Weights in Inferential Tests?


New Member

I've conducted a survey and noticed that males were under-represented. To "fix" this, I weighted sex to reflect the true proportions in the population.

Can I run inferential tests such as regressions or chi-square, with weighted sex? Or is it better to remove the weights when conducting these tests?
I don't really know this subject, but my supervisor told me that if I was going to do weighted analysis I should do so in all analysis, but not when making tables. The last part makes sense since then for e.g. sex, those would give a descriptive of 50% males and 50% females when this wasn't really the reality.

Just throwing out what my intution tells me it should be perfectly fine to do a weighted analysis of chi-square, but why not try to do both and see if there is any difference?