Quantitative Method to Model the Underreporting of Safety Incidents

Abstract

Conventional statistical and regression analyses of safety incidents ignore the prospect for underreporting. In settings where underreporting is anticipated, the use of conventional models may lead to incorrect conclusions. This note summarizes a regression method that disentangles the incidence and reporting phenomena. The regression produces coefficient estimates for each of the underlying functions and thus should lead to more accurate conclusions regarding safety and reporting phenomena, and thus more efficient policy recommendations and resource allocation.