Can We Drop a Group of Study Participants From Analyses and Modeling of Behavioral Scale Data? A Note on an Application of the Benjamini-Hochberg Method

This note discusses a procedure for evaluating whether a group of subjects in a larger study could be dropped from analyses of data obtained with a behavioral measuring instrument when pursuing research questions of initial concern. Recent decades have witnessed notable facilitation of data collection processes in educational and psychological research, leading at times to comprehensive data sets containing one or more groups of respondents that may not be relevant to substantive queries regarding a population of interest. The outlined approach helps scholars to avoid potentially seriously biased parameter and reliability estimates, standard errors, and hypothesis testing results in empirical studies, which yield misleading substantive conclusions. The described procedure is readily employed with widely available software and is illustrated using data from a child development questionnaire.