This post continues the discussion of the Scottish Community Engagement Trial (ScotCET). In four posts (you can find them here, here, here, and here) I discussed the apparent failure of implementation over the summer. I concluded that despite the unexpected direction of the findings (i.e., those who received the procedurally just treatment had more negative opinion of the police), the treatment effect appeared to be consistent, was not heterogeneous, and thus it could be actually attributed to the research design.
However, it is still difficult to pin down a good interpretation of the treatment effect. Because there might have been a certain level of variation in how police officers delivered the treatment, it is probably best to stick with a purely descriptive definition of the treatment, noting that the people in the treatment and control group had systematically different experiences with the police.
Those of you, who read the outline of my PhD might hark back, that one of the major presumptions in the procedural justice literature is that previous experiences with the police shape people’s views regarding the procedural justice of the police, which in turn influence police legitimacy (e.g., normative alignment). This can be depicted the following way:
This figure provides a classic example of mediation where previous experiences with the police has a direct effect on moral alignment, and an additional indirect one through procedural justice. Even though we have only an imperfect understanding of the nature of the direct effect, we can shift the focus of the analysis to the indirect effect, or in other words the effect of procedural justice on normative alignment. Therefore, the current experiment provides an apt test of this core hypothesis of the procedural justice literature. The next post will outline causal mediation analysis as a statistical technique that can be used to address this particular problem.