Hi! I am trying to analyze the labor market dynamics of subsets of occupations (e.g., teachers, teacher assistants) using the CPS data. My current goal is to use weighted tabulation to calculate how the number of teachers holding multiple jobs changed over the study period (e.g., 1990-2000), using basic monthly CPS data. In this case, can I simply use the WTFINL (final basic weight)? Or should I construct separate weights that account for the fact that I select subsets of occupations?
Thank you very much.
You should be able to simply use WTFINL without any additional modifications. WTFINL provides the number of people that a given individual represents from the surveyed population. Even though you’re selecting a subset of the population (i.e. teachers), the sum of their individual weights should approximate the total number of teachers in the given sample year.
I recommend paying particular attention to your sample sizes and standard errors when analyzing small subsets of the population (e.g. when crossing detailed occupation and jobholder status codes). With small sample sizes, you will have a lot of random sampling variation from year to year. If you find this to be an issue, you can combine similar occupation or multiple years of data to increase your sample sizes.