The application of a number of standard statistical tests to the Estep T-model for security selection raises questions about its general efficacy. In particular, the model's results are not uniformly significant; nor are they all significant in the "right" direction. There is also a question about whether the model's components are estimated in the most suitable form.