An Alternative Framework for Nonexperimental Cross-Sectional Mediation Studies: Associational Variable Analysis

Abstract

Despite strong evidence of its logical and empirical problems, there remains a veritable eruption of mediational studies using nonexperimental cross-sectional data (NECSD). At best, such studies can say little about true mediation and at worse can create an inaccurate picture of the conclusions from the data. This article overviews the reasons to not frame NECSD studies as mediational and articulates reasons why it may continue despite solid reasons to stop. Associational variable analysis is introduced as an alternative framework to help articulate a more accurate depiction of findings. Articulation of the steps, examples of conducting the steps, and framing findings are provided from an empirical data set. Associational variable analysis is an alternative conceptual framework drawing on the analytic techniques developed for mediation but provides a more accurate framework for the articulation of findings from NECSD studies with mediational-type goals. Translational Abstract Mediators are variables that serve as a putative mechanism, cause, or theoretical process linking independent and dependent variables. Mediation studies using nonexperimental cross-sectional data do not allow causal or directional hypotheses. Framing such studies as “mediation” can create an inaccurate picture of the conclusions from the data. This article overviews reasons to not frame studies using nonexperimental crosssectional data as testing mediation. An alternative framework termed “associational variable analysis” is introduced to help articulate a more accurate depiction of findings from studies using nonexperimental cross-sectional data with mediational-type goals. Articulation of the steps, examples of conducing the steps, and framing findings are provided from an empirical data set. Associational variable analysis is an alternative conceptual framework drawing on the analytic techniques developed for mediation but provides a more accurate articulation of findings.