Regression for comparing two groups on one variable? And t-test with survey data?

#1
Hi -

I have two populations / groups that I need to compare on one variable. Now usually I would use a t-test or anova but ive been told that these are unable to be used with survey data in stata. Is this correct advice - can I not use a t-test or anova in stata with survey data? And would a regression work for my goal, to compare two groups on an interval DV?

Thanks for any help.
 

noetsi

Fortran must die
#2
If you are doing certain types of samples, cluster samples for example or one where weights are involved, you might get invalid results if you ran a t test (or any other type of regression). But I suspect Stata has a module to deal with that, much as SAS does.

There could be lots of reasons that Stata, or any software, might not work. You have to provide more details to address this. What specifically about the survey data do you think makes a t test not work (or not work in a specific software, in which case this question really should be in the Stata forum rather than here].
 

hlsmith

Omega Contributor
#3
I wonder if you are referring to the never ending debate on whether you can treat integer data as continuous data and this may not be a program issue - but a statistical issue?