Separation-proof random effects logistic regression - a sketch of some ideas

Event Date: 
Friday, 19 October, 2018 - 13:00
Location: 
IBMI
Lecturer: 
prof. dr. Georg Heinze

In binary response models, it may happen that iterative parameter estimation diverges because binary outcomes are perfectly predictable for some or all subjects. However, in contrast to what data analysis might suggest, we assume that the measured covariates are not sufficient to predict the outcome status perfectly. This phenomenon, termed 'separation', is rather considered a small-sample problem. Firth's correction to logistic regression has become a standard procedure to deal with separation in binary response models with fixed effects. We will first review Firth's and other methods to deal with separation (Mansournia, Geroldinger, Greenland, Heinze, Am J Epi 2018). For random effects models, satisfactory paralleling solutions have been repeatedly called for, but are still missing in the statistical literature. In this presentation we would like to share some of our ideas on how to tackle the problem. We will first define desirable properties of possible solutions. Then, starting at the abovementioned approaches, we will explore some suitable extensions for mixed models. In particular, Bayesian approaches, solutions based on ridge regression and data augmentation could be promising candidates to address the problem. A colonoscopy study comparing purgatives for bowel preparation will serve as an illustrative example.

About IBMI

Institute for Biostatistics and Medical Informatics (IBMI), formerly Institute for BioMedical Informatics (so still IBMI) was founded by the Faculty of Medicine as a result of a need for a unit which would perform, or coordinate, tasks related to data analysis and providing information, relevant for research in medicine. The programme of the institute, and its development, have been adjusting thorugh time to changes in financing and technological progress, but the basic aim remain the same: to support research in medicine. This is achieved through the following tasks:

Contact

Institute for Biostatistics and Medical Informatics
University of Ljubljana, Faculty of Medicine
Vrazov trg 2, 1000 Ljubljana
Slovenia

tel: +386 1 543-77-70
fax: +386 1 543-77-71
email: ibmi (at) mf.uni-lj.si