Mediators & Process
Developmental psychologists study the causal process through which disorder develops. The identification of a risk factor does not necessarily imply a causal process, because (1) a risk factor might be a proxy for a causal factor and empirically related to a disorder only because of its correlation with this causal factor (the so-called third-variable problem), (2) a risk factor might occur as an outcome of a process that is related to a disorder rather than the antecedent of the disorder, or (3) a risk factor might play a causal role in a more complex, multivariate process. Therefore, developmental psychologists often attempt to understand the process through which risk factors are related to eventual disorder. The factors that are identified as intervening variables in this process are called mediators, which are defined as variables that account for (or partially account for) the statistical relation between a risk factor and a disorder.
Four empirical steps are required to demonstrate at least partial mediation (Figure 1-2). First, risk factor A must be empirically related to outcome C (ie, there must be a phenomenon to be mediated, called the total effect). Second, A must be related to mediator B. Third, B must be related to C. Finally, in a stepwise regression or structural equation model analysis, when B has been added to the prediction of C, the resulting relation between A and C (called the direct effect) must be reduced significantly from the original bivariate relation. The difference between the total effect and the direct effect is called the indirect effect, and it is the magnitude of the mediation of the effect of A on C by B.
An example (depicted in (Figure 1-2)) is the biased pattern of social information processing that often results from an early history of physical abuse. Early abuse (A) is a known risk factor for the development of conduct disorder (C) (ie, it is statistically correlated with later conduct disorder, but many abused children do not develop this disorder nor do all persons who have conduct disorder show a history of abuse). Early abuse is also correlated with the development of a social-information-processing pattern (B) of hypervigilance to threatening cues, perceiving the social world as a hostile and threatening place, and poor social-problem-solving skills. These mental factors (B) are associated statistically with later conduct disorder (C) and account for about half of the effect of early abuse (A) on later conduct disorder (C).