CPS sample size in April and June 2001

There seem to be a sample size adjustment in 2001, which causes it to jump from around 120,000 to 140,000. However, it does not happen just once. In April it increases, decreases in May, increases in June and then stays at a consistant level from july on. (sample sizes taken from here)


March – 116,663

April – 140,268

May – 118,323

June – 140,743

July – 142,178

However, it does not seem that the WTFINL or EARNWT variables account for the level changes in April and June. This issue was been addresed in this post from July 2015. However, I am seeing results similar to the ones outlined by the author of that post.

Is there a way around this issue? I have found that the NBER basic monthly files do not show a similar increase in sample size in April and June of 2001.


The issue seems to be that the IPUMS-CPS source data has different record counts than what is currently on the US Census Bureau’s CPS FTP site. On the CPS FTP site, the record count increases in July 2001, as is expected due to the survey expansion. But, you are correct in pointing out that, in the IPUMS-CPS data, there is a 20% increase in observations in April 2001, followed by a 15% decrease in May 2001, followed again by a 18% increase in June 2001 that then stays consistent. As it happens, the IPUMS-CPS Team is already looking into the source of this deviation and will be making any necessary corrections (if necessary) soon.