Final -2019 AHS - Supporting Statement B

Supporting Statement for Paperwork Reduction Act Submission

American Housing Survey

OMB Number 2528-0017

B. Collection of Information Employing Statistical Methods

1. Universe and Respondent Selection

The 2019 AHS sample will consist of 117,537 sample housing units. The table below describes the different parts of the sample.

Sample Part Description	Sample Size (number of housing units
National cases representative of the U.S.	34,906
Top 15 metro oversample that include approximately 3,000 units each. Some metros will have a bit less than 3,000 housing units while some will have a bit more than 3,000 housing units	46,154
Subsidized-renter oversample	5,185
10 Metropolitan areas that include approximately 3,000 units each. Most metropolitan areas have slightly more than 3,000 housing units due to the accumulation of the new growth of housing since 2015	31,292
TOTAL	117,537

Please refer to the attached 2019 AHS Sample Design and Weighting document for detailed statistical methodology.

The Housing Insecurity Research Module follow-on will draw upon the 117,537 sample housing units expected for the AHS regular production sample. The target sample size for the Housing Insecurity Research Module follow-on is 4,000 housing units. Sample size and eligibility criteria were designed to capture low-income households more likely to experience policy-relevant housing problems, account for differences in tenure distribution by income, account for variation in population size and tenure across Census divisions, and produce statistically-supported options for developing a housing insecurity index. The table below details the sample size targets.

	Tenure	Renter			Owner
	Poverty Threshold	0-2x	2-3x	Total	0-2x	2-3x	Total
Census Division	1	74	13	87	54	20	74
	2	177	30	207	133	48	181
	3	315	54	369	255	92	347
	4	98	17	115	74	27	101
	5	330	56	386	236	85	321
	6	123	21	144	79	29	108
	7	236	41	277	159	57	216
	8	140	24	164	107	39	146
	9	385	66	451	225	81	306
	Poverty x Tenure n=	1,878	322		1,322	478
	Poverty x Tenure %=	85%	15%		73%	27%
	Tenure n=			2,200			1,800
	Tenure %=			55%			45%
	Total n=						4,000

These sample targets are adjusted to account for Census Division sample sizes, and desired poverty and tenure splits. We expect that our sample targets by poverty level will be met. If during AHS regular data collection, we find some cases especially hard to recruit for this follow-on survey (for instance, owners with household incomes 0-2X the poverty rate in the Northeast), we will adjust the 0-2X and 2-3X split to recruit our target number of renters and owners in our sample by division (these targets are in the “Total” column).

2. Procedures for Collecting Information

We expect that a majority of regular AHS interviews will be conducted through personal visit. However, we will conduct a phone match to retrieve phone numbers that match the sample addresses and attempt to collect the data by telephone interview, when possible.

The Housing Insecurity Research Module follow-on data will be collected via telephone. Up to four telephone numbers will be collected during the regular AHS interview, which will be used to attempt to contact potential follow-on survey respondents.

The Housing Insecurity Research Module questionnaire will consist of four sections, a screener, plus three topical sections on affordability, stable occupancy, and decent and safe.

3. Methods to Maximize Response

Based on the 80.1 percent weighted response rate of the 2017 AHS (80.3 percent unweighted), the Census Bureau expects the 2019 AHS response rate to be 80 percent. If an occupant is reluctant to participate, the FR informs the regional office staff, who sends a follow-up letter explaining the survey in detail and urging the occupant's cooperation. A Census Bureau FR or his/her supervisor will contact the occupant again.

Census will begin contacting potential Housing Insecurity Research Module follow-on respondents within two to three weeks after completion of the regular AHS interview to increase the likelihood that the respondent who opted-in is still in the household and the telephone numbers provided are still valid. The Housing Insecurity Research Module follow-on survey will also utilize a $40 gift card incentive to help reach response targets.

Several research findings informed the decision to set the gift card amount at $40. In research on the Consumer Expenditure Survey, Goldenberg et al. (2009) found that $40 incentives were effective at increasing response rates and reducing field costs by reducing the number of contacts required to complete interviews. In research on the Survey of Income and Program Participation (SIPP), Creighton et al. (2007) found that $20 incentive worked with the lowest-income groups, but $40 incentives were more effective with groups with higher incomes. Our sample targets renters and owners with household incomes under 300% of the poverty level, with a majority under 200% of the poverty level. As our sample includes a range of lower incomes, the $40 incentive is an effective incentive level for our sample. Other research has found that there may be diminishing returns in the effectiveness of incentives at higher levels. Croft et al. (2007) found that $40 had the same effect on response rates as $70. Horrigan and Herz (2004) in research on the BLS American Time Use Survey, found that respondents reported that $20 was too low of an incentive and $50 was too high.

4. Testing Procedures

The Census Bureau conducted a test of the interview forms and procedures in September 1983 and again in 1994 and 1995. Based on the results of the tests, HUD made modifications and changes.

In 2004, HUD and the Census Bureau conducted a major review of the survey questions for neighborhood quality, income, utility costs, and renter subsidies, making changes to each to improve the quality of this information. In October 2004, we tested these changes in a live hothouse test held in Pittsburgh, Pennsylvania.

In 2010, questions in the updated Mortgage module were cognitively tested by the Center for Survey Measurement (CSM) at the Census Bureau. We determined that more work needed to be done on the module, so we reverted to the 2009 AHS version and flow of questions where the 2011 version fell short.

In 2014, HUD and the Census Bureau conducted an exhaustive review of the core questionnaire content. We revised questions, changed the order, and eliminated some questions based on nonresponse rates and respondent burden. Because of these changes, questions in some of the core modules, e.g., Breakdown, Neighborhood Quality, Equipment, Home Improvement, Recent Movers, Income, and Mortgage, needed to be tested for implementation in the 2017 questionnaire.

In 2018, along with minor revisions to the core content, questions in the new supplemental modules, Post-Secondary Education and Home Accessibility, were cognitively reviewed and tested by CSM. The Census Bureau and HUD reviewed and discussed all of CSM’s recommendations.

In 2018, the Center for Behavioral Science Methods (CBSM) at the Census Bureau cognitively reviewed and tested the Housing Insecurity Research Module questions to look for problems with proposed questions and to measure these new concepts. The Census Bureau and HUD reviewed and discussed all of CBSM’s recommendations.

5. Contacts for Statistical Aspects and Data Collection

HUD consulted the following individuals on the statistical data collection and analysis operation:

Tamara Cole

Associate Directorate for Demographic Programs

U.S. Census Bureau

(301) 763-4665

Statistical Design, Analysis, and Collection

Stephen Ash

Demographic Statistical Methods Division

U.S. Census Bureau

(301) 763-4294

Statistical Design

Attachments