ABSTRACT
European Organization for Research and Treatment of Cancer, Brussels, Belgium
In this chapter, the effect of misspecifying the parametric response model on dose effect estimation, hypothesis testing and safe dose determination for the clustered binary data models is investigated. In Section 11.1 the impact on testing for dose effect under a possibly incorrectly specified model is investigated. A robust alternative to the classical parametric approaches is proposed in Section 11.2. The impact of misspecifying a model on the determination of a safe dose is studied in Section 11.3, while profile score-based alternatives are introduced in Section 11.4.
