Controlling nuisance variables by using a matched-groups design

In this article, the authors provide an introduction to the use of a matched-groups design as a strategy for controlling nuisance variables. Building on the insightful comments of Campbell and Kenny and of Meehl, the authors utilized Monte Carlo simulations to highlight three major limitations of this control strategy (i.e., regression toward different means, systematic unmatching, and the generalizability of results) that they believe have received insufficient attention by researchers. Particular attention is given to how the effect of regression toward different means can result in researchers' drawing erroneous conclusions from their data. Recommendations for researchers considering the use of a matched-groups design are provided.