FNS - FDPIR Study Response

July 15, 2013

Questions received from OMB July 1, 2013 on FDPIR:

  1)      A core research question is to understand why participation in the FDPIR has declined. Yet, the representative portion of this study does not include a) persons who are eligible but are not enrolled and b) persons who were previously enrolled and are no longer enrolled. So, I see a potentially big coverage issue. To some extent, these groups might be recruited in the discussion groups proposed, but that recruiting methodology is not explicit in the package and would not yield estimates that would be robust enough to be reported for policy purposes. I would want FNS to discuss this a bit more with us. They do have access to the universe of SNAP participants, and potentially persons who are eligible for FDPIR but choosing SNAP instead could be surveyed. Even better, ERS’ Food Aps survey is nationally representative and includes both SNAP participants and SNAP eligible participants who are not enrolled in SNAP. That is another potential resource to discuss.  Also, HUD is drawing a nationally representative survey of AIAN from tribal lists. They are developing a frame using USPS, tribal, and enumeration methods, and perhaps that could be instructive for FNS. Jennifer Stoloff is the project officer.

To address reasons for declining participation, we will focus on the combination of analysis of Census data, TRIM analysis, and qualitative information from FDPIR-eligible persons and administrative staff.

There is no single source of data that identifies the universe of individuals that are potentially eligible for SNAP.

For the discussion groups, we will work with Tribes to identify and recruit FDPIR participants and well as potentially eligible non-participants. This starts with a discussion with the FDPIR director (see Supporting Statement Part A, Attachment B4c: Site Visit Planning Call Script) to explain the purpose of the focus group and focus group logistics. Eligible non-participants will be identified with the assistance of FDPIR Directors and other Tribal programs and organizations that are likely to serve populations who are eligible for FDPIR. Examples include elder programs, Tribal WIC programs, Tribal TANF programs, and food banks. In the previous study that included discussion groups with eligible nonparticipants (Finegold et al., 2009), FDPIR Directors were in contact with elder programs and with nutritionists in other food assistance and health programs and were able to work with these programs to recruit discussion group participants. The research team will suggest a number of strategies for recruiting potentially eligible nonparticipants, such as providing a toll-free number for people to call if they would like to participate or if they have questions, and providing flyers that can be posted at various programs, food banks, etc.

Because households switch between FDIR and SNAP, distinctions between participants and nonparticipants are somewhat arbitrary and accurate only for a selected point in time. It is likely that some participating households have chosen not to participate in FDPIR at certain times, and vice versa. Hence, the survey questions asked of participating households and the discussion groups that include participants and nonparticipants will provide important information concerning the reasons for declining participation. The 2009 study cited above identified the following factors that influence participation choices: 1) ease in enrollment and continued participation; 2) milieu/cultural compatibility; 3) choice in food selection, 4) access to stores with competitive pricing and wide selection, with access including both distance and availability of transportation; and 5) flexibility/choice in food selection. These topics are included in the participant survey and in the discussion groups. The survey data and discussion groups will be particularly useful in understanding barriers to FDPIR and other food assistance program participation and reasons people chose to leave FDPIR.

In recent years, SNAP has employed broad-based categorical eligibility (i.e. higher income limit and no asset limit), and this may have attracted greater numbers of households eligible for FDPIR to participate in SNAP. On the other hand, changes among the eligible FDPIR participant population may have occurred. Secondary data sources will be used to consider these issues.

We will use TRIM3, a microsimulation model operated by the Urban Institute, to assess how changes in FDPIR policy, changes in household composition and characteristics, and economic factors may affect eligibility. We will use TRIM3 to model eligibility for FDPIR and SNAP for two years corresponding to relatively high and relatively low FDPIR enrollment. For each year, we will produce estimates for total eligibles for FDPIR only, SNAP only, and both FDPIR and SNAP and compare these numbers considering breaks by at least one relevant characteristic, to the extent possible based on sample size constraints. By drawing comparisons of eligibility in both programs over time, we hope to identify whether changes in program rules or characteristics of the population have had an impact on eligibility.

We will also use simulated SNAP benefit data from TRIM3 and assumptions based on the latest valuations for FDPIR benefits to draw comparisons over time. TRIM3 models the detailed rules for determining SNAP benefits taking into account state variations in rules. We will combine this with the latest information on the estimated retail value of the average FDPIR package to evaluate the likelihood that an eligible person chooses to participate in one program over the other based solely on the value of benefits offered by each. Comparing changes in this likelihood across years will provide insight into whether the incentive for participating in one program or the other over time based solely on the value of the benefits offered has changed.

Another question is whether FDPIR is used more by a subgroup in the population that has been declining.  We will use three sources of data to respond to this research question: survey data of current participants, data from the 2010 census, and data from the 2005-09 American Community Survey (ACS).  We will use the census and ACS to conduct analyses of the tribal areas in which FDPIR has been located over the past decade to determine what groups have experienced declines in their population shares. 

OMB suggests that there is access to the universe of SNAP participants from which to draw a sample of FDPIR eligible but SNAP participating households who could be surveyed to understand why they chose SNAP. However, the tribes do not administer SNAP; SNAP is administered by the States. The nationally representative sample of tribes in this study spans 12 States. Recruiting 12 States to provide confidential SNAP household identification information would be a daunting task. FDPIR tribal reservations do not fit precisely into the usual State geographical areas (counties) thereby requiring complex address searches to find tribal members in SNAP. In addition, the ideal sample would require SNAP participants prior to April 2009 when the ARRA SNAP maximum allotment increase was implemented. The declining FDPIR participation trend started before that SNAP benefit increase yet the predominant “common knowledge” reason for recent conversions is the increase. The mixed analyses described above attempt to look for those longer term reasons.

OMB mentions the ERS nationally representative FoodAPS survey. When consulted ERS explained that their survey is address-based and it is theoretically possible that FoodAPS includes households residing on tribal reservations. However, only 10 households in the sample of 4800 reported FDPIR receipt and ERS expects there is a very small sample of households located on reservations. It’s possible but likely complex to identify other sampled households living on reservations to aggregate household characteristics.

The HUD survey suggested by OMB seems a very promising survey with its reservation-based sample. By sorting on reservations running FDPIR the research could focus on low-income FDPIR eligible households. Unfortunately, the survey does not ask about SNAP or FDPIR to enable sorting the sampled households into the two groups of interest for the FDPIR Study.

2)      Food Aps also asks questions about food access; how does the content of that survey overlap/complement/inform this survey? FNS could speak to Mark Denbaly at ERS.

FoodAPS overlaps slightly with this FDPIR Study because it has 10 responding households who participate in FDPIR. Food access  will be measured more analytically in FoodAPS by using household address, store locations gathered through grocery receipts and examining store types and size in the surrounding area. What FoodAPS finds about shopping patterns in reservation-like areas may inform the FDPIR study. The FDPIR Study asks several questions in the participant, the FDPIR Director, and the Tribal Leader surveys regarding food store access and available transportation. However, the survey participants are probably much different because FoodAPS is nationally representative of all low-income households and the FDPIR Study limits its survey participants to those living on tribal reservations. Further, FDPIR provides commodities and access may focus more on picking up the commodities.

3)      A minor note, but sometimes 17 site visits are noted and sometimes 19 site visits are noted. Should this be adjusted?

There will be 17 site visits. The only time the number 19 is noted is in reference to 19 discussion groups, because 2 discussion groups will be held in each of the two largest of the 17 sites. No changes are required to Supporting Statement.

4)      The power analysis cited does not indicate if they would have sufficient sample for subgroup comparisons named on pages 10-11. Perhaps add? Question clarified as follows: Page 4 of Part B states, “This design allows for computation of national estimates of the characteristics of participating households as well as estimates for large subgroups, such as households with elderly participants.”  Also, the list on pages 10-11 of Part B suggests that comparisons may be drawn between sites that fall into various categories of, for example, program administration.  Are comparisons based on the bulleted list planned?

 

Page 4 of part B pertains to the sample design for the case record review and the household survey and refers to participant characteristics, such as age and household composition. The list on pages 10-11of part B is specifically a list of program characteristics that will be used to select a purposive sample of sites for the site visits so that the site visits can provide a picture of the variety of local experiences across the country. Our analysis by site characteristics is essentially qualitative, giving a sense of how particular characteristics may influence participation or participant satisfaction.

  5)      A. 9. The first sentence of the first paragraph reads as though the participants will receive $25 cash; however, later in the paragraph, there is clarification that the participants will receive a gift card with a value of $25.  We recommend clarifying this in the first sentence.

 We have revised the sentence to read as follows:

Each survey respondent and discussion group participant will receive cash, a gas voucher, or a store or VISA gift card valued at $25 as a token of our appreciation.