Estimating the number of serious road injuries per vehicle type in the Netherlands by using multiple imputation of latent classes

Statistics that are published by official agencies are often generated by using population registries, which are likely to contain classification errors and missing values. A method that simultaneously handles classification errors and missing values is multiple imputation of latent classes (MILC). We apply the MILC method to estimate the number of serious road injuries per vehicle type in the Netherlands and to stratify the number of serious road injuries per vehicle type into relevant subgroups by using data from two registries. For this specific application, the MILC method is extended to handle the large number of missing values in the stratification variable ‘region of accident’ and to include more stratification covariates. After applying the extended MILC method, a multiply imputed data set is generated that can be used to create statistical figures in a straightforward manner, and that incorporates uncertainty due to classification errors and missing values in the estimate of the total variance.