Wood et al. (2020) studied the rollout of a procedural justice training program in the Chicago Police Department and found large and statistically significant impacts on complaints and sustained complaints against police officers and police use of force. This document describes a subtle statistical problem that led the magnitude of those estimates to be inflated. We then re-analyze the data using a methodology that corrects for this problem. The re-analysis provides less strong conclusions about the effectiveness of the training than the original study':' although the point estimates for most outcomes and specifications are negative and of a meaningful magnitude, the confidence intervals typically include zero or very small effects. On the whole, we interpret the data as providing suggestive evidence that procedural justice training reduced the use of force, but no statistically significant evidence for a reduction in complaints or sustained complaints.