Tuesday, January 8, 2019
Sometimes, hindsight is 20/20
When I was involved in writing this article about three years ago, I did not yet know that my recently published meta-analysis would require a considerable rethink about the short term effects of weapons on aggressive behavior. At least one person who tweeted about the review published in 2016 appeared to have a skeptical take (PlumX was fairly useful), and the commentary from that tweet was about spot on. Narrative reviews have their limitations. Of course, so do meta-analyses. That said, based on recent meta-analytic evidence it appears that some things from 2016 do still seem somewhat relevant: there is still tentative evidence that weapons influence accessibility of aggressive thoughts (even adjusting for publication bias) and primary appraisals of threat (also even when adjusting for publication bias). Where things get challenging is when we examine behavioral outcomes, where much of that work was conducted with small samples several decades ago, and where estimates for publication bias are to some degree all over the place, albeit trending toward negligible average effect size. That finding alone requires me to adopt a skeptical stance. In the meantime, let's see what evidence makes its way public in the next few years. That should help us make some educated decisions about how to view this particular phenomenon. In the meantime, the sort of caution I wish I would have advocated a few years ago is very much in order.