On the role of test sequence length, model refinement, and test coverage for reliability