Hello,
I have very basic knowledge in Statistics. I need to use sample weights for survey data consisting of 388 cases. I have Design weights and Analysis weights. From what I understood, when I want to produce frequency tables, there is no problem to use the analysis weights and get totals for the population size. However, when trying to measure significance, this poses a problem because SPSS is led to believe the population size is the sample size, and even small differences are shown as significant. I can then turn the analysis weights into proportional (multiplying them by the sample size and dividing by the population size) but since the sample design involved clustering, this could show misleading results as well. Please, advise me on what is the best course of action! (I apologize if the explanation is not adequate; I do not know much about statistics). Any advice is appreciated! Thank you!
Reply With Quote
Posting Permissions