The Effect of Labor Unions on Occupational Safety: Evidence from the U.S. Steel Industry
This study advances industrial relations research by documenting the relationship of union representation and industrial injury. I argue that the conventional analyses of a union injury differential are flawed and introduce new methods to determine a casual effect. Initial OLS regression analyses of the total 2018 population of U.S. iron and steel mills show no association, negative or positive, between union representation and industrial injury. To address covariate imbalance in the data, I use propensity score methods to draw a matched sample of 118 establishments from the total 2018 population of U.S. iron and steel mills. Subsequent multivariate regression analyses reveal that union representation is negatively associated with industrial injury. Specifically, union representation is associated with a reduction of nearly 2 injuries per 100 full time-employees in a given calendar year. The relationship of union representation and industrial injury is moderated by establishment size, such that the effect of union representation declines as establishment size increases. The final predictive model explains 42.4 percent of variance in industrial injury among iron and steel mills.