Regression Analysis for Categorical Moderators
Have you ever wondered if birth order effects vary across ethnic groups? Whether a particular clinical intervention is likely to yield dissimilar outcomes for men and women? Or if the effectiveness of a sales promotion is dependent on a market segment? Questions like these can be effectively answered by using a statistical tool known as moderated multiple regression (MMR). This book provides practical guidance for using MMR to better assess whether the relationship between two quantitative variables is moderated by group membership. Included are discussions and fully worked-out examples of how to conduct and interpret MMR analysis, as well as descriptions of computer programs that allow investigators to check whether their MMR test for moderation can be trusted. Using examples from a variety of different disciplines--from psychology and education to management and political science--Herman Aguinis first shows readers how to distinguish between moderated and mediated relationships. Next, he presents a description of the MMR procedure and its basic statistical assumptions, explains how to conduct an MMR analysis using computer packages, and demonstrates how to interpret the resulting output.
What people are saying - Write a review
We haven't found any reviews in the usual places.
What Is a Moderator Variable
Importance of A Priori Rationale in Investigating Moderating
Performing and Interpreting Moderated Multiple
The Homogeneity of Error Variance Assumption
MMRs LowPower Problem
Computing Statistical Power
Complex MMR Models
Further Issues in the Interpretation
The Signed Coefficient Rule for Interpreting Moderating
Summary and Conclusions
APPENDIX A Computation of Bartletts 1937 M Statistic
APPENDIX E TheoryBased Power Approximation
affect power African Americans Aguinis assess binary moderator Bobko categorical moderator centered cluding coding scheme coefficient associated Cohen computer programs conclusions conducted data set DeShon difference dummy coding effect coding error rates error variance assumption estimate ethnicity example F test fect Figure first-order effects gender groups homogeneity of error homoscedasticity impact implement interaction effect intercept interpretation Journal Latinos measurement error MMR analysis MMR model MMR test MMR users MMRPWR model including moderating effect moderator variable moderator-based subgroups multiple regression null hypothesis observed effect sizes Perf performance score population power analysis power of MMR Predicted Salary predictor variable procedure product term program MMRPOWER provides Psychology quadratic quantitative predictor regarding regression coefficient salary increase sample size sample sizes science journals shows simulation situations slope social science specific SPSS standard deviation statistical power statistically significant Stone-Romero tenure status tion total sample Type I error untenured faculty variance explained
Page 176 - Conditions under which stronger job performance-job satisfaction relationships may be observed: a closer look at two situational contingencies'.