We explore the in- and out- of sample robustness of tests for choice inconsistencies based on parameter restrictions in parametric models, focusing on tests proposed by Ketcham, Kuminoff and Powers (KKP). We argue that their non-parametric alternatives are inherently conservative with respect to detecting mistakes. We then show that our parametric model is robust to KKP's suggested specification checks, and that comprehensive goodness of fit measures perform better with our model than the expected utility model. Finally, we explore the robustness of our 2011 results to alternative normative assumptions highlighting the role of brand fixed effects and unobservable characteristics.