Merging Jan-Dec IPUMS CPS files: How to set up weights correctly?

CPS
1

Hello, I’d like to estimate the number of young adults enrolled in higher education for the nation overall and top states. To do so, I’d like to pool monthly files (Jan to Dec 2019) to produce an annual average (and boost the sample size for the state level analysis). I’m using STATA.

Since I’m merging 12 monthly files into one dataset, I have to adjust the person-level final weight variable (wtfinl). Is it correct to divide wtfinl by 12 and round the outcome? (STATA doesn’t work with noninteger frequency weights) Is my approach correct:

gen perwt12=(wtfinl)/12
replace perwt12=round(perwt12)

Thank you,
Jeanne

2

Your approach is approximately correct. A conceptually more precise approach is to multiply the sampling weight in sample x by (sample size of sample x)/(pooled sample size). Because the sample sizes of each basic monthly sample within a given year is roughly similar, these two approaches should be approximately equivalent.

1 Like
3

Thank you!

4

Thank you!