Hi everyone,
I am trying to implement a meta-analytic mixed-effects model in SPSS, which uses two polytomous categorical predictors with 4 categories each. I am especially interested in interaction effects. For the time being the best thing to do seems to use the metareg.sps macro by Wilson (
http://mason.gmu.edu/~dwilsonb/ma.html), dummy-code the predictors and compute interaction terms to enter in the regression equation. However, this leaves me with a far too high number of predictors for the number of cases I have (which is 30).
Does anyone have an idea how I could solve this? Are there ANOVA-type meta-analytic approaches which can handle more than one predictor at a time and compute interactions between the predictors? Is it even recommendable to use a regression if I don't have any continuous predictors at all? If nothing else works: Would dummy-coding or effect-coding be preferrable here?
Thanks in advance,
Nicole
Locked
