EJ index that takes into account MOE's

I am researching methods to create a new environmental justice index for my organization.  The current one uses 7 demographic variables for vulnerable communities and simply flags each one as 1 if it is above the regional average and 0 if it is not.  The sum of the 1's is the index score.

I am hoping to use a more refined methodology this time around.  There are lots out there to choose from, but I haven't found any that consider MOE's in the calculation. Has anyone created an index that took MOE's into account?  I am mostly concerned with finding a way to take significant difference into account.

Any suggestions?

  • Have you seen the statistical testing spreadsheet for ACS data?
    www.census.gov/.../statistical-testing-tool.html

    You might be able to use the "Statistical Testing for Two Estimates" sheet, with the regional average in the first column and individual geography values in the second column to determine whether they are statistically different. If your demographic variables come straight from ACS, I think this would be straightforward. If you are working with derived estimates (e.g. summing up multiple income bins or race/ethnicity groups) it might be more difficult to get reasonable estimates of the MOEs for the derived data.
  • SACOG's environmental justice analysis conducted in 2015 used a similar index to further identify vulnerable communities (beyond race/ethnicity and income) but used quintiles over the regional average. We tested about 10 different variables at both Census Tract and Block Group level and calculated CV's for all to help in our determination of which variables were reliable 'enough'. I believe our cut off was 30% for use of any single geography and dropping the variables that had the highest number of geographies that were above 30%. We didn't necessarily have a system in place for taking the MOE's into account for variables that were chosen.