Hello,

I have a question about complex sampling (analysis preparation wizard) when working with large publicly available datasets. 

I am at the step where I select "what sampling design should be assumed for estimation," and I am not sure how to determine if WR (sampling with replacement), equal WOR (equal probability sampling without replacement), or unequal WOR (unequal probability sampling without replacement. The survey that I am using uses both a disproportionate stratified sample design AND a simple random sample design. 

Thank you for your help!

Hi,

The question is  difficult to answer without knowing more about the design. However, for most public use datasets one would use WR despite the fact that the actual design might be something else. Unequal WOR, which is often the actual design, has rather stringent demands on the sampling information available. Disproportionate allocation to strata is not relevant in the context of choosing between WOR/WR because sampling is independent in each stratum. 

Best

Jon

Hello,

I have a question about complex sampling (analysis preparation wizard) when working with large publicly available datasets. 

I am at the step where I select "what sampling design should be assumed for estimation," and I am not sure how to determine if WR (sampling with replacement), equal WOR (equal probability sampling without replacement), or unequal WOR (unequal probability sampling without replacement. The survey that I am using uses both a disproportionate stratified sample design AND a simple random sample design. 

Thank you for your help!