Parametrization, Linear Hypotheses, Marginality, and "Types" of Tests in Linear, Generalized Linear, and Similar Statistical Models John Fox Department of Sociology McMaster University Judging, for example, by questions asked repetitively on the r-help mailing list, there is a great deal of confusion surrounding the parametrization of models with linear predictors (e.g., linear models, generalized linear models) and related issues, such as the principle of marginality in the formulation of statistical models and hypothesis tests. This paper seeks to clarify and illuminate these issues with specific reference to the manner in which statistical models are implemented in R.