So a while back, I had a blog on the dynamics of a consultation. Since that time, I have a new perspective on it. There are multiple ways to solve a problem. Statistics has many approaches and methodologies- you only need the justification for using it. So one statistician may perform a Mann-Whitney while another chooses a two sample t-test. Both are admissible…there is nothing wrong with choosing a non-parametric approach. The key is to realizes that some tests are stronger than others. In the world of dissertations, you must be able to defend the approach. Period. If you can defend it to the committee, then you are in a good spot. If the explanation is not clear, do not present it to anyone in public. Please do not embarrass yourself. Especially in an oral defense. If you are in front of an audience and clueless- cancel the meeting.. You should be able to explain the statistical methodology. PLLLLEAASEE question the reasoning behind the consultant’s approach. Explanations help you learn the material. And lastly, substantiate the claims of your understanding with references. If the best approach is a linear mixed model, please make sure you know the assumptions of the test. There are a few sites which list the assumptions of the test. You should know whether your data met all of the requirements to run a particular model.
-Moore to Follow Amy