GEE

[MAALKAN]

Enhancement: Generalised Estimating Equations Purpose: Analysing repeated measures binary data Use-case: PhD Thesis

I have performed a Generalised Estimating Equations (GEE) analysis in SPSS but I am keen to use JASP exclusively. My problem is that I want to examine group differences in accuracy (i.e., binary outcome: correct/not) for three different test measures and trial order which would need to be entered as within-subject factors and group as between-subject factor.

I would also like to use the GZLM procedure (for non-repeated data) in JASP but as it requires random effects (and i wish to model fixed effects) I can't use it. Would fixed effects also be an option you would consider adding please?

[FBartos]

Hi @MAALKAN ,

thanks for the suggestion. We will look into it, but I wouldn't expect that we would implement gee very soon.

In the meantime, I think that the Generalized Linear Mixed Model analysis (with binomial family) from the Mixed Models module for the repeated measures and Logistic Regression from the Regression module for no repeated measures should be able to do the job.

Cheers, Frantisek

[MAALKAN]

Thank you @FBartos for the feedback.

[tomtomme]

@MAALKAN I read here: https://github.com/statsmodels/statsmodels/wiki/Examples#generalized-estimating-equations-gee about many possible implementation of such models. The last mentions ARIMA for time series analysis, now available in jasps time series module. This may however not fit your request for RM binary data.

@FBartos Possible R packages: R (packages glmtoolbox,[14] gee,[15] geepack[16] and multgee[17])

[mlotinga]

Would really like to see GEE implemented in JASP too - useful for estimating population averaged effects from within-subjects binary data, which GLMMs don't offer. Hubbard et al, 2010 - To GEE or not to GEE