Ten Ways to Improve the Use of Statistical Mediation Analysis in the Practice of Child and Adolescent Treatment Research

Abstract

Despite guidelines and repeated calls from the literature, statistical mediation analysis in youth treatment outcome research is rare. Even more concerning is that many studies that have reported mediation analyses do not fulfill basic requirements for mediation analysis, providing inconclusive data and clinical implications. As a result, after more than five decades of research, it is still largely unknown through which processes youth treatment works and what the effective treatment components are. In this article, we present ten ways in which the use of statistical mediation analysis in youth treatment outcome research may be improved. These ten ways are related both to conceptual and methodological issues. In discussing how youth clinical researchers may optimally implement these directions, we argue that studies should employ the strongest research designs possible. In so doing, we describe different levels of a mediation evidence ladder. Studies on each step of the ladder contribute to an understanding of mediation processes, but the strongest evidence for mediation is provided by studies that can be classified at the highest level. With the help of the ladder of mediation evidence, results from youth mediation treatment outcome research can be evaluated on their scientific as well as clinical impact.