oversampling and undersampling with weights

[Tracy Lam]

Hi all,

I am comparing means of some demographic characteristics with and without the use of survey weights. If it is the case that the means with the sample weights are higher than the means without the sample weights, does that mean that this group was under- or over-sampled?

I have a general idea about the code for the sampling weights and estimation techniques, but I don't understand what the output is telling me. Please advise!

[Joao Santos Silva]

Dear Tracy,

You cannot see whether a group was over- or under-sampled just by comparing the means. To get any decent help here you need to give us more information about what you want to do and the data you have.

Joao

[Steve Samuels]

You want us to tell you what your output means, but you don't show it! Please reread FAQ 12. Whatever it shows, don't ignore the weights. They are intended to make the sample results more closely match the population. A researcher might sample subgroups at different rates for any number of reasons, for example: cost (sample fewer high cost areas; time (sample one person in a HH); importance (sample larger sampling units at a higher rate than smaller ones, because the larger ones represent more people); special interest. While there are situations in which ignoring the weights is justified, estimation of demographic characteristics is not one of them.