Why do unionized workers have more nonfatal occupational injuries?

Most empirical studies have estimated a positive union-nonunion "injury gap," suggesting that unionized workers are more likely to have a nonfatal occupational injury than their nonunion counterparts. Using individual-level panel data for the first time, I study several explanations for this puzzling result. I find that controlling for time-invariant individual fixed effects already reduces the gap by around 40%. Some of the explanations that I study contribute in reducing this gap even further. I, however, do not find evidence of the gap becoming negative and the impact of unions on nonfatal injuries appears to be insignificant at best.