Subscribe Bookmark RSS Feed

effect sizes for mixed model anova with nesting= need effects for each term in model


Community Member


Jul 21, 2014

I'm looking for a way to calculate an appropriate effect size measure using a mixed model with subjects as a nested variable. When I set random effects in the fit model function I can't get sums of squares. I think I've successfully calculated f^2 independent of SS but I'm not sure (ps, what are the conventional measures for f and f^2). Anyone know of a quick and easy way to do this in JMP? Just keep in mind I don't have a random factor, just subjects are nested. I'm also using least squares estimation. When I go into SPSS and try to fit fixed and random effects I don't get an option to choose an effect size or an estimation procedure (other than MLE which I don't want) and I'm not convinced it is actually taking nested variables into account appropriately ( I tried to not explicitly specify them as I was told SPSS takes care of this on its own but hmm), nor can I request partial eta for each term in the model. I also don't know how to structure my covariate (subjects)- I just left it unstructured. Is SPSS running this model using GLM? That really confuses me. I feel like when I work with SPSS I'm submitting my data into some kind of unspecified blackhole- what it does in there I don't know! Don't feel that way about JMP. Help! haha