This paper develops a statistical modeling method for estimating several indicators for sickness absence by region and subbranch within the Human Health and Social Work Activities branch.
The Netherlands Working Conditions Survey is an annual survey that measures, among many other variables, the amount of sickness leave of employees. In the context of the Arbeidsmarkt Zorg en Welzijn (AZW) program we carry out a feasibility study for estimating sickness absence for a detailed breakdown of the AZW subpopulation of employees in the Human Health and Social Work Activities branch. The breakdown is with respect to a subdivision into 28 regions, a subdivision into 14 AZW subbranches, as well as the combination of both. For this purpose, a small area estimation approach is developed based on different types of unit-level multilevel models that account for all classifications of interest as well as for selectivity of the survey response with respect to the target population.