some much needed horn-tooting

I would just like to say that there is something supremely satisfying about teaching myself some fairly advanced statistical concepts.

Perhaps I should preface this by saying that yesterday was rather demoralizing, because my new and improved dissertation experiment procedure still didn't really work.  So I have to go back to the drawing board - again - and revamp things.  And I also think there's no possible way to do what I was planning/hoping to do.  That doesn't mean I can't accomplish the same end using different means.  It's just that... now I have to step waaaaay out of the box, and given the constraints I now know to exist, having done two very illuminating "pilot experiments", shall we call them, I have to rethink how to get at what I want within those somewhat stringent constraints.  It can be done.  I think.  But I don't want to think about it any more right now.

Instead, I am trying to work on my LSA conference talk, which is coming up in about two weeks, and it has come to my attention that I don't actually understand the math behind logistic regression very well.  I haven't worked with logistic regression very much, because I'm usually using normal ol' linear regression.  And now I'm getting these weird error messages that I don't really know how to interpret, because I don't intuitively grasp the math behind the logit transform.

While this could easily cause me to throw up my hands in exasperation and burrow into a little hole ("nothing is going my way this week!"), it instead feels a bit empowering.  I'm reading some tutorials on logistic regression and actually starting to understand things better, and I feel like this problem is eminently solvable.  Solvable problems are truly necessary sometimes.  And once I solve this problem, I will have some results, and that will be seriously gratifying after fighting with my dissertation procedures for the past month or so.

Now I need to take a shower and do some stuff!