CHAS Data Documentation

Dataset/CHAS Data Documentation

• Other Datasets

The CHAS special tabulation is a count of the number of households (or housing units) that have certain combinations of HUD-specified characteristics, summarized for HUD-specified geographies.

2017-2021 CHAS (released September 2024):
For the data dictionary for the 2017-2021 data, click here: Data Dictionary 17-21.

2016-2020 CHAS (released August 2023):
For the data dictionary for the 2016-2020 data, click here: Data Dictionary 16-20.

2015-2019 CHAS (released September 2022):
For the data dictionary for the 2015-2019 data, click here: Data Dictionary 15-19.

2014-2018 CHAS (released September 2021):
For the data dictionary for the 2014-2018 data, click here: Data Dictionary 14-18.

2013-2017 CHAS (released August 2020):
For the data dictionary for the 2013-2017 data, click here: Data Dictionary 13-17.

2012-2016 CHAS (released August 2019):
For the data dictionary for the 2012-2016 data, click here: Data Dictionary 12-16.

2011-2015 CHAS (released June 2018):
For the data dictionary for the 2011-2015 data, click here: Data Dictionary 11-15.

2009-2013 CHAS (released July 2016):
For the data dictionary for the 2009-2013 data, click here: Data Dictionary 09-13.

2008-2012 CHAS (released June 2015):
For the data dictionary for the 2008-2012 data, click here: Data Dictionary 08-12.

Disability data table based on 2009-2011 CHAS (released May 2014):
For the data dictionary of disability table based on the 2009-2011 data, click here: Disability Data Dictionary 09-11.

2007-2011 CHAS (released May 2014):
For the data dictionary for the 2007-2011 data, click here: Data Dictionary 07-11.

Disability data table based on 2008-2010 CHAS (released May 2013):
For the data dictionary of disability table based on the 2008-2010 data, click here: Disability Data Dictionary 08-10.

2006-2010 CHAS (released May 2013):
For the data dictionary for the 2006-2010 data, click here: Data Dictionary 06-10.

2005-2009 CHAS (released January 2012):
For the data dictionary for the 2005-2009 data, click here: Data Dictionary 05-09.

2006-2008 CHAS (released December 2010):
For the data dictionary for the 2006-2008 data, click here: Data Dictionary 06-08.

2005-2007 CHAS (released December 2009):
For the data dictionary for the 2005-2007 data, click here: Data Dictionary 05-07.

Underlying Data:
The American Community Survey (ACS), from which the CHAS are now derived, has a smaller sample size than the Decennial Census (which was the basis of the 2000 CHAS). As a result, the Census Bureau cannot produce data using only one year of survey responses, except in very populous areas. For areas with population 65,000 or greater, ACS estimates are available each year using only the most recent year’s survey responses (known as "1-year data"). For areas with population 20,000 or greater, ACS estimates are available each year based on averages of the previous three years of survey responses ("3-year data"). For areas with population less than 20,000—including all census tracts, and many places, counties, and minor civil divisions—the only ACS estimates available are based on averages of the previous five years of survey responses ("5-year data").

HUD can purchase special tabulations of 1-year data, 3-year data, or 5-year data, subject to the same population thresholds. In 2009 and 2010, HUD only requested a special tabulation of 3-year data (2005-07 and 2006-08). In 2011, HUD requested a special tabulation of 5-year data (2005-2009). In future years, HUD expects to rotate to balance the timeliness of the data and its geographic precision.

Available Geographies:
HUD has obtained, or will obtain, the CHAS tabulations at the following geographic summary levels:

• 040 = State
• 050 = State - County
• 060 = State - County - County Subdivision
• 070 = State - County - County Subdivision - Place/Remainder
• 080 = State - County - County Subdivision - Place/Remainder - Tract
• 155 = State - Place - County
• 160 = State - Place
• 170 = State - Consolidated City

These are census definitions, which may be different than terminology used at the local level. For instance, St. Louis city and Baltimore city are treated as counties by the Census Bureau, and are geographically exclusive of their surrounding counties. Minor civil divisions are a census term for cities in certain states, while "place" is the Census designation for the vast majority of cities. For instance, Chicago is both a place and a minor civil division within Cook County; the CHAS data for Cook County includes Chicago households, but Chicago's data can be separately viewed in the MCD files or place files.

More explanation of Census geography is available here: Understanding Census Bureau Geography.

Other issues:
As with the CHAS 2000 and all other special tabulations of Census data, the Census Bureau requires that the CHAS data be rounded. The rounding scheme is as follows: 0 remains 0; 1-7 rounds to 4; 8 or greater rounds to nearest multiple of 5. This causes discrepancies when adding up smaller geographies and when adding up data within CHAS tables. Consider a city where the CHAS data indicate that there were 4 renter households with extremely low income and 4 owner households with extremely low income. One might be tempted to conclude that there are 8 total households with extremely low income. If another CHAS table indicates that there are actually a total of 15 extremely low income households, that would appear to be contradictory. This situation is the result of rounding. The city could have 6 renter households with extremely low income and 7 owner households with extremely low income, which is a total of 13 extremely low income households; but all of these numbers would be rounded, to 4, 4, and 15. As a result, HUD advises: 1. Use the largest geographies possible (rather than adding up smaller units of geography); 2. Use the total and subtotals published in the CHAS data (rather than creating your own totals and subtotals); and 3. If you must create a derived estimate by adding multiple CHAS estimates, understand that rounding will cause the resulting number to be less accurate.