Do I need to rescale sampling weights after selecting cases?


#1

I am interested in downloading an extract of the 2009-2011 ACS that only includes individuals living in metropolitan areas between the ages of 25-34. Are the sample weights in the data extract adjusted accordingly, or do I need to adjust the sample weights myself to account for dropping so many cases? If so, how do you manually adjust the sample weights?

Previously, I downloaded a full extract of the 2009-2011 ACS and then dropped individuals living in metropolitan areas who were not aged 25-34. This way I could run tabulations and regressions much faster. But would using the orginal sample weights produce means and tabulations (e.g. % unemployed, % white, etc.) that are inaccurate because the inverse probability of selection is different in the smaller data set?


#2

The sample weights (PERWT) in IPUMS-USA are copied directly from the Census Bureau’s PUMS files. The Census Bureau provides a detailed explanation of weights in the Handbook series. Because the weighted frequencies are meant to represent the entire U.S. population (accounting for sampling error, etc.), each person within your sample will represent the same proportion of the U.S. population regardless of the inclusion or exclusion of others.

I hope this helps.