Title: Top-down Transformation Choice

Abstract: Simple models are preferred over complex models, but oversimplistic models may lead to errorenous interpretations. The classical approach is to start with a simple model whose shortcomings are assessed in residual-based model diagnostics. Eventually, one increases the complexity of this initial too simple model and obtains a better fitting model. We illustrate how transformation analysis can be used to implement an alternative approach to model choice. Instead of adding complexity to simple models, we demonstrate how step-wise complexity reduction can help to identify simpler and better interpretable models. We aim at modelling body mass index distributions in Switzerland by means of transformation models and try to understand the impact of sex, age, smoking and other lifestyle factors on a persons' body mass index. In this process, we try to find a compromise between model fit and model interpretability. Special emphasis is given to the understanding of the connections between transformation models of increasing complexity. The models used in our analysis range from evergreens, such as the normal linear regression model with constant variance, to novel models with extremely flexible conditional distribution functions, such as transformation trees and transformation forests.