Do I need to weight my sample before I select cases?


I need to select North Africans and Middle Easterners in each MSA if there are at least 10 of them. However, I need help with regard to weighting process. Am I supposed to weight my sample first and select the cases later or select my cases from the unweighted sample and use weight while analyzing my data? For instance, if I don’t use weight in the first place, I would have 58 MSAs that have at least 10 North Africans. If I weight the sample first, the number of MSAs that contain North Africans triples! If anyone could clarify this problem for me, I would appreciate it. Thanks in advance.

If you wish to select MSA’s based on the total estimated number of North Africans living within the MSA, then you should apply the weights first to identify MSA’s with a sufficiently large North African Population. The weights multiply each respondent by the number of actual people in the population that respondent represents. So if you want to know how many North Africans actually lived in the Chicago MSA, you would need to apply the weights to get that estimate.

If, however, you are more interested in ensuring that your MSA’s have at least 10 North African response cases (to prevent 0 cell issues) then you should not apply the weights before selecting the MSA’s.

I hope this helps.