How to integrate Area Median Income with Table B19001 (household income brackets by race)



For my current research project, I am working on quantifying the number and percentage of lower-income households in wildfire-prone census tracts in the San Francisco Bay Area.  I am considering using the commonly-implemented Area Median Income (AMI) metric to group households into categories. That being said, AMI requires household size for computation, and Table B19001 (household income brackets by race), one of the most used pieces in urban displacement demography research, does not include this. I am curious if there are any workarounds possible or alternative metrics that may prove better.


Thank you, and have a wonderful day.


  • Isaac

    I have done exactly what you want to accomplish using the ACS PUMS data. If you are unfamiliar with this resource I suggest taking at look at the IPUMS USA website ( where you can generate and download data extracts from PUMS or run on line analyses. I will warn you that breaking households by income into ranges based on percent of AMI is a bit tedious but it is certainly doable.
  • In reply to Cliff Cook:

    Thank you Cliff, this information is really helpful.

    For the data selection on IPUMS USA, I have added the total household income, race, and Hispanic origin variables to my cart. Since much of the data corresponds to census tracts, do you have any recommendations for what geographical variable I should use?

  • In reply to Isaac Gendler:


    If it's tract-level data you want, I'm afraid you're limited to the Comprehensive Housing Affordability Strategy (CHAS) data, a special tabulation from the ACS that has income data already expressed in terms of AMI (so no need for your own calculations).

    You can access it at Table 11 has the most detail on income levels; other tables provide a bit more detail on the usual groupings (at or below 30%, 31%-50%, 51%-80%, 80%+).

    Unfortunately, because it's a special tabulation, it lags about 1.5 years behind the ACS release schedule. Right now, the most current data is for the 2012-2016 period; if previous years are any indication, the 2013-2017 data will be out this summer. If current data is more important than the specific measure of "lower-income," you could always use poverty data (which also accounts for differences in household size).

    Good luck!
  • In reply to Matt Schroeder:

    I did not address the issue of geography and I apologize for that. The smallest geography available PUMS data are PUMAs which are about 100,000 people in size.

    Matt is correct that the CHAS data might be a better source for your project.