CPS-ATUS merge dropping a large number of observations

I am trying to merge CPS data to ATUS data for years 2003-2019. As mentioned in this workshop, I first dropped all observations that weren’t MIS 8 and then merged to ATUS using CPSIDP (1:1 merge). However, only 16,773 (out of 651749) observations merged. Should I be able to merge more observations? If yes, how?

Could you please clarify which CPS samples you are using? If you are only using ASEC data, your month-in-sample restriction criteria are too narrow; for within-year ASEC to ATUS links you can retain individuals with IPUMS CPS MISH values of 5-8 without repeat observations of the same individual. However, if you are using basic monthly data, restricting to a single MISH value (e.g., the eight household interview) is appropriate. Without knowing which CPS samples you are using, it is harder to comment on the number of cases you are merging. However, I will note that I would not expect all cases to match as only a subsample of outgoing CPS households are selected for inclusion in the ATUS sample; you can see sample sizes from IPUMS ATUS and review Chapter 3 of the BLS ATUS User’s Guide for more details on sample size and strata, as well as information on response rates.