Combining pweights and aweights in Stacked Difference-in-Differences Analysis

[marie carmel]

Hello Statalist members,

I am conducting a stacked Difference-in-Differences (DID) analysis using survey data that includes sampling weights (pweights). Now,I follow Wing et al. (2024) and create a stacked_weight variable to ensures that each stack contributes appropriately to the overall estimate.

Given that Stata does not permit the simultaneous use of pweights and aweights, I am seeking advice on alternative methods to appropriately incorporate both types of weights in my analysis.
Any guidance, references, or examples from similar analyses would be greatly appreciated.

Thank you for your assistance.

Maire

[Dirk Enzmann]

It depends on what the analytic weights actually represent: If they do not represent the number of cases used to study aggregate statistics (see the manual for help weight) but are normalized population weights (i.e. their sums are the number of cases and their means are 1) and if you are not interested in the population number itself, I believe (but may be wrong here and if so would like to be corrected!) that you can simply multiply your population and analytic weights and treat their product as population weights.