Suppressor Variables Versus Suppression Effects

• In Event: Topics in Multiple Linear Regression

Abstract

Correlated predictors are commonplace in applied social science research. The extent to which they are correlated will influence the estimates and statistics associated with the other variables they are modeled along with. These effects may include enhanced or diminished regression coefficients for the other variables. The former case implies suppressor effects; the latter implies mediator effects. This paper examines the history, definitions, and design implications and interpretations when variables are tested as suppressors versus when variables are found to exhibit suppression effects. Empirical data from a single study illustrate the different approaches to studying potential suppressors and the interpretations of their results. Implications for teaching about suppressors and suppression effects are discussed.

Authors

• Kelsey Klein, Boston College

• Larry H. Ludlow, Boston College