Friday, October 21, 2011My background in statistics isn't as strong as it should be. As a consequence I tend to be very insecure when working on quantitative projects, doing a lot of research and trying to figure out what assumptions I might be violating and how to fix them. When I really feel out of my depth I've taken to going to university-sponsored statistical consulting centers, hoping they'd be able to a) understand my problem, and b) propose a workable solution.
Unfortunately, not once has one of these centers been very helpful. I walk away each time persuaded that these grad students know less about statistics than I do. Today was the worst example. I'm trying to compare logistic regression coefficients across groups and know for a fact that you can't use t-tests for reasons I won't go into. When I met with said grad student he told me to use....t-tests.
So this is the last time I'm going to a stats consulting center. However spotty my statistics training is, at least I know when I don't know something.