Revisiting Mediation in the Social and Behavioral Sciences

Abstract

The process of mediation is of critical importance to the social and behavioral sciences and to evolutionary social psychology in particular. As with the concept of evolutionary adaptation, however, one can argue that causal mediation is in need of explicit theoretical justification and empirical support. Mainstream evolutionary social psychology proposes, for example, that organisms are “adaptation executers”, and not “fitness maximizers”. The execution of adaptations is triggered by fitness-relevant ecological contingencies at both ultimate and proximate levels of analysis. This logic is essentially equivalent to what methodologists refer to as the process of mediation; the adaptations to be executed (or not, depending upon the prevailing environmental circumstances) causally mediate the effects of the ecological contingencies upon the fitness outcomes. Thus, the process of mediation can be generally conceptualized as a causal chain of events leading to a given outcome or set of outcomes. If a predictor variable operates through an intervening variable to affect a criterion variable, then mediation is said to exist. Nevertheless, it does not appear that some psychologists (particularly evolutionary-social psychologists) are sufficiently well-versed in the fundamental logic and quantitative methodology of establishing causal mediation to support such claims. In the current paper, we set out to review the ways researchers support their use of mediation statements and also propose critical considerations on this front. We start with more conventional methods for testing mediation, discuss variants of the conventional approach, discuss the limitations of such methods as we see them, and end with our preferred mediation approach. DOI:10.2458/azu_jmmss_v04i1_figueredo3

The development of the child unpredictability schema: Regulation through maternal life history trade-offs.

“…Type I sums of squares (sequential sums of squares; SAS Institute, Inc., 2002) were estimated for each predictor variable, and then used to test each predictor for statistical significance. Sequential sums of squares give predictor variables a hierarchical order of preference, whereby the first variable is allowed to “predict” the criterion variable, with each successive variable predicting the residual sums of squares left over from the prior variables and only retained if they increase the model sums of squares by a statistically significant amount (see Cohen & Cohen, 1983; Figueredo, Garcia, Cabeza de Baca, Gable, & Weise, 2013 for a brief explanation of Type I sequential sums of squares). In other words, sequential sums of squares estimate the significance of each predictor variable predicting the criterion, after the prior variable has predicted the criterion variable.…”

<p>Big Five Personality Traits and Gratitude: The Role of Emotional Intelligence</p>

“… 47 Therefore, we assumed that emotional intelligence could be a potential mediator between dimensions of personality according to the Big Five Model, and gratitude (Hypothesis 3). Although there are well-established dilemmas with the implementation of cross-sectional mediation analyses to gather evidence for causal processes, 48 , 49 mediation models are useful with strong prior theoretical justification and empirical support 50 for the relationships between the independent variables, dependent variable, and mediator. In the following section, we have attempted to provide the main rationale that regards the crucial assumption of the temporal precedence in the mediational scheme.…”

Neighborhood Social Cohesion and Mobility Limitations Among Community-dwelling Older Americans: The Mediating Roles of Depressive Symptoms and Mastery

“…Nonetheless, this potential issue can be compensated by our sample size. Moreover, the Sobel test can suggest the presence of an indirect effect but does not provide the magnitude of the indirect effect (Figueredo et al, 2013). We used the product of the coefficients method to calculate the magnitude of the indirect effect.…”