The AHS is a longitudinal housing unit survey that asks questions about the quality of housing in the United States. Returning to the same housing units every other year to gather data until a new sample of housing units is drawn, the AHS allows users the unique opportunity to analyze housing and household changes over time. In gathering information, Census Bureau interviewers visit or telephone the household occupying each housing unit in the sample. For unoccupied units, they obtain information from landlords, rental agents, or neighbors. The survey is redesigned from time to time to make sure it meets current needs and new topics are introduced for specific survey years.

The universe of interest for the AHS consists of the residential housing units in the United States that exist at the time the survey is conducted. The universe includes both occupied and vacant units but excludes group quarters, businesses, hotels, and motels. Geographically, the survey covers 50 states and the District of Columbia.

Survey Design

The AHS was designed to include two samples, the National sample and the independent Metropolitan area sample. From 1973 to 2005, the AHS was two surveys conducted independently of one another. The National survey was enumerated every other odd-numbered year, while the Metropolitan survey occurred in selected areas on a rotating basis. Starting in 2007, the National and Metropolitan surveys were conducted in the same time-period to reduce costs.

The 2015 American Housing Survey underwent a major redesign – a new sample was redrawn for the first time since 1985 and new households were asked to participate in the survey, the questionnaire was redesigned, variables were dropped, added, or modified, recodes and imputation methods were streamlined, and the weighting methodology changed. As a result, tables were redesigned and some estimates became incomparable with previous years.

Geography

Overview

While national data are collected each survey year, typically no more than 30 metropolitan areas are sampled in any given survey year. Prior to 2015, the AHS sampled the Big 6 metropolitan areas (New York, Northern New Jersey, Philadelphia, Chicago, Detroit, and Los Angeles) at least every 4 years, with all other available metro areas surveyed more sporadically. Los Angeles was an exception, as it was scheduled to be sampled in 2009, but interviews were postponed until 2011 for budgetary reasons. From 2015 onward, however, the same most populous “Top 15” metropolitan areas will be oversampled in every survey cycle, as will an additional 10 metros (referred to as the ‘dynamic 10’). The additional 10 metros will be selected from a rotating sample of twenty other Metropolitan areas, which will be in sample every other survey year.

Comparability over time

Because geographic boundaries change over time, AHS metropolitan areas are not always consistent over time and do not always match OMB definitions. In the AHS Table Creator, detailed metropolitan areas names display below the table title to reduce confusion. When comparing a metropolitan area between different survey years, use these detailed metropolitan area names to determine if they are comparable. For PUF data, see below for more information:

2015 and Later

For 2015 and later surveys, the AHS metro area definitions match the official Office of Management and Budget (OMB) delineations. AHS metros match Core Based Statistical Areas (CBSA) definitions of that year by the same name. Starting with 2019, some AHS metro area definitions line up with OMB’s original 2010-based CBSA definitions, created in 2013, because OMB changed some CBSA definitions in late 2018. For more information, see https://www.census.gov/programs-surveys/metro-micro.html.

Crosswalk between 2013 and 2015

Due to changes in geographic boundaries, use caution when comparing 2015 metro estimates to prior years. Please see Summary of the Differences between the 2015 and 2013 AHS Metro Areas for more information.

1985 to 2013

For more on differences in metro area definitions from 1985 to 2013, see Metropolitan Area Histories and Geography, Public Use File: 1985- 2013.

Sample Selection and Size

Housing units participating in the AHS have been scientifically selected to represent all housing units in the United States. The same National sample of housing units is interviewed every two years until a new sample is selected. From 2015 onward, this includes the “Top 15” most populous metropolitan areas. A rotating sample of twenty other Metropolitan areas are in sample every four years, ten in each survey year (referred to as the “Dynamic 10”). The U.S. Census Bureau updates the sample by adding newly constructed housing units.

Each housing unit in the AHS national sample is weighted and represents other housing units in the United States. The weighting is designed to minimize sampling error and utilize independent estimates of occupied and vacant housing units. For information regarding the sample size and response rate, see Accuracy of the Data.

From 2015 onward, the AHS sample has been comprised of an integrated national longitudinal sample and an integrated metropolitan longitudinal sample. The integrated national longitudinal sample includes three parts: (1) national cases representative of the U.S. and 9 Census divisions, (2) an oversample of housing units from the Top 15 metropolitan areas, and (3) housing units that are part of a HUD-assisted renter oversample. The integrated metropolitan longitudinal sample represents 10 additional metropolitan areas.

Because of the split-sample approach described in the Survey Design section above, three separate weights were developed for the 2021 AHS:

One weight is applicable to all characteristics except those pertaining to the six topical modules mentioned on the About page,
A second “split sample” weight is applicable to the intent to move and housing search modules.
A third “split sample” weight is applicable to the pets, secondhand smoke, and wildfire risk topical modules.

Sampling Errors

The data in this report are subject to error from sampling and other causes, such as incomplete data and wrong answers. For a complete description of the types of errors and formulas for constructing confidence intervals see Accuracy of the Data.