The next generation of moderator research in personality psychology.

The concept of a moderator variable has a long and checkered history in personality psychology. In this article I review the methodological and conceptual advances in recent moderator variable research. I then report the results of a number of empirical analyses of three newly proposed idiographic moderator variables: scalability, metatraits, and construct similarity, as well as the ipsatized variance index. The results of these analyses indicate that moderator effect sizes in personality can be expected to correspond to a correlation of about .10. I then consider two recent prescriptions, (a) aggregating trait-specific moderators into general ones, and (b) evaluating joint (e.g., triple interaction) moderator effects. Following these prescriptions did not increase the size of the moderator effects in these data. I conclude that moderator effects in personality will be small, and will thus be most reliably detected in very large samples with predictors that are strongly related to the criteria. Even small moderator effects can be important and interesting in the context of theory disconfirmation. However, moderator variables generally will not serve to transform weak relations among personality variables into strong ones.