The other day, we met with a second statistics consultant to find out whether we did our “logistic mixed model regression” correctly; or rather, to find out if he could see any problems or issues with how we’d specified the model, and how we’re interpreting it. We were feeling concerned about the analysis because neither of us have ever done anything like this before, and we taught ourselves how to do it from a (really great) book that I bought from Amazon one day after reading a couple of online reviews. I mentioned the book in an earlier blog entry: An Introduction to Categorical Data Analysis (Second Edition) by Alan Agresti. I can’t say this enough times… money spent on stats books is never wasted.
Anyway, we were both a little nervous about this meeting because if we found out there were problems with our analysis, it would mean not shooting for a CHI paper, and doing a lot more work to fix the problems. Plus, we really like the way the analysis turned out! Fortunately, the person we talked with didn’t find any problems, and even seemed really impressed. That was definitely a relief.
After having these meetings, I think I might really enjoy being a statistics consultant. I don’t know enough stats, of course, to actually do it. But it seemed like it would be a lot of fun to get to hear about and think through all the different kinds of research problems that would come across the desk of a stats consultant. It also seemed like it would take a lot of practice to develop the skills to listen to a ~10 min summary of someone’s project, and ask all the right questions to learn enough about the research to be able to help them.
