Sick at work: methodological problems with research on workplace presenteeism

Research on determinants of workplace presenteeism is based on the assumption that presenteeism is the outcome of a decision process with two options: taking sick leave, or attending work despite illness. However, frequency measures of sickness presenteeism are usually employed as the dependent variables in previous research on the determinants of the decision process. Monte Carlo simulations show that the analytical strategies used in previous research are likely to identify determinants of vulnerability, but will underestimate the determinants that affect the decision process under consideration. As a solution to this shortcoming, an alternative analytical strategy is presented, which uses an empirical estimator of the individual presenteeism propensity.