Testing moderator variable hypotheses: A reply to stone and hollenbeck

E. F. Stone and J. R. Hollenbeck (1984, Organizational Behavior and Human Performance, 34, 195–213) argue strongly in favor of the use of moderated regression analysis as the appropriate technique in testing for the presence of “moderator variables.” The primary thrust of the Stone and Hollenbeck article is to present criticisms of positions on tests for moderator variables taken by H. J. Arnold (1982, Organizational Behavior and Human Performance, 29, 143–174) and by M. R. Blood and G. M. Mullet (1977, Where Have All the Moderators Gone: The Perils of Type II Error, College of Industrial Management, Georgia Institute of Technology). The Stone and Hollenbeck critique of the Blood and Mullet position is well placed, consisting essentially of a restatement (with special reference to Blood and Mullet) of arguments previously put forward by H. J. Arnold and M. G. Evans (1979, Organizational Behavior and Human Performance, 24, 41–59), J. Cohen (1978, Psychological Bulletin, 85, 858–866) and J. Cohen and P. Cohen (1975, Applied multiple regression/correlation analysis for the behavioral sciences, Hillsdale, NJ: Erlbaum). Their critique of Arnold's (1982) position is without foundation and results from a failure to recognize the different types of information carried by correlation coefficients and regression coefficients when moderator variables are being analyzed.