MODERATORS AND SUBGROUPS1

Abstract
It is argued that an alternative preferable to the use of moderators is subgrouping subjects on the basis of the patterns or profiles of their scores across several dimensions of significance. This means that the variance between subgroups is of primary concern. Evidence is presented to suggest that, as implied, the simple fact of subgroup membership will often significantly enhance both meaning and prediction.

This publication has 4 references indexed in Scilit: