mann-whitney U test

[Wang Weiping]

Hi there,

After the Mann-Whitney U test, how to see whether there is any difference between the two groups at various levels of explained variables?
For me , my dependent variable is education expectations(its a ordinal variable , 1= high school 2 = bachelor 3 = master 4= phd ) ,
explaining variable is area (its a dummy variable 0= rural 1= urban).
So , if I want to know that is there difference between rural and urban students for their education expectation is bachelor? What method I can use ? Which stata command will be useful?
Hope to see your reply .
Thanks !

[Nick Cox]

I would not even start with Mann-Whitney U test here. I would start with a chi-square test (for all that it ignores the ordering) and in terms of modelling proceed to say ologitor oprobit

[Wang Weiping]

I would not even start with Mann-Whitney U test here. I would start with a chi-square test (for all that it ignores the ordering) and in terms of modelling proceed to say ologitor oprobit

Hi , professor Cox,

Thanks for you timely reply . Because I want to know if is there difference between rural and urban students for their education expectation is bachelor? And if their education expectation is mater degree (or phd degree)?

There is a stata command tabchi I found in this forum . I just wondering this command is useful for my research design?



weiping wang

[Nick Cox]

It really doesn't matter much how you do the chi-square test. tabulate will do it for you. Otherwise I don't understand that you are asking anything different from #1.

[Wang Weiping]

It really doesn't matter much how you do the chi-square test. tabulate will do it for you. Otherwise I don't understand that you are asking anything different from #1.

Hi , professor Cox,
Let me reframe my question. My explained variable is education expectations(Its a ordinal variable , 1= high school 2 = bachelor 3 = master 4= phd ) , explaining variable is area (Its a dummy variable 0= rural 1= urban).
Using chi-square test , tabulate will tell us if there is an statistically significance for the all group .
But ,for me , I want to know that for every education expectations level , is there an statistically significance between rural and urban students. (eg. when their education expectation is bachelor is there an statistically significance between rural and urban students)

Best Regards,


weiping wang

[Nick Cox]

That’s why you need a model not only to ask but also to answer focused questions.