New resource for mixed modeling with SAS

[Will Hopkins]

Go to https://sportsci.org for an article on "How to Identify and Specify Fixed and Random Effects for Linear Mixed Models in the Statistical Analysis System." It was motivated by the need for a fresh look at mixed modeling to help a colleague understand and use it.

If you don't use SAS, you might find the first few sections useful, because I go back to basics by explaining data, variables, linear models, fixed effects, and random effects. 

If you do use SAS, then the 15 SAS programs linked to the article are bound to be useful. Each program has code to simulate relevant data for a wide range of models. You can modify the values and/or the code and re-run the programs to enhance your understanding.

The simulations and programs are all for general linear mixed models, where the dependent variable is continuous, and the effects are differences or changes in means. I have not provided simulations and programs for generalized linear mixed models, where the dependent variable is a count or a proportion and the effects are ratios thereof, but I do explain how generalized models work.

This paragraph didn't make the cut, but it's worth stating here: "Notwithstanding the recent huge advance of artificial intelligence (AI), I don't think it will provide you with trustworthy programs for mixed modeling, not yet, anyway. The reviewer of this article tried it, and although the resulting program looked amazingly good, it didn't do what it was supposed to. I imagine its programs will work properly, if AI ever truly understands its store of information."

Will