- Description
- Codes
- Comparability
- Universe
- Availability
- Questionnaire Text
- Flags
- Source Variables
- Editing Procedure

## Description

DENSITY reports the average local population density among residents of each Public Use Microdata Area (PUMA) in persons per square mile. Specifically, DENSITY gives the population-weighted geometric mean of the population densities of census tracts in each PUMA.

DENSITY uses tract population densities from the nearest decennial census. In 2000 samples, DENSITY uses 2000 census tract densities. In 2005-2021 ACS and 2010 decennial census samples, DENSITY uses 2010 densities. In 2022-onward ACS samples, DENSITY uses 2020 densities.

DENSITY reports a population-weighted average density rather than the density of the whole PUMA (total PUMA population divided by PUMA area) because the population-weighted density better represents the typical local density among PUMA residents. For example, in a PUMA in southern Florida, most of the population may reside in dense developments near the coast, but if most of the PUMA's area is comprised of unpopulated interior wetlands, the whole PUMA's density would be much lower than the high urban density where most PUMA residents live. This PUMA's average tract density, weighted by tract populations (so each PUMA resident's local density is given equal weight), would be appropriately high, corresponding with a typical PUMA resident's local context.

Using a geometric mean corresponds to measuring the average density on a logarithmic scale, which is suitable because population densities generally have a log-normal distribution (highly concentrated at the lower end of the distribution with a long positive tail). For such distributions, the geometric mean is appropriately less sensitive to large outliers, more sensitive to variations among small values, and generally closer to the median than is the arithmetic mean. In practical terms, a logarithmic scaling makes sense because a difference between densities of 10 and 100 is about as significant for the character of a place as any other factor-of-10 difference (e.g., 1,000 and 10,000), and it is clearly more significant than an equal absolute difference of 90 at high densities (e.g., 10,010 and 10,100).

The specific steps to compute DENSITY are 1) multiply each tract's population by the logarithm of its density (population divided by land area), 2) sum these products for all tracts in each PUMA, 3) divide the sum for each PUMA by the total PUMA population, and 4) exponentiate the results to return to a linear scaling of population densities. (In the first step, if a PUMA boundary subdivides a tract, we use the whole tract's density, but we limit the population weight to the portion that also resides in the PUMA.)

For a detailed explanation and demonstration of the DENSITY measure (as well as the METPOP00 and METPOP10 variables), see:

DENSITY is a 7-digit numeric variable with 1 implied decimal. The units are persons per square mile.

## Comparability

The comparability of this variable is limited by two factors. First, PUMA boundaries change after each decennial census. PUMA extents are consistent only within each of these sets of samples: 1) 2000 census and 2005-2011 ACS, 2) 2010 census and 2012-2021 ACS, and 3) 2020 census and 2022-onward ACS.

Second, the source years for tract densities vary across samples. Specifically, in 2000 samples, DENSITY uses 2000 census tract densities. In 2005-2021 ACS and 2010 decennial census samples, DENSITY uses 2010 densities. In 2022-onward ACS samples, DENSITY uses 2020 densities.

## Universe

- All households and group quarters.

## Availability

- 2022: All samples
- 2021: All samples
- 2020: All samples
- 2019: All samples
- 2018: All samples
- 2017: All samples
- 2016: All samples
- 2015: All samples
- 2014: All samples
- 2013: All samples
- 2012: All samples
- 2011: All samples
- 2010: All samples
- 2009: All samples
- 2008: All samples
- 2007: All samples
- 2006: All samples
- 2005: All samples
- 2004: --
- 2003: --
- 2002: --
- 2001: --
- 2000: 5%; 1% unwt
- 1990: --
- 1980: --
- 1970: --
- 1960: --
- 1950: --
- 1940: --
- 1930: --
- 1920: --
- 1910: --
- 1900: --
- 1880: --
- 1870: --
- 1860: --
- 1850: --

- 2022: All samples
- 2021: All samples
- 2020: All samples
- 2019: All samples
- 2018: All samples
- 2017: All samples
- 2016: All samples
- 2015: All samples
- 2014: All samples
- 2013: All samples
- 2012: All samples
- 2011: All samples
- 2010: All samples
- 2009: All samples
- 2008: All samples
- 2007: All samples
- 2006: All samples
- 2005: All samples
- 2000: PR 5%
- 1990: --
- 1980: --
- 1970: --
- 1930: --
- 1920: --
- 1910: --

## Flags

This variable has no flags.## Editing Procedure

There is no editing procedure available for this variable.