Title: Exploration of the variability of variable selection based on distances between bootstrap sample results
Authors: Christian Hennig - UCL (United Kingdom) [presenting]
Willi Sauerbrei - Universitaet Freiburg (Germany)
Abstract: It is well known that variable selection in multiple regression can be unstable and that the model uncertainty can be considerable. The model uncertainty can be quantified and explored by bootstrap resampling. We will present approaches that use the results of bootstrap replications of the variable selection process to obtain more detailed information about the data. Analyses will be based on distances between the results of the analyses of different bootstrap samples. The distances are used to map the bootstrap results by mutidimensional scaling and to cluster them. Clusters are of interest because they could point to substantially different interpretations of the data that could arise from different selections of variables supported by different bootstrap samples. These and further issues will be illustrated by some data examples including a study on ozone effects in children.