文摘
Tests of mediation hypotheses (i.e., X ➔ M ➔ Y) are widespread in social psychology In many cases, "X ➔ Y ➔ M" is a plausible alternative model A common practice is to compare both indirect effects (reverse mediation testing) Monte Carlo simulations show that this approach often leads to wrong conclusions This especially applies when "M" is measured less reliably than "Y"