Emsi’s ZIP- and tract-level demographics data are created by mapping Emsi county-level demographic data down to more granular geographies.
We start by collecting census tract population demographic data from the American Community Survey (ACS) 5-year Public Use Microdata Sample (PUMS) dataset. ACS’s population data are available by age, gender, and race/ethnicity. Since we already have county-level Emsi demographics created from the Census’ Population Estimates program, the ACS data is only used to create proportions to apply to the county-level data. The proportions inform how much of each county’s population is applied to the Census tracts within it. Applying the proportions gives us population data at the tract level.
Next, the tract-level data is aggregated to ZIP-level data using a tract-to-ZIP mapping provided by the US Deparment of Housing and Urban Development (HUD). Those files can be found here. The tract-to-ZIP mapping provides proportions and mapping assignments, allowing Emsi to map populations in appropriate proportions to the ZIP codes that touch the census tract. The proportions are necessary because a census tract can have population in multiple ZIP codes. The HUD mapping shows what proportion of the population of the tract to put in each ZIP code that touches the tract.
To summarize, Emsi begins with county-level data, uses proportioning to bring the data down to the tract level, and then uses a tract-to-ZIP mapping to reaggregate the data up to the ZIP level.