# CPS basic monthly weights and unemployment (in R)

**URL:** <https://forum.ipums.org/t/cps-basic-monthly-weights-and-unemployment-in-r/3756>\
**Category:** CPS\
**Created:** [September 16, 2020, 2:32pm UTC](https://forum.ipums.org/t/cps-basic-monthly-weights-and-unemployment-in-r/3756 "2020-09-16T14:32:20Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Chris\_Wodicka](https://avatars.discourse-cdn.com/v4/letter/c/7993a0/32.png) [@Chris\_Wodicka](https://forum.ipums.org/u/Chris_Wodicka)\
**Post date:** [September 16, 2020, 2:32pm UTC](https://forum.ipums.org/t/cps-basic-monthly-weights-and-unemployment-in-r/3756/1 "2020-09-16T14:32:20Z")

</div>

Hello,

I couldn’t find this specifically addressed anywhere, so apologies if I missed something.

In the past, I have used srvyr to generate estimates using replicate weights from other data products, but I have a question about doing something similar with the CPS basic monthly.

I’m working with several months of CPS basic monthly data to estimate state-level unemployment rates by race.

I am using compwt (and then divided by number of months in my sample) as my weight because it’s my understanding that is the correct weight to use for analyses related to monthly employment.

If I want to generate the correct SE and MOE estimates in R using srvyr, do I just need to run something like

svy\_employment ← as\_survey(cps\_data, weights = compwt)

or are there other arguments, etc., that need to be specified?

Thank you!

---

<div class="post-metadata">

**Author:** ![Matthew\_Bombyk](https://avatars.discourse-cdn.com/v4/letter/m/34f0e0/32.png) [@Matthew\_Bombyk](https://forum.ipums.org/u/Matthew_Bombyk)\
**Post date:** [September 22, 2020, 9:08pm UTC](https://forum.ipums.org/t/cps-basic-monthly-weights-and-unemployment-in-r/3756/2 "2020-09-22T21:08:39Z")

</div>

For basic monthly CPS samples, there are no replicate weights or sampling design information included in the public use microdata. You can still calculate standard errors by treating it as a simple weighted sample (as in your example code), but they will likely be slightly underestimated. Unfortunately that’s the best we can do with basic monthly samples. The correct person-level weight to use for most analyses is [WTFINL](https://cps.ipums.org/cps-action/variables/WTFINL#description_section). We recommend COMPWT only be used for replicating published unemployment figures from BLS.

---

<div class="post-metadata">

**Author:** ![Chris\_Wodicka](https://avatars.discourse-cdn.com/v4/letter/c/7993a0/32.png) [@Chris\_Wodicka](https://forum.ipums.org/u/Chris_Wodicka)\
**Post date:** [September 23, 2020, 4:10pm UTC](https://forum.ipums.org/t/cps-basic-monthly-weights-and-unemployment-in-r/3756/3 "2020-09-23T16:10:21Z")

</div>

That’s helpful. Thank you!