A cautionary note on interpreting research findings in the presence of statistical suppression.

Abstract

In regression analyses predictor variables can suppress the effects of other predictor variables, sometimes even resulting in "flipped" relations relative to their zero-order relations (i.e. negative suppression). Drawing on research examining the relations between religion and prejudice, and between ideology and desiring "tall poppies" (successful people) to fall, we highlight examples where researchers appear to have made inappropriate or confusing interpretations of their findings. We compare these examples to a best practice illustration involving associations between psychopathy and counter-productive work behavior. Finally, we provide practical guidelines for thinking about suppression effects in research programmes.