Weights for ASEC and JTS linked file

I have linked the ASEC and JTS files using CPSIDP. When using the merged data, should I use the JTS weight since I lose around 50% of the JTS observations after linking?


If you are analyzing variables from the JTS, you’ll want to use JTSUPPWT. Ideally you’d want to adjust this weight to account for potential differential attrition. This is what the IPUMS CPS longitudinal weights try to accomplish. However these are based on the basic monthly weights (WTFINL). You could construct an adjusted JTSUPPWT based on those JTS respondents who can link to the following ASEC (see the Stata replication files next to “Create Your Own” in the Longitudinal Weights table on the IPUMS CPS longitudinal weights page). This will make the analysis more representative of the full group of JTS respondents. Please note that using this program will generate weights that sum to the number of individuals who are in the JTS sample and are eligible to link to the ASEC. This is about half of the sample, as you noted (you can also check this using the RoPES tool), so although you can use these weights to calculate means and other statistics, they shouldn’t be used to calculate totals.

If the amount of attrition is low (especially for any subgroups of interest), adjustment of the weights is probably not necessary, but that’s up to you to decide.