Chapter 1 There is no exercise for Chapter 1.
Chapter 2 1. Bivariate regression. 2. One-way MANOVA. 3. Path analysis. 4. t test of independent samples. 5. Bivariate correlation. 6. Discriminant analysis or logistic regression. 7. One-way ANCOVA. 8. One-way MANCOVA. 9. Multiple regression. 10. One-way ANOVA. 11. Factorial MANOVA. 12. Logistic regression. 13. Factorial MANCOVA. 14. Factor analysis.
Chapter 3 1. (a) No missing values. (b) No. Frequencies in groups are equivalent. (c) Below median: outliers would be those ≥ 22.5; above median: outliers would be those ≥ 17. (d) Below median: 64; above median: 26 and 57. (e) Distributions display moderate positive skewness and would be transformed by taking the square root. (f) No. Levene’s test is significant, which indicates unequal variance between groups. 2. Taking the inverse produced the most normal distribution. 3. (a) All variables should be transformed: grad94 is moderately positive (SQRT); loinc93 is moderately negative (SQRT[K-X]); read94me is substantially positive (LG10[X]); math94me is substantially positive (LG10[X]). (b) Zero subjects exceeded the chi-square critical value of 18.467. (c) Some variables display slight curvilinear tendencies. (d) Homoscedasticity is questionable because residuals do not cluster around the center; however, there is a nice horizontal spread throughout the plot.
Chapter 4 1. (a) Are there significant mean differences for hours worked per week between those satisfied and dissatisfied with job? Are there significant mean differences for hours worked per week among general 1
happiness categories? Is there significant interaction of hours worked per week between job satisfaction and general happiness? (b) Factors intersect slightly. (c) Most likely because lines differ so much. Not too happy may be significantly different from very happy and pretty happy. (d) Probably not. 2. (a) Factor interaction is not significant [F(2, 891) = 1.682, p = .187]. (b) Main effect for satjob2 [F(1, 891) = .147, p = .701] and happy [F(2, 891) = 1.597, p = .203] are not significant. (c) Student responses may vary depending on the accuracy of the estimates they made in response to Questions 1b, 1c, and 1d. 3. (a) Are there significant mean differences in current salaries between males and females? Are there significant mean differences in current salaries between minority status? Is there significant interaction in current salaries between gender and minority status? (b) Outliers equal to or greater than 26,000 were eliminated based on minority status. Because group distributions of salnow are substantially positively skewed, it was transformed by computing its logarithm. Levene’s test indicates that homogeneity of variances cannot be assumed. Line graph reveals no factor interaction. (c) & (d) Factor interaction is not significant [F(1, 431) = 1.05, p = .306]. Gender significantly affects current salary [F(1, 431) = 92.69, p < .001, partial η2 = .177]. Minority status significantly affects current salary [F(1, 431) = 27.56, p < .001, partial η2 = .060]. Results reveal that gender accounts for 17.7% of variance in current salary.
Chapter 5 1. Does income differ by gender among employees when controlling for differences in hours worked per week? Does income differ by age category when controlling for differences in hours worked per week? Does the relationship between income and gender differ by age category among employees when controlling for differences in hours worked per week? 2. Recode all values in rincom91 that are greater than or equal to 22 as sysmis. 3. (a) Tests for normality indicate nonnormal distributions. Histograms reveal that income distributions for all groups are moderately negatively skewed. The reader may select to transform income by taking its reflection and square root in order to create more normal distributions; however, the authors have selected not to transform it. (b) No. Interaction between the factors and covariate is not significant [F(7, 691) = 1.11, p = .353]. Yes. Tests for homogeneity of regression slopes indicate fulfillment of assumption. (c) Using Levene’s test of equal variances, homogeneity of variance can be assumed [F(7, 696) = 1.43, p = .189]. 4. Line plot reveals slight factor interaction. 5. (a) No. Interaction between agecat4 and sex is not significant [F(3, 695) = .827, p = .479, partial η2 = .004]. (b) Yes. Main effects for age [F(3, 695) = 21.73, p < .001, η2 = .086] and gender [F(1, 695) = 28.09, p < .001, partial η2 = .039] are significant. (c) Yes. The covariate of hrs1 significantly influences the DV [F(1, 695) = 25.71, p < .001, partial η2 = .036]. (d) Although main effects for each factor are significant, effect sizes indicate that each factor accounts for a small percentage of variability in the DV (age: 8.6%, gender: 3.9%). 6. A 2 × 4 analysis of covariance was conducted on income. Independent variables consisted of gender and age category. The covariate was hours worked per week. Initial data screening led to the transformation of income by eliminating all values greater than or equal to 22. After significant adjustment by the covariate 2