Bias in cross-sectional analyses of longitudinal mediation.

Abstract

Most empirical tests of mediation utilize cross-sectional data despite the fact that mediation consists of causal processes that unfold over time. The authors considered the possibility that longitudinal mediation might occur under either of two different models of change: (a) an autoregressive model or (b) a random effects model. For both models, the authors demonstrated that cross-sectional approaches to mediation typically generate substantially biased estimates of longitudinal parameters even under the ideal conditions when mediation is complete. In longitudinal models where variable M completely mediates the effect of X on Y, cross-sectional estimates of the direct effect of X on Y, the indirect effect of X on Y through M, and the proportion of the total effect mediated by M are often highly misleading.

Temperament, Family Environment and Anxiety in Preschool Children

“…For example, temperament might lead to more overinvolved parenting which might, in turn, lead to increased anxiety. Examination of mediation necessitates longitudinal research, preferably with three time points (Maxwell & Cole, 2007).…”

Self-Defining Memories of Body Image Shame and Binge Eating in Men and Women: Body Image Shame and Self-Criticism in Adulthood as Mediating Mechanisms

“…The crosssectional design of our study hinders the possibility of determining temporal causality between the examined variables (Maxwell & Cole, 2007). In fact, our study aim was to examine the plausibility and significance of a hypothesized theoretical model testing the structural links among our variables of interest.…”

Cognitive therapy and eye movement desensitization and reprocessing for reducing psychopathology among disaster-bereaved individuals: study protocol for a randomized controlled trial

“…Therefore, we are only able to compare the treatment condition to the waiting list controls at the one week post-treatment assessment and not at the 12 weeks and 24 weeks follow-up assessments. Accordingly, the potential working mechanisms of change and the outcome variables will be measured concurrently, we will therefore be unable to examine mediation effects (Maxwell & Cole, 2007). Lastly, we will use an outreach method to recruit participants, which may limit the generalizability of future results to treatment-seeking bereaved people in general.…”