Fahrmeir, Ludwig; Gieger, Christian; Heumann, Christian
An application of isotonic longitudinal marginal regression to monitoring the healing process.
Collaborative Research Center 386, Discussion Paper 89
This paper discusses marginal regression for repeated ordinal measurements that are isotonic over time. Such data are often observed in longitudinal studies on healing processes where, due to recovery, the status of patients only improves or stays the same. We show how this prior information can be used to construct appropriate and parsimoniously parametrized marginal models. As a second aspect, we also incorporate nonparametric fitting of covariate effects via a penalized quasi-likelihood or GEE approach. We illustrate our methods by an application to injuries from sporting activities.