Never Will I Ever: Multilevel Mixture Cure Modeling

Despite the development and adoption of time-to-event/survival analysis techniques in the research fields of management and organizational behavior, the adoption of more advanced survival techniques is lagging behind. Some subjects might never experience the event-of-interest, however long they are followed-up, which has broadly been neglected in our fields. In our field, this study is the first to introduce the mixture cure model and its multilevel generalization. More specifically, we introduce a multilevel mixture cure model with correlated random effects. Using a large simulation study to test the accuracy of parameter estimates and the reliability of statistical inference clearly demonstrates the analytical and empirical superiority of the multilevel mixture cure model, compared to the traditionally used survival analysis techniques. Therefore, we hope that researchers studying time-to-event data will add the multilevel mixture cure methodology to their methodological toolbox and, thus, ensure our empirical research is evidenced by analytical techniques at the cutting edge of the most rigorous and advanced methodologies.status: Published onlin