TPP OMB ICR Supporting Statement - Part B (1-24-18)

SUPPORTING STATEMENT - PART B:

COLLECTION OF INFORMATION EMPLOYING STATISTICAL METHODS

OMB Control Number 0584-NEW

Study of Third-Party Processor Services, Fees, and Business Practices

Contract Officer Representative: Rosemarie Downer

USDA, Food and Nutrition Service

3101 Park Center Drive

Alexandria, Virginia 22302

December 2017

Contents

B1. RESPONDENT UNIVERSE AND SAMPLING METHODS 1

B2. PROCEDURES FOR THE COLLECTION OF INFORMATION 9

B3. METHODS TO MAXIMIZE THE RESPONSE RATES AND TO DEAL WITH NONRESPONSE 14

B4. TEST OF PROCEDURES OR METHODS TO BE UNDERTAKEN 16

B5. INDIVIDUALS CONSULTED ON STATISTICAL ASPECTS & INDIVIDUALS COLLECTING AND/OR ANALYZING DATA 18

Tables

Table 1: Sampling Methodology Overview for SNAP Retailers, TPPS and ISOs. 2

Table 2: Sampling Frame Characteristics- Affiliation to a Chain, Urbanicity, and Sales 4

Figures

Figure 1: Process to Identify, Validate, Recruit and Interview ISOs 6

Figure 2: SNAP Retailer Satisfaction Survey Data Collection Overview 8

Attachments

Appendix A: Study Objectives and Research Questions

Appendix B: Legal AuthoritY

APPENDIX C: SNAP RETAILER SURVEY SCREENSHOTS

APPENDIX D: Data collection instruments- pRETEST results

Appendix E: Data Confidentiality Agreement

Appendix F: SNAP RETAILER Survey Invitation Letter

Appendix G: SNAP RETAILER Survey Invitation Letter- 2^nd CONTACT

APPENDIX H: SNAP RETAILER SURVEY invitation PHONE script- 3^rd CONTACT

Appendix I: SNAP RETAILER Survey in English

Appendix J: SNAP RETAILER Survey in Spanish

Appendix K: EBT Vendor Interview Invitation Letter

Appendix L: EBT Vendor Interview Guide

APPENDIX M: Estimated Total Annual Burden Table (Excel)

APPENDIX N: Estimated Cost of Time burden table (Excel)

PART B: Collection of Information Employing Statistical Methods

B1. RESPONDENT UNIVERSE AND SAMPLING METHODS

Describe (including a numerical estimate) the potential respondent universe and any sampling or other respondent selection method to be used. Data on the number of entities (e.g., establishments, State and local government units, households, or persons) in the universe covered by the collection and in the corresponding sample are to be provided in tabular form for the universe as a whole and for each of the strata in the proposed sample. Indicate expected response rates for the collection as a whole. If the collection had been conducted previously, include the actual response rate achieved during the last collection.

The goal of the Study of Third-Party Processor Services, Fees, and Business Practices is to understand the business practices of third-party processors (TPPs) and independent sales organizations (ISOs) that provide Electronic Benefits Transfer (EBT) processing services and equipment to Supplemental Nutrition Assistance Program (SNAP) authorized retailers to: (1) describe the contractual agreements between SNAP retailers and ISOs and with TPPs and (2) describe the SNAP retailers’ level of satisfaction with the ISOs and TPPs. The study results will provide the U.S. Department of Agriculture’s (USDA) Food and Nutrition Service (FNS) with the information needed to inform future FNS policies regarding requirements for vendors providing EBT equipment and services to SNAP authorized retailers and the TPP services-related guidance for retailers.

The study will collect information from two groups of respondents: (1) SNAP-authorized retailers through a SNAP Retailer Survey (Appendix C, I, J) and (2) SNAP vendors - TPPs, and ISOs that sell EBT services or equipment to the SNAP-authorized retailers through an EBT Vendor Interview Guide (Appendix L).

Table 1 details the parameters of the sample used for each target audience: SNAP retailers, TPPs, and ISOs.

Table 1: Sampling Methodology Overview for SNAP Retailers, TPPS and ISOs.

N (Universe)	~250,000	5–10	n/a*
Sample Size:	1,500	5–10	50–55
Expected # of Respondents:	1,200	5–10	40–45
Expected Response Rate:	80%	100%	80%
Sample Data Source:	FNS	FNS, Primary Research	Primary and Secondary Research
Duration:	24 minutes	93 minutes	93 minutes
Data Collection Method:	Online or interactive voice response (IVR). Telephone interview for follow-up with nonrespondents	Telephone interview	Telephone interview

* The research team will use snowball sampling to identify ISOs until a maximum of 45 ISO interviews are achieved.

The SNAP Retailer Survey will focus on the nature of SNAP retailers’ relationships with TPPs and ISOs and their satisfaction with the products and services received, including exploration of the factors and decision points related to procurement of products and services needed to accept EBT payments. The researchers will select a sample of 1,500 non-exempt retailers¹ from among approximately 250,000 SNAP retailers² in the universe. Selected retailers will be active SNAP-authorized retailers that support EBT transaction processing using either a standalone EBT device or through their credit/debit card processing equipment.

The researchers performed a preliminary analysis of the SNAP retailers based on the data sets provided by FNS. The goals of the analysis were to: (1) assess the quality of the SNAP retailer data available on the FNS Store Tracking and Redemption System (STARS), particularly regarding the indicators needed for sample stratification: sales volume and store location (rural/urban); (2) determine the sample frame; and (3) ensure the data available provide the appropriate contact information on owners (or decision makers) for retailers in the sampling frame.

An analysis of the distribution of annual retail sales indicates that the researchers can classify stores, using the Food Stamp Program (FSP) Entity ID,³ into two categories: less than $1.5 million (52.8%), and $1.5 million or more (47.2%). The researchers also assessed the stores’ affiliation to a chain based on the presence of a “store chain number” in the stores data sets received from FNS: 54.3% of stores have a store chain number and were classified as being part of a chain. However, a quick review of the data sets indicated an underestimation of their incidence, considering that a few stores with “Walmart”, “CVS”, or “7-Eleven” in their name do not have a store chain number. Lastly, the researchers used the store location’s zip code to classify each store as rural or urban: 18.3% of the stores are in non-metropolitan areas (referred to herein as rural) and 81.7% are in metropolitan areas (or urban).

Considering that stores are our primary sampling units, the researchers can use the distribution of stores based on their affiliation to a chain, location, and annual sales to stratify the population for sampling purposes. Table 2 provides a summary of the sampling frame.

Table 2: Sampling Frame Characteristics- Affiliation to a Chain, Urbanicity, and Sales

Sales Revenue	Not Affiliated to a Chain		Affiliated to a Chain
	Rural	Urban	Rural	Urban
Sales < $1.5M	13,184	71,366	10,567	35,985
	66.9%	76.2%	41.0%	33.0%
Sales >= $1.5M	6,535	22,232	15,214	73,105
	33.1%	23.8%	59.0%	67.0%
Total	19,719	93,598	25,781	109,090
	100.0%	100.0%	100.0%	100.0%

To ensure 1,200 complete responses at the envisioned 80 percent response rate, the researchers will pull a sample of 1,500 stores from among the 248,188 qualified. The researchers will use non-proportionate stratification, under-sampling chain stores to account for their underestimation in the universe of stores. The researchers will consider resampling to assess the incidence of these stores before pulling the final sample for the survey. The objective is to ensure that the distribution of stores in the sample from the eight groups reflected in the table above is similar to the distribution in the general population.

Resampling here means pulling, with replacement, a random sample multiple times in order to estimate the true proportion of chain stores in the universe of 248,188 qualified stores. The researchers will pull a random sample of stores from the database and calculate the proportion of chain stores (including chain stores without chain store number) in that sample. The researchers will then “put that sample back in the database”.  We will pull another random sample from the database and calculate the proportion of chain stores (including chain stores without chain store number), then put that sample back in the database. We will repeat the process several times. The distribution of the proportion of chain stores, calculated from the different samples, will be used to determine the true proportion of chain stores in the target population with a given confidence level. This will help mitigate the risk of non-representativeness of chain stores in the survey sample, a non-representativeness that can occur because not all chain stores in the target population can be identified as such.

In effect, stores that are affiliated to a chain but not identified with a chain number will be sampled in the non-chain stratum; they will then be manually identified and reclassified as chain stores (the limited sample size makes that possible versus working on the total population), increasing the sample size of the chain stratum. Therefore, the non-proportionate stratification recommended here will mitigate that risk, especially for non-chain stores in small strata (e.g., the stratum non-affiliated, rural, sales >= $1.5M in Table 2 above).

The EBT vendor interviews in this study include TPPs and ISOs. The researchers define TPPs as entities that connect directly to the State EBT processor or an EBT gateway. Thus, to identify the TPPs in operation, the researchers will contact the State EBT processors (FIS, Conduent (formerly Xerox), and Solutran) to provide a list of TPPs for the study. While there may be some fluctuations in the number of TPPs over time, FNS has indicated that there are between 5 and 10 entities with direct access to State EBT processors. The combination of all TPPs identified by the three State EBT processors will constitute the TPP interview universe. The researchers will conduct a census of all TPPs.

The researchers define ISOs as entities that provide direct and indirect services to retailers seeking to process EBT payments. To identify ISOs for the study, the researchers will use a snowball approach. Snowball sampling is a non-probability sampling technique. It operates as a participant recruitment method in which research participants are asked to assist in identifying other potential respondents. Snowball approaches are recommended to access “hidden” or unknown populations. ⁴ In this study, the number and identity of all entities serving in an ISO role is not known. Figure 1 depicts the process to identify, validate, and recruit ISOs.

Figure 1: Process to Identify, Validate, Recruit and Interview ISOs

As a starting point, the researchers will ask TPP participants to indicate ISOs they interface with to process EBT payments. The researchers will note these organizations, validate the names, and contact them to conduct ISO interviews. In the ISO interviews, the researchers will ask participants if they work directly with SNAP retailers or if their services and equipment may be offered to retailers using another organization. By doing so, the researchers will identify additional ISOs and contact them for interviews. Alternatively, if an ISO works directly between the retailer and a TPP only, that interview will not yield additional ISOs.

The researchers will identify ISOs to compose the study sample frame. Including the small number of TPPs (ranging from 5 to 10 organizations), the study universe for the EBT vendor interviews will a total of 55–65 TPPs and ISOs. Because the ISO interviews will be completed after the TPP interviews, the sample of ISOs will be the total number of interviews planned for the study (N=55–65) less the number of TPP interviews completed. When that number of ISO interviews is reached (ranging from 50 to 55), data collection from the ISOs will stop.

The researchers assembled an initial list of ISOs using information publicly available on FNS’s website, internal team members’ relationships, secondary research, and other sources such as State agencies or State EBT processors. The researchers will use this list to select participants for the ISO instrument pretesting and to cross-check and possibly supplement the sample achieved using the snowball sampling technique.

B2. PROCEDURES FOR THE COLLECTION OF INFORMATION

Describe the procedures for the collection of information including:

• Statistical methodology for stratification and sample selection,

• Estimation procedure,

• Degree of accuracy needed for the purpose described in the justification,

• Unusual problems requiring specialized sampling procedures, and

• Any use of periodic (less frequent than annual) data collection cycles to reduce burden.

A survey of the SNAP retailers and the EBT vendor interview are a one-time data collection effort. The procedure for conducting the EBT vendor interviews involves the research team (FNS and or its contractor) contacting EBT vendors to schedule a phone interview. These participants will receive via email an EBT Vendor Interview Guide (Appendix L) and an EBT Vendor Interview Invitation Letter (Appendix K) describing the purpose of the study and how the study team will use the data. The team will record and transcribe the interviews for analytical purposes. The research team will write a summary of each interview, which will serve as the foundation for the data analysis and final report. The EBT vendor interviews will yield qualitative data.

The information gleaned from these interviews will be used to describe the relationships between different types of ISOs and TPPs and how these relationships affect the products and services offered to retailers. This information cannot be obtained from the retailers, as they may not know how the services they procure connect them to EBT. The organizations interfacing with SNAP retailers to provide products and services related to EBT will be identified using a nationally representative survey of SNAP retailers.

The SNAP Retailer Survey will use statistical techniques. After the study team receives Office of Management and Budget (OMB) approval, the researchers will work with FNS data to identify retailers active in SNAP at that point in time. Figure 2 depicts the steps for conducting the data collection process for the SNAP Retailer Survey.

Figure 2: SNAP Retailer Satisfaction Survey Data Collection Overview

STEP 1: Send Letter Inviting Retailers to Participate in Online and IVR Survey. The SNAP Retailer Invitation Letter (Appendix F) will provide participants with information about the study. It will note USDA efforts to understand how the changes brought about by the 2014 Farm Bill impacted retailers’ ability to serve their SNAP customers and the importance of participation. The invitation letter will include information necessary for informed consent and will encourage participation. The invitation letter, in a window envelope for cost-effective personalization, will: (1) state the purpose of the study and provide persuasive rationale for participating in the study; (2) clearly indicate that the survey is sponsored by FNS to establish credibility and improve response rate (the letter will be printed on FNS letterhead or include the agency logo and be signed by the responsible FNS project official); (3) identify contractors with contact information; and (4) assure privacy for the respondents, explain how the study team will use the data, and provide an estimate of the response burden.

• Invitation to Online or IVR. The SNAP Retailer Invitation Letter (Appendix F) will present participants with three options for completing the survey: (1) a URL for the online questionnaire and individualized log-in credentials (user ID and password); (2) a QR code that participants can scan with a mobile device to enter and complete the survey online; and (3) an IVR with a toll-free number and individual passcode to enter the system.⁵ QR codes in particular can further facilitate participation rates by making access to the survey easy and immediate, as log-in credentials are built in for each individual. The IVR option enables data collection from respondents who may have spotty Internet coverage. Technically, completion of the survey is signified by clicking on the submit button at the end of the questionnaire. We will apply similar procedures to the IVR modality. Participants will receive recorded, pertinent informed consent information and will need to indicate agreement either orally or by entering a designated telephone key to proceed to the survey.

STEP 2: Mail Reminder to Nonresponders. Within 10 days of sending the invitation letter, we will prepare a mail reminder (Appendix G) to participants who did not start the survey. This time period accounts for three days for mail delivery and gives participants five days to complete the survey online. The content of the mail reminder will reflect that of the invitation letter.

STEP 3: Conduct Telephone Follow-Up with Nonresponders. Telephone follow-ups are a necessary step to maximize the response rate. A copy of the telephone script is available in Appendix H. The researchers will conduct a telephone follow-up only with participants who did not start an online or IVR survey beginning 10 business days after the reminder mail drop. This allows enough time for mail delivery and for participants to complete the survey online or through the IVR platform.⁶ Telephone follow-ups include five attempts to each number in the sample at different times of the day and days of the week from 9 a.m. through 9 p.m., with some customization by retailer type for times and to avoid busy periods. An attempt is defined as either leaving a message or nonresponse after 10 rings. The interviewers will record disconnected, changed, and fax numbers. If the changed number message provides an alternative number, we will use that number to contact the respondent. We will make five call attempts at the new number. Interviewers experienced in conducting business-to-business research will conduct follow-up calls. In addition to fully briefing the interviewers on the objectives of the study and administration of the questionnaire, the interviewer training will stress flexibility in scheduling interviews at the times that are convenient to the respondents.

We will analyze the survey of SNAP retailers’ data using statistical techniques. The survey seeks to provide a national picture of TPP and ISO products and services being procured by retailers, their selection of vendors, and satisfaction with services received. The study will test subgroup differences that are meaningful to the goals of the study. Subgroups of interest include store location (urban vs. rural), whether the store is part of a chain, and the size of the store as measured by revenues. To examine whether any observed differences between subgroups are statistically significant, the research team will conduct subgroup analyses of the survey data, using techniques appropriate given the variable types and sampling distribution, to test difference levels between two subgroups. These techniques may include t-test, Wilcoxon-Mann-Whitney test, ANOVA, chi-square, and the Kruskal-Wallis test to gauge differences in satisfaction levels across multiple groups (e.g., large, medium, and small retailers). Additionally, we will explore the effect of the interaction between retailer size and location on the variables of interest (e.g., how small retailers in rural areas differ from large retailers in urban/suburban areas in their satisfaction of TPPs and ISOs).

The study design for the SNAP Retailer Survey estimated the sample size needed to detect reasonable differences across subgroups if they exist. Assuming a survey response rate of 80 percent, the study needs to contact 1,500 retailers to obtain 1,200 completed responses. Of the 1,200 responding retailers, assuming one-third are in subgroup 1 (e.g., rural) and two-thirds are in subgroup 2 (e.g., urban), the minimal detectable effect size (MDES) will be 0.17 assuming a two-tail statistical test with an expected statistical power of 0.8. This means, with a sample of 1,200 retailers, the study will be able to detect a group difference of 0.17 standard deviation between the two subgroups if the sample size ratio of the two subgroups is 1:2. If the sample ratio of the subgroups is 1:1 (e.g., northern vs. southern states) or 1:3 (large vs. small retailers), the MDES will be 0.16 and 0.19, respectively.

In the SNAP Retailer Survey, one issue to consider is whether it is feasible to identify an owner or corporate officer (vs. store-level manager) as the prospective survey participant and direct the survey request to that person. This is not a simple process as the information may be missing from the FNS data set or outdated. As a result of this challenge and because chain stores decisions are made at a central office, the research team proposes the non-proportionate stratification sampling approach with an under-sampling of chain stores.

B3. METHODS TO MAXIMIZE THE RESPONSE RATES AND TO DEAL WITH NONRESPONSE

Describe methods to maximize response rates and to deal with issues of non-response. The accuracy and reliability of information collected must be shown to be adequate for intended uses. For collections based on sampling, a special justification must be provided for any collection that will not yield "reliable" data that can be generalized to the universe studied.

We are aware of the impending challenges in achieving an 80 percent response in surveys where the response is voluntary. Because people receive a significant amount of mail, telephone calls, and online requests, growing numbers of people react by ignoring all rather than risk responding to requests that are neither legitimate nor well intentioned. For smaller food retailers, these concerns are magnified because owners and managers are very busy managing the store. Thus, we will sample in two steps:

1. The sample will include the expected sample size of 1,500 retailers. If we meet the desired response rate and obtain the 1,200 respondents necessary to achieve the MDES, we will stop. In the event we do not reach the target sample size, we will access a second reserved replacement sample (developed following the same procedures from those not previously sampled).

2. The researchers will replace nonrespondents from those with like characteristics in the reserved sample and continue interviewing until the desired number of completes is achieved. The replacement sample will be selected from the total population of SNAP retailers excluding those selected in the initial sample. We will consider the same stratification used in the initial sampling as well as the response from each strata in pulling the replacement sample. For example, if the initial sample specifies 160-complete in Stratum # 1 yet we get only 120-complete, considering the response rate in that stratum with the initial sample, the size of the replacement sample for Stratum # 1 should be such to yield an additional 40-complete. In other words, we will replace nonrespondents from the initial sample with like-characteristics in the reserved sample and continue interviewing until we achieve the desired number of completes for each stratum.

• Encouraging response rates. The invitation letter to sampled retailers will emphasize that participants are randomly selected and thus are given an opportunity to represent their peers — in this case, other small businesses facing the same issues. The letter will emphasize the potential for respondents to impact future procedures. Lastly, the letter will appeal to social norms, stating that many of the respondents’ peers will take this opportunity to make their opinions heard. In our experience, personalization of the letter (i.e., addressing the letter to the individual selected rather than using a generic greeting) also contributes to one’s willingness to participate.

• Invitation signed by FNS. An invitation, signed by FNS, will also be mailed to SNAP retailers along with instructions on how to access the SNAP Retailer Survey. The invitation will note USDA efforts to understand how the changes brought about by the 2014 Farm Bill impact retailers’ ability to serve their SNAP customers. The researchers will draft the invitations to be approved and signed by FNS.

• Nonresponse bias analysis. If response rates fall below the minimum required 80 percent, the researchers will perform a nonresponse bias analysis as required by OMB. The nonresponse analysis will compare characteristics of responding and nonresponding retailers using retailer variables included in the database supplied by FNS to determine whether there are significant differences between the two groups that may impact the generalizability of findings. Variables of interest include, for example, store location in rural or urban areas, and store size in annual sales, number of sites, or number of registers of respondents. We should note that depending on the specific data provided by FNS, we may have to derive some variables (e.g., characteristics of the neighborhood where the store is located) by combining the FNS data with external data sources (e.g., Census data). We will then assess the type of nonresponse and determine the presence of nonresponse bias using chi-square tests, t-tests, analysis of variance, Bayesian analysis, or any other pertinent techniques. In the event of any nonresponse bias, the research team would weight survey respondents to mirror the target population. Data weighting will ensure that subgroups and characteristics of interest are properly represented in the survey results and that the distribution of the survey data is representative of the target population.

B4. TEST OF PROCEDURES OR METHODS TO BE UNDERTAKEN

Describe any tests of procedures or methods to be undertaken. Testing is encouraged as an effective means of refining collections of information to minimize burden and improve utility. Tests must be approved if they call for answers to identical questions from 10 or more respondents. A proposed test or set of tests may be submitted for approval separately or in combination with the main collection of information.

The contractor, on behalf of FNS, pretested the SNAP Retailer Survey and the EBT Vendor Interview Guide with seven retailers and two vendors, respectively. See Appendix D for a memorandum of the pretest data collection tools and results of the pretest analysis. The interviews were intended to focus on the following areas:

• Burden: Approximately how much time did it take the respondent to complete the survey?

• Comprehension: Does the respondent understand the questions being asked?

• Retrieval: Can the respondent recall or retrieve pertinent information needed for the answer or do they need to do further research to answer the questions?

• Judgement: Is the respondent motivated to fully respond to the questions?

• Response: Can the respondent provide the answer in the format requested?

• Branch Logic: Does the order and/or skip pattern of the questions flow properly?

Selection and Recruitment of Pretest Participants. In the SNAP Retailer Survey, the research team collaborated with a partner that provides consulting services to small retailers and convenience stores on point of sale systems, the researchers identified a set of criteria for recruiting retailers to participate in the survey pretest. The researchers determined that the pretest should include stores located in rural and urban areas. Additionally, the researchers indicated that small retailers and those affiliated with chains should also be represented in the pretest. The researchers used these criteria to identify retailers and contacted retailers in different parts of the country to determine their availability and willingness to participate in the pretest. After this initial contact, the researchers followed up with each retailer to provide the link to the survey and schedule the interview. The interviewer completed three pretest interviews via telephone and four pretest interviews in person.

Before interviewing the retailers, the researchers conducted initial cognitive interviews with internal SMEs who previously owned SNAP-authorized stores. Three former store owners completed the survey and participated in a cognitive interview. By having internal SMEs complete the survey and participate in a cognitive interview, the researchers augmented the number of pretest cases critical to refining the instrument without exceeding the nine interviews allowed by OMB.

In the EBT vendor interviews, the combined sample size for TPP and ISO vendors supporting SNAP retailers is 60. Preliminary research estimates that there are 5 to 10 TPPs and 50 to 55 ISOs supporting the EBT market. This small sample demanded an approach for pretesting that does not jeopardize the data collection by using too many potential respondents during the pretest. For this reason, researchers recruited two participants for the pretest interviews. The researchers selected two ISOs from the list of prescreened organizations provided in the study plan. Because the data collected in the pretest cannot be used in the final analysis, the researchers selected vendors with a smaller share of the market to pretest the EBT Vendor Interview Guide. Additionally, researchers interviewed one of the study’s SMEs, a former EBT operations lead for a large TPP. The SME helped the researchers perform a cognitive assessment of the EBT Vendor Interview Guide without impacting the OMB limit while maintaining the integrity of the sample size.

Pretest and Cognitive Interviews. The burden estimates for this study are developed based on feedback from participants in the pretest and cognitive interviews. Each interview lasted 30–60 minutes for the pretest with the retailers and 60–90 minutes for the pretest with the EBT vendors. During the interviews, the researchers used a semi-structured interview protocol and took notes on participants’ responses.

After the interviews, the researchers synthesized feedback from interviewees and conducted a thematic analysis of the responses to identify areas for improvement or clarification. The SNAP Retailer Survey (Appendix C, I, J) and EBT Vendor Interview Guide (Appendix L) were revised based on these recommendations.

B5. INDIVIDUALS CONSULTED ON STATISTICAL ASPECTS & INDIVIDUALS COLLECTING AND/OR ANALYZING DATA

Provide the name and telephone number of individuals consulted on statistical aspects of the design and the name of the agency unit, contractor(s), grantee(s), or other person(s) who will actually collect and/or analyze the information for the agency.

The following contractor staff designed the study and/or will collect and analyze information for the Food and Nutrition Service:

Daniel Geller, Ph.D., Director of Evaluation Services, Manhattan Strategy Group

Email: [email protected]

Office: 301-828-1348

Shirley Eng, M.S., Project Director, Manhattan Strategy Group

Email: [email protected]

Office: 917-859-8968

Marilia Mochel, A.B.D., Project Manager, Manhattan Strategy Group

Email: [email protected]

Office: 301-828-1512

Alex Majchrowicz, Agricultural Economist Food Assistance, Economic Research Service

Email: [email protected]

Office: 202-694-5355

Evan Schulz, Mathematical Statistician, Department of Agriculture, National Agricultural Statistics Service

Email: [email protected]

Office: 202-690-8640

The following FNS staff member is responsible for receiving and approving contract deliverables:

Rosemarie Downer, Ph.D., Social Science Researcher, SNAP Research and Analysis Division, Office of Policy Support, Food and Nutrition Service

Email: [email protected]

Office: 703.305.2129

1 Retailers that are not exempt from the 2014 Farm Bill mandate and must pay 100 percent for EBT equipment and services. These retailers include Convenience Stores (CS), Combination Grocery/Other (CO), Super Store, (SS), Supermarket (SM), Smaller Grocery Store (SG), Medium Grocery Store (MG), Large Grocery Store (LG), Meat/Poultry Specialty (ME), Bakery Specialty (BB), Seafood Specialty (SE), and Fruits/Vegetables Specialty (FV).

2 The number of SNAP-authorized retailers is provided by FNS.

3 Food Stamp Program Entity ID (FSP Entity ID) is a seven-digit number assigned to each retailer store location. For example, store XYZ has three locations. Each location has a unique FSP entity ID.

4 Snowball samples are traditionally seen as samples of convenience, where no statistical inference is possible (Goodman 1961; TenHouten 1992). In the last decade, researchers have shown that under certain conditions chain-referral sampling can be used for statistical inference with demonstrated success (Heckathorn 1997, Heckathorn 2002, Heckathorn and Jeffri 2001, Salganik and Heckathorn 2004, Heckathorn 2004, Salganik 2006, Volz and Heckathorn 2008, Heckathorn 2011). The research questions relying on information from the sample created using the chain-referral (snowball) do not require statistical inference regarding the prevalence of ISOs, which is gained from the national survey of SNAP retailers. While we will not be making statistical claims based on the interviews, SMEs consulted for this project anticipate that there are 50-60 organizations in the United States that provide processing services with TPPs. Our study expects to interview a similar total number of organizations which will provide a comprehensive view of the EBT services market available to retailers.

5 We considered adding a paper and pencil option as well. However, in our analysis, the relatively small gain in response using paper surveys does not justify the added costs of printing and processing returned surveys.

6 The IVR system will record open-ended responses as voice messages. Voices messages will be transcribed to text for analysis.