Hello Deborah,

If population parameters are known, then you may compare sample attributes to population via the chi-square goodness-of-fit test. Sample values will furnish the observed frequencies (e.g., N of married, single, widowed, divorced participants) whereas the population parameters would provide the expected proportions (if the goal was to test the hypothesis that your sample was no different than the population on the attribute of interest).

So, don't worry about "overlap." You're testing sample distribution vs. a population distribution.

With non-probability samples, even a substantial sample size (e.g., 40% of the population) can still yield biased estimates of population parameters.

Good luck with your work.

Hello Deborah, I refer to Armitage and Berry; Statistical Methods in Medical Research. Go to pages 411 - 430 approx. I have edition 3, which I've had for probably 15 years thereabouts. The glue has disintegrated, so all the pages are falling out, but I know where the book is. It would be recommended to answer your questions. If you want me to find my book that is falling apart, let me know, as I said all the pages are falling out so I'll be a bit slower. However you should have it in your library reference collection. Regards.