As of yet, the October 2025 CPS has not been released.
This creates a problem when doing longitudinal studies: What to do about weights when linking repondents’ appearances in the CPS? For example, suppose I wanted to trace out the employment status on those who first entered the CPS in August of 2024 and left the CPS in November of 2025. LNKFW8WT can’t be used because observations for October 2025 are not available. What are the recommended alternatives?
Thanks.
The October 2025 Current Population Survey data were not collected due to the U.S. federal government shutdown. Read our blog post on the missing October 2025 data to learn more about the implications for IPUMS CPS users.
Thanks for pointing out the Blog Post. Unfortunately, the post did not mention the effect the missing October CPS has on longitudinal weights.
Am I correct that the missing October data means that the weights LNKFW8WT, LNKWMIS14WT, and LNKWMIS58WT are not available for households who entered the CPS in July, August, September, and October of 2024 or 2025? What alternatives are suggested if I want to provide population estimated based on my linked samples?
Also, am I correct , if I’m using the validation codes provided in the CPS 2018 Summer Workshop (long.txt and wide.txt) all I need to do is to let the code know how many periods I will be looking at and not have to change anything in the body of the code?
Thank you.
—Stuart
I apologize for any confusion created by my post yesterday. Here is a revised summary of which weights are missing for which respondents. Since the October 2025 sample does not exist, longitudinal weights are missing when the weight would have required the October 2025 sample in order to be created. Longitudinal weights are only included in the first time period of the time periods being linked. For example, LNKFWMIS58WT is only available when respondents are in month in sample 5.
LNKFW1YWT is missing in the October 2024 sample for people in month in sample (MIS, see MISH variable) 1-4.
LNKFW1MWT is missing in the September 2025 sample for people in MIS 1-7.
LNKFWMIS14WT is missing in the July 2025, August 2025, and September 2025 samples for people in MIS 1.
LNKFWMIS58WT is missing in the July 2025, August 2025, and September 2025 samples for people in MIS 5.
LNKFWMIS45WT is missing in the January 2025 sample for people in MIS 4.
LNKFW8WT is missing for all cohorts who would have been in MIS 2-8 in October 2025.
The Census Bureau has not provided any guidance on how to apply weights to longitudinal analyses of CPS data when the panel would have included October 2025. Note that these longitudinal weight variables are created by IPUMS, not by the Census Bureau. The Census Bureau may not come out with any methodological guidance on weighting these types of longitudinal analyses, given that they do not provide these weights to begin with. While IPUMS researchers work to produce new information on methodology for working with the data we provide, we do not currently have any definitive guidance to share on how you might weight a longitudinal analysis that is missing the October 2025 sample.
We do provide replication code that may be helpful in creating your own longitudinal weights in this context. See our page on linking and the CPS and scroll down to find the file named “Stata replication files for creating LNKFW1MWT for January 2012.” You would need to modify this code considerably but may find it a useful starting point. Longitudinal weights account for the differential probability of being linked from one sample to another, or across all samples in the panel. Constructing weights that account for the missing October 2025 data is not straightforward. I would suggest looking to other researchers in your field to see how people are dealing with the missing part of the panel and how they are weighting longitudinal analyses. There may be some imputation methods of interest as well.
Regarding the code for creating validated longitudinal identifiers, yes, you can use that code as is if you do not want to modify the validation requirements we use. You should review the code to ensure it meets your needs in terms of which variables you want to validate based on, how much change to age you would want to allow, etc.
Thank you Isabel. –Stuart