US K12 teacher race over time

Hi. I have a novice-level inquiry. I am interested to document the local (United States context) distribution of K12 teacher race and gender over the last 25 years. I am, for example, interested to compare teacher race in the city of Detroit vs. that in other cities. To the extent possible, it is very important not to conflate teacher race in urban school districts with the suburban school districts. I would very much appreciate your recommendation as to how to do this in general. My additional specific questions are:

  1. Can I confirm that 3-digit occupation is the most precise way to define occupation?
  2. What is required to properly measure only active teachers: NOT unemployed, on leave, career hiatus, or retired?
  3. What is the appropriate approach to unit of geography?

It sounds like the US Census and ACS data, available via IPUMS USA, is likely the best data source for this work. I will answer each of your questions one at a time.

(1) Yes, in IPUMS USA, the OCC variable identifies the occupation of each individual. Since occupational coding schemes change over time and it sounds like you are going to be analyzing data over time, you might want to use the temporally harmonized occupation variables: OCC2010, OCC1990, or OCC1950. Each of these variables aims to aid comparisons of occupation over time by keeping a given coding scheme constant (e.g., 2010, 1990, or 1950) and applying that scheme to adjacent years.

(2) If you look at the universe statement for the OCC variable, you will notice that OCC identifies the occupation of both individuals who are currently working and individuals who have worked in the last five years. The EMPSTAT variable will identify current working status for each observation and will help identify active teachers.

(3) Sub-state level geographic identification is a challenge with public use microdata. This is due to restrictions on the data that protect the confidentiality of respondents. Although the identification of cities and metropolitan areas is limited CITY, MET2013, or METAREA may be useful variables for you.

Thank you!