Commentary: on the application of potential outcomes-based methods to questions in social psychiatry and psychiatric epidemiology

The authors of this paper seem somewhat ambivalent about the causal status of the effect estimated. On the one hand, there is a disclaimer that the paper only estimated an association, but on the other they interpret the results of the decomposition model as estimating the effect that was due to race alone, due to SES alone, and due to the intersection of race and SES. This certainly seems like a causal attribution. Indeed, the authors clearly state that the study results are applicable when “disparities are taken to represent causal effects of race and SES, when proper adjustment has been made for confounding variables”. Therefore we are assuming that the interpretation was for an estimate where confounder control was in effect, and the association disclaimer is just a “technical” issue in this particular analysis.