Test for the equality of two paired means and proportions with weighted data

[alexgonzalez]

Hello,

How can I perform hypothesis testing for means and proportions between two paired samples with weighted data?

I have data from a survey with a complex design. Therefore, in the dataset I have survey weights for each individual and bootstrap weights to estimate variances.

I already outputted the distributions of and outcome of interest on the same group of individuals before and after some intervention. Therefore, I already have the estimate values and standard errors of the values I want to compare (i.e. means and % of individuals within certain cut-off values along with theirs standard errors).

I'm using SAS EG 7.1.

Thanks a lot,

A.G.

[Reeza]

Look into PROC SURVEYMEANS, specifically work through the examples.
If you still have issues, search on LexJansen.com for specific examples closer to yours.

Or do you already have the means and distributions, accounting for the weights?

[alexgonzalez]

Hi Reeza,
Great suggestion, I'll do that.
To answer your question, I do already have the distributions (i.e. mean, percentiles, and % individuals within cutoffs) along with the standard errors accounting for the weights and bootstrap weights. Any example in particular you may have in mind?
Thanks a lot,
A.G.

[Reeza]

Just a note that if you used PROC MEANS you'll want to ensure you used the right VARDEF option to account for your weights.