The analysis and identification of homologizer/moderator variables when the moderator is continuous: An illustration with anthropometric data

Human biological data often contain homologizers, that is, variables which moderate the strength rather than the form of the relationship between two other variables. Current methods for the identification and analysis of continuous homologizer variables [Z] (variables which moderate the strength rather than the form of a relationship between other variables [X,Y]) recommend dividing the sample into subgroups on the basis of the homologizer variable (Z) and testing whether correlations (X,Y) are significantly different among subgroups. We propose an alternative strategy which avoids the use of subgroups and which offers greater power to detect homologizer effects. In addition, it allows the identification of homologizer effects in data sets where multiplicative as well as additive terms are included in the model. The proposed strategy is validated through Monte Carlo simulations and an example of its application to a set of anthropometric data is given. © 1992 Wiley‐Liss, Inc.