Assessing moderator effects, main effects, and simple effects without collinearity problems in moderated regression models

It is common to assess moderation effects with moderated regression analysis. If the interaction effect is detected by moderated regression analysis, one may subsequently assess the simple effect of the focal predictor with simple slopes analysis. However, the two traditional analyses may suffer from considerable correlations among predictors. These correlations lead to correlations among the individual effects of predictors and thus may make it difficult to detect the interaction effect, the simple effect, and the main effects of the focal predictor and the moderator with the two traditional analyses. This paper suggests alternative analyses assessing the various effects without the collinearity problem. The alternative analyses provide the statistics for the various effects derived from the confidence-interval estimate for the overall effect size of predictors. In addition, this paper presents a practical guideline for assessing the various effects with the traditional analyses as well as the alternative analyses.