should be examined in future research.

The reduction observed under negative framing aligns with prior work on negativity bias, which suggests that negative information carries greater psychological weight than positive information (Baumeister et al., 2001; Rozin & Royzman, 2001). Individuals tend to

Note. Attitude ratings on a scale from 1–6, where 1 represents negative attitudes toward graduate school and 6 represents positive attitudes toward graduate school, are shown here for each framing condition: positive, neutral, and negative. Error bars represent standard error.

remember and give greater influence to negative experiences or messages, making them more susceptible to discouraging information when making significant life decisions.

Interestingly, there was no statistically significant difference between the neutral and positive conditions. This suggests that reinforcing already favorable attitudes with additional positive messaging may be redundant or ineffective. One possible explanation is that students with inherently positive views of graduate education may be more influenced by external structural factors, such as financial feasibility, program accessibility, or career goals, than by internal appeals to optimism (Hearn & Rosinger, 2014).

Prior research has also highlighted demographic influences on framing susceptibility. For instance, Fan (2017) found that female students in China were more susceptible to framing effects in decisionmaking scenarios modeled on the original Tversky and Kahneman (1981) paradigm. This finding underscores the importance of considering individual differences, such as gender, in framing research. Our study aligned with classic work on framing (Tversky & Kahneman, 1986) but extended it by applying the paradigm to higher education, a context that has received limited empirical attention.

These findings carry several implications for higher education professionals and institutions. First, the way institutions, faculty, and advisors communicate about graduate school can significantly shape student attitudes. Overemphasis on challenges such as debt, stress, or limited job prospects may unintentionally discourage applications if not balanced with equally important information such as available resources, funding opportunities, and career benefits.

Second, the results suggest that neutral, fact­based messaging may be sufficient to maintain students’ naturally positive views. Rather than relying on overtly optimistic rhetoric, institutions may benefit from providing accurate, balanced, and transparent information. This approach can foster informed decision­making without creating skepticism or disengagement. In more specialized contexts, however, positive framing can still be useful. For example, Jeno and colleagues (2020) found that positively framed intrinsic goals delivered via a need­supportive mobile app increased undergraduate students’ effort, achievement, and engagement in science learning. These findings highlight the potential of technology­enhanced framing strategies to strengthen motivation. Future research could explore how digital tools might frame information about graduate education in ways that promote realistic but positive engagement.

Finally, these findings are especially relevant for

first­generation and underrepresented students, who often face additional barriers to graduate education (Posselt, 2016). If these students are disproportionately exposed to negative or discouraging messages, whether in advising or informal settings, existing disparities in graduate school access may worsen. Providing balanced, supportive guidance tailored to these students’ needs is therefore essential. Future research should also investigate whether framing effects vary across demographic groups, including socioeconomic status and first­generation college status.

Limitations and Future Directions

Although this study offers novel insights into how framing affects undergraduate students’ attitudes toward graduate school, several limitations must be acknowledged.

Demographics

The most substantial limitation is the lack of demographic information beyond the participants’ status as college students. Without details on age, gender, race/ ethnicity, socioeconomic status, year in school, or firstgeneration college status, our ability to interpret the findings or generalize them across student populations is constrained. This limitation is particularly critical given that graduate school decision­making is often shaped by identity­based experiences and structural inequities. For example, first­generation students or those from underrepresented backgrounds may interpret messages about graduate school differently than continuing­generation students or those with greater financial resources. Thus, although our results provide an important proof of concept for the role of framing, we cannot meaningfully speak to how these effects vary across demographic subgroups. Future studies should incorporate a comprehensive set of demographic measures such as the list suggested above. Beyond descriptive reporting, these measures could also serve as individual difference variables, allowing researchers to test how framing interacts with participant variables.

Our rationale for not collecting these data aligns with trends in cognitive psychology research (see Roberts et al., 2020), where demographic reporting is often limited unless directly relevant to the hypothesis. However, because our study specifically framed the topic of graduate school, demographic factors are clearly relevant and should have been included. Additionally, we acknowledge that demographic data is valuable in all research as it aids in transparency and generalizability. Therefore, we view this omission as an oversight in study design and will prioritize demographic data collection in future research.

Measurement Validity

The present study did not use an established validated scale to measure attitudes toward graduate school, and we did not report reliability statistics. Our design emphasized exposure to framing statements followed by a single global evaluation of graduate school. Although this allowed us to capture the immediate framing effect, reliance on a single ­ item measure limits reliability. Future work should incorporate validated, multi­item scales that would permit assessment of internal consistency and enhance measurement validity. Additionally, our neutrally framed global attitude item might have introduced acquiescence bias, as agreement ­ based wording can encourage endorsement regardless of condition. Although intended as a neutral baseline, its structure might have biased responses. Future studies should use both positively and negatively worded items or alternative formats to reduce this bias.

Sampling

Another limitation concerns sample size. Although our power analysis indicated sufficient power to detect a medium effect, our total sample of 92 students represents a relatively modest cohort. Larger samples will be necessary to more precisely estimate effect sizes, increase confidence in the robustness of results, and test for possible interactions between framing and demographic factors. Expanding sample size will also strengthen the generalizability of findings beyond the particular institutional context of the present study.

Design

The present study relied exclusively on quantitative methods within a controlled experimental setting. Although this allowed for precise testing of the framing effect, it may not fully capture the complexity of how students form attitudes about graduate school. Future research should utilize a mixed­methods approach to provide more nuanced and holistic understanding of student attitudes. Focus groups, interviews, or openended survey items could illuminate how students perceive and interpret messages about graduate school, offering qualitative context that could complement our quantitative findings.

Ecological Validity

A further limitation is the study’s use of preconstructed, brief statements presented in isolation via an online platform. In real­world settings, students encounter information about graduate school through a variety of channels including advising conversations, peer interactions, institutional messaging, and social media over extended periods of time. Our findings therefore capture only an immediate, one­time effect of framing.

Future studies should examine whether framing effects persist across repeated exposures, whether they vary depending on the source of information, and whether they ultimately influence students’ behaviors, such as seeking advising, applying to graduate school, or enrolling in advanced programs.

Conclusion

Altogether, our study reveals that framing significantly affects undergraduate students’ attitudes toward graduate school. Specifically, negative framing diminishes enthusiasm while positive framing does not significantly enhance attitudes compared to a neutral baseline. These findings emphasize the importance of how information about graduate school is presented and communicated to students. It is essential for educators, policymakers, and academic advisors to consider the potential impact of negative framing and accordingly strive to provide students with a balanced and comprehensive view of graduate education. By doing so, we can better support students in making informed, confident, and realistic decisions regarding their academic futures and careers.

References

Baker, A. R., Lin, T.-J., Chen, J., Paul, N., Anderson, R. C., & Nguyen-Jahiel, K. (2017). Effects of teacher framing on student engagement during collaborative reasoning discussions. Contemporary Educational Psychology, 51, 253–266. https://doi.org/10.1016/j.cedpsych.2017.08.007

Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001). Bad is stronger than good. Review of General Psychology, 5(4), 323–370. https://doi.org/10.1037//1089-2680.5.4.323

Blackwell, L. S., Trzesniewski, K. H., & Dweck, C. S. (2007). Implicit theories of intelligence predict achievement across an adolescent transition: A longitudinal study and an intervention. Child Development, 78(1), 246–263. https://doi.org/10.1111/j.1467-8624.2007.00995.x

Carnevale, A. P., Strohl, J., & Melton, M. (2011). What’s it worth? The economic value of college majors. Georgetown University Center on Education and the Workforce. https://cew.georgetown.edu/wp-content/uploads/2014/11/whatsitworthcomplete.pdf

Espinosa, M. P., & Gardeazabal, J. (2023). Framing effects on risk-taking behavior: Evidence from a field experiment in multiple-choice tests. Experimental Economics, 26(2), 535–567. https://doi.org/10.1007/s10683-023-09797-7

Fan, W. (2017). Education and decision-making: An experimental study on the framing effect in China. Frontiers in Psychology, 8, 744. https://doi.org/10.3389/fpsyg.2017.00744

Faul, F., Erdfelder, E., Lang, A.-G., & Buchner, A. (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39, 175–191.  https://doi.org/10.3758/bf03193146

Guo, X., Li, R., Ren, Z., & Zhu, X. (2024). Examining the effect of nudging on college students’ behavioral engagement and willingness to participate in online courses. Journal of Health Psychology, 30(10), 2708–2718. https://doi.org/10.1177/13591053241281588

Hearn, J. C., & Rosinger, K. O. (2014). Socioeconomic diversity in selective colleges: Implications for student and institutional outcomes. In M. Bastedo & P. G. Altbach (Eds.), The state of college access and completion: Improving college success for students from underrepresented groups (pp. 71–86). Routledge. Jeno, L. M., Dettweiler, U., & Grytnes, J. A. (2020). The effects of a goal-framing and need-supportive app on undergraduates’ intentions, effort, and achievement in mobile science learning. Computers & Education, 159, 1–14. https://doi.org/10.1016/j.compedu.2020.104022

Kahneman, D., & Tversky, A. (1981). The framing of decisions and the psychology

of choice. Science, 211(4481), 453–458. https://doi.org/10.1007/978-1-4613-2391-4_2

Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not created equal: A typology and critical analysis of framing effects. Organizational Behavior and Human Decision Processes, 76(2), 149–188. https://doi.org/10.1006/obhd.1998.2804

Mullen, A. L., Goyette, K. A., & Soares, J. A. (2003). Who goes to graduate school? Social and academic correlates of educational continuation after college. Sociology of Education, 76(2), 143–169. https://doi.org/10.2307/3090274

OpenAI. (2024). ChatGPT. [Large language model]. https://openai.com/chatgpt

Perna, L. W. (2004). Understanding the decision to enroll in graduate school: Sex and racial/ethnic group differences. The Journal of Higher Education, 75(5), 487–527. https://doi.org/10.1080/00221546.2004.11772335

Posselt, J. R. (2016). Inside graduate admissions: Merit, diversity, and faculty gatekeeping. Harvard University Press. https://doi.org/10.4159/9780674915640

Posselt, J. R., & Grodsky, E. (2017). Graduate education and social stratification. Annual Review of Sociology, 43, 353–378. https://doi.org/10.1146/annurev-soc-081715-074324

Roberts, S. O., Bareket-Shavit, C., Dollins, F. A., Goldie, P. D., & Mortenson, E. (2020). Racial inequality in psychological research: Trends of the past and recommendations for the future. Perspectives on Psychological Science, 15(6), 1295–1309. https://doi.org/10.1177/1745691620927709

Rozin, P., & Royzman, E. B. (2001). Negativity bias, negativity dominance, and contagion. Personality and Social Psychology Review, 5(4), 296–320. https://doi.org/10.1207/s15327957pspr0504_2

Santana-Monagas, E., da Costa Ferreira, P., Simao, A. M. V., & Nunez, J. L. (2024). How (de) motivating teaching styles shape message framing outcomes on students’ self-efficacy, emotions, and grades. Learning and Individual Differences, 110, 102420. https://doi.org/10.1016/j.lindif.2024.102420

Shah, A. K., Shafir, E., & Mullainathan, S. (2016). Scarcity frames value. Psychological Science, 26(4), 402–412. https://doi.org/10.1177/0956797614563958

Suroyo, S., & Putra, B. M. (2022). The impact of framing effect: How framing effect affects students in choosing university’s major. Jurnal Analisa Sosiologi, 11(2), 167–191. https://doi.org/10.20961/jas.v11i2.60238

Tversky, A., & Kahneman, D. (1986). Rational choice and the framing of decisions. Journal of Business, 59(4), S251–S278. https://doi.org/10.1017/cbo9780511803475.013

Yeager, D. S., & Dweck, C. S. (2012). Mindsets that promote resilience: When students believe that personal characteristics can be developed. Educational Psychologist, 47(4), 302–314. https://doi.org/10.1080/00461520.2012.722805

Author Note.

Ralph G. Hale  https://orcid.org/0000­0001­5026­8417 We have no known conflict of interest to disclose.  Hannah V. Hyman played a lead role in conceptualization, research design, data collection, data analysis and interpretation, and substantive original writing. Iris R. Wright played a lead role in conceptualization, research design, data collection, data analysis and interpretation, and substantive original writing. Ralph G. Hale played a lead role in conceptualization, research design, and supervision and a supporting role in data collection, data analysis and interpretation, writing, and editorial assistance. Correspondence concerning this article should be addressed to Ralph G. Hale, University of North Georgia, 3820 Mundy Mill Rd, Oakwood, GA 30566. Email: ralph.hale@ung.edu

Psi Chi Journal of Psychological Research: 2025 Reviewers

We sincerely appreciate the hard work on the part of the following individuals who each completed at least one review in 2025. Without the assistance of such dedicated professionals, Psi Chi Journal would not be able to function.—Rob Wright (Interim Editor)

Julio I. Agbanyim Grand Canyon University

Lara K. Ault Saint Leo University

Brittany Avila University of Nevada, Reno

Karl Bailey Andrews University

Mark E. Basham Regis University

Michelle Beddow University of Michigan–Dearborn

Chris Bjornsen Longwood University

Britani Black Marshall University

Olivia Bomester Helderberg College of Higher Education

Sandra A. Bosch Northcentral University

Karen Brakke Spelman College

Sheila Brownlow Catawba College

Candice Burkett University of Indianapolis

Lorinda B. Camparo Whittier College

Mary Jo Carnot Chadron State College

Kevin R. Carriere Washington and Jefferson College

Bradley J. Caskey Birmingham-Southern College

Shawn R. Charlton University of Central Arkansas

Patrick J. Cooper Lynn University

Sarah Cronin

Bemidji State University

Teddi S. Deka

Missouri Western State University

Jacinta Dickens University of Pittsburgh

Urska Dobersek University of Southern Indiana

Martin J. Downing Lehman College

Gregory S. Drury

Stephen F. Austin State University

Jennifer Dyer-Seymour California State University, Monterey Bay

Ryan C. Ebersole Albany Medical Center

Jorie H. Edwards

Southwestern Oklahoma State University

Carlos A. Escoto

Eastern Connecticut State University

Warren Fass

University of Pittsburgh, Bradford

Frank Ferraro

Midland University

Tifani Fletcher

West Liberty University

April Fugett Marshall University

Azenett A. Garza Caballero

Weber State University

Elizabeth Gennari

Florida Southern College

James Gigliello

St Joseph’s University

Rebecca Gilbertson

University of Minnesota Duluth

Jennifer M. Gonder Farmingdale State College

Sangeeta Gupta Hood College

Heather Haas University of Montana Western

Ralph Hale University of North Georgia

Jonathan J. Hammersley

Western Illinois University

Thomas F. Harlow

University of Maryland Global Campus

Karen Y. Holmes

Norfolk State University

Marya T. Howell-Carter Farmingdale State College

Robert Hymes

University of Michigan–Dearborn

Robert C. Intrieri

Western Illinois University

Tammy J. Jechura

Albion College

Mary E. Jenson

South Dakota State University

Maria Jorgensen Farmingdale State College

Allen H. Keniston University of Wisconsin—Eau Claire

Laura B. Koenig Kutztown University

Samuel Kohlenberg University of Denver

Lisa M. Korenman

United States Military Academy

David S. Kreiner University of Central Missouri

Michael Langlais

Baylor University

Antonio V. Laverghetta

Saint Leo University

Timothy J. Legg

Legg Counseling & Consulting Services, LLC

Lisa K. Lockhart

University of the Incarnate Word

Amy Maslowski University of Minnesota Duluth

Chandra D. Mason

Mary Baldwin University

Alexander Matthews

Metropolitan State University of Denver

Tara Mitchell

Lock Haven University

Betsy L. Morgan

University of Wisconsin–La Crosse

Victoria A. Muller Ewald

St. Mary’s College of Maryland

Walter Murphy

Texas A&M University–Central Texas

Zoann Murphy National University

Beverly J. Myers Stillman College

Eyad J. Naseralla

Buffalo State University

Marisol Parra-Tatge University of Central Florida

Aleksandria Perez Grabow

California State University

San Marcos

Marilyn Petro

Nebraska Wesleyan University

(continued on next page)

Psi Chi Journal of Psychological Research: 2025 Reviewers (continued)

Julee S. Poole

Purdue University Global

Amy Posey Benedictine College

Dee Posey

Washington State University

Mark G. Rivardo Saint Vincent College

Alan Roberts

Indiana University

Lisa Rosen Texas Woman’s University

Raylene Ross University of South Carolina

Marcia J. Rossi

Alabama State University

Christopher A. Sanders Old Dominion University

Michelle Santiago Moravian University

Donald Saucier

Kansas State University

Yassir Semmar Empire State University

Christina S. Sinisi

Charleston Southern University

KatieAnn R. Skogsberg Centre College

Kathryn Sperry Weber State University

Charles M. Stapleton University of Indianapolis

Crystal N. Steltenpohl Center for Open Science Programs

David B. Strohmetz

University of West Florida

Natashia Swalve

Grand Valley State University

Jill M. Swirsky

Holy Family University

Vishal Thakkar

University of Texas Southwestern Medical Center

Kelli R. Thompson Texas A&M University–Kingsville

Leonell Torres-Pagan

Pace University

Maria A. Turkson Penn State Harrisburg

Mary Utley Drury University

Scott VanderStoep Hope College

Karen L. Vittengl

Truman State University

Jordan Wagge Avila University

Olivia Williams Eastern Kentucky University

Caitlin Wong Wantagh UFSD

William D. Woody University of Northern Colorado

Jill Yamashita

California State University, Monterey Bay

Maaly Younis The American University in Cairo

Evan L. Zucker Loyola University New Orleans

CORRECTION

Correction to Abrams et al. (2025)

Regarding the article, “Perceptions of Sex Work: What Drives Opposition?” by Tiffany R. Abrams, Lauren M. Banicki, and Amanda G. Pirlott (Psi Chi Journal of Psychological Research, 2025, Vol. 30, No. 4, pp. 391–402. https://doi.org/10.24839/23257342.JN30.4.391),the authors have corrected terminology throughout the manuscript, including in the article’s title and abstract. Specifically, the term “sex work” has been replaced with “prostitution” to accurately reflect the original materials used in the study. The corrected version of the article, “Perceptions of Prostitution: What Drives Opposition?” is now available on the Psi Chi Journal website and should be used in all citations and references moving forward. https://doi.org/10.24839/2325-7342.JN2026.006

Publish Your Research in Psi Chi Journal

Undergraduate, graduate, and faculty submissions are welcome year round. Only one author (either first author or coauthor) is required to be a Psi Chi member. All submissions are free. Reasons to submit include

• a unique, doctoral­level, peer­review process

• indexing in PsycInfo, EBSCO, Web of Science's Emerging Sources Citation Index, and Crossref databases

• free access of all articles at psichi.org

• our efficient online submissions portal

View Submission Guidelines and submit your research at https://www.psichi.org/page/JN_Submissions

Become a Journal Reviewer

Doctoral­level faculty in psychology and related fields who are passionate about educating others on conducting and reporting quality empirical research are invited become reviewers for Psi Chi Journal. Our editorial team is uniquely dedicated to mentorship and promoting professional development of our authors—Please join us!

To become a reviewer, visit https://www.psichi.org/page/JN_BecomeAReviewer

Resources for Student Research

Looking for solid examples of student manuscripts and educational editorials about conducting psychological research? Download as many free articles to share in your classrooms as you would like.

Search past issues, or articles by subject area or author at https://www.psichi.org/journal_past

Learn About Psi Chi

Psi Chi is the International Honor Society in Psychology. Membership is primarily open to undergraduates, graduate students, transfer students, full­time and part­time faculty members, and alumni.

See membership benefits and a link to apply at https://www.psichi.org/page/member_benefits

Register an account: https://pcj.msubmit.net/cgi-bin/main.plex