In SPSS, you can adjust summary statistics for stratified samples using
Weight by ..., and it allows you to then do a chi-squared test. I found a lot of examples of people doing a chi-squared like this, but nobody mentioning whether SPSS actually accounts for this in the chi square calculation.
Does standard SPSS adjust the chi squared test of independence for the stratified sample after using Weight by ...?
The R "survey" package uses Rao-Scott correction. Similarily, SAS has a Rao-Scott Chi-square test. I'm aware of the SPSS complex surveys extension - while I'd be curious how well that one works, here I'm specifically interested in whether base SPSS does this correctly.
Thanks a lot for your help!
The answer is No. Base SPSS does not adjust standard errors for stratified samples.
All calculations in SPSS on weighted samples that use the standard error (p-values, confidence intervals, ...) will be wrong without the Complex Surveys extension.
Confimed by IBM, UCLA, PARE and Thomas Lumley.
All sources recommended by Thomas Lumley.