Should I use person weight (PERWT) or replicate weights (REPWTP) for calculating standard errors for estimates?


#1

I am trying to obtain estimates related to poverty by race from IPUMS-USA based on ACS 2012-2016 5-year sample (e.g., the percentage of African American children below the poverty level).

I would like to calculate standard errors for my estimates. When I tried the following SAS code, I got an error that my repwtp2 is negative.

proc surveyfreq data=set1;

table race*poverty;

weight PERWT;

repweights REPWTP1-REPWTP80 / jkcoefs= 0.05 ;

run ;

Is the above code correct? And if so, how do I address the issue of the negative rep weights?


#2

Detailed documentation for specifying the use of replicate weights for various statistical software, is available on this page. Following this documentation should help identify the issue. As long as I am understanding your task correctly, I don’t think using replicate weights will make too much of a difference. Your standard errors may increase in size slightly, but given that your tabulations should include a relatively large number of observations, this change shouldn’t be overwhelming.