Supplementary Supporting Statement for FSS Split Panel Test Part A

SUPPLEMENTAL SUPPORTING STATEMENT Part A

THE FOOD SECURITY SUPPLEMENT

TO THE CURRENT POPULATION SURVEY

Substantive Change

OMB Control Number 0536-0043

April 13, 2020

1. JUSTIFICATION

1. Circumstances making collection of information necessary.

USDA’s Economic Research Service (ERS) is requesting a substantive change to the Food Security Supplement to the Current Population Survey (CPS) for a one-time split panel test to assess new survey items. The Food Security Supplement (FSS) has been used annually to collect data on food security, food spending, and participation in Federal and community nutrition assistance programs since 1995, with the latest data collection in December 2019. The survey questions have been largely unchanged in those 25 years. In an effort to ensure the relevance of the survey questions ERS contracted with the U.S. Census Bureau to review the current FSS, develop changes, and conduct cognitive testing of the revised survey instrument. To begin, ERS researchers undertook a review of the survey items in collaboration with research staff at USDA’s Food and Nutrition Service (FNS) and survey methods staff at the Census Bureau’s Center for Behavior Science Methods in the Research and Methodology Directorate. ERS also considered previous comments about the survey received from academic food security experts. Through this review, some wording used in the survey items was perceived to be outdated (for example, questions asking about food spending referred to spending at produce stands but did not ask about spending at farmers markets or online food purchases). Thus, some survey items were revised. Some new questions (regarding food assistance) were added. In addition, reordering of questions was proposed. The food security items were moved earlier in the survey interview with a possible effect of reducing non-response to the food security items from break-offs part way through the survey. A new questionnaire was developed for testing that incorporated revisions to some of the survey items and a reordering of survey items (Attachment A). The revised questionnaire went through three rounds of cognitive testing led by the U.S. Census Bureau in 2019. The findings and recommendations from the cognitive test are included in Attachment D.

ERS is requesting to add a one-time split panel test to assess the revised FSS survey that resulted from the review and cognitive testing process. This split panel test will be a separate data collection from the regular annual FSS data collection used to monitor the Nation’s food security. We are proposing to add a split panel test because we do not want the test instrument to impact annual food security estimates in such a way that would cause an unnecessary break in the long time series. The data from this one-time split panel test will be used to assess the functioning of the redesigned survey and the effect of revising some survey items on key outcomes of interest, like the measure of food security. We will use the data and analysis in planning for how to revise and implement changes in the regular annual food security supplement collection. We do not plan to use the one-time split panel test data for regular monitoring of food security prevalence.

2. How, by whom, and for what purpose information is to be used.

The data collected in the split panel test will be used by ERS to assess the statistical impact of revising the survey instrument. The food security measure is based on a Rasch measurement model. ERS researchers will conduct psychometric analyses to determine if the food security data collected with the test instrument continues to conform to the assumptions and measurement properties of the measurement model. ERS will also compare the prevalence and severity of food insecurity between the half of the sample receiving the test instrument and the half of the sample receiving the original instrument. Other survey items on food spending and use of nutrition assistance programs will also be examined to understand the extent to which changes in the survey impacted the responses. Directions of changes will be considered. Some changes in responses will be expected when individual survey items changed substantially. In addition to analyzing the differences in the survey instruments, ERS may also use the split panel test data to answer other research questions.

3. Use of improved information technology.

There are no changes in the use of information technology from the original approval. The split panel data will be collected with Computer Assisted Personal Interviewing (CAPI) and Computer Assisted Telephone Interviewing (CATI), just as other administrations of the annual FSS have been collected.

4. Efforts to identify duplication.

There are no changes from the original approval related to duplication reduction. The split panel data collection is a one-time data collection that does not precisely mirror any other data collections. Evaluation of the revised survey is not possible with any existing data.

5. Methods to minimize burden of small businesses.

There are no changes from the original approval. The collection does not involve small businesses or other small entities.

6. Consequences if information collection were less frequent.

The requested change is for a one-time split panel test data collection. If this one-time data collection is not approved, ERS will be unable to test the revised instrument. Not testing the instrument could result in either not updating the instrument for fear of negative impacts on data collection or fielding the revised instrument with no way to fully assess the impacts of the change of the instrument on data collection and the resulting food security estimates.

7. Special circumstances.

There are no changes from the original approval.

8. Federal register notice and consultation with outside persons.

Notice of Intent to Request Revision of the Current Population Survey Food Security Supplement – A Currently Approved Information Collection was published in the Federal Register on February 10, 2020 (Volume 85, Number 27, Pages 7529-7530). A copy of the notice is included as Attachment B. Two comments were received in response to the Federal Register Notice and are included as Attachments C-1 and C-2 with the ERS reply included in each attachment. Mark Rifkin, of the Academy of Nutrition and Dietetics, contacted ERS Social Science Analyst Alisha Coleman-Jensen, via phone to ask about the content of the revised questionnaire and purpose of the test (Attachment C-1). Coleman-Jensen explained the reason for the test and gave an overview of changes. She followed-up via email and provided the test questionnaire (Attachment A) to Mr. Rifkin. Marie Gualtieri, a Health and Aging Policy Fellow with the American Political Science Association Congressional Fellows for Representative Rosa DeLauro, contacted ERS Social Science Analyst Alisha Coleman-Jensen, via email to request a phone call about USDA’s household food security questionnaire (Attachment C-2). Coleman-Jensen set up a call with Gualtieri and a member of USDA’s Office of Congressional Relations. On the call Gualtieri indicated she had seen the Federal Register Notice and asked Coleman-Jensen to describe what would be tested and the purpose of the test. Gualtieri followed up with a letter from Representative DeLauro. ERS provided a response letter from Marca Weinberg (acting administrator for ERS). As explained in the letter, the issues raised by Rep. DeLauro are important and point to potential future research areas but are not directly relevant to the content of this test. ERS has already addressed some of the issues raised by Rep. DeLauro with other research, as noted in the ERS response.

Over the years, many experts from a broad range of disciplines and institutions have been consulted on food security measurement and the FSS. Some concerns from those earlier reviews have been incorporated into this test instrument. For example, a professor, and user of the CPS-FSS, raised a question previously as to whether responses to the food security items may be influenced by an individual’s responses to the questions about participation in Federal nutrition assistance programs that occurred just before the food security items in the original instrument. In the test instrument we move the food security items before the questions about Federal nutrition assistance in response to this query and also to possibly reduce non-response to the food security items.

In developing this new test instrument, ERS research staff consulted and collaborated with multiple staff at USDA’s Food and Nutrition Service and the U.S. Census Bureau. Contact information for a point person at each agency is provided below:

Kathleen Kephart, Survey Statistician

Center for Behavioral Science Methods

Research and Methodology Directorate

U.S. Census Bureau

Phone: 301-763-8891

Email: [email protected]

Anita Singh, Ph.D., Chief, SNAP Evaluation Branch

Office of Policy Support

Food and Nutrition Service, U.S. Dept. of Agriculture

Phone: 703-305-2152

Email: [email protected]

9. Payments or gifts to respondents.

There are no changes from the original approval.

10. Confidentiality provided to respondents.

There are no changes from the original approval.

11. Questions of a sensitive nature.

There are no changes from the original approval.

12. Provide estimates of the hour burden of the collection of information. The statement should indicate the number of respondents, frequency of response, annual hour burden, and an explanation of how the burden was estimated.

An estimated 39,000 households will be interviewed for this one-time only split panel test. This estimated total sample of 39,000 is based on using of the total CPS sample for the test. Census estimates 52,000 households per monthly sample of the CPS. We are proposing to use three-quarters of the regular monthly CPS sample for the test. These households will be interviewed only once for food security data collection. We are proposing to use only three-quarters of the CPS sample because we do not want to interview households in the test that may also be interviewed in the regular FSS. Pending approval, we are planning to conduct the split panel test in September 2020. CPS months-in-sample groups 2, 3, 4, 6, 7, and 8 will be interviewed. We do not include CPS months-in-sample 1 and 5 (one-quarter of the total sample) because they will be interviewed in the regular December 2020 FSS data collection.

The estimated one-time respondent burden for the total sample is 4,729 hours; an average respondent burden of 7.3 minutes for each of the 39,000 households expected to be in the Supplement universe. The estimate assumes an 80 percent response rate to the supplement and is based on the average proportion of sampled households that were asked each question in recent survey years and typical reading and response times for the questions.

An estimated 31,200 households will complete the split panel test (respondents). The average respondent burden is 8.8 minutes for each of the split panel completions. This estimate takes into account screening within the supplement and the average proportion of respondent households that answered each section based on their screening status and presence of children. Half of the 31,200 households will receive the original instrument and half will receive the test instrument. Burden estimates were calculated separately for the original and test instrument based on the number of questions, and the total burden represents the average of the two. The burden estimates also account for screening within the instrument based on income and reports of food hardship, and presence of children.

An estimated 7,800 households will not complete the supplement (non-respondents). The average burden is 1 minute for supplement non-respondents. This estimate assumes that non-respondents only hear the one-minute introduction to the supplement.

The table below provides a breakdown of the number of households expected to complete different parts of the split panel test questionnaire and average and total response time for households in each category based on recent CPS-FSS data collections.

Therefore, during the one year when both regular CPS-FSS data collection and split panel test will be conducted, an estimated 92,802 households will be interviewed and the estimated annual respondent burden for the total sample is 11,194 hours.

In this one-time split panel, one respondent will answer for the entire household. No cost other than the respondents’ time is incurred. The cost of the respondents’ time spent in answering the Supplement questions is estimated to be $112,881.23. Cost of respondents’ time is estimated based on the average hourly earnings of production and non-supervisory private-sector employees ($23.87 per hour in January 2020, as estimated by the U.S. Bureau of Labor Statistics; https://data.bls.gov/timeseries/CES0500000008) multiplied by the total response time for all respondents and non-respondents (4,729 hours).

Part A, Question 12, Projected Buren for Split Panel Test
Current Population Survey Split Panel Test of Revised Food Security Supplement Questionnaire (Burden estimates for three-quarter sample of CPS)	Sample Size	Freq.	Responses				Non-response				Total Burden Hours
			Resp. Count	Freq. x Count	Min./ Resp.	Burden Hours	Nonresp. Count	Freq. x Count	Min./ Nonr.	Burden Hours
Split Panel Test Total Sample (Respondents and Nonrespondents)	39,000	1	31,200	31,200	8.844	4,599	7,800	7,800	1	130	4,729 hours (7.3 minutes per household)
Detailed Burden Estimates by Screening Status of Respondents:
1a (Original instrument). Higher-income households screened out after food spending questions (average 12 questions)			8,502	8,502	6	850
1b (Test instrument). Higher-income households screened out after food spending questions (average 12 questions)			8,502	8,502	6	850
2a (Original instrument). Households with children and either lower-income or screened in after food spending questions (average 33 questions)			2,537	2,537	14	592
2b (Test instrument). Households with children and either lower-income or screened in after food spending questions (average 35 questions)			2,537	2,537	15	634
3a (Original Instrument). Households with no child and either lower-income or screened in after food spending questions (average 25 questions)			4,561	4,561	11	836
3b (Test Instrument). Households with no child and either lower-income or screened in after food spending questions (average 26 questions)			4,561	4,561	11	836
Total Split Panel Test Respondents			31,200			4,599

13. Provide an estimate of the total annual cost burden to respondents or record keepers resulting from the collection of information.

There is no cost burden to respondents.

14. Annualized costs to Federal government.

The one-time split panel test will cost ERS $705,000.

15. Reasons for changes in burden.

The total burden for the one-time split panel test (4,729 hours) is lower than the burden for the regular annual food security supplement (6,465 hours) because the split panel test will only use three-quarters of the CPS sample. The minutes per respondent for the split panel test (7.3 minutes) is up slightly from the regular annual food security supplement (7.2 minutes) because the test instrument will test some additional new survey items.

16. Tabulation, analysis, and publication plans.

The data from this one-time split panel test will be used to assess the functioning of the redesigned survey and the effect of revising some survey items on key outcomes of interest, like the measure of food security. We will compare survey responses between the two panels, consider differences in item non-response between the two panels, and conduct Rasch analysis of the food security scale in the test instrument to ensure the items continue to fit the measurement model. We plan to publish the results of the analysis of the split panel data in an ERS technical bulletin. We will use the data and analysis in planning for how to revise and implement changes in the regular annual food security supplement collection. We do not plan to use the one-time split panel test data for regular monitoring of food security prevalence. We will make the one-time split panel test data publicly available.

17. Request for approval of non-display of expiration date.

There are no changes from the original approval.

18. Exceptions to certification statement.

There are no changes from the original approval.