Which method to use for analysis of order/differences in measures / repeated measures analysis?

[Jasmin Mayer]

I have a cross-sectional data set and I have three evaluations of political parties. Now I want to analyse if these evaluations differ between East and West Germans, with the assumption that in both groups, party A is evaluated more positively than party B which is evaluated more positively than party C (hypothesis 1a) and test if a specific attitudes has an effect on this order (hypothesis 1b). However, I also assume that party C is evaluated more positively by East than by West Germans (hypothesis 2). For the last hypothesis, I would simply use a t-test.

Now, I am a bit at loss which test/method to use for the first hypothesis. I similar paper from a social psychological background uses multivariate repeated measures analysis. I never used that before. Is there something within the framework of regression analysis I can use?

Thank you in advance!

[Clyde Schechter]

If your data are cross-sectional rather than longitudinal, repeated-measures analysis sounds like it would not be applicable in any case. It sounds to me like you can just do your ordinary regression model, with party##east_vs_west interaction get everything you need from there. If the ## notation is not familiar to you, stop and learn about it with -help fvvarlist- and then use the -margins- command to interpret your results.

The -margins command is a bit complicated, but, fortunately, the excellent Richard Williams has written a remarkably clear introduction to it that is available at https://www3.nd.edu/~rwilliam/stats/Margins01.pdf. That file includes some worked examples that are reasonably similar to what you are doing.