how to run mcnemar-bowker tests for 2x3 analyses with pweights (or aweights)?

[Dawne Vogt]

I am looking for some advice on how to run a weighted mcnemar-bowker test for a 3x2 analysis of change over time in category membership. We figured out how to run the regular mcnemar test for a 2x2 analysis (although results say it is a wald test?). However, we haven't figured out how to do it when you have three response categories. Based on our reading, we think the test we need to do is McNemar, Stuart-Maxwell, or a Symmetry test. We have weighted data so we need to do this in a way that allows us to apply pweights (or aweights). Our sample size is large (N over 7500), in case that is relevant to selecting the most appropriate strategy. Thanks in advance for any advice!

[Amanda Aubel]

Hi Dawne,

Did you ever find a solution to this question? I too am trying to test for symmetry of a 4x4 table with pweights. It looks like the symmetry command is able to handle more than 2x2, but it only accepts fweights (not pweights). Did you find another solution?

Thanks,
Amanda

[Rashidul Azad]

You can use Cochran Q test. Command: "Cochran" in Stata. It may serve your purpose. But I am not sure whether mcnemar-bowker test and Cochran Q test are same.