PM effect size estimation for mediation analysis: a cautionary note,alternate strategy,and real data illustration

As research on mediation has grown, so too has interest in identifying ways to assess the size of indirect effects in a mediation analysis. One such estimate – the ratio of the indirect effect to the total effect (P_M) – was tested in a sample of 21,297 children from the Early Childhood Developmental Study. Results showed that the two independent variables – low self-control and externalizing behavior – correlated equally well with a common mediator (mathematical thinking). Moreover, the two pathways shared the same mediator and dependent variable (subsequent externalizing behavior). Despite this, P_M was three times larger in the self-control-initiated pathway than in the externalizing-initiated pathway, based on a stronger correlation between prior and subsequent externalizing behavior than between low self-control and externalizing behavior. The comparison pathways approach, where pathways mediated by mathematical reasoning are contrasted with pathways mediated by a control variable (general knowledge), is offered as an alternative to effect size measurement in mediation research.