Measurement error and the effects of unions on the compensating differentials for fatal workplace risks

With essentially the same estimation methods and time period but using different UK data sets and risk measures, Sandy and Elliott (1996) found that nonunion male manual workers had a substantially higher fatal risk premium than union workers while Siebert and Wei (1994) found the opposite. This paper attempts to reconcile these conflicting results using the different measures of fatal risk employed in these two studies together on the same data set. The conflicting results are due to several types of measurement error. We find substantial miscoding in the employee-reported industry affiliation as well as larger errors-in-variables bias in the industry-based risk than in the occupation-based risk.