Title:Selective Inference for Additive and Linear Mixed Models

Abstract:This work addresses the problem of conducting valid inference for additive and linear mixed models after model selection. One possible solution to overcome overconfident inference results after model selection is selective inference, which constitutes a post-selection inference framework, yielding valid inference statements by conditioning on the selection event. We extend recent work on selective inference to the class of additive and linear mixed models for any type of model selection mechanism that can be reapplied to new data in a bootstrap-like manner. We investigate the properties of our proposal in simulation studies and apply the framework to a data set in monetary economics. Due to the generality of our proposed approach, it is particularly suitable for the given application for which the final additive mixed model is selected using a hierarchical selection procedure based on the conditional Akaike information criterion and involves varying data set sizes.