|
|
small (250x250 max)
medium (500x500 max)
Large
Extra Large
Full Size
Full Resolution
|
|
oolt-A-ol s jfft.xifyyi USDA It of Agriculture Food and Nutrition Offloaof An*y*md EwfcwMon Synthesis Report for Food Stamp Trafficking Study May 1998 ^ rt-ottte ¥3 (L- Contract No.: MPR Reference No.: 53-3198-3-013 8171-091 FINAL REPORT FOR THE FOOD STAMP PARTICIPANT TRAFFICKING STUDY March 1998 Authors: Anne Ciemnecki Lara Hulsey James Ohls Irving Piliavin Mercer Sullivan Josh Rossol Submitted to: Office of Analysis and Evaluation USDA Food and Consumer Service 3101 Park Center Drive Alexandria, VA 22302 Project Officer: Sharron Cristofar, Program Analyst Submitted by: Mathematica Policy Research, Inc. P.O. Box 2393 Princeton, NJ 08543-2393 (609) 799-3535 Project Director: James Ohls Survey Director: Anne Ciemnecki USDA non-discrimination Statement: The U.S. Department of Agriculture (USDA prohibits discrimination in all its programs and activities on the basis ofrace, color, national origin, gender, religion, age, disability political beliefs, sexual orientation, or marital or family status. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA's TARGET Center at (202) 720-2600 (voice and TDD). To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room 326-W, Whitten Building, 14th and Independence Avenue, SW, Washington, DC 20250-9410 or call (202) 720-5964 (voice and TDD). USDA is an equal opportunity provider and employer. u ACKNOWLEDGMENTS Sharron Cristofar, the United States Department of Agriculture project officer for the study, has provided continuing important input throughout the design, survey, and analysis of the project. Useful input has also been obtained from her colleagues, Ted Macaluso, Steven Carlson, and Margaret Andrews. At Mathematica Policy Research, Inc., Linda Mendenko contributed extensively to every phase of the study and efficiently managed the field data collection. Lynn Mackenzie served as an effective field supervisor. John Burghardt reviewed the report and provided valuable suggestions and help with revisions. The text was much improved through the editing of Roy Grisham, Jane S. Miller, and Marjorie Mitchell. Cathy Harper, Cynthia Castro, Monica Capizzi. and Jennifer Baskwell produced the final document. We are indebted to two field ethnographers who collected valuable data for the project Steven Koester and Owen Murdoch, as well as to RIVA Market Research, Inc., for carrying out the post-survey focus groups. We also thank the survey respondents for their time in participating in the data collection. 111 M»»&Mf USDA UnM*d AfrieuNura Food and Consumer Service 3101 Park Center Drive Alexandria. VA 22302-1500 Final Report for the Food Stamp Participant Trafficking Study Jury 1998 Enclosed for your information is a copy of the "Final Report for the Food Stamp Participant Trafficking Study." Food stamp trafficking occurs when someone sells their coupons for cash at a discount, rather than using them to buy food. The Food and Nutrition Service (FNS) treats trafficking as a serious violation of program integrity both because it is illegal and because it undermines the nutritional goals of the program. FNS fights trafficking through undercover investigations of food retailers that are suspected of engaging in trafficking. The reason for this is that authorized food retailers are the only ones who can redeem coupons for cash from the Government. FNS also works with State Food Stamp Agencies, who are responsible for investigating trafficking by participants, and has received requests to conduct survey research studies on the trafficking behavior of participants. The enclosed report presents the results of a methodological test of whether it is possible to obtain reliable and valid information about participant food stamp trafficking through a survey of active participants. We reviewed procedures used in other survey research studies of illegal behavior and tested the approach which appeared to work best in those other surveys. However, we found that the method does not work reliably for food stamp trafficking. The report is a valuable methodological contribution that will be of interest to individuals who study program integrity and law enforcement issues. In addition to the feasibility test of survey research methods, the report presents results of focus groups with food stamp participants. The focus group results discuss the terminology participants use to refer to trafficking and suggest hypotheses that might be tested in future research. Ifyou have questions regarding this report, please contact Dr. Theodore F. Macaluso, Chief, Food Stamp Evaluation Research Branch, Office of Analysis, Nutrition and Evaluation at (703)305-2019. Enclosure AM EQUAL OPPORTUNITY EMPLOYER W CONTENTS Chapter Page EXECUTIVE SUMMARY ix I INTRODUCTION 1 A. LIMITATIONS OF THE STUDY 2 B. OVERVIEW OF REPORT 3 II DATA COLLECTION ACTIVITIES CONDUCTED FOR THE STUDY 5 A. PRE-SURVEY FOCUS GROUPS 5 1. Material Covered 5 2. Sampling and Recruiting 6 3. Fielding 6 B. THE PARTICIPANT SURVEY 7 1. Survey Content 7 2. Approach to Data Collection g 3. Sampling 10 4. Fielding Procedures 11 5. Fielding Results 12 C POST-SURVEY FOCUS GROUPS 12 1. Material Covered 12 2. Fielding 13 D. ETHNOGRAPHIC RESEARCH 14 1. Approach 14 2. Content 15 3. Fielding 15 CONTENTS (continued) Chapter Page HI ASSESSMENT OF WHETHER THE PARTICIPANT SURVEY ELICITED ACCURATE SELF-REPORTS OF TRAFFICKING 17 A. THE SURVEY SELF-REPORT DATA ON TRAFFICKING 18 B. COMPARISON TO OTHER AVAILABLE INFORMATION 20 1. Comparisons with Survey Data on Neighborhood Prevalence and of Trafficking by Friends 20 2. Pre-Survey Focus Groups 23 3. Post-Survey Focus Groups 23 4. An External Estimate of Trafficking Incidence 24 IV OTHER HYPOTHESES SUGGESTED BY STUDY RESULTS 25 1. Many Buyers in Food Stamp Trafficking Transactions May Buy the Coupons for Their Own Use; Sometimes, They Themselves May Be Food Stamp Program Participants 25 2. There Is No Clear "Profile" of Characteristics of People Who Traffic 26 3. "Middlemen" May Not Be a Major Factor in Trafficking 27 REFERENCES 31 APPENDIX A: SURVEY METHODOLOGY 33 APPENDIX B: FOOD STAMP SURVEY 55 vi TABLES Table Page 111.1 REPORTED PARTICIPATION IN TRAFFICKING 19 111.2 PERCEPTIONS ABOUT THE EASE AND PREVALENCE OF FOOD STAMP TRAFFICKING 21 A. 1 RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE AND ELIGIBILITY RATES, BY SITE 46 A.2 RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE ELIGIBILITY RATES, BY STRATUM 48 Vll /HI mmp« EXECUTIVE SUMMARY Exchanging food stamps for cash or for nonfood goods and services (food stamp trafficking) is a serious federal policy concern. Unauthorized use offood stamps weakens the ability of the Food Stamp Program (FSP) to accomplish its objective of encouraging nutritious food use by program participants, and trafficking undermines confidence in the program among the general population. Most investigations of food stamp trafficking focus on food retailers as they are the only ones who can obtain cash from the government for food stamps. A recent U.S. Department of Agriculture (USDA), Food and Nutrition Service (FNS) study of authorized food stores suggests that between three and four percent of food stamps are exchanged for cash or nonfood items (Macaluso 1995). Because of extensive anecdotal evidence suggesting that food stamp trafficking is common, the USDA contracted with Mathematica Policy Research, Inc. (MPR) to conduct an exploratory study of food stamp trafficking, focusing on the motivations and dynamics oftrafficking from the point ofview of FSP participants. This report summarizes the findings from that study. STUDY OBJECTIVES The present study was conceived as an exploratory research project, designed to obtain preliminary information about trafficking by food stamp participants and to examine the feasibility of further research. In that vein, one key objective of the project was to test the feasibility of obtaining accurate survey information from FSP participants about their trafficking experiences. A second goal was to obtain preliminary data on the characteristics and motivations of people who exchange their food stamps for cash or for goods, so as to help devise deterrence strategies that the government could employ to reduce trafficking. DATA COLLECTION Several different data collection activities were undertaken as part ofthe research: A series of focus groups was conducted with FSP participants in three metropolitan areas in different parts ofthe country. The goal was to obtain information that would help in designing a survey of participants, asking them about their attitudes toward trafficking, their trafficking behavior, and the dynamics of trafficking in their neighborhoods. The focus groups included an explicit discussion of whether the participants would be willing to share information about their trafficking behavior in an interview. A survey of720 FSP participants was conducted in those same three metropolitan areas and in nearby rural areas. This survey obtained information about respondent household characteristics, household income and expenditures, attitudes and opinions about ix trafficking; respondents' social support systems; and respondents' actual experiences with trafficking. One part of the interview tested an innovative approach to obtaining survey data about trafficking, under which respondents listened to questions on an audio tape and wrote down the answers on a one-page answer sheet. Based on a review of methods employed with similar difficult survey situations, this method had seemed the most promising available, because it had been used with apparent success in survey applications involving illegal behavior. It was hoped that by ensuring that the interviewer would not know their answers, respondents would answer honestly. Post-survey focus groups wen; conducted at two of the three survey sites with respondents in the survey. These focus groups discussed the survey findings (see below) with participants in an attempt to gain additional insight into the response patterns seen. Ethnographic research was conducted at two of the survey sites. In this work, researchers with extensive "street-lever contacts in low-income areas attempted to identify and interview buyers of food stamps, to discuss their experiences with trafficking and to obtain additional insight into the dynamics of trafficking and the motivations of food stamp sellers. SAMPLING In light of the exploratory nature of the research, we decided not to allocate the resources that would have been required to obtain a nationally representative sample of FSP participants. However, the three metropolitan areas selected for the field work were chosen through a combination of purposive and random methods, so that a reasonable cross-section of sites would be ensured in terms of area of the country, size of the metropolitan area, and use of electronic benefit transfer (EBT) food stamp issuance methods. For each of the three metropolitan areas, we chose four different survey locations to achieve a mix of poverty densities and urban/rural locations. We chose respondents randomly from among FSP participants living in those areas. FINDINGS CONCERNING THE FEASIBILITY OF OBTAINING ACCURATE SELF-REPORTS OF TRAFFICKING Despite success elsewhere, the method of using audio tapes to ask questions about food stamp trafficking does not appear to measure the phenomenon accurately. In the current survey, the weight of the evidence suggests that there was significant underreporting of trafficking behavior. Overall, approximately 5.1 percent of the survey respondents reported selling food stamps in the previous month. Together with survey data on the average amount of benefits sold per transaction, this implies that approximately two percent of dollar benefits were sold. While there is no "gold standard" comparison with which to test the validity of these estimates, they appear to be low when compared to the following sources: Respondents were asked a series of questions about how common they thought trafficking was in their neighborhoods. While the answers reflect only opinions and are difficult to quantify, the pattern of responses suggests a considerably higher prevalence of trafficking than the survey indicates. Similarly, during the pre-survey focus groups, participants were asked how common they believed trafficking was. The general discussion suggested greater incidence of trafficking than the survey answers showed. During the post-survey focus groups, which were conducted with survey respondents, the participants were asked whether they had felt "safe" in replying honestly to the questions on trafficking. Several reported feeling that being fully candid during the survey was unsafe. It was apparent that they had replied negatively to many questions when the truthful answer was affirmative. The survey-based estimates are lower than would be expected, according to the findings of a 1995 FNS analysis of trafficking. The FNS analysis, which did not include all forms of trafficking, estimated that at least 3.8 percent of coupons were trafficked in 1993, well above the estimates from the survey. None of these sources is conclusive. The totality of the evidence, however, suggests that it is quite likely that there is significant underreporting in the survey. HYPOTHESES ABOUT TRAFFICKING SUGGESTED BY THE STUDY The evidence from the focus groups, the survey, and the ethnographic research, taken together, gives rise to a number of interesting hypotheses about trafficking. Because ofthe limited scope of this preliminary study, with data collection in only three areas of the country, none of these findings can be viewed as conclusive. All, however, appear quite likely from the evidence generated by the study. Following are key hypotheses: Many buyers in food stamp trafficking transactions may purchase the a»tponsfor their owt use; sometimes, they themselves may be FSPparticipants. Participants in the pre-survey focus groups repeatedly voiced their belief that the buyer in a food stamp trafficking transaction is often a food stamp recipient, and that many buyers use the stamps themselves at the grocery store. Focus group members reported that, as a way of stretching their food budgets, they and their friends often bought f od stamps from people who wanted to sell them, with several viewing this practice as simply prudent household management. They also reported that it was common for a recipient to be both a buyer and a seller over the course of a month, perhaps selling coupons initially to obtain cash for some high-priority use and then, if cash became available later, using it to buy coupons at a discount and gain access to food more cheaply. XI There is no clear "profile "ofcharacteristics of people who traffic. During preliminary analysis of the survey data, we examined whether self-reports of trafficking were correlated with demographic, social, or economic characteristics of respondents. While some weak possible correlations were identified, no single characteristic or set of characteristics stood out as being highly predictive of trafficking behavior. To be sure, as noted earlier, there is evidence that trafficking was substantially underreported in the survey, and this weakens our ability to identify significant correlations, if in fact they exist. Electronic benefit transfer (EBT) may change the dynamics ofthe transaction. One of the three study sites had been issuing food benefits under an EBT system for several months prior to the study. A number of respondents at that site reported trafficking, and the focus group discussions yielded insights as to how this trafficking took place. The dynamics of the trafficking under EBT are apparently quite different from trafficking in food coupons. The buyer and seller reportedly often go to the store together, because the buyer wants to avoid (1) prepaying for an EBT card on which benefits have been depleted, and (2) prepaying for an EBT card that has been reported as lost or stolen. The general feeling about EBT as it relates to trafficking was that "where there is a will there is a way." Focus group respondents reported that the cycle with which benefits were posted to the EBT cards contributed to selling benefits. In the EBT site, the AFDC benefit was credited early in the month. At this time, recipients needed food and spent the welfare benefit on groceries. The food benefit was credited four or five days later. By the time the food benefit was credited, food had been purchased and recipients needed cash. "Middlemen ' may not be a majorfactor in trafficking. One issue of considerable interest in the current study was the role of "middlemen" who buy coupons and then sell them at a profit, either to people who plan to use them for food or to stores that can "launder" diem. It was believed at the jutset of the study that such middlemen might represent a significant share of buyers. However, no evidence emerged from the research to suggest this. During the pre-survey focus groups, participants were asked to talk about what categories of people bought food stamps. The group participants tended to focus on two types of buyers: (1) retail stores that cashed the coupons at banks, and (2) low-income people who intended to use the coupons to purchase food (as discussed in Section 1, above). The existence of middlemen who bought the coupons to make money by reselling them was seldom mentioned by the focus group participants, even after direct probing. Similar findings were obtained during the ethnographic research. The ethnographers probed extensively about middlemen and found virtually no evidence that they exist. None of the 10 respondents interviewed, all ofwhom had themselves been buyers of food stamps tot various reasons, reported ever having bought coupons and reselling them for cash. xn I. INTRODUCTION The Food Stamp Program (FSP) represents a key component of America's "safety net" for low-income households. With annual outlays in benefits of more than $19 billion in 1997, it has an average caseload of more than 22 million people each month. The program is thus an important source of support for America's poorest households. Furthermore, by distributing benefits either as coupons or through electronic benefit transfer (EBT) accounts, both of which can legally be used only to purchase food, the FSP helps assure that America's poor have access to nutritious meals. In light of the importance of the program, food stamp trafficking-particularly program participants exchanging food stamps for cash-represents a serious concern. Trafficking weakens the ability of the FSP to encourage nutritious food use by program participants, and it undermines confidence in the program among the general public. Unfortunately, the very nature of trafficking makes the practice difficult to study. In most contexts, asking respondents in a survey to report on their socially unacceptable behavior runs the risk of underreporting, since respondents do not want the interviewers to think badly of them. In the case of trafficking, this problem is greatly exacerbated by the illegality of the actions being asked about. Because trafficking is illegal, FSP participants are understandably reluctant to discuss any trafficking activities in which they may be involved. In particular, they are often skeptical about confidentiality pledges and fearful that admitting to selling coupons could lead to loss of benefits or even criminal prosecution. This report summarizes the results of a research study that attempted to overcome these barriers and to learn more about trafficking, with a particular focus on the participant side of the trafficking transaction. Recognizing that obtaining accurate information from participants was likely to be 1 difficult, the study was designed with several different components. Data collection activities used both qualitative and quantitative techniques. One component of the study-a survey of a total of more than 700 program participants living in three different parts ofthe country-was designed to test a promising method for obtaining self-reports from participants concerning whether and how they had engaged in trafficking. In addition, the survey was designed to obtain preliminary information about trafficking with which to develop hypotheses that could be examined in later research. Supporting the survey were several other types of data collection: (1) pre-survey focus groups, conducted to help develop the survey instrument by obtaining information about trafficking from recipients; (2) post-survey focus groups of respondents to probe further about bow they answered the survey questions and ascertain the motivations that determined their responses; and (3) ethnographic interviews with purchasers of food stamps to ask them about trafficking activity and about the characteristics and apparent motivations of traffickers. This report summarizes the findings ofthe study. A. LIMITATIONS OF THE STUDY Several significant methodological limitations, many of which were recognized from the outset of the study, must be kept in mind when assessing the results in this report. The project was designed as an exploratory study, not as a definitive analysis of participant trafficking. Because at the outset so little information was available about either participant trafficking behavior or about how to obtain information from participants about trafficking, it was felt that a full, detailed study oftrafficking on a national basis was not warranted. Rather, the goals ofthe study were more modest and focused on obtaining information that could help shape further studies. In light ofits objectives, the study was limited to three areas ofthe country. While an effort was made to choose geographically separate areas with different characteristics, the clustering ofdata collection into three areas precludes making valid national generalizations ofthe results. While the results may be suggestive about hypotheses concerning trafficking for the country as a whole, we cannot assess the representativeness of the three areas chosen for the study and therefore cannot formally generalize the results. Furthermore, as discussed in Chapter III, there is considerable evidence that techniques used in the survey were less successful than had been hoped in eliciting accurate reports oftrafficking. It is likely that a number of respondents in fact engaged in trafficking but did not admit during the survey to having done so. Furthermore, we are not able to determine whether the traffickers who did say they trafficked are systematically different from the traffickers who did not admit to trafficking. It is thus not possible to make reliable generalizations about traffickers from the survey results, even leaving aside the high degree of clustering in the survey. B. OVERVIEW OF REPORT Chapter II provides details about the data collection activities undertaken for the study. Chapter III assesses the degree to which it was possible to elicit accurate self-reports of trafficking during the participant survey. Chapter IV highlights a number of other hypotheses about participant trafficking that have emerged from one or more of the data collection activities. Appendix A provides details about the data collection work, and Appendix B reproduces the data collection instrument used in the participant survey. < "O y mmm D. DATA COLLECTION ACTIVITIES CONDUCTED FOR THE STUDY It was felt that a multifaceted data collection strategy would best serve the objectives of the study. Details of the methods used follow. A. PRE-SURVEY FOCUS GROUPS Two pre-survey focus groups were conducted with FSP participants in each ofthe three primary areas where the survey was to be done (see below).1 These sessions were intended to help die project team learn more about the trafficking process and about how participants thought about and talked about trafficking. It was anticipated that this would be useful both in suggesting topics to be covered during the surveys and in wording items as clearly as possible to respondents. 1. Material Covered The exact topics covered in the focus groups evolved slighUy over time, as we drew from the outcomes ofthe earlier groups in planning later ones. In general, however, the following seven areas were included: 1. Introductory Material The opening material was designed to explain the purposes of and ground rules for the focus groups. Also included was an exercise to facilitate self-introductions by participant group members and to get them started talking to one another. 2. Using Food Stomps. The first substantive part of the focus groups elicited information about how respondents use their food stamps, including their experience at stores. This provided a nonthreatening way of opening the discussion ofalternative ways coupons can be used. 'Details about how the three areas for data collection were selected and about the other aspects ofthe data collection—particularly, the household survey-are presented in Appendix A. 3. Coping Experiences. This discussion focused on how food stamp recipients cope with unexpected expenses that may arise during the month. It provided a context for the discussion oftrafficking to follow. 4. Trafficking. The discussion asked how commonly trafficking was believed to occur in the neighborhoods of the group participants. It also asked about the mechanics of trafficking, in terms of how buyers and sellers find each other, what types of people are buyers, and how much coupons sell for. This discussion also focused on what types of goods and services could be bought with food stamps and participants' general attitudes toward trafficking. Motivations for trafficking also were discussed. 5. Trial of Taped Interview. To see how well it would work, we gave respondents the taped interview that was planned for the general survey. While the sample sizes were too small to test the success of the method in eliciting accurate response, it was possible to test the logistics of the self-administered audiotape-based interviewing process. 6. Discussion of Willingness to Share Personal Information About Trafficking. A discussion was initiated concerning whether and under what circumstances focus group participants would be willing to provide accurate information about their trafficking experiences. 7. Closing. In closing, group participants were thanked for their help and were given their payments for participating. 2. Sampling and Recruiting The focus group participants were randomly sampled from lists of active participants provided by the state food stamp offices and were recruited in advance from MPR's survey telephone center. The focus groups had approximately 12 members each. 3. Fielding Each focus group was held in a "neutral" location, such as a community center or a hotel, which was convenient for the group participants to reach. Each was moderated by a senior MPR staff member with extensive experience in conducting focus groups. At least one assistant was also present at each group. The interview were taped, and transcripts were prepared. B. THE PARTICIPANT SURVEY The participant survey had two overall objectives. The first was to test a strategy for obtaining accurate self-reports oftrafficking behavior from participants. (See Section 2.) The second was to obtain information with which to develop hypotheses about participant trafficking, such as the characteristics of traffickers, their motivations, and the particular circumstances that lead them to traffic. To some degree, accomplishing the second of these objectives depended on accomplishing the first, since accurate self-reports about who trafficks are necessary to describe the respondent characteristics associated with trafficking. 1. Survey Content The data collection instrument consisted ofa series of modules designed to obtain information about the respondents themselves and their knowledge of and participation in trafficking. The following information was obtained: • Household Information and Food Stamp Benefit Receipt Module A collected background information on the respondent's age and marital status. It also covered number and ages of other people in the household, number of people covered by the food stamp benefit, availability of adequate cooking and food storage appliances, date and amount of most recent food stamp benefit, responsibility for food stamp benefit spending, and typical food stamp spending pattern. • Household Income and Expenditure. Module B collected information about the respondent's income and expenses, food shopping trips and expenses over the past week and month, employment status and earnings ofhousehold members, participation in Aid to Families with Dependent Children (AFDC) and other government programs, recent hardships experienced, and food sufficiency.2 • Attitudes, Opinions, and Beliefs About Foot* Stamp Trafficking. Module C collected information about the respondent's beliefs and opinions in several areas, including 2AFDC has since been replaced by the Temporary Assistance For Needy Families (TANF) program. outlook for the future, the public assistance system, consequences of lying to an AFDC worker, and the consequences of selling food stamps. Social Support. Module D collected information about the respondent's connection to friends, family, and other support networks in the community. Some questions were designed to elicit the degree to which the respondent was stable and established in the community; others probed whether and how respondents were recently victimized by crimes. Buying and Selling Food Stamps. Module E contained questions about the respondent's general knowledge and perception oftrafficking activity in the community: the ease or difficulty with which trafficking occurs, which store types and which people are involved in trafficking, ethical and moral views on trafficking, consequences of selling food stamps, and items that are exchanged for food stamps. At the end of this module, the respondent used an audiotape to answer a self-administered series of questions about whether he or she had bought or sold food stamps during the previous month and, if so, the value of the coupons transacted. Demographic Characteristics. Module F collected information on the respondent's ethnicity and racial background, gender, educational attainment, and history ofpublic assistance participation. It also solicited the respondent's opinions for ways in which the FSP might be improved and invited the respondents to possibly participate in a post-survey focus group. Interviewer Observation. The final section of the questionnaire required the interviewer's observation of the physical environment. 2. Approach to Data Collection Most of the data collection instrument was administered using standard in-person interviewing methods, with the interviewer reading each question, probing as necessary, and then recording the answer on a questionnaire. However, 14 questions-those concerning the respondent's own experiences buying and selling food stamps-were administered using an audio recording and headphones. The respondent listened to the prerecorded questions through headphones and then recorded his or her answers on a self-administered answer sheet, which was then sealed inside an envelope. The interviewer then signed the back of the envelope over the seal, which was not to be broken until the survey center of Mathematica Policy Research, Inc. (MPR) received the envelope for data processing. The decision to use audiotape recordings reflected two major considerations. First, the survey literature contains considerable evidence that respondents are more likely to report sensitive or illegal behaviors in a self-administered questionnaire format than in an oral interview. Apparently, an audiotape interview provides a sufficient feeling of anonymity to respondents that they answer honestly about any involvement in illegal activities. In a study of alternative ways of measuring drug use, Turner et al. (1992) found that "the self-administered questionnaire yielded higher estimated prevalence rates. As predicted, examination ofthe ratios indicates that the advantage of the self-administered questionnaire increases with the presumed sensitivity ofthe drug in question." Similarly, Gfroerer and Hughes (1992) conclude that "for sensitive question such as those on the use of illicit drugs, it is likely that people will be more willing to reveal their drug use on a self-administered answer sheet than in a verbal response—whether by telephone or in person—to an interviewer." Hay (1990) compared standard interview and self-administered techniques and found that the frequency distributions indicated that a significantly higher percentage of the questionnaire respondents reported ever having more than a sip or taste of an alcoholic beverage. Similarly statistically significant differentials were observed between the interview and questionnaire respondents on reported smoking." However, because the study population for the current research was expected to have low levels of reading ability, there was concern that a traditional self administered instrument might lead to considerable confusion on the part cf respondents, thus reducing our ability to obtain accurate replies.3 Thcretv re, we decided to have respondents listen to a self-administered audiotape and then mark their response on a simple answer sheet. A study by Camburn et al. (1991) provided evidence that this can be an effective approach. In particular, this study had good success in using audiotapes to ask respondents who were 12 to 21 years old a set of questions about such sensitive issues as drug use, alcohol use, sexual practices, and eating disorders. 3. Sampling Because the study was exploratory, the survey was limited to three areas of the country. A combination of random and purposive sampling was used to assure diversity among the areas in terms ofsize ofpopulation, geographical location, and food stamp issuance methods. The following areas were selected for data collection: Baltimore, Maryland; Denver, Colorado; and Spokane, Washington. Within each of these three primary areas, urbanized zip code areas were stratified into four groups, based on the percentage ofhouseholds below the poverty level in the zip code areas.4 One zip code was then randomly selected from three of the four strata (The stratum with the lowsst concentration ofpoverty was not sampled.) Furthermore, for each of the three primary areas initially chosen (Baltimore, Denver, and Spokane), the rural counties contiguous to the urban area were listed. One of the counties in each primary sampling area was then randomly chosen. 3About 10 to 20 percent of U.S. adults and a higher percentage of food stamp recipients have reading difficulties. The percentage ofU.S. adults who completed high school is 71 percent Fifty-six percent ofthe food stamp recipients in die San Diego Food Stamp Cash-Out Demonstration and 41 percent in the Alabama Food Stamp Cash-Out Demonstration completed high school. 4These were defined as zip code areas where at least 80 percent ofthe population was in areas classified by the Census Bureau as urban. 10 Overall, therefore, four subareas were chosen from each ofthe primary sampling areas. Three ofthese were urbanized, with varying poverty concentrations; the fourth was rural but adjacent to the urban area. Once the local sampling areas had been defined, either in terms of zip codes or in terms of the designated rural counties, MPR obtained, from the state food stamp offices, lists offood stamp cases in these local areas. Respondents were then randomly sampled from these case lists. Because many ofthe study objectives required having a substantial representation oftraffickers in the sample, and because it was believed that the likelihood oftrafficking might be correlated with the amount ofthe food stamp benefit, households were sampled with probabilities of selection proportional to the amount of the benefit, as listed in the sampling frames. 4. Fielding Procedures Interviewers were hired and trained for the project at each of the three primary sites. The selection criteria for hiring interviewers included strong verbal skills, an ability to work effectively with low-income respondents, and good organizational skills. All but 2 of the 26 interviewers hired had interviewing experience. Interviewer training began with advance study and completion of written training exercises. In-person training, held at local hotels, then lasted eight hours. Trainers explained the background and purpose of the study, reviewed the screener and the interviewer- and self-administered questionnaires, provided instructions for asking each question, discussed methods for contacting respondents, addressed sensitivities about criminal behavior, and went over administrative procedures. After the main session, interviewers finished their training by completing mock interviews with a supervisor. 11 The interviewing was conducted between December 1995 and April 1996. An experienced supervisor managed the interviewers by telephone. The interviewers returned their data to MPR's main survey operation center near Princeton, New Jersey, for logging, data entry, and verification. 5. Fielding Results A total of 720 interviews were completed: 245 in Baltimore, 221 in Denver, and 254 in Spokane.3 The overall response rate among eligible respondents was 77 percent. The main reasons for nonresponse were interviewer inability to locate the respondent (seven percent); refusal on the part of the respondent (six percent); and interviewers' inability to complete cases, despite multiple attempts (six percent). Other reasons for nonresponse included language barriers (two percent) and poor health ofrespondents (two percent). C. POST-SURVEY FOCUS GROUPS An additional round of focus groups was conducted several months after the survey was completed. Based on resource considerations, we decided to conduct these focus groups at two of the three survey sites (Denver and Baltimore). The purposes ofthese post-survey focus groups were (1) to examine apparent discrepancies between data obtained in parts of the survey and other available information, and (2) to provide more-detailed followup about some ofdie survey findings. The participants in these focus groups were recruited from among the earlier survey respondents. 1. Material Covered The seven main areas covered by the post-survey focus groups were the following: 'One interview was subsequently lost in the mail. 12 1. Introduction. The sessions began with a discussion of their purposes and of the ground rules to be followed. In addition, self-introductions were made. 2. FoodStamps-Baseline Knowledge. The first set of topics was designed to provide a nonthreatening way to get respondents to talk about the use of their food stamps. To stimulate the discussion, questions were asked about participants' perceptions of the mechanics of how the program works and about how food stamps relate to their food budgeting. Toward the end of this module, participants were asked to give their reactions to a number of situations in which food stamp recipients were faced with financial problems that might lead them to traffic. 3. Survey Data: Incidence of Trafficking. The preliminary survey results showing a relatively low reported incidence oftrafficking were summarized for the members ofthe focus groups. These participants were then asked whether they thought the survey results were accurate. They were also asked how they had reacted to the relevant survey questions and whether, if they had been trafficking, they would have reported having done so in the interviews that used the recorded tapes. 4. Survey Data: Profiles of Traffickers. In part because no clear profile of trafficker characteristics had emerged from the preliminary analysis of the survey data, focus group participants were asked to indicate what characteristics, if any, they would associate with trafficking. 5. Survey Data: Urban Versus Rural Trafficking. Based on preliminary survey results, it appeared that trafficking was more common in urban than in rural areas. The focus group moderator probed for reasons. 6. Survey Data: About the Trafficking Process. Focus group participants were asked to comment on survey data about the ease with which buyers and sellers of food stamps could be located in their neighborhoods. 7. Additional Suggestions. Focus group participants were asked for their suggestions as to how it might be possible to reduce trafficking. 2. Fielding Two post-survey focus groups were conducted in each of the three primary sampling areas. Each was moderated by a senior member of MPR's subcontractor, RIVA Market Research, and observed by an MPR staff member. MPR recruited focus group members from lists of survey respondents who had expressed an interest in participating in subsequent focus ^groups. 13 D. ETHNOGRAPHIC RESEARCH It was believed that food stamp intermediaries could provide an additional perspective on FSP participants who trafficked their benefits; therefore, the research design called for interviews of people who acted as brokers in recipient trafficking exchanges. This work is described here. 1. Approach Because ofpotential difficulties locating food stamp buyers and getting them to cooperate with the study, we decided to enlist the help of ethnographic researchers who already had street-level contacts in the areas where the survey had been conducted. In particular, we identified urban ethnographers working in both Baltimore and Denver, and we asked them to locate food stamp buyers and interview them in detail. These two researchers had spent years conducting drug-related studies in their respective areas and had contacts among residents of neighborhoods with high poverty rates. In reviewing this work, it is important to note that the researchers' past experience had focused on studying drug use. Because of the difficulty of identifying appropriate respondents and the intensive nature of the data collection, ethnographic research, by its very nature, seldom involves random samples of subjects. In the current case, the fact that the two researchers involved had previously studied drug use may have skewed the profile ofrespondents toward people with drug problems. In developing this line of data collection, we designed a protocol that described die research issues to be covered in the interviews and identified die types of respondents we were interested in. Based on this protocol, each ethnographer drew upon his local contacts to identify suitable respondents and then conducted the interviews. 14 2. Content Because of the unstructured nature of ethnographic research, it was not appropriate to develop detailed specifications as to the exact content of the interviews. Rather, the interviews were allowed to proceed in whatever directions seemed most fruitful to accomplish the basic research objectives. In general, however, the interviews covered topics similar to those covered in the household survey but did so from the perspective of food stamp buyers rather than sellers. Major topics included (1) the respondents' experiences with trafficking; (2) their perceptions of the motivations of traffickers; and (3) the logistics of trafficking, including locations where trafficking take place, what prices coupons sell for, and how buyers and sellers come together. Respondents were asked to report both their own experiences and their perceptions of common practices regarding trafficking in their neighborhoods. 3. Fielding Five interviews were conducted in Baltimore and five in Denver. When field conditions permitted and respondents were willing, the interviews were taped, and recordings were made of six of the interviews. In the four instances where taping was not possible, the analysis was based on interviewer notes from the discussions with respondents. The results of the individual interviews were then summarized by the two field ethnographers who conducted them, and a further synthesis was then conducted. o-^. I ® // l\\\ n HI. ASSESSMENT OF WHETHER THE PARTICIPANT SURVEY ELICITED ACCURATE SELF-REPORTS OF TRAFFICKING As discussed earlier, a key objective of the project was to assess the feasibility of obtaining accurate reports oftrafficking by using a survey-based data collection procedure. Based on evidence from earlier research done on other illegal activities, it seemed reasonable to expect that fairly accurate responses could be obtained if the respondent listened to a recording ofthe questions using a headphone and then marked his or her responses on an anonymous answer sheet. This approach was intended to make the respondent feel "safe" in reporting illegal activities and to reduce respondents' tendencies to give answers that they think their interviewers viewed as socially acceptable. However, based on the survey responses to the questions about whether respondents trafficked, coupled with other information on the prevalence of trafficking, we believe that there is a high likelihood that the survey reports significantly understated the prevalence of trafficking. In particular, it appears that the number of respondents who reported having engaged in trafficking behavior is lower than we would have expected based on any of the following sources of information: (1) other questions from the survey related to the general perceived prevalence of trafficking; (2) information on trafficking from the focus groups conducted for the project; and (3) an independent estimate of certain types of trafficking made by USDA for the year 1993. To be sure, these alternative sources ofcomparison information about trafficking prevalence are themselves subject to error, and as a result our inferences are not conclusive. Nevertheless, the weight of the evidence strongly suggests underreporting in the survey. The discussion below 17 provides details of these arguments. First, we present the relevant survey self-report data. Then we examine these data in light of the comparison information. A. THE SURVEY SELF-REPORT DATA ON TRAFFICKING Three percent of the survey respondents said that they had sola food stamps for cash the previous month, and two percent had exchanged them for goods or services (Table 111.1). The overall reported rate of selling or exchanging food stamps was 5.1 percent The reported cash value of food stamps sold ranged from $20 to $180. The mean value of benefits sold was $61.08, which is approximately 41 percent of average household benefits. These data suggest that about two percent of benefits were sold.' In contrast, two percent of respondents said that they had bought food stamps; that is, they obtained food stamps either for cash or in exchange for goods or services. Of these, about one percent said they received food stamps in exchange for cash and one percent said that they received food stamps in payment for goods or services. The value of food stamps bought varied from $5 to $352, with an average of$77. Most respondents who reported trafficking indicated that they either bought or sold, but very few reported both. Only about one-half of one percent of the sample bought and sold during the month. 'Calculated as 5.1 times 0.41, where 5.1 is the probability of selling and 0.41 is the average percentage sold if the recipient sells. 18 TABLE HI. 1 REPORTED PARTICIPATION IN TRAFFICKING (Percentages Unless Otherwise Noted) Full Sample (N = 719) Selling Food Stamps Sold Food Stamps for Cash 3.4 Received Goods or Services in Exchange for Food Stamps 2.1 Received Cash, Goods, or Services for Food Stamps 5.1 Buying Food Stamps Bought Food Stamps for Cash 1.0 Received Food Stamps in Exchange for Goods or Services 1.0 Received Food Stamps for Cash, Goods, or Services 2.0 Any Trafficking 6.5 Mean Value of Food Stamps Sold (in Dollars)* 61.08 Minimum 20.00 Maximum 180.00 Mean Value of Food Stamps Bought (in Dollars)* 77.05 Minimum 5.00 Maximum 352.00 SOURCE: Food Stamp Trafficking Survey, weighted data. NOTE: N indicates sample size. Not all sample members responded to all questions. For each question on this table, between 0 and 9 respondents did not provide responses. 'The value of food stamps bought and sold are calculated based on responses from those who reported buying or selling food stamps. 19 B. COMPARISON TO OTHER AVAILABLE INFORMATION Unfortunately, there are no definitive data on the incidence oftrafficking by participants with which to compare the survey results. Furthermore, even if definitive national data were available, our ability to assess the accuracy of the survey data would be significantly limited by the fact that they are drawn from only three areas of the country, and we don't know how nationally representative these areas are. Despite these problems, however, a number of comparisons between the self-report data and other sources are of interest. We present them below. 1. Comparisons with Survey Data on Neighborhood Prevalence and of Trafficking by Friends Besides being asked directly about their own trafficking behavior, respondents in the survey were asked several questions about their perceptions of the prevalence of trafficking in their communities One survey question, for instance, asked how many food stamp recipients in the respondents' neighborhoods sold food stamps for cash in any one month; choices included "most people," "some," "a few," and "nobody." (Note that this question focused specifically on selling for "cash.") Approximately 2.8 percent of respondents answered that "most people" that were FSP participants in their neighborhoods trafficked, and another 7.6 percent and 8.5 percent, respectively, responded "some" and "a few." (See Table MX) About 9.7 percent said "nobody." Thus the number of respondents answering "most people" or "some people" exceeded the number selecting the "nobody" category. How these results are assessed depends greatly on what is assumed about the 71.3 percent of respondents who gave a "don't know" response to the question. At one extreme, it is possible that those answering "don't know" did so because there is no significant trafficking in their 20 TABLE mi PERCEPTIONS ABOUT THE EASE AND PREVALENCE OF FOOD STAMP TRAFFICKING (Percentages) Full Sample (N-719) In any one month in your neighborhood, how many food stamp recipients sell some or all ofthem for cash? Most people 2.8 Some 7.6 A few 8.5 Nobody 9.7 Don't know 71.3 Refused/missing 0.3 If someone in your neighborhood had cash and wanted food stamps, would it be easy to find someone to sell them? Very easy 23.1 Somewhat easy 13.3 Somewhat difficult 8.1 Very difficult 15.6 Don't know 39.5 Refused/missing 0.4 How many of the three food stamp recipients you know best sold food stamps for cash in the past year? None 34.0 One 8.2 Two 4.2 Three 5.1 Don't know three food stamp recipients 26.4 Don't know how many sold food stamps 21.9 Missing 0.3 How often did these friends sell benefits in the past year? Every month 4.9 Nearly every month 2.5 Once in a while 4.3 Question not applicable 82.3 Missing 1.0 SOURCE: Food Stamp Trafficking Survey, weighted data. NOTE: N indicates sample size. 21 neighborhoods. On the other hand, it is plausible that, despite their answers, the "don't know" respondents lived in neighborhoods that were quite similar in terms of trafficking to those of the people who did answer the question. If any significant number ofthe "don't know" respondents live in neighborhoods where substantial amounts oftrafficking do occur, this would suggest a prevalence rate higher than that implied by the small number (3.4 percent) of respondents who reported having themselves sold coupons for cash trafficked in the previous month. A related survey question asked how easy it would be in the respondent's neighborhood for someone who had cash and wanted to buy food stamps to find a seller. As shown in Table III.2,36.4 percent of respondents replied that finding a seller would be either "very easy" or "somewhat easy," compared to 23.7 percent who indicated that it would be "somewhat difficult" or "very difficult." (About 40 percent indicated that they didn't know how easy it would be.) A third relevant questioning sequence asked respondents to think about their three best friends on food stamps and estimate how many trafficked during a 12-month period. Approximately SO percent of respondents said that they knew three other food stamp recipients and had an opinion as to whether the people they knew trafficked. Approximately 17.5 ofrespondents (continuing to use the overall sample as the base) thought that at least one of the people they knew trafficked. Furthermore, of those who thought their friends trafficked, most believed that the friends did so "every month" or "nearly every month." (In these two categories, 7.4 percent responded, compared to only 4.3 percent responding "once in a while.") While not conclusive, these findings suggest that the food stamp recipients who responded to the survey perceive that trafficking occurs frequently in their communities. This is in contrast to their direct self-reports of trafficking. 22 2. Pre-Survey Focus Groups As described in Chapter II, the design work for the study included conducting focus groups of FSP participants in each of the planned survey sites. While it is not possible to quantify the focus group responses precisely, the general impression gained from both the sessions and the transcripts was that several of the participants in each session had engaged in trafficking activity and that virtually all participants were aware of trafficking taking place among their acquaintances. Respondents were able to speak with some confidence about such details as the cash price of coupons, the locations in their communities where trafficking often occurred, and what items could and could not be easily bought with food stamps. As with the written survey, these focus group sessions provide evidence that trafficking is a much more common phenomenon than suggested by the self-reports discussed in Section III.A. 3. Post-Survey Focus Groups To further explore the likely accuracy of the survey responses and to learn respondents' perceptions about the questions and our interviewing approach, we held focus groups in two ofthe survey sites several months after the survey. During this second round of focus groups, samples of the respondents to the earlier survey were assembled and asked to describe their reactions to the survey, particularly the self-administered taped interview. In addition, the focus group members were shown the survey results concerning the incidence of trafficking and asked whether they thought them believable. The responses strongly suggest that some of the respondents did not provide accurate self-reports concerning their trafficking activities. Several indicated that they had distrusted the survey and suspected that it was a "sting" operation to detect food stamp fraud. Furthermore, several focus group participants explicitly said that the use of the tape recorder did not significantly increase their 23 confidence that their replies would be kept confidential. These focus group responses provide significant evidence that the incidence oftrafficking is significantly underreported in the survey data. 4. An External Estimate of Trafficking Incidence A USDA study (Macaluso, 1995) used evidence drawn from federal trafficking investigations to estimate the approximate level oftrafficking in the United States in 1993. The data include only trafficked food stamps that were sold directly to stores and therefore were not ultimately spent for food; they thus exclude sales to people who then used the coupons to obtain food in a store. The USDA study estimated that food stamp trafficking in 1993 amounted to about $800 million, approximately 3.8 percent ofall food stamps issued during the year. As noted earlier, the tabulations from the current survey imply that about two percent ofdollar benefits are trafficked. This is further evidence that trafficking is significantly underreported in the survey. In addition, the questions in the current survey asked about all transactions involving selling food stamps (including selling them to a person who was planning to use them to buy food), whereas the USDA study, because of the methodology it used, was able to assess sales of food stamps only to buyers who then "laundered" them through stores. Thus, if the numbers could be put on a comparable basis, the gap between the survey numbers and the USDA estimates would further widen. For instance, if, as the survey data suggest, considerable selling takes place to buyers who buy for their personal use, then adjusting the survey data to exclude mis portion oftrafficking would reduce the survey-based estimates ofthe value ofcoupons trafficked to well below 2 percent, further increasing the discrepancy with the USDA estimate of 3.8 percent. 24 IV. OTHER HYPOTHESES SUGGESTED BY STUDY RESULTS As discussed earlier, the overall project was an exploratory study designed both to assess an apparently promising technique for surveying food stamp recipients about their trafficking activities and to identify which other hypotheses about trafficking behavior might warrant further study. As it turned out, the "headphone" interviewing approach we tested appears not to have been successful in eliciting reliable reports of trafficking activities. However, the various research activities conducted for the project, including the survey, the focus groups, and the ethnographic research, have yielded a number of interesting hypotheses that may warrant further study. Several of these hypotheses are discussed below, together with their supporting evidence. 1. Many Buyers in Food Stamp Trafficking Transactions May Buy the Coupons for Their Own Use; Sometimes, They Themselves May Be Food Stamp Program Participants Participants in the pre-survey focus groups repeatedly voiced their beliefthat the buyer in a food stamp trafficking transaction is often a food stamp recipient, and that many buyers use the stamps themselves at the grocery store. Focus group members reported that, as a way of stretching their food budgets, they and their friends often bought food stamps from people who wanted to sell them, with several viewing this practice as simply prudent household management. They also reported that it was common for a recipient to be both a buyer and a seller over the course of a month, perhaps selling coupons initially to obtain cash for some high-priority use and then, if cash became available later, using it to buy coupons at a discount and gain access to food more cheaply. This theme of food stamp recipients buying food stamps for their own use but also selling them came up in the focus groups in all three cities in which the groups were held. 25 The ethnographic research conducted for the study also found evidence that many buyers of food stamps are purchasing them for their own use, rather than intending to resell them to stores that can "launder" them through the regular banking system. In both cities where these interviews were conducted, respondents indicated that most purchasers they were aware ofwere buying the coupons because they represented a cheap way ofobtaining food. In Baltimore, for instance, the ethnographic researchers interviewed a respondent who lived in the suburbs but regularly came into the city to purchase food stamps at a discount so that she could use them to obtain food. These transactions were made easier because she had previously lived in the city and retained ties to her old neighborhood. In Denver, one ofthe respondents in the ethnographic study was a crack dealer wl occasionally traded crack for food stamps (about S65 ofcoupons per month) for her own use in buying food but would not accept more food stamps than she needed for food. In another Denver interview, the household (a couple) were drug addicts. Usually, they bought food stamps for use only in buying food, though occasionally they used the stamps to buy drugs. 2. There Is No Clear "Profile" of Characteristics of People Who Traffic During preliminary analysis ofthe survey data, we examined whether self-reports of trafficking were correlated with demographic, social, or economic characteristics of respondents. While some weak possible correlations were identified, no single characteristic or set of characteristics stood out as being highly predictive oftrafficking behavior. To be sure, as noted earlier, there is evidence that trafficking was substantially underreported in the survey, and this weakens our ability to identify significant correlations, if in fact they exist. (Indeed, the apparent underreporting oftrafficking is the reason that we have not formally analyzed the survey data in greater detail.) However, the fact 26 that no strong correlations were present in the survey data is at least consistent with the conclusion that traffickers do not fali into readily identifiable groups. We explored this issue further during the post-survey focus groups, because the preliminary analysis of the survey data had shown a surprising lack of regularity or patterns in trafficker characteristics. Focus group participants were asked to discuss among themselves what characteristics they would expect a person selling food stamps to have and were given several specific sets of choices (for example, men/women, younger people/older people; people with kids/people without kids). In general, no strong patterns emerged from these discussions. While there was some tendency for the groups to focus on certain characteristics, such as being young and being a woman, it appeared that the discussion group members were really focusing more on the profile of the food stamp recipients they knew rather than on the characteristics of traffickers from within that group. Overall, no clear consensus about a pattern of characteristics emerged. This would be consistent with a hypothesis that many different types of people traffic for many different reasons. 3. "Middlemen" May Not Be a Major Factor in Trafficking One issue of considerable interest in the current study was the role of "middlemen" who buy coupons and then sell them at a profit, either to people who plan to use them for food or to stores who can "launder" them. It was believed at the outset of the study that such middlemen might represent a significant share of buyers. However, as discussed below, no evidence emerged from the research to suggest this. During the pre-survey focus groups, group members were asked to talk about what groups of people bought food stamps. The group participants tended to focus on two types of buyers: (1) retail stores who cashed the coupons in at banks, and (2) low-income people who intended to use the 27 coupons to purchase food (as discussed in Section 1, above). The existence of "middlemen," who bought the coupons to make money by reselling mem, was seldom mentioned by the focus group participants. Furthermore, even after direct probing, most of them said that they did not know of such people. In addition, in one of the questions on the participant survey (QE7), respondents were asked what type ofpeople might be likely to buy food stamps. The answer categories, which were all read by the interviewer, included such responses as "people you live with," friends," "storekeepers," and, most important for the current discussion, "a middleman who buys food stamps and sells them to someone else." (Multiple responses were allowed.) Fewer than five percent of the respondents chose this "middleman" response as one of their replies. Similar findings were obtained during the ethnographic research. The ethnographers probed extensively about middlemen and found virtually no evidence that they exist. None of the 10 respondents interviewed, all of whom had themselves been buyers of food stamps for various reasons, had ever bought coupons and resold them for cash. Furthermore, they were not aware of people (other than retailers who cashed the coupons in at banks) who did so. Indeed, several ofthese respondents indicated that they were not able to figure out how a middleman could make money. In their experience, the discount price of coupons was fixed in any given neighborhood at one rate (usually SO percent); they felt that, if somebody bought coupons at SO percent, it would be hard to convince another person or a store to repurchase them at a higher rate. In their experience, coupons could be bought at the SO percent rate easily, and they didn't see why anybody would pay a middleman a higher rate. The fact that these respondents, who were located through street contacts, were not themselves major middlemen is not surprising, since such people, if they exist, would probably be harder to 28 interview. However, these respondents were being interviewed not only about their own experiences but also for their perspective on their local social scenes, and the fact that they had not heard of middlemen is reasonably telling. 4. Electronic Benefit Transfer (EBT) Does Not Completely Prevent Recipient Trafficking bat Does Change the Dynamics of the Transaction One of the three study sites had been issuing food benefits under an EBT system for several months prior to the study. A number of respondents at that site reported trafficking, and the focus group discussions yielded insights as to how this trafficking took place. The dynamics of the trafficking under EBT appeared to be quite different from trafficking in food coupons. Focus group participants indicated that, under EBT, the benefit is usually sold to someone the seller knows and trusts, since if the transaction is with a stranger, it is riskier and more time-consuming for both buyer and seller. In general, sellers were said to "spend down" the amount of the food benefits they needed for themselves before turning t!se EBT card over to a buyer. If the seller knows the buyer, the buyer may pay in advance and the seller may give out an identification (PIN) number with the card. If the buyer and seller do not trust each other, the buyer often accompanies the seller to the store. In this situation, the seller shows his or her own identification and pays for the groceries as if he or she is shopping; payment is made after the pair leave the store. The buyer and seller were reported to go to the store together often because the buyer wants to avoid (1) prepaying for an EBT card on which benefits have been depleted, and (2) prepaying for an EBT card that has been reported as lost or stolen. The general feeling about EBT as it relates to trafficking was that "where there is a will there is a way." Focus group respondents reported that the cycle with which benefits were posted to the EBT cards contributed to selling benefits. In the EBT site, the AFDC benefit was credited early in the 29 month. At this time, recipients needed food and spent the welfare benefit on groceries. The food benefit was credited four or five days later. By the time the food benefit was credited, food had been purchased and recipients needed cash. 30 REFERENCES Camburn, Donald, Marcie Cynamon, and Yossi Hard. "The Use of Audio Tapes and Wrinen Questionnaires to Ask Sensitive Questions During Household Interviews." Paper presented at the National Field Directors Conference and National Field Technologies Conference, San Diego, May 1991. Gfroerer, Joseph C, and Arthur L. Hughes. "Collecting Data on Illicit Drug Use by Phone." Survey Measurement ofDrug Use: Methodological Studies. Rockville, MD: National Institute on Drug Abuse, 1992. Hay, David A. "Does the Method Matter on Sensitive Survey Topics?" Survey Methodology, vol. 16, no. 1, June 1990, pp. 131-136. Macaluso, Theodore F. "The Extent of Trafficking in the Food Stamp Program." Washington, DC: U.S. Department ofAgriculture, Food and Consumer Service, August 1995. Turner, Charles F., Judith T. Lester, Barbara J. George, Michael L. Hubbard, and Michael B. Witt. "Effects of Mode of Administration and Wording on Data Quality." Survey Measurement of Drug Use: Methodological Studies. Rockville, MD: National Institute on Drug Abuse, 1992. 5 AU 31 M uro mm APPENDIX A SURVEY METHODOLOGY 3>3 di This appendix presents the methods used to select the sample, conduct the Food Stamp Recipient Trafficking Survey, and process the data. A. METHODS FOR SELECTING AM) LOCATING RESPONDENTS 1. Sample Design A combination ofpurposive and probability sampling was used to select three states from which the survey sample would be drawn. FCS's specifications for the work were that each state be "stratified by level of urbanization to include one local area food stamp office considered to be central city, one in a metropolitan area not considered to be central city, one in a suburban area, and one in a sparsely populated (rural) area." Random probability methods were used to select respondents from each local office area. Mathematica Policy Research, Inc. (MPR) used a three-stage sample design to implement this requirement Stage One involved selecting three metropolitan areas in different parts ofthe nation. Stage Two involved selecting a range ofurban to rural strata within and near each metropolitan area. Stage Three involved selecting Food Stamp Program (FSP) participants within strata. a. Stage One, Selecting Metropolitan Areas • The sample frame consisted of all 134 Metropolitan Statistical Areas for the 48 contiguous states and Washington, DC. The areas were stratified into three geographic strata based on FCS regions (Mid-Atlantic, Northeast, and Southeast; Midwest, Mountain Plains, and Southwest; and West) and three population size groups (more than 3 million, 1 to 3 million, and 250,000 to 1 million). Areas also were classified as issuing benefits using or not using Electronic Benefit Transfer (EBT) technology. • A random number between 0 and 1 was assigned to each area. • We used the random number to pick die EBT site first Baltimore, Maryland, had tin. lowest random number and was selected. Baltimore, in the Eastern region, was the largest stratum. 35 • Next we randomly selected L*,* third region as the second site. Using the lowest random number, we selected Spokane, Washington, among all the areas in the third region that were not in the largest size stratum. Spokane was in the smallest size stratum. • Denver, Colorado, represented the midsized population stratum in the Central region. b. Stage Two, Selecting Strata Within Metropolitan Areas • We listed every zip code in each of the three selected metropolitan areas, along with its total population, population in poverty, and population living within die urban area. • After excluding zip codes that were less than 80 percent urban, we sorted the remaining zip codes by the percentage of their households living below the federal poverty level. • We formed four strata in each area, based on the percentage living in poverty. Thestrata ranged from the very highest poverty density to no poverty. • After excluding zip codes in the lowest poverty density strata, we randomly selectedone zip code from each stratum. The selected zip codes were: Baltimore Denver Spokane 20 percent or more poverty 10 to 20 percent poverty 5 to 10 percent poverty 20 percent or more poverty 10 to 20 percent poverty 5 to 10 percent poverty 20 percent or more poverty 10 to 20 percent poverty 5 to 10 percent poverty 21231 21237 21210 80223 80207 80228 99202 99205 99016 To select a rural area, we listed every rural county contiguous to the metropolitan area and selected one at random. The rural counties selected were: Baltimore Kent County Denver Cleark Creek County Spokane Pend Oreille County • The three metropolitan sites and four subsites formed 12 strata. 36 c Stage Three, Selecting Respondents • MPR asked states for lists ofFSP participants in each stratum (that is. zip code or rural county). The states sent machine-readable data sets, which an MPR programmer edited. • The programmer sorted cases in each stratum by size of food stamp benefit. She then used interval sampling to select a sample proportionate to the size of the monthly benefit. She selected 150 cases in the Baltimore and Denver strata and 200 cases in the Spokane strata.' Cases that would have entered the sample with certainty were selected before applying the interval method. • In the "less than 10 percent" stratum in Baltimore and Denver, there were not enough FSP participants to yield 60 completed interviews. For these strata, we supplemented by adding randomly selected zip codes from the same stratum. We added zip codes 21206 and 21220 to 21210, and zip codes 99223 to 99016. Samples were drawn on the combinations of zip codes. 2. Obtaining Contact Information for Respondents Maryland and Colorado sent names and addresses of FSP participants to MPR. Washington State mailed the sample frame to MPR without names and addresses. MPR selected the survey sample and sent the results back to the state for the consent mailing. Washington sent a letter to potential sample members, offering them a two-week period in which to decline to participate. Only 2.3 percent of those to whom letters were mailed objected. 3. Screening Criteria To be eligible to participate in the study, individuals had to be active food stamp participants and be living in the stratum from which they were selected. To determine eligibility, interviewers administered a short screener before conducting the interview. Eligibility rates are discussed in Section E. 'More cases were needed in Spokane because their Human Research Review Board required a passive consent procedure before releasing names to MPR. r 37 B. SELECTION AND TRAINING OF FIELD STAFF 1. Hiring MPR hired and trained 26 interviewers for this project; 11 worked in Baltimore, 9 in Denver, and 6 in Spokane. To the extent possible, we hired interviewers who lived in the communities in which they interviewed. All but two interviewers had prior interviewing experience. The number of interviewers hired per site depended on the hours per week each could work. In Spokane, where each interviewer could work more hours per week, we hired fewer interviewers. In Baltimore, where experienced interviewers had other project commitments, we needed more. The goal was to hire enough interviewers to have at least 120 hours of labor available each week. 2. Training Interviewer training had three components. The first was a two-hour self-study module to prepare for the main training session. We expected interviewers to spend two hours reviewing the questionnaire and related materials a few days prior to the main training session. We required them to complete written training exercises to prove that they had mastered the materials. (New inter, iewers had additional time to view MPR's video, "General Interviewer Training," and to review lessons with their supervisor.) The main training session lasted eight hours. We held sessions at a local hotel. Trainers explained the background anc purpose of the study, reviewed the screener, went over the interviewer- and self-administered questionnaires, provided instructions for asking each question, discussed methods for contacting respondents, and addressed sensitivities about criminal behaviors associated with interviewing. Interviewers had ample time for role-playing, practice interviewing, and administrative procedures. After the main session, interviewers finished their training by completing mock interviews with a supervisor. 38 MPR trained the Baltimore interviewers on December 2. The Baltimore training session served as "Trainers* Training." Anne Ciemnecki, the survey director, was the lead trainer. Sharron Christofar, the Contracting Officer's Technical Representative, discussed the importance of the project to FCS. Linda Mendenko, MPR's field coordinator, covered administrative materials. Lynne MacKenzie, MPR's field manager, led the role-playing and mock interviewing. Denver and Spokane main trainings were held on December 9 and 16. Lynne MacKenzie was the lead trainer for the Denver and Spokane sessions. C. METHODS FOR COLLECTING THE DATA 1. Timing of the Field Period Interviewing began immediately after training in Baltimore and Denver. The Spokane sample was not available until December 29. We completed interviewing by April 6. Following are the numbers of interviews completed weekly at each site: Baltimore Denver Spokiine Total Week Ending Number Percent Number Percent Number Percent Number Percent December 9 23 9.4 0 0.0 0 0.0 23 32 December 16 35 14.2 13 5.9 0 0.0 48 6.7 December 23 21 1.5 3 1.4 0 0.0 24 3.3 December 30 22 9.0 7 32 1 0.4 30 42 January 6 22 9.0 12 5.4 19 7.5 53 7.4 January 13 9 3.7 16 7.2 38 15.0 63 8.8 January 20 20 L2 23 10.4 26 10.2 69 9.6 January 27 17 6.9 16 12 22 8.7 55 7.6 February 3 11 4J 23 10.4 19 7.5 53 73 February 10 13 53 18 8.1 32 12.6 63 8.8 February 17 18 7.3 20 9.0 23 9.0 61 8.5 39 Baltimore Denver Spokane Total Week Ending Number Percent Number Percent Number Percent Number Percent February 24 9 3.7 14 6.3 22 8.7 45 63 March 2 11 4.5 9 4.0 13 5.1 33 4.6 March 9 0 0.0 10 4.5 5 2.0 15 2.1 March 16 4 1.6 4 1.8 3 \2 11 IJ March 23 5 2.0 5 22 6 2.4 16 22 March 30 4 1.6 17 7.7 15 5.9 36 5.0 April 6 1 0.4 11 5.0 10 3.9 22 3.0 Total 245 100.0 221 100.0 254 100.0 720 1M.0 To encourage interviewer productivity at the end ofdie project, MPR offered field interviewers a bonus of$20 for every interview and $5 for every screener completed after March 22. This kept interviewers' enthusiasm high when sample was sparse. It also kept interviewers motivated to finish their assignment rather than to move to easier, newer projects. 2. Ensuring Respondent Confidentiality Because trafficking in food stamps is illegal, MPR took two important steps to ensure respondents' confidentiality. First was documenting contacts with the respondents. Usually, we document information about the respondent (case identification number, name, address, telephone number, and directions to the home) on a contact sheet detailing the interviewer's attempts to reach the sample member. For this study, we separated the contact information onto two forms, a Record of Contacts Form and a Contact Notes Form. The Contact Notes Form contained identifying information used to locate the sample member and schedule the interview. The Contact Notes Form was left with the respondent at the end of the interview to assure the respondent that his or her name would not be linked with data. The Record of Contacts Form contained the case-tracking number 40 and a record ofattempts to reach the respondent (that is, number and time ofcalls, interim case status codes). Interviewers mailed the form to MPR along with the completed interview. They wrote no names, addresses, or other information that could identify the sample member in any way on the Record of Contacts Form. The second step in ensuring confidentiality was to use an audiotaped interview and self-administered answer sheet to ask about the sample member's direct involvement in food stamp trafficking. The respondent used audio headphones to listen to 14 questions about personal buying or selling of food stamps. He or she recorded answers on a self-administered answer sheet and sealed them in an envelope. The interviewer signed his or her name across the back ofthe envelope over the seal. An unbroken signature implied that the envelope had not been opened before receipt at MPR. The audiotaped interview provided all the advantages ofa self-administered, pencil-paper questionnaire in increasing reports of illegal or socially disapproved behavior. It was intended to overcome the suspected barrier of poor reading skills in the food stamp population. Ninety-six percent ofthe survey respondents used the self-administered audiotape and answer sheet. Those who did not fell into three categories: 1. People with print impairments (visual disorders, innumeracy, and attention disorders) who could not focus on the answer sheet well enough to fill it out 2. People whose primary language was not English or Spanish who needed an interpreter for the interview 3. Those who refused MPR produced a Spanish language version ofthe audiotape. 41 3. Field Management and Reporting Procedures Field interviewers received most of their assignments at the beginning of the field period. Distribution of assignments early allowed interviewers to plan their time and travel efficiently. Adjustments to assignments were made as data collection progressed. Field interviewers reported progress to their field supervisor weekly by telephone at prearranged times: hours worked, expenses, and case-by-case progress. During the reporting session, the supervisor provided interviewers with feedback from MPR's quality review process, handled administrative needs such as supply orders, and answered nonurgent questions. Interviewers were encouraged to call supervisors immediately for urgent matters. The supervisor reported summaries of field progress and expenses to MPR's field coordinator weekly to provide management with information that guided staffing and sample-release decisions. 4. Problems Faced During the Field Period Two serious problems arose during data collection. Six weeks after interviewing began, an interviewer was robbed at gunpoint at the Highlander Ridge Apartment complex in Baltimore. Highlander Ridge is a gated public housing project. Because of this incident, MPR mandated that escorts accompany interviewers working in the Highlander Ridge complex. Escorts may be other interviewers but are often friends or relatives of the interviewers who travel with interviewers. Escorts added to the cost of the data collection, but no new incidents were reported after we established the policy. The second problem was in the Denver stratum with 10 to 20 percent of the households living below the federal poverty level. Here, a child was killed in a drive-by shooting during the third week of interviewing. We advised interviewers to stay out of the area for a few weeks until emotions cooled and interviewers felt more comfortable working there. 42 D. QUALITY ASSURANCE AND DATA PROCESSING PROCEDURES Interviewers mailed completed work to MPR weekly. As quality control clerks received field interviewers' packages, they logged completed work into a questionnaire tracking system. Each completed interview had: • A Record of Contacts Form • The interviewer-administered main questionnaire • The self-administered answer sheet from the audiotaped interview The clerks entered the date they received each ofthe three components of a completed interview. Weekly reports listed cases reported by the interviewer as complete but not received at MPR, and cases where one or more components were missing. MPR used these reports to track missing documents. Most of the time, when paper was missing, the work was completed but not mailed. Two interviewers reported work completed that was not, in fact, done; MPR terminated those interviewers and reassigned their cases to other staff members. Usually, MPR validates field interviewer work by calling respondents or mailing a thank-you note with a prepaid return postcard for respondent to acknowledge that the interview was completed. To protect respondent confidentiality, we were unable to link contact information with interview data for this study. Therefore, quality control callbacks and the usual means of verification were not possible. Instead, we verified interview completion by comparing the sample member's birth date (and age) and monthly benefit amount from the administrative files provided by the states with the data collected during the interview. The questionnaire tracking file contained these two variables. As the clerk logged completed work into the tracking file, she compared the age and monthly benefit amounts 43 recorded in the interview with data on the administrative record. (Record data were not available to the interviewer.) She assigned codes to indicate the closeness ofthe match. For example: Birth Date: Codel Code 2 Code 3 Code 4 Very strong match Probable match Weak match No match Exact date Month and year correct; day may be incorrect; or day and year correct; month off by one month or two months; or month and day correct, year offby up to three years Two out ofthree elements do not match, but not too far off None ofthe three elements match Age: Codel Code 2 Code 3 Code 4 Very strong match Probable match Weak match No match Exact age Off by 1 to 2 years Offby3to4years Offby 5 or more years Benefit Amount: Codel Code 2 Code 3 Code 4 Very strong match Probable match Weak match No match Respondent amount within S percent of amount on file Respondent amount within 10 percent of amount on file Respondent amount within 25 percent of amount on file Difference greater than 25 percent We produced weekly reports of verification codes by interviewer and looked for patterns of Codes 3 and 4 in both age/birth date and benefit amount fields. We uncovered no fraudulent interviews using this procedure. 44 Besides operating the tracking system, quality control staff reviewed each questionnaire prior to data entry. They determined whether problems, such as strings of missing data or incorrect recording of responses, would make data entry impossible. The quality control staff followed a set ofspecifications set up by the survey director to ensure that recorded values were within a reasonable range, response codes had been assigned correctly, and the questionnaire had internally consistent responses. Any errors, missing data, or inconsistencies could not be double-checked with the respondent because of anonymity requirements, but they were reported to the interviewer. This feedback made the interviewer aware of his or her errors and helped prevent them from recurring. After the questionnaires were manually edited and open-ended items were coded, they were batched and sent to data entry clerks who used programmable key-to-disk data-entry equipment. An editing program to perform checks on value ranges, questionnaire skip logic, and internal consistency for the full sample was prepared. Errors were flagged automatically during data entry and corrected by a senior member of the quality control staff. She reviewed the file and located error codes flagged by the quality control program. The data editor reviewed the questionnaire and corrected the error. A second operator rekeyed all the data. As the data were reentered, they were compared automatically with the first data file. Discrepancies were brought to the operator's attention. Finally, the survey data were transmitted to the research database. E. RESPONSE RATES AND OTHER FINAL STATUSES MPR interviewers completed a total of720 interviews. One was lost in the mail. The remaining 719 interviews were processed for analysis. The overall response rate was 76.7 percent. 45 TABLE A I RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE AND ELIGIBILITY RATES. BY SITE All Sites Baltimore Denver Spokane Number rcfccntamgc Number Pcfcentaft Number IVTCftitty Number FtfccnUnfc ENgibtWy Smnty Status Completed Interviews Cannot Locate 720 65 76.7 6.9 245 33 78.3 10.5 221 20 75.7 6.8 254 12 76.0 3.6 Rcfusal/Breakofrs 56 6.0 10 3.2 17 SJ 29 8.7 Multiple Unsuccessful Attempts Language Barriers Health Problem Other S3 20 19 6 56 2.1 2.0 0.6 21 0 3 6.7 0.0 19 0.3 IS 9 10 0 51 3.1 3.4 0.0 17 II 6 5 5.1 3.3 1.8 1.5 Total Eligible Saapk 939 199.0 313 100J 292 109J 334 100.0 Eligibility Rates Eligible Sample Ineligible Sample 939 375 71.5 28.5 313 87 21.8 292 182 61.6 384 334 106 75.9 2-t.l Total Saaipk Released 1.314 100.9 400 109.0 474 1MJ 440 1M.0 Reason for laebgiMHty Moved from Stratum 154 11.7 63 15.8 S3 17.5 68 15.4 No Longer Participating 214 7 16.3 0.5 22 2 5.5 0.5 96 3 20J 06 36 2 U 0.5 Total Ineligible Sample 373 28J •7 21J 182 ■at 196 24.1 NOTE: Due to rounding, percentages may not equal 100 46 1. Eligibility Rates and Reasons for Ineligibility MPR released 1,314 cases to field interviewers (see Table A.1). Of these, 375 cases, or 28.5 percent, were ineligible to participate in the survey. Cases were deemed ineligible for one of three reasons. The most common reason was that the sample member had movedfrom the stratum in which he or she was selected. Moves accounted for 214 cases, or 16.3 percent of the sample. The next most common reason for ineligibility was that the sample member no longer participated in the FoodStamp Program; 154 cases, or 11.7 percent ofthe sample, were not participating in the FSP when contacted for an interview. (Depending upon the site, contact occurred between 2 and 10 months after the survey sample was drawn from lists of FSP participants.) The third reason for a case being considered ineligible was death. Seven program participants died before we could interview them. 2. Eligibility by Site and Strata Although the overall ineligible rate was 28.5 percent, ineligibility across sites was 21.8 percent in Baltimore, 38.4 percent in Denver, and 24.1 percent in Spokane. The mobility rate was remarkably stable across the sites (15.8 percent in Baltimore, 17.5 percent in Denver, and 15.5 percent in Spokane). Differences in eligibility were related to the percentage ofprogram participants still receiving food stamps when interviewed. In Denver, where the frame consisted of all active cases in July 1995,20.3 percent of the sample were no longer receiving food stamps. In Spokane and Baltimore, where the sample was drawn from lists of cases active in October 1995, the percentages of the sample no longer receiving food stamps were 8.2 and 5.5, respectively. In the Baltimore site, ineligible rates were remarkably similar across three ofthe four strata (see Table A.2). The ineligibility rate was 19 percent in Stratum 11 (where more than 20 percent of households were under the federal poverty level), 21 percent in Stratum 12 (where 10 to 20 percent of households lived under the federal poverty level), and 19 percent in the rural stratum (Kent 47 TABLE A.2 RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE AND ELIGIBILITY RATES. BY STRATUM Stratum II More than 20 Percent of Poverty Stratum 12 10 to 20 Percent of Poverty Stratum 13 Less than 10 Percent of Poverty Stratum 14 Kent County (Rural) Baltimore Total Number Percentage Number Percentage Number Percentage Number Percentage Number Percentage ':■:■: -." . ■ Eligibility Survey Statas Completed Interviews 61 75J 61 77.2 54 75.0 69 85.2 245 78.3 Cannot Locale 10 113 10 12.7 11 15.3 2J 33 10.5 Refusal/Breakoffs 2 2.5 4 5.1 3 42 1-2 10 3.2 Multiple Unsuccessful Attempts 8 9.9 4 5.1 3 42 7.4 21 6.7 Language Barriers 0 0.0 0 0.0 0 0.0 0.0 0 0.0 Hearth Problem 0 0.0 0 0.0 0 0.0 3.7 3 1.0 Other 0 0.0 0 0.0 1 1.4 0.0 1 0.3 Total Eligible Sample SI 100.0 n ItM 72 100.0 SI 100.1 313 IM.t Eligibility Rates Eligible Sample Ineligible Sample 81 19 81.0 19.0 79 21 79.0 21.0 72 28 72.0 280 SI 19 81.0 190 313 87 78.2 21.8 Total Saaiple Released 1M MM 100 MM 100 MM 1M It* m ito.t Reasons for • ■eligibility Moved from Stratum 12 No Longer Participating 5 Deceased 2 12.0 16 16.0 26 26.0 9 9.0 63 15.8 5.0 5 5.0 2 2.0 10 10.0 22 5.5 2.0 0 0.0 0 0.0 0 0.0 2 0.5 Total Ineligible Sample 19 19.0 21 21.0 28 28.0 19 I9J V 2IJ 48 TABLE A2 (commutd) Stranm2l Stratum 22 Stratum 23 Stratum 24 More than 20 Percent 10 to 20 Percent of Less than 10 Percent of Clear Creek Countv Denver ofPoverty Poverty Poverty (Rural) Total Number Percentage Number Percentage Number Percentage Number Percentage Number Perec nuge Ptaim Eligibility Survey Statuses Completed Interviews 63 Cannot Locate 9 Refusal/Breakofrs S Muhipie Unsuccessful Attempts S Language Barriers 5 Health Problem 3 Other 0 656 94 8.3 8.3 5.2 3.1 0.0 62 81.6 5 66 6 7.9 2 2.6 0 0.0 1 IJ 0 0.0 55 833 0 00 1 15 5 7.6 3 4.5 2 3.0 0 0.0 41 75.9 221 6 111 20 2 37 17 0 00 15 1 19 9 4 74 10 0 00 0 75.7 68 5.7 5.1 31 34 00 Total Eligible Saatptt IN.0 76 100.0 66 100.0 54 IMA 292 Eligibility Rates Eligible Sample Ineligible Sample 96 41 70.1 299 76 49 60.8 39.2 66 52 559 44 1 54 40 574 426 292 182 616 38 4 Total Sample Released 137 100.0 125 100.0 US 180)4 94 100.0 474 ISM Moved from Stratum No Longer Participating Deceased 33 8 0 24.1 58 0.0 25 24 0 20.0 19.2 0.0 13 39 0 11.0 33.1 0.0 12 25 3 12.8 266 3.2 S3 96 3 17.5 203 06 Tctal ladigiMe Saapie 41 29.9 49 39.2 52 44.1 40 424 IS2 38.4 49 TABLE A 2 (ctmliimtd) Stratum3l More than 20 Percent of Poverty Stratum 32 10 to 20 Percent of Poverty Stratum 33 Less than 10 Percent of Poverty Stratum 34 Pend Oreille Counts (Rural) Spokane Total Number Percentage Number Percentage Number Percentage Numbe r Percentage Number rtrccnttfc aMm Eligibility Sarvty Statues Completed Interviews Cannot Locale Rcfusal/Breakoffs Multiple Unsuccessful Attempts Language Barriers Health Problem Other 61 1 7 2 2 1 0 824 1.4 9.5 2.7 2.7 1.4 0.0 62 3 9 5 4 4 0 71.3 3.4 I0J 5.7 46 46 0.0 58 13 70.7 2.4 159 37 6.1 IJ 0.0 73 6 0 7 0 0 5 80.2 66 0.0 7.7 0.0 0.0 5.5 254 12 29 17 II 6 5 76.0 3.6 1.7 5 1 3.3 1.8 0.5 Total Eligible Sample 74 IMJ S7 IMJ 82 IMJ 91 IMJ 334 IMJ Eiigibihty Rales Eligible Sample Ineligible Sample 74 29 71.8 28 2 •7 31 73.7 263 12 32 719 28 1 91 14 86.7 133 334 106 76.0 240 Total Sample Released 163 ion.* US IMJ 114 IMJ 1#5 IMJ 4M IMJ Reasons for Ueligibilir) Moved from Stratum No Longer Participating Deceased 19 10 0 114 9.7 00 It 12 1 15.3 10 2 08 22 10 0 193 8.8 00 9 4 1 86 3.8 10 68 36 2 15.5 u 0.5 Total Ineligible Sample 29 28.1 31 26J 32 28.1 14 13.4 IM 24.1 NOTE: Percentages may not equal 100 due to rounding 50 County). In Stratum 13 (where less than 10 percent of the households lived below the federal poverty level), 28 percent of the sample were ineligible (26 percent had moved from the stratum between the time the sample was drawn and the interview was attempted). At the Denver site, rates of and reasons for ineligibility differed across the strata. The highest ineligibility rate was 44 percent in Stratum 23 (where fewer than 10 percent ofthe households lived below the federal poverty level). One-t rd ofthe sample in this stratum was ineligible because they were no longer receiving food stamps on the interview day. In Denver's rural stratum (Clear Creek County), 43 percent were ineligible. One-quarter were no longer participating in the FSP when interviewed. The county food stamp fraud investigator told MPR that Clear Creek was a highly mobile county; thus, many nonparticipants may have moved. In Stratum 22 (where 10 to 20 percent of the households lived under the federal poverty level), the ineligibility rate was 39 percent, split almost evenly between those who left the FSP (19 percent) and those who moved (20 percent). The ineligibility rate for Stratum 21 (the poorest of the Denver strata) was 30 percent; one-quarter had moved out of the stratum. In the Spokane site, Strata 31 and 33 (more than 20 and fewer than 10 percent of households living under the federal poverty level, respectively), ineligibility accounted for 28 percent of the sample released. In both strata, the number of movers was almost double the number ofthose who no longer participated in the FSP. Ineligibility rates were lower, 26 percent, in Stratum 33 (10 to 20 percent living under the federal poverty ' vel) and much lower. 13 percent, in rural Pend Oreille County. Despite the lower frequency than in other Spokane strata, the ratio of movers to nonparticipants in Pend Oreille County was the same-cases were screened out because of moving twice as often as because of nonparticipation. 51 3. Response Rates and Reasons for Nonresponse Overall, 77 percent of sample members eligible to participate in the survey completed interviews. There were three important reasons for nonresponse. First the sample member could not be located (seven percent). Second, the sample member refused to be interviewed (six percent). For another six percent of the sample, interviewers could not complete the case despite multiple telephone and in-person contacts. Other reasons for nonresponse were language barriers (two percent) and respondents' poor health (two percent). 4. Response Rates and Nonresponse by Site and Strata Response rates were quite even across the three sites. Baltimore's response rate was 78 percent, whereas Denver's and Spokane's were 76 percent Baltimore had the highest nonlocatable rate (10 percent) and the lowest refusal rate (3 percent). In Spokane, the nonlocatable rate was six percent, but the refusal rate was nine percent This pattern of refusal rates increasing with the proportion of sample located may imply that sample members not easily located for this survey were, in reality, reluctant respondents. FSP offices in Denver and Spokane helped MPR with locating by checking administrative records for address updates during the field period. In the Baltimore site, however, only Kent County updated addresses. In Baltimore County, where intensive trafficking investigations were under way independently ofthis study, we were concerned that asking for address updates might hurt sample members by focusing attention on them; therefore, we did not request address updates in Baltimore County. In the Baltimore site, the average response rate in the urban strata was 75.8. The response rate in the rural county was 85.2 percent. Both refusal rates and the percentage of nonlocatable sample were lower in the rural area. 52 In Denver, where the overall response rate was 75.7 percent, two ofthe urban strata were well bove average and one was well below. The response rates for Stratum 22 (10 to 20 percent living under the federal poverty level) and Stratum 23 (less than 10 percent living under the federal poverty level) were 81.6 and 83.3 percent, respectively. The lowest response rate in the Denver site. 65.6 percent, occurred in the poorest stratum. In that area, all the components of nonresponse were higher than average. In the rural stratum, the response rate was 75.9 percent A remarkably high proportion of rural stratum sample members were either nonlocatable (11.1 percent) or unable to complete the interview because ofhealth problems (7.4 percent). In Spokane, the response rates were highest in the most urban and rural strata (82.4 and 80.2 percent, respectively) and lower in the strata with 10 to 20 percent and less than 10 percent of the households living under the federal poverty level (71.3 and 70.7 percent, respectively). In the three urban strata, refusal rates increased as the percentage of households living under the federal poverty level decreased. The rural strata had no refusals, however. a f1 OUBBBMH APPENDIX B FOOD STAMP SURVEY / 4 LU! CASE ID: I—I—l—l_!—!_! INTERVIEWER ID: I I I I I I • FOOD STAMP STUDY A. INTROOUCnOl: Hello, My name Is and I aa from Mathematica Policy Research, Inc., a research company in Princeton, New Jersey. We are conducting a research study for the United States Department of Agriculture. The study is about (food stamps/Independence food stamp benefits). You nay have recently received a letter about this study. The purpose of the study 1s to see how and where people use their (food stamps/Independence food stamp benefits). We will ask questions about you and your household, shopping for food with food stamps, and other uses of food stamps. You have been selected to be part of this very Important study. He would l?ke you to help us by participating 1n a 45-m1nute In-person interview. I can conduct the interview in your home or we can meet at any other place that is convenient for you. B. INFORMATION]: What is this study about? The purpose of the study is to see how and where people use their (food stamps/Independence food stamp benefits). He will ask questions about you and your household, shopping for food with food stamps, and other uses of Who is sponsoring this study? Who is conducting this study? The United States Department of Agriculture Food and Consumer Service is sponsoring this study. Hathematica Policy Research, Inc., an independent research and survey company located in Princeton, New Jersey is conducting the study for them. Why Is Hathematica calling me now? Hathematica would like you to take part in a 45-minute in-person interview. The interviewer can come to your home or will meet you any place that is convenient and comfortable for you. By discussing how and where you use your (food stamps/Independence food stamp benefits), you will help the researchers better understand how to improve the Food Stamp Program. 57 How was I selected? You havt boon randomly selected from (food stamp/Independence food stamp benef\tj clients In the (Denver/Baltimore/Spokane) area. ■t Do I ttmmtto participate? What If I decide not to? AlthoutpB-your participation is vary Important, it is voluntary. Your participation will not affect any benefits you get now or in the future. If I do participate, will the information be confidential? Yes. All information collected will be kept strictly confidential and used only for research purposes with no names attached. The results of the study are for research purposes only. C. ELIGIBILITY SCREEN: 1. Before I tell you more about this study, I need to confirm that I am speaking to the person I'm supposed to speak to. Are you SPEAKING WITH CORRECT SAMPLE MEMBER....01 NOT SPEAKING WITH CORRECT SAMPLE MEMBER - BtD. THANK RESTONOOrr 2. Did ydfi receive food stamps during the month of THIS».MOMW? IYES 01 - GQ mjt%4 NO 00 3. Did you receive food stamps during the month of LAST MONTH? YES 01 NO 00 - I'm sony. This aunray to aaJy for paopto «me raeaivad food thto month w toot. Thai* you for your Unto. 58 4. Oo you live in the READ ZIP CODE OR COUNTY FROM CONTACT SHEET (zip code/county)? YES 01 - NO - I'M SORRY. Wo aro wiry interviewing poopia who two in «M READ (ELKBSLE CODE/COUNTY) FROM CONTACT SHEET. Thank you for your ttano. RECORD SR CODE ORCOUNTY: !_!_:_i_;_; OR KENT 01 CLEAR CREEK 02 RENO d'ORERlE 03 5. IS THIS A TELEPHONE OR IN-PERSON CONTACT? TELEPHONE - SCHEDULE TINE TO CONDUCT IN-PERSON INTERVIEW IN-PERSON - GO TO SECTION D D. AGREEMENT TO PARTICIPATE: MARYLAND AND COLORADO: Before we begin, I would like to tell you how this interview will work. For the next 45 minutes, I will be is ing questions about you and how you use your (food stamps/Independence food stamp benefits). The information you share will be used to improve the Food Stamp Program. All the answers you give me will be kept completely confidential and will never be associated with your name. That means no one at the Food Stamp Office or any other program or agency will know the way you answered these questions. Some questions will be private. When it is time to answer the private questions, I will give you a tape in a Walkman and an answer sheet to fill out by yourself. No one, not ever. I, will know how you answered the private questions. When we are all finished, I will ask you to seal your questionnaire in an envelope. Then, I will leave this sheet [INTERVIEWER: SHOW CONTACT NOTE FORM] with your name and address with you so that you know hat your name and answers are separate. WASHINGTON: HAND STUDY DESCRIPTION TO SAMPLE MEMBER. ASK SAMPLE MEMBER TO FOLLOW ALONG AS YOU READ ALOUD. We are asking you to be in a study that Mathematica Policy Research, an independent research company, is doing for the United States Department of Agriculture (USDA). The USDA is the agency that runs the Food Stamp Program. The study is about how people use food stamps, as well as how they misuse them, for example, by selling them. The USDA will use the study findings to improve the Food Stamp Program. 59 If you agree to be in the study, we will ask questions a bout you and your household, and about how you use you food stamps. Some of the questions are about illegal uses of food staops. You do not have to answer personal questions. The interview will take about 45 minutes to complete, and can be done in your own home or someplace else that you choose. All of your answers will be completely confidential and will never be linked with your name. That means no one at the Food Stamp Office or at any other program or agency will know the way you answered the questions. Your participation in this study is completely voluntary. You can decide not to be interviewed, or you can stop the interview at any time, without any penalty or loss of any benefits you receive now or in the future. Please understand the following: • Nothing you say in the interview can or will be used against you by the State of Washington, the U.S. Department of Agriculture, or any other legal authority. The only exception is that if you give us information about the abuse or neglect of a child, the interviewer must report this to the Department of Social and Health Services. • We will not connect your name, address, or telephone number with your answers. The interviewer will not keep a record of your name, address, or telephone number. • The interviewer has signed a confidentiality pledge that prevents her from linking your answers with you as an individual. There will be a report of all the interviews conducted for this study (about 720), but it will be a summary with no names mentioned. Some que is time tape in not even question envelope name and separate stions about your own use of food stamps will be private. When it to answer the private questions, the interviewer will give you a a Walkman and an answer sheet to fill out by yourself. No one, the interviewer, will know how you answered the private s. When you finish, you should seal your answer sheet in an Then, the interviewer will leave the Contact Sheet with your address with you so that you know that your name and answers are You may also keep a copy of this Study Description. We hope you will agree to be interviewed. If you have any questions about the study, now or in the future, you may call the researchers listed on this form at 1-800-777-0085. Hay we begin the interview now? YES NO. .01 ,00 - RESCHEDULE 60 DATE: | | | - ! | | TBifli "Bay START TINE; | | |:| | | AM 01 PM....02 Al. I'd like to begin this interview by finding out about you and your household. First, when is your birthday? When were you born? "WJNTTT -HAT- -9 '^H A2. Are you now . Married, 01 Living with someone as married, 02 Widowed 03 Divorced, 04 Separated, 05 Or have you never been married? 06 A3. Which of the following best describes where you live now . . . A pi ace you own, 01 A place you rent, 02 A place where you live rent free, 03 You are homeless, 04—] You live in a shelter, group home, U SKIP TO A13, PAGE 7 or treatment facility, or 05J You live in some other type of place? (SPECIFY) 06 61 A4. Counting yourrelf, how many ptoplt live In your housthold. INTERVIEWER: COUNT PEOPLE WHO LIVE IN THE HOUSEHOLD ON A REGULAR BASIS FOR SIX NORTHS OF THE YEAR OF NORE. LIVE ALONE 01 - SKIP TO AlO | | | TOTAL PEOPLE IN HOUSEHOLD A5. How many of these people are children under the age of 16? I | | CHILDREN UNDER 16 INTERVIEWER: AS SOON AS YOU BELIEVE ALL HOUSEHOLD NOSERS ARE ACCOUNTED FOR, YOU NAY CONFIRM A6 AND/OR A7 WITH RESPONDENT. A6. How many of the people in your household are people age 16 and older who are related to you? INTERVIEWER: DO NOT COUNT RESPONDENT. I I | RELATED PEOPLE AGE 16 AND OLDER A7. How many of the people in your household are people age 16 and older who are net related to you? I | | UNRELATED PEOPLE AGE 16 AND OLDER A8. INTERVIEWER CHECK: DOES A5 ♦ A6 ♦ A7 * 1 • A4? YES 01 NO - FIX A4- - A7 A9. Including yourself, how many of the FILL A4 people in your household are covered by your (food stamps/Independence food stamp benefits)? I | | COVERED BY FOOD STAHP BENEFIT 62 A10. Dots the place where you live have a working refrigerator or freezer (that you are allowed to use)? V" 01 W 00 - SKIP TO A12 All. Is the refrigerator or freezer large enough to meet your needs? YK 01 • 00 A12. Does the place where you live have a working stove, toaster oven, or microwave oven (that you are allowed to use)? «■ 01 m 00 A13. On what date did you receive your most recent (food stamps/ Independence Food Stamp benefit)? PROBE: When was your most recent food benefit credited to your Independence Card? '^Tr1"1^1--19 hdk-1 AM. How much did you receive in (food stamps/Independence Food Stamp benefit) on FILL DATE FROM A13. ^ A15' fuftStt SSi J?]1!" 0f lour !Ind PILL DATE FROM A13 do you have left?«P«nd«"«) food stamp benefits from PROBE: That is, what is the value of your food stamp benefits that you have not yet spent? A16. During the past year, for how many months did you receive (food stamps/Independence Food Stamp benefits)? I | | MONTHS OUT OF 12 63 A17. Who takes responsibility for your food stamp benefits each month? That is, who keeps the (food steeps/Independence card) and decides how and where the (food stamps/benefits) will be spent? CIRCLE/CODE ALL THAT APPLY SAMPLE MEMBER 01 SPOUSE/PARTNER 02 CHIL0/STEPCHILD/6RAM0CHIL0 03 PARENT OR STEPPARENT 04 BROTHER OR SISTER 05 GRANDPARENT 06 OTHER RELATIVE 07 - NONRELATIVE 08 J ADMINISTRATOR AT A SHELTER, GROUP HOME, OR TREATMENT FACILITY 09 A18. INTERVIEWER: IS ONLY THE SAMPLE NOSER COOED IN A17? YES, SAMPLE MEMBER ONLY 01 - SKIP TO A20 NO, SAMPLE MEMBER ANO/OR OTHERS 00 SAMPLE MEMBER NOT COOED AT ALL -4 - SKIP TO A20 A19. Do you and your FILL RELATIONSHIP(S) FROM A17 agree about how and where to use the (food stamps/Independence food benefits) . . . All of the time 01 Most of the time 02 Some of the time, or 03 Hardly ever? 04 64 A20. Within how »any days after you rtctlvt your (food stains/Independence food stamp benefits) do you begin to spend them? SAME DAY 01 WITHIN | | | DAYS Oft BETWEEN I I I AND I I I DAYS A21. Do you usually spend the entire amount in one or two days or do you save some for later in the month? SPEND ALL WITHIN ONE OR TWO DAYS 01 - «or TO SECTION • OW MM 10. SAVE SOME FOR LATER IN THE MONTH 02 VARIES/DEPENDS 03 A22. About how much do you usually save for later in the month? PROBE: On average, what is the value of the (food stamps/Independence food stamp benefits) that you save for later in the month? A23. Typically, how long do your (food stamps/Independence food stamp benefits) last each month? IF RESPONDENT SAYS BETWEEN X AND Y, CODE Y. ONE WEEK 01 TWO WEEKS 02 THREE WEEKS 03 ENTIRE MONTH 04 65 Bl. The next questions are about your income and expenses. Let's begin with housing. What was your monthly (rent/mortgage payment) last month? B2. My next questions are about how often your household shopped for food in the last 7 days at supermarkets, neighborhood grocers, convenience stores, and specialty stores. I would like you to include all trips-trips to pick up a few items, as well as major shopping trips by all household members. SHOW CARD A How often in the last 7 days, did your household go to: B3. B3a. B4. a. Supermarkets? b. Neighborhood stores? c. Convenience stores such as 7-11 or stores that sell gas anc groceries? d. Specialty stores such as bakeries, vegetable stands, farme s markets, dairy stores, meat markets, health food stores, c- othc similar places. ASK B2a-d BEFORE GOING TO B3. 00 NOT ASK B3-B4 FOR ANY TYPES • STORES WHERE THE NUMBER OF TRIPS EQUALS ZERO. During the last 7 days, about how much did your household spenc t (TYPE OF STORE)? Include all purchases you made, whether you pe z f: them by check, cash, or (food stamps/your Independence card). K ,'NC TO NEAREST DOLLAR. Does the AMOUNT FROM B3 include the dollar value of food purchasec with (food stamps/your Independence card)? About how much of this AMOUNT FROM B3 was for non-food items such as cleaning or paper products, commercially prepared pet food, or tobacco products? INTERVIEWER NOTE: ALCOHOL IS A FOOD ITEM. Suoenaarkets b. Metghoornood norn e. Convenient* storw a. Specialty stern of Tripe LAST 7MTJ !_U is Tetet LAST 7MTS » U_l_l « l_l_LJ « U_I_J * l-l-t l IfKtl TIS....01 ■0 00 - Tft 01 m oo - m—oi ■0 00 - rts 01 »0 00 - u •J u 11 • l-J_ _LJ l_l 66 B5. Would you say that the amount your household spent on food in the last 7 days was more, less, or about the same as usual? MORE 01 LESS 02 ABOUT THE SAME AS USUAL 03 B6. During the last 7 days, what was the dollar value of purchases made (with food stamps/with your Independence food benefits) by your household? Consider all (food stamp purchases/Independence food benefit purchases) by your household, even for just a few items. B7. Now I'd like to ask about month!v expenditures. First, I am going to ask about the amount your household spent on food in LAST MONTH. During LAST MONTH, how much did your household spend in total at supermarkets, neighborhood grocery stores and convenience stores? Include all purchases you made, whether you paid for them by cash, check, or (food stamps/Independence card). Do not include any expenses for meals eaten away from home or for home-delivered or carry-out meals. B7a. About how much of this AMOUNT FROM B7 was for non-food items such as cleaning and paper products, commercially prepared pet food, or tobacco products? B8. During LAST MONTH, how much did your household spend on food at specialty stores such as bakeries, delicatessens, vegetable stands, farmers* markets, dairy stores, meat markets, health food stores, and other similar places? $11.1111 67 B9. During LAST MONTH, did anyone in your household buy and cat at home ready-to-eat meals like Chinese food, pizza or "fast food" from delivery services, carry out places or pick-up windows? YES. NO.. .01 .00 - SKIP TO 112 BIO. How much did these cost? Include food, beverages, any delivery charges, tax and tips. PROBE FOR ALL INSTANCES OF TAKE OUT « HOME DELIVERED FOOO IN THE PAST MONTH. Bll, B12 B13 Would you say that the amount your household spent on take out and delivered foods in LAST MONTH was more, less, or about the same as usual? MORE 01 LESS 02 ABOUT THE SAME AS USUAL 03 Are you currently working at a job for pay? Include any self-employment. YES 01 NO 00 - SKIP TO R15 How many hours do you usually work per week (at all jobs)? | | | HOURS PER WEEK B14. How much do you earn (at all jobs), before taxes and other deductions? RECORD HOURLY MARE UNLESS RESPONDENT CAN ONLY PROVIDE SALARY FOR A DIFFERENT PAY PERIOD. IF HOURLY WAGE IS NOT RECORDED, RECORD BOTH SALARY AJD PAY PERIOO. PER HOUR OR PER MEEK 01 DAY 02 EVERY TWO WEEKS....03 TWICE A MONTH 04 MONTHLY 05 YEARLY 06 B15. INTERVIEWER: REFER TO A4 ON PACE 6. DOES SAMPLE MEMBER LIVE ALONE? YES. NO.. HOMELESS OR IN GROUP HOME. .01 - SKIP TO BIB .00 -4 - SKIP TO BIB B16. Is anyone else in your household currently working at a job for pay? YES 01 NO 00 - SKIP TO BIB B17. How much (do they/does (he/she)) earn? Please tell me (the combined) wages earned before taxes and deductions by all of the others in your household. PER HOUR OR BVLAM on SHOW CALCULATIONS PER WEEK 01 OAY 02 EVERY TWO WEEKS....03 TWICE A MONTH 04 MONTHLY 05 YEARLY 06 69 818. During last Month, did you receive AFDC, that is Aid to Families With Deptndtnt Childrtn/FILL LOCAL NAMES FOR DENVER/SPOKANE/BALTIMORE. YES 01 NO 00 - SKIP TO B20 B19. How much did you receive from AFDC/LOCAL NAME last month? B20. How much other income did people in your household (including yourself) receive last month. Tell me about income from all other sources such as General Assistance, Unemployment, Social Security, SSI, retirement benefits or any other income you have. Oo not include the amount of the food stamp benefit. PROBE: Please tell me your other income before taxes and other deductions. NO OTHER INCOME 00 B21. Generally, over the past six months, did your income change from month to month? YES 01 NO 00 - SKIP TO B22 70 B21a. What caused your income to change? EARNINGS FLUCTUATED 01 CHILD SUPPORT PAYMENTS FLUCTUATED...02 GIFTS 03 OTHER (SPECIFY) 04 B21b. Was your Income last month more, less, or about the sane as usual? MORE ,...01 MS 02 ABOUT THE SAME 00 B22. Now I am going to ask about various events and conditions that happen to people. I'm interested in those that happened to you during the last 12 months, that is since NEXT MONTH, 1995. As I ask about the specific events, please think carefully so I can record things accurately. First, think about financial matters. Did any of the following happen to you since NEXT MONTH, 1995? a. Car, household appliance, or fumrture inniiiiiid 01 00 b. Pawned or sold off valuables to make ends meet 01 00 c Pressured to pay bills by stores, creditors, or Ml collectors 01 00 d. Maior worsening of your financial condroon Q1 00 Now I am going to ask about specific hardships. Did any of the following happen to you since NEXT MONTH, 1995. e. Fell behind in paying your rent or mortgage 01 00 f. Evicted from your aoartment/house 01 00 a- Had your utilities (water, heat, or electncrrvl shut off 01 00 h. Unable to purchase needed food 01 00 • Unable to afford needed medical care oi 00 j. Had to temporarily live with others or m a shelter or on the street* g^ QQ 71 B23. These next questions are about the foods eaten 1n your household. Which of the following statements best describes the amount of food eaten in your household—enough food to eat, sometimes not enough to eat, or often not enough to eat? ENOUGH FOOO TO EAT 01 SOMETIMES NOT ENOUGH TO EAT... 02—i OFTEN NOT ENOUGH TO EAT 03_T SKIP T0 B25 B24. Do you have enough of the kinds of food you want to eat, or do you have enough but no*, always the kinds of food you want to eat? ENOUGH OF THE KINDS YOU WANT 01 ENOUGH BUT NOT ALWAYS THE KINDS YOU WANT 02 B25. In the last 12 months, did you ever eat less than you felt you should because there wasn't enough money to buy food? YES 01 NO 00-, DOr T KNOW -ij-* *■ T° B28 B26. Did this happen in the last 30 days? YES 01 NO 00-, DON'T KNOW .lJ-* SMP T0 BM B27. In the last 30 days, how many days did you eat less than you felt you should because there wasn't enough money to buy food? I I I NUMBER OF DAYS 72 B28. People do different things when they ere running out of money for food In order to nuke their food money go further. In the last 30 days, did you or anyone 1n your household . . . a. get food or borrow money for food from friends or relatives? b. put off paying a bill so you would have money to buy food? c. get emergency food from a church, food pantry, or food bank? d. eat meals at a soup kitchen? IK NO 01 00 01 00 01 00 01 00 73 raa Cl. These next questions are about your beliefs. Please tell me if you strongly agree, agree, disagree, or strongly disagree with the following statements. SHOW CARD B a. I worry that I won't be abta to do the kind of work that I want to do because I don't have enough education b. I have had as much opportunity to succeed cs people from other neighborhoods c. If a parson works hard, s/hs get ahead STRONGLY d. As I get older, things will get better e. rt is okay to lie ;f it keeps your friends out of trouble f. It is okay to break the law if it helps put food on your family's table g. I like to take chances h. I get upset when I have to wart for something i. I act without stooping to think I l get bored easily 01 01 01 01 01 UNDEODEP DISAGREE STRONGLY DISAGREE 02 02 02 02 02 03 03 03 03 03 04 04 04 04 04 05 05 05 05 05 01 02 03 04 OS 01 02 03 04 05 01 02 03 04 05 01 02 03 04 05 01 02 03 04 05 74 C2. Now I'd like to get your opinions about a variety of issues. In the next set of questions I will read a statement and ask you to tell me if you strongly agree with the statement, agree with the statement, disagree with the statement, or strongly disagree with the statement. SHOW CARD B a. Public assistance workers (or AFOC workers) try to give everybody an even break b. it is sometimes all right to get around the r -lies if vou can get away with it. c. To get ahead, you sometimes have to do something which may be aoainst the law d. Most successful people probably used illegal means to become successful e. People who leave things lying around outside their house should expect that some of their things might be taken or stolen f. It's okav to steal from someon who is rich and can easily replace it B. It's okay to steal from the government since tne government has so much money that it won't hurt them miONQLY DISAGREE STRONGLY DISAGREE 01 01 01 01 01 01 01 02 02 02 02 02 02 02 03 03 03 03 04 04 04 04 05 05 05 05 03 04 05 03 04 05 03 04 05 C3. Now I'd like to ask you how important some things are tell me if they are very important, pretty important, important, not too important, or not important at all, is it . . . to you. Please somewhat How important SHOW CARD C a. ... for you to have a lot VERY IMPORTANT RRETTV rMRORTANT SOMEWHAT NWOftTANT NOT TOO MPORTAMT NOT BMPORTAMT AT AU m n? oa na. n« b. ... for you to have a good reputation in the community? c. ... to plan ahead? 01 01 02 02 03 03 04 04 05 05 75 C4. Please tell me how much you agree with the following statements Would you sty definitely ys, probably yes, probably no, or definitely no. a. If you lit to your public assistance (or AFDC) worker, you w« wind up IOW>Q your benefits b. If you Ml to your public assistance (or AFDC) worker, you will wind up being charged by the police c. If you lie to your public assistance (« AFDC) worker, you win wind uji being tern to court SHOW CARD D vis 01 01 01 02 02 02 UNDeCDCD MMMAMLV NO M—W NO 03 04 05 03 04 05 03 04 05 C5. Now imagine that you get caught selling food stamps. How big a problem would that cause for you? Would it cause 1 . . . SHOW CARD E Very big problem, 01 Big problem, 02 Small problem 03 Very small problem, or 04 No problem at all? 05 C6. Now tell me how much of a problem would it be to have *he following incidents occur in your life. ... If your friends find out that you had done something that was against the law? Would it be a . . . SHOW CARD E Very big problem 01 Big problem 02 Smal 1 probl em 03 Very small problem, or 04 No pro1-1 em at all? 05 76 C7. ... If your relatives find out that you had done something that was against the law? Would it be a. . . SHOW CARD E Very big problem, 01 Big problem 02 Smal 1 problem, 03 Very small problem, or 04 No problem at all? 05 NOT APPLICABLE (NO RELATIVES OR NO CONTACT WITH RELATIVES) -4 C8. ... If your children find out that you had done something that was against the law? Would it be a. . . SHOW CARD E Very big problem 01 Big problem, 02 Smal 1 probl em 03 Very small problem, or 04 No problem at all? 05 NOT APPLICABLE (NO CHILDREN) -4 77 01. My next questions ire about getting together with friends end relatives. Please think about the past year. In a typical week during the past year, how often did you talk on the telephone with friends or relatives? Did you talk on the telephone . . . IF NO REGULAR ACCESS TO A PHONE, CODE LESS THAN THAT. Every day, 07 Five or six times a week, 05 Three or four times a week, 03 Once or twice a week, or , 01 Less than that? 00 D2. In a typical week during the past year, how often did you get together in person with friends, neighbors, or relatives? By get together I mean going out together or visiting in each other's homes? Did you get together ... iv%ry day 07 Five or six times a week, 05 Three or four times a week, 03 Once or twice a week, or 01 Less than that? 00 D3. How often do you attend church or religious services? Do you go Once a week or more, 01 At least once a month, 02 Once every two or three months 03 Three or four times a year 04 Once or twice a year, or 05 Less than that? 00 D4. Do you belong to any clubs or organizations such as church groups, unions, tenant associations, athletic grojps or school groups? ^S 01 m 00 - SKIP TO 06 78 05. How often do you attend the meetings of the clubs or organizations to whi-h you belong? Do you attend meetings. . . Once a week or more, 01 At least once a month, 02 Once tvry two or three months, 03 Three or four times a year, 04 Once or twice a year, or 05 Less than that? 00 06. How long have you lived in this (city/town)? RECORD TO NEAREST YEAR. IF LESS THAU CUE YEAR, RECORD MONTHS. I I I YEARS OR | | | MONTHS D7. How long have you lived at your current address? RECORD TO NEAREST YEAR. IF LESS THAN ONE YEAR, RECORD MONTHS. |_|_| YEARS OR |_|_| MONTHS Please think about the (city/town) where you have lived the longest. D8. How long did you live there? RECORD TO NEAREST YEAR. IF LESS THAN ONE YEAR, RECORD MONTHS. I l__l YEARS OR | I I MONTHS D9. In the past year, has soreone used a weapon, force, or strong arm methods to get money, food stamps or something else from you? YES. .01 m 00 - SKIP TO 012 010. How many times in the past year did this happen? I I | TIMES IN PAST YEAR 79 Oil. In the most recent event, what was the person trying to get from you? CIRCLE ALL THAT APPLY MONEY 01 DRUGS 02 FOOD STAMPS/OTHER BENEFITS 03 S" 04 OTHER (SPECIFY) 05 012 (In addition to that,) in the past year, have you had your pocket picked or your purse or wallet sn
Click tabs to swap between content that is broken into logical sections.
Title | Final report for the food stamp participant trafficking study |
Date | 1998 |
Contributors (individual) | Ciemnecki, Anne. |
Contributors (group) |
United States Dept. of Agriculture Food and Consumer Service Office of Analysis and Evaluation. Mathematica Policy Research, Inc. |
Subject headings |
Food stamp fraud--United States Food stamps--United States |
Type | Text |
Format | Pamphlets |
Physical description | xii, 111 p. ;28 cm. |
Publisher | Alexandria, Va. : U.S. Dept. of Agriculture, Food and Consumer Service, Office of Analysis and Evaluation, |
Language | en |
Contributing institution | Martha Blakeney Hodges Special Collections and University Archives, UNCG University Libraries |
Source collection | Government Documents Collection (UNCG University Libraries) |
Rights statement | http://rightsstatements.org/vocab/NoC-US/1.0/ |
Additional rights information | NO COPYRIGHT - UNITED STATES. This item has been determined to be free of copyright restrictions in the United States. The user is responsible for determining actual copyright status for any reuse of the material. |
SUDOC number | A 98.2:F 73/19 |
Digital publisher | The University of North Carolina at Greensboro, University Libraries, PO Box 26170, Greensboro NC 27402-6170, 336.334.5304 |
Full-text | oolt-A-ol s jfft.xifyyi USDA It of Agriculture Food and Nutrition Offloaof An*y*md EwfcwMon Synthesis Report for Food Stamp Trafficking Study May 1998 ^ rt-ottte ¥3 (L- Contract No.: MPR Reference No.: 53-3198-3-013 8171-091 FINAL REPORT FOR THE FOOD STAMP PARTICIPANT TRAFFICKING STUDY March 1998 Authors: Anne Ciemnecki Lara Hulsey James Ohls Irving Piliavin Mercer Sullivan Josh Rossol Submitted to: Office of Analysis and Evaluation USDA Food and Consumer Service 3101 Park Center Drive Alexandria, VA 22302 Project Officer: Sharron Cristofar, Program Analyst Submitted by: Mathematica Policy Research, Inc. P.O. Box 2393 Princeton, NJ 08543-2393 (609) 799-3535 Project Director: James Ohls Survey Director: Anne Ciemnecki USDA non-discrimination Statement: The U.S. Department of Agriculture (USDA prohibits discrimination in all its programs and activities on the basis ofrace, color, national origin, gender, religion, age, disability political beliefs, sexual orientation, or marital or family status. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA's TARGET Center at (202) 720-2600 (voice and TDD). To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room 326-W, Whitten Building, 14th and Independence Avenue, SW, Washington, DC 20250-9410 or call (202) 720-5964 (voice and TDD). USDA is an equal opportunity provider and employer. u ACKNOWLEDGMENTS Sharron Cristofar, the United States Department of Agriculture project officer for the study, has provided continuing important input throughout the design, survey, and analysis of the project. Useful input has also been obtained from her colleagues, Ted Macaluso, Steven Carlson, and Margaret Andrews. At Mathematica Policy Research, Inc., Linda Mendenko contributed extensively to every phase of the study and efficiently managed the field data collection. Lynn Mackenzie served as an effective field supervisor. John Burghardt reviewed the report and provided valuable suggestions and help with revisions. The text was much improved through the editing of Roy Grisham, Jane S. Miller, and Marjorie Mitchell. Cathy Harper, Cynthia Castro, Monica Capizzi. and Jennifer Baskwell produced the final document. We are indebted to two field ethnographers who collected valuable data for the project Steven Koester and Owen Murdoch, as well as to RIVA Market Research, Inc., for carrying out the post-survey focus groups. We also thank the survey respondents for their time in participating in the data collection. 111 M»»&Mf USDA UnM*d AfrieuNura Food and Consumer Service 3101 Park Center Drive Alexandria. VA 22302-1500 Final Report for the Food Stamp Participant Trafficking Study Jury 1998 Enclosed for your information is a copy of the "Final Report for the Food Stamp Participant Trafficking Study." Food stamp trafficking occurs when someone sells their coupons for cash at a discount, rather than using them to buy food. The Food and Nutrition Service (FNS) treats trafficking as a serious violation of program integrity both because it is illegal and because it undermines the nutritional goals of the program. FNS fights trafficking through undercover investigations of food retailers that are suspected of engaging in trafficking. The reason for this is that authorized food retailers are the only ones who can redeem coupons for cash from the Government. FNS also works with State Food Stamp Agencies, who are responsible for investigating trafficking by participants, and has received requests to conduct survey research studies on the trafficking behavior of participants. The enclosed report presents the results of a methodological test of whether it is possible to obtain reliable and valid information about participant food stamp trafficking through a survey of active participants. We reviewed procedures used in other survey research studies of illegal behavior and tested the approach which appeared to work best in those other surveys. However, we found that the method does not work reliably for food stamp trafficking. The report is a valuable methodological contribution that will be of interest to individuals who study program integrity and law enforcement issues. In addition to the feasibility test of survey research methods, the report presents results of focus groups with food stamp participants. The focus group results discuss the terminology participants use to refer to trafficking and suggest hypotheses that might be tested in future research. Ifyou have questions regarding this report, please contact Dr. Theodore F. Macaluso, Chief, Food Stamp Evaluation Research Branch, Office of Analysis, Nutrition and Evaluation at (703)305-2019. Enclosure AM EQUAL OPPORTUNITY EMPLOYER W CONTENTS Chapter Page EXECUTIVE SUMMARY ix I INTRODUCTION 1 A. LIMITATIONS OF THE STUDY 2 B. OVERVIEW OF REPORT 3 II DATA COLLECTION ACTIVITIES CONDUCTED FOR THE STUDY 5 A. PRE-SURVEY FOCUS GROUPS 5 1. Material Covered 5 2. Sampling and Recruiting 6 3. Fielding 6 B. THE PARTICIPANT SURVEY 7 1. Survey Content 7 2. Approach to Data Collection g 3. Sampling 10 4. Fielding Procedures 11 5. Fielding Results 12 C POST-SURVEY FOCUS GROUPS 12 1. Material Covered 12 2. Fielding 13 D. ETHNOGRAPHIC RESEARCH 14 1. Approach 14 2. Content 15 3. Fielding 15 CONTENTS (continued) Chapter Page HI ASSESSMENT OF WHETHER THE PARTICIPANT SURVEY ELICITED ACCURATE SELF-REPORTS OF TRAFFICKING 17 A. THE SURVEY SELF-REPORT DATA ON TRAFFICKING 18 B. COMPARISON TO OTHER AVAILABLE INFORMATION 20 1. Comparisons with Survey Data on Neighborhood Prevalence and of Trafficking by Friends 20 2. Pre-Survey Focus Groups 23 3. Post-Survey Focus Groups 23 4. An External Estimate of Trafficking Incidence 24 IV OTHER HYPOTHESES SUGGESTED BY STUDY RESULTS 25 1. Many Buyers in Food Stamp Trafficking Transactions May Buy the Coupons for Their Own Use; Sometimes, They Themselves May Be Food Stamp Program Participants 25 2. There Is No Clear "Profile" of Characteristics of People Who Traffic 26 3. "Middlemen" May Not Be a Major Factor in Trafficking 27 REFERENCES 31 APPENDIX A: SURVEY METHODOLOGY 33 APPENDIX B: FOOD STAMP SURVEY 55 vi TABLES Table Page 111.1 REPORTED PARTICIPATION IN TRAFFICKING 19 111.2 PERCEPTIONS ABOUT THE EASE AND PREVALENCE OF FOOD STAMP TRAFFICKING 21 A. 1 RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE AND ELIGIBILITY RATES, BY SITE 46 A.2 RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE ELIGIBILITY RATES, BY STRATUM 48 Vll /HI mmp« EXECUTIVE SUMMARY Exchanging food stamps for cash or for nonfood goods and services (food stamp trafficking) is a serious federal policy concern. Unauthorized use offood stamps weakens the ability of the Food Stamp Program (FSP) to accomplish its objective of encouraging nutritious food use by program participants, and trafficking undermines confidence in the program among the general population. Most investigations of food stamp trafficking focus on food retailers as they are the only ones who can obtain cash from the government for food stamps. A recent U.S. Department of Agriculture (USDA), Food and Nutrition Service (FNS) study of authorized food stores suggests that between three and four percent of food stamps are exchanged for cash or nonfood items (Macaluso 1995). Because of extensive anecdotal evidence suggesting that food stamp trafficking is common, the USDA contracted with Mathematica Policy Research, Inc. (MPR) to conduct an exploratory study of food stamp trafficking, focusing on the motivations and dynamics oftrafficking from the point ofview of FSP participants. This report summarizes the findings from that study. STUDY OBJECTIVES The present study was conceived as an exploratory research project, designed to obtain preliminary information about trafficking by food stamp participants and to examine the feasibility of further research. In that vein, one key objective of the project was to test the feasibility of obtaining accurate survey information from FSP participants about their trafficking experiences. A second goal was to obtain preliminary data on the characteristics and motivations of people who exchange their food stamps for cash or for goods, so as to help devise deterrence strategies that the government could employ to reduce trafficking. DATA COLLECTION Several different data collection activities were undertaken as part ofthe research: A series of focus groups was conducted with FSP participants in three metropolitan areas in different parts ofthe country. The goal was to obtain information that would help in designing a survey of participants, asking them about their attitudes toward trafficking, their trafficking behavior, and the dynamics of trafficking in their neighborhoods. The focus groups included an explicit discussion of whether the participants would be willing to share information about their trafficking behavior in an interview. A survey of720 FSP participants was conducted in those same three metropolitan areas and in nearby rural areas. This survey obtained information about respondent household characteristics, household income and expenditures, attitudes and opinions about ix trafficking; respondents' social support systems; and respondents' actual experiences with trafficking. One part of the interview tested an innovative approach to obtaining survey data about trafficking, under which respondents listened to questions on an audio tape and wrote down the answers on a one-page answer sheet. Based on a review of methods employed with similar difficult survey situations, this method had seemed the most promising available, because it had been used with apparent success in survey applications involving illegal behavior. It was hoped that by ensuring that the interviewer would not know their answers, respondents would answer honestly. Post-survey focus groups wen; conducted at two of the three survey sites with respondents in the survey. These focus groups discussed the survey findings (see below) with participants in an attempt to gain additional insight into the response patterns seen. Ethnographic research was conducted at two of the survey sites. In this work, researchers with extensive "street-lever contacts in low-income areas attempted to identify and interview buyers of food stamps, to discuss their experiences with trafficking and to obtain additional insight into the dynamics of trafficking and the motivations of food stamp sellers. SAMPLING In light of the exploratory nature of the research, we decided not to allocate the resources that would have been required to obtain a nationally representative sample of FSP participants. However, the three metropolitan areas selected for the field work were chosen through a combination of purposive and random methods, so that a reasonable cross-section of sites would be ensured in terms of area of the country, size of the metropolitan area, and use of electronic benefit transfer (EBT) food stamp issuance methods. For each of the three metropolitan areas, we chose four different survey locations to achieve a mix of poverty densities and urban/rural locations. We chose respondents randomly from among FSP participants living in those areas. FINDINGS CONCERNING THE FEASIBILITY OF OBTAINING ACCURATE SELF-REPORTS OF TRAFFICKING Despite success elsewhere, the method of using audio tapes to ask questions about food stamp trafficking does not appear to measure the phenomenon accurately. In the current survey, the weight of the evidence suggests that there was significant underreporting of trafficking behavior. Overall, approximately 5.1 percent of the survey respondents reported selling food stamps in the previous month. Together with survey data on the average amount of benefits sold per transaction, this implies that approximately two percent of dollar benefits were sold. While there is no "gold standard" comparison with which to test the validity of these estimates, they appear to be low when compared to the following sources: Respondents were asked a series of questions about how common they thought trafficking was in their neighborhoods. While the answers reflect only opinions and are difficult to quantify, the pattern of responses suggests a considerably higher prevalence of trafficking than the survey indicates. Similarly, during the pre-survey focus groups, participants were asked how common they believed trafficking was. The general discussion suggested greater incidence of trafficking than the survey answers showed. During the post-survey focus groups, which were conducted with survey respondents, the participants were asked whether they had felt "safe" in replying honestly to the questions on trafficking. Several reported feeling that being fully candid during the survey was unsafe. It was apparent that they had replied negatively to many questions when the truthful answer was affirmative. The survey-based estimates are lower than would be expected, according to the findings of a 1995 FNS analysis of trafficking. The FNS analysis, which did not include all forms of trafficking, estimated that at least 3.8 percent of coupons were trafficked in 1993, well above the estimates from the survey. None of these sources is conclusive. The totality of the evidence, however, suggests that it is quite likely that there is significant underreporting in the survey. HYPOTHESES ABOUT TRAFFICKING SUGGESTED BY THE STUDY The evidence from the focus groups, the survey, and the ethnographic research, taken together, gives rise to a number of interesting hypotheses about trafficking. Because ofthe limited scope of this preliminary study, with data collection in only three areas of the country, none of these findings can be viewed as conclusive. All, however, appear quite likely from the evidence generated by the study. Following are key hypotheses: Many buyers in food stamp trafficking transactions may purchase the a»tponsfor their owt use; sometimes, they themselves may be FSPparticipants. Participants in the pre-survey focus groups repeatedly voiced their belief that the buyer in a food stamp trafficking transaction is often a food stamp recipient, and that many buyers use the stamps themselves at the grocery store. Focus group members reported that, as a way of stretching their food budgets, they and their friends often bought f od stamps from people who wanted to sell them, with several viewing this practice as simply prudent household management. They also reported that it was common for a recipient to be both a buyer and a seller over the course of a month, perhaps selling coupons initially to obtain cash for some high-priority use and then, if cash became available later, using it to buy coupons at a discount and gain access to food more cheaply. XI There is no clear "profile "ofcharacteristics of people who traffic. During preliminary analysis of the survey data, we examined whether self-reports of trafficking were correlated with demographic, social, or economic characteristics of respondents. While some weak possible correlations were identified, no single characteristic or set of characteristics stood out as being highly predictive of trafficking behavior. To be sure, as noted earlier, there is evidence that trafficking was substantially underreported in the survey, and this weakens our ability to identify significant correlations, if in fact they exist. Electronic benefit transfer (EBT) may change the dynamics ofthe transaction. One of the three study sites had been issuing food benefits under an EBT system for several months prior to the study. A number of respondents at that site reported trafficking, and the focus group discussions yielded insights as to how this trafficking took place. The dynamics of the trafficking under EBT are apparently quite different from trafficking in food coupons. The buyer and seller reportedly often go to the store together, because the buyer wants to avoid (1) prepaying for an EBT card on which benefits have been depleted, and (2) prepaying for an EBT card that has been reported as lost or stolen. The general feeling about EBT as it relates to trafficking was that "where there is a will there is a way." Focus group respondents reported that the cycle with which benefits were posted to the EBT cards contributed to selling benefits. In the EBT site, the AFDC benefit was credited early in the month. At this time, recipients needed food and spent the welfare benefit on groceries. The food benefit was credited four or five days later. By the time the food benefit was credited, food had been purchased and recipients needed cash. "Middlemen ' may not be a majorfactor in trafficking. One issue of considerable interest in the current study was the role of "middlemen" who buy coupons and then sell them at a profit, either to people who plan to use them for food or to stores that can "launder" diem. It was believed at the jutset of the study that such middlemen might represent a significant share of buyers. However, no evidence emerged from the research to suggest this. During the pre-survey focus groups, participants were asked to talk about what categories of people bought food stamps. The group participants tended to focus on two types of buyers: (1) retail stores that cashed the coupons at banks, and (2) low-income people who intended to use the coupons to purchase food (as discussed in Section 1, above). The existence of middlemen who bought the coupons to make money by reselling them was seldom mentioned by the focus group participants, even after direct probing. Similar findings were obtained during the ethnographic research. The ethnographers probed extensively about middlemen and found virtually no evidence that they exist. None of the 10 respondents interviewed, all ofwhom had themselves been buyers of food stamps tot various reasons, reported ever having bought coupons and reselling them for cash. xn I. INTRODUCTION The Food Stamp Program (FSP) represents a key component of America's "safety net" for low-income households. With annual outlays in benefits of more than $19 billion in 1997, it has an average caseload of more than 22 million people each month. The program is thus an important source of support for America's poorest households. Furthermore, by distributing benefits either as coupons or through electronic benefit transfer (EBT) accounts, both of which can legally be used only to purchase food, the FSP helps assure that America's poor have access to nutritious meals. In light of the importance of the program, food stamp trafficking-particularly program participants exchanging food stamps for cash-represents a serious concern. Trafficking weakens the ability of the FSP to encourage nutritious food use by program participants, and it undermines confidence in the program among the general public. Unfortunately, the very nature of trafficking makes the practice difficult to study. In most contexts, asking respondents in a survey to report on their socially unacceptable behavior runs the risk of underreporting, since respondents do not want the interviewers to think badly of them. In the case of trafficking, this problem is greatly exacerbated by the illegality of the actions being asked about. Because trafficking is illegal, FSP participants are understandably reluctant to discuss any trafficking activities in which they may be involved. In particular, they are often skeptical about confidentiality pledges and fearful that admitting to selling coupons could lead to loss of benefits or even criminal prosecution. This report summarizes the results of a research study that attempted to overcome these barriers and to learn more about trafficking, with a particular focus on the participant side of the trafficking transaction. Recognizing that obtaining accurate information from participants was likely to be 1 difficult, the study was designed with several different components. Data collection activities used both qualitative and quantitative techniques. One component of the study-a survey of a total of more than 700 program participants living in three different parts ofthe country-was designed to test a promising method for obtaining self-reports from participants concerning whether and how they had engaged in trafficking. In addition, the survey was designed to obtain preliminary information about trafficking with which to develop hypotheses that could be examined in later research. Supporting the survey were several other types of data collection: (1) pre-survey focus groups, conducted to help develop the survey instrument by obtaining information about trafficking from recipients; (2) post-survey focus groups of respondents to probe further about bow they answered the survey questions and ascertain the motivations that determined their responses; and (3) ethnographic interviews with purchasers of food stamps to ask them about trafficking activity and about the characteristics and apparent motivations of traffickers. This report summarizes the findings ofthe study. A. LIMITATIONS OF THE STUDY Several significant methodological limitations, many of which were recognized from the outset of the study, must be kept in mind when assessing the results in this report. The project was designed as an exploratory study, not as a definitive analysis of participant trafficking. Because at the outset so little information was available about either participant trafficking behavior or about how to obtain information from participants about trafficking, it was felt that a full, detailed study oftrafficking on a national basis was not warranted. Rather, the goals ofthe study were more modest and focused on obtaining information that could help shape further studies. In light ofits objectives, the study was limited to three areas ofthe country. While an effort was made to choose geographically separate areas with different characteristics, the clustering ofdata collection into three areas precludes making valid national generalizations ofthe results. While the results may be suggestive about hypotheses concerning trafficking for the country as a whole, we cannot assess the representativeness of the three areas chosen for the study and therefore cannot formally generalize the results. Furthermore, as discussed in Chapter III, there is considerable evidence that techniques used in the survey were less successful than had been hoped in eliciting accurate reports oftrafficking. It is likely that a number of respondents in fact engaged in trafficking but did not admit during the survey to having done so. Furthermore, we are not able to determine whether the traffickers who did say they trafficked are systematically different from the traffickers who did not admit to trafficking. It is thus not possible to make reliable generalizations about traffickers from the survey results, even leaving aside the high degree of clustering in the survey. B. OVERVIEW OF REPORT Chapter II provides details about the data collection activities undertaken for the study. Chapter III assesses the degree to which it was possible to elicit accurate self-reports of trafficking during the participant survey. Chapter IV highlights a number of other hypotheses about participant trafficking that have emerged from one or more of the data collection activities. Appendix A provides details about the data collection work, and Appendix B reproduces the data collection instrument used in the participant survey. < "O y mmm D. DATA COLLECTION ACTIVITIES CONDUCTED FOR THE STUDY It was felt that a multifaceted data collection strategy would best serve the objectives of the study. Details of the methods used follow. A. PRE-SURVEY FOCUS GROUPS Two pre-survey focus groups were conducted with FSP participants in each ofthe three primary areas where the survey was to be done (see below).1 These sessions were intended to help die project team learn more about the trafficking process and about how participants thought about and talked about trafficking. It was anticipated that this would be useful both in suggesting topics to be covered during the surveys and in wording items as clearly as possible to respondents. 1. Material Covered The exact topics covered in the focus groups evolved slighUy over time, as we drew from the outcomes ofthe earlier groups in planning later ones. In general, however, the following seven areas were included: 1. Introductory Material The opening material was designed to explain the purposes of and ground rules for the focus groups. Also included was an exercise to facilitate self-introductions by participant group members and to get them started talking to one another. 2. Using Food Stomps. The first substantive part of the focus groups elicited information about how respondents use their food stamps, including their experience at stores. This provided a nonthreatening way of opening the discussion ofalternative ways coupons can be used. 'Details about how the three areas for data collection were selected and about the other aspects ofthe data collection—particularly, the household survey-are presented in Appendix A. 3. Coping Experiences. This discussion focused on how food stamp recipients cope with unexpected expenses that may arise during the month. It provided a context for the discussion oftrafficking to follow. 4. Trafficking. The discussion asked how commonly trafficking was believed to occur in the neighborhoods of the group participants. It also asked about the mechanics of trafficking, in terms of how buyers and sellers find each other, what types of people are buyers, and how much coupons sell for. This discussion also focused on what types of goods and services could be bought with food stamps and participants' general attitudes toward trafficking. Motivations for trafficking also were discussed. 5. Trial of Taped Interview. To see how well it would work, we gave respondents the taped interview that was planned for the general survey. While the sample sizes were too small to test the success of the method in eliciting accurate response, it was possible to test the logistics of the self-administered audiotape-based interviewing process. 6. Discussion of Willingness to Share Personal Information About Trafficking. A discussion was initiated concerning whether and under what circumstances focus group participants would be willing to provide accurate information about their trafficking experiences. 7. Closing. In closing, group participants were thanked for their help and were given their payments for participating. 2. Sampling and Recruiting The focus group participants were randomly sampled from lists of active participants provided by the state food stamp offices and were recruited in advance from MPR's survey telephone center. The focus groups had approximately 12 members each. 3. Fielding Each focus group was held in a "neutral" location, such as a community center or a hotel, which was convenient for the group participants to reach. Each was moderated by a senior MPR staff member with extensive experience in conducting focus groups. At least one assistant was also present at each group. The interview were taped, and transcripts were prepared. B. THE PARTICIPANT SURVEY The participant survey had two overall objectives. The first was to test a strategy for obtaining accurate self-reports oftrafficking behavior from participants. (See Section 2.) The second was to obtain information with which to develop hypotheses about participant trafficking, such as the characteristics of traffickers, their motivations, and the particular circumstances that lead them to traffic. To some degree, accomplishing the second of these objectives depended on accomplishing the first, since accurate self-reports about who trafficks are necessary to describe the respondent characteristics associated with trafficking. 1. Survey Content The data collection instrument consisted ofa series of modules designed to obtain information about the respondents themselves and their knowledge of and participation in trafficking. The following information was obtained: • Household Information and Food Stamp Benefit Receipt Module A collected background information on the respondent's age and marital status. It also covered number and ages of other people in the household, number of people covered by the food stamp benefit, availability of adequate cooking and food storage appliances, date and amount of most recent food stamp benefit, responsibility for food stamp benefit spending, and typical food stamp spending pattern. • Household Income and Expenditure. Module B collected information about the respondent's income and expenses, food shopping trips and expenses over the past week and month, employment status and earnings ofhousehold members, participation in Aid to Families with Dependent Children (AFDC) and other government programs, recent hardships experienced, and food sufficiency.2 • Attitudes, Opinions, and Beliefs About Foot* Stamp Trafficking. Module C collected information about the respondent's beliefs and opinions in several areas, including 2AFDC has since been replaced by the Temporary Assistance For Needy Families (TANF) program. outlook for the future, the public assistance system, consequences of lying to an AFDC worker, and the consequences of selling food stamps. Social Support. Module D collected information about the respondent's connection to friends, family, and other support networks in the community. Some questions were designed to elicit the degree to which the respondent was stable and established in the community; others probed whether and how respondents were recently victimized by crimes. Buying and Selling Food Stamps. Module E contained questions about the respondent's general knowledge and perception oftrafficking activity in the community: the ease or difficulty with which trafficking occurs, which store types and which people are involved in trafficking, ethical and moral views on trafficking, consequences of selling food stamps, and items that are exchanged for food stamps. At the end of this module, the respondent used an audiotape to answer a self-administered series of questions about whether he or she had bought or sold food stamps during the previous month and, if so, the value of the coupons transacted. Demographic Characteristics. Module F collected information on the respondent's ethnicity and racial background, gender, educational attainment, and history ofpublic assistance participation. It also solicited the respondent's opinions for ways in which the FSP might be improved and invited the respondents to possibly participate in a post-survey focus group. Interviewer Observation. The final section of the questionnaire required the interviewer's observation of the physical environment. 2. Approach to Data Collection Most of the data collection instrument was administered using standard in-person interviewing methods, with the interviewer reading each question, probing as necessary, and then recording the answer on a questionnaire. However, 14 questions-those concerning the respondent's own experiences buying and selling food stamps-were administered using an audio recording and headphones. The respondent listened to the prerecorded questions through headphones and then recorded his or her answers on a self-administered answer sheet, which was then sealed inside an envelope. The interviewer then signed the back of the envelope over the seal, which was not to be broken until the survey center of Mathematica Policy Research, Inc. (MPR) received the envelope for data processing. The decision to use audiotape recordings reflected two major considerations. First, the survey literature contains considerable evidence that respondents are more likely to report sensitive or illegal behaviors in a self-administered questionnaire format than in an oral interview. Apparently, an audiotape interview provides a sufficient feeling of anonymity to respondents that they answer honestly about any involvement in illegal activities. In a study of alternative ways of measuring drug use, Turner et al. (1992) found that "the self-administered questionnaire yielded higher estimated prevalence rates. As predicted, examination ofthe ratios indicates that the advantage of the self-administered questionnaire increases with the presumed sensitivity ofthe drug in question." Similarly, Gfroerer and Hughes (1992) conclude that "for sensitive question such as those on the use of illicit drugs, it is likely that people will be more willing to reveal their drug use on a self-administered answer sheet than in a verbal response—whether by telephone or in person—to an interviewer." Hay (1990) compared standard interview and self-administered techniques and found that the frequency distributions indicated that a significantly higher percentage of the questionnaire respondents reported ever having more than a sip or taste of an alcoholic beverage. Similarly statistically significant differentials were observed between the interview and questionnaire respondents on reported smoking." However, because the study population for the current research was expected to have low levels of reading ability, there was concern that a traditional self administered instrument might lead to considerable confusion on the part cf respondents, thus reducing our ability to obtain accurate replies.3 Thcretv re, we decided to have respondents listen to a self-administered audiotape and then mark their response on a simple answer sheet. A study by Camburn et al. (1991) provided evidence that this can be an effective approach. In particular, this study had good success in using audiotapes to ask respondents who were 12 to 21 years old a set of questions about such sensitive issues as drug use, alcohol use, sexual practices, and eating disorders. 3. Sampling Because the study was exploratory, the survey was limited to three areas of the country. A combination of random and purposive sampling was used to assure diversity among the areas in terms ofsize ofpopulation, geographical location, and food stamp issuance methods. The following areas were selected for data collection: Baltimore, Maryland; Denver, Colorado; and Spokane, Washington. Within each of these three primary areas, urbanized zip code areas were stratified into four groups, based on the percentage ofhouseholds below the poverty level in the zip code areas.4 One zip code was then randomly selected from three of the four strata (The stratum with the lowsst concentration ofpoverty was not sampled.) Furthermore, for each of the three primary areas initially chosen (Baltimore, Denver, and Spokane), the rural counties contiguous to the urban area were listed. One of the counties in each primary sampling area was then randomly chosen. 3About 10 to 20 percent of U.S. adults and a higher percentage of food stamp recipients have reading difficulties. The percentage ofU.S. adults who completed high school is 71 percent Fifty-six percent ofthe food stamp recipients in die San Diego Food Stamp Cash-Out Demonstration and 41 percent in the Alabama Food Stamp Cash-Out Demonstration completed high school. 4These were defined as zip code areas where at least 80 percent ofthe population was in areas classified by the Census Bureau as urban. 10 Overall, therefore, four subareas were chosen from each ofthe primary sampling areas. Three ofthese were urbanized, with varying poverty concentrations; the fourth was rural but adjacent to the urban area. Once the local sampling areas had been defined, either in terms of zip codes or in terms of the designated rural counties, MPR obtained, from the state food stamp offices, lists offood stamp cases in these local areas. Respondents were then randomly sampled from these case lists. Because many ofthe study objectives required having a substantial representation oftraffickers in the sample, and because it was believed that the likelihood oftrafficking might be correlated with the amount ofthe food stamp benefit, households were sampled with probabilities of selection proportional to the amount of the benefit, as listed in the sampling frames. 4. Fielding Procedures Interviewers were hired and trained for the project at each of the three primary sites. The selection criteria for hiring interviewers included strong verbal skills, an ability to work effectively with low-income respondents, and good organizational skills. All but 2 of the 26 interviewers hired had interviewing experience. Interviewer training began with advance study and completion of written training exercises. In-person training, held at local hotels, then lasted eight hours. Trainers explained the background and purpose of the study, reviewed the screener and the interviewer- and self-administered questionnaires, provided instructions for asking each question, discussed methods for contacting respondents, addressed sensitivities about criminal behavior, and went over administrative procedures. After the main session, interviewers finished their training by completing mock interviews with a supervisor. 11 The interviewing was conducted between December 1995 and April 1996. An experienced supervisor managed the interviewers by telephone. The interviewers returned their data to MPR's main survey operation center near Princeton, New Jersey, for logging, data entry, and verification. 5. Fielding Results A total of 720 interviews were completed: 245 in Baltimore, 221 in Denver, and 254 in Spokane.3 The overall response rate among eligible respondents was 77 percent. The main reasons for nonresponse were interviewer inability to locate the respondent (seven percent); refusal on the part of the respondent (six percent); and interviewers' inability to complete cases, despite multiple attempts (six percent). Other reasons for nonresponse included language barriers (two percent) and poor health ofrespondents (two percent). C. POST-SURVEY FOCUS GROUPS An additional round of focus groups was conducted several months after the survey was completed. Based on resource considerations, we decided to conduct these focus groups at two of the three survey sites (Denver and Baltimore). The purposes ofthese post-survey focus groups were (1) to examine apparent discrepancies between data obtained in parts of the survey and other available information, and (2) to provide more-detailed followup about some ofdie survey findings. The participants in these focus groups were recruited from among the earlier survey respondents. 1. Material Covered The seven main areas covered by the post-survey focus groups were the following: 'One interview was subsequently lost in the mail. 12 1. Introduction. The sessions began with a discussion of their purposes and of the ground rules to be followed. In addition, self-introductions were made. 2. FoodStamps-Baseline Knowledge. The first set of topics was designed to provide a nonthreatening way to get respondents to talk about the use of their food stamps. To stimulate the discussion, questions were asked about participants' perceptions of the mechanics of how the program works and about how food stamps relate to their food budgeting. Toward the end of this module, participants were asked to give their reactions to a number of situations in which food stamp recipients were faced with financial problems that might lead them to traffic. 3. Survey Data: Incidence of Trafficking. The preliminary survey results showing a relatively low reported incidence oftrafficking were summarized for the members ofthe focus groups. These participants were then asked whether they thought the survey results were accurate. They were also asked how they had reacted to the relevant survey questions and whether, if they had been trafficking, they would have reported having done so in the interviews that used the recorded tapes. 4. Survey Data: Profiles of Traffickers. In part because no clear profile of trafficker characteristics had emerged from the preliminary analysis of the survey data, focus group participants were asked to indicate what characteristics, if any, they would associate with trafficking. 5. Survey Data: Urban Versus Rural Trafficking. Based on preliminary survey results, it appeared that trafficking was more common in urban than in rural areas. The focus group moderator probed for reasons. 6. Survey Data: About the Trafficking Process. Focus group participants were asked to comment on survey data about the ease with which buyers and sellers of food stamps could be located in their neighborhoods. 7. Additional Suggestions. Focus group participants were asked for their suggestions as to how it might be possible to reduce trafficking. 2. Fielding Two post-survey focus groups were conducted in each of the three primary sampling areas. Each was moderated by a senior member of MPR's subcontractor, RIVA Market Research, and observed by an MPR staff member. MPR recruited focus group members from lists of survey respondents who had expressed an interest in participating in subsequent focus ^groups. 13 D. ETHNOGRAPHIC RESEARCH It was believed that food stamp intermediaries could provide an additional perspective on FSP participants who trafficked their benefits; therefore, the research design called for interviews of people who acted as brokers in recipient trafficking exchanges. This work is described here. 1. Approach Because ofpotential difficulties locating food stamp buyers and getting them to cooperate with the study, we decided to enlist the help of ethnographic researchers who already had street-level contacts in the areas where the survey had been conducted. In particular, we identified urban ethnographers working in both Baltimore and Denver, and we asked them to locate food stamp buyers and interview them in detail. These two researchers had spent years conducting drug-related studies in their respective areas and had contacts among residents of neighborhoods with high poverty rates. In reviewing this work, it is important to note that the researchers' past experience had focused on studying drug use. Because of the difficulty of identifying appropriate respondents and the intensive nature of the data collection, ethnographic research, by its very nature, seldom involves random samples of subjects. In the current case, the fact that the two researchers involved had previously studied drug use may have skewed the profile ofrespondents toward people with drug problems. In developing this line of data collection, we designed a protocol that described die research issues to be covered in the interviews and identified die types of respondents we were interested in. Based on this protocol, each ethnographer drew upon his local contacts to identify suitable respondents and then conducted the interviews. 14 2. Content Because of the unstructured nature of ethnographic research, it was not appropriate to develop detailed specifications as to the exact content of the interviews. Rather, the interviews were allowed to proceed in whatever directions seemed most fruitful to accomplish the basic research objectives. In general, however, the interviews covered topics similar to those covered in the household survey but did so from the perspective of food stamp buyers rather than sellers. Major topics included (1) the respondents' experiences with trafficking; (2) their perceptions of the motivations of traffickers; and (3) the logistics of trafficking, including locations where trafficking take place, what prices coupons sell for, and how buyers and sellers come together. Respondents were asked to report both their own experiences and their perceptions of common practices regarding trafficking in their neighborhoods. 3. Fielding Five interviews were conducted in Baltimore and five in Denver. When field conditions permitted and respondents were willing, the interviews were taped, and recordings were made of six of the interviews. In the four instances where taping was not possible, the analysis was based on interviewer notes from the discussions with respondents. The results of the individual interviews were then summarized by the two field ethnographers who conducted them, and a further synthesis was then conducted. o-^. I ® // l\\\ n HI. ASSESSMENT OF WHETHER THE PARTICIPANT SURVEY ELICITED ACCURATE SELF-REPORTS OF TRAFFICKING As discussed earlier, a key objective of the project was to assess the feasibility of obtaining accurate reports oftrafficking by using a survey-based data collection procedure. Based on evidence from earlier research done on other illegal activities, it seemed reasonable to expect that fairly accurate responses could be obtained if the respondent listened to a recording ofthe questions using a headphone and then marked his or her responses on an anonymous answer sheet. This approach was intended to make the respondent feel "safe" in reporting illegal activities and to reduce respondents' tendencies to give answers that they think their interviewers viewed as socially acceptable. However, based on the survey responses to the questions about whether respondents trafficked, coupled with other information on the prevalence of trafficking, we believe that there is a high likelihood that the survey reports significantly understated the prevalence of trafficking. In particular, it appears that the number of respondents who reported having engaged in trafficking behavior is lower than we would have expected based on any of the following sources of information: (1) other questions from the survey related to the general perceived prevalence of trafficking; (2) information on trafficking from the focus groups conducted for the project; and (3) an independent estimate of certain types of trafficking made by USDA for the year 1993. To be sure, these alternative sources ofcomparison information about trafficking prevalence are themselves subject to error, and as a result our inferences are not conclusive. Nevertheless, the weight of the evidence strongly suggests underreporting in the survey. The discussion below 17 provides details of these arguments. First, we present the relevant survey self-report data. Then we examine these data in light of the comparison information. A. THE SURVEY SELF-REPORT DATA ON TRAFFICKING Three percent of the survey respondents said that they had sola food stamps for cash the previous month, and two percent had exchanged them for goods or services (Table 111.1). The overall reported rate of selling or exchanging food stamps was 5.1 percent The reported cash value of food stamps sold ranged from $20 to $180. The mean value of benefits sold was $61.08, which is approximately 41 percent of average household benefits. These data suggest that about two percent of benefits were sold.' In contrast, two percent of respondents said that they had bought food stamps; that is, they obtained food stamps either for cash or in exchange for goods or services. Of these, about one percent said they received food stamps in exchange for cash and one percent said that they received food stamps in payment for goods or services. The value of food stamps bought varied from $5 to $352, with an average of$77. Most respondents who reported trafficking indicated that they either bought or sold, but very few reported both. Only about one-half of one percent of the sample bought and sold during the month. 'Calculated as 5.1 times 0.41, where 5.1 is the probability of selling and 0.41 is the average percentage sold if the recipient sells. 18 TABLE HI. 1 REPORTED PARTICIPATION IN TRAFFICKING (Percentages Unless Otherwise Noted) Full Sample (N = 719) Selling Food Stamps Sold Food Stamps for Cash 3.4 Received Goods or Services in Exchange for Food Stamps 2.1 Received Cash, Goods, or Services for Food Stamps 5.1 Buying Food Stamps Bought Food Stamps for Cash 1.0 Received Food Stamps in Exchange for Goods or Services 1.0 Received Food Stamps for Cash, Goods, or Services 2.0 Any Trafficking 6.5 Mean Value of Food Stamps Sold (in Dollars)* 61.08 Minimum 20.00 Maximum 180.00 Mean Value of Food Stamps Bought (in Dollars)* 77.05 Minimum 5.00 Maximum 352.00 SOURCE: Food Stamp Trafficking Survey, weighted data. NOTE: N indicates sample size. Not all sample members responded to all questions. For each question on this table, between 0 and 9 respondents did not provide responses. 'The value of food stamps bought and sold are calculated based on responses from those who reported buying or selling food stamps. 19 B. COMPARISON TO OTHER AVAILABLE INFORMATION Unfortunately, there are no definitive data on the incidence oftrafficking by participants with which to compare the survey results. Furthermore, even if definitive national data were available, our ability to assess the accuracy of the survey data would be significantly limited by the fact that they are drawn from only three areas of the country, and we don't know how nationally representative these areas are. Despite these problems, however, a number of comparisons between the self-report data and other sources are of interest. We present them below. 1. Comparisons with Survey Data on Neighborhood Prevalence and of Trafficking by Friends Besides being asked directly about their own trafficking behavior, respondents in the survey were asked several questions about their perceptions of the prevalence of trafficking in their communities One survey question, for instance, asked how many food stamp recipients in the respondents' neighborhoods sold food stamps for cash in any one month; choices included "most people," "some," "a few," and "nobody." (Note that this question focused specifically on selling for "cash.") Approximately 2.8 percent of respondents answered that "most people" that were FSP participants in their neighborhoods trafficked, and another 7.6 percent and 8.5 percent, respectively, responded "some" and "a few." (See Table MX) About 9.7 percent said "nobody." Thus the number of respondents answering "most people" or "some people" exceeded the number selecting the "nobody" category. How these results are assessed depends greatly on what is assumed about the 71.3 percent of respondents who gave a "don't know" response to the question. At one extreme, it is possible that those answering "don't know" did so because there is no significant trafficking in their 20 TABLE mi PERCEPTIONS ABOUT THE EASE AND PREVALENCE OF FOOD STAMP TRAFFICKING (Percentages) Full Sample (N-719) In any one month in your neighborhood, how many food stamp recipients sell some or all ofthem for cash? Most people 2.8 Some 7.6 A few 8.5 Nobody 9.7 Don't know 71.3 Refused/missing 0.3 If someone in your neighborhood had cash and wanted food stamps, would it be easy to find someone to sell them? Very easy 23.1 Somewhat easy 13.3 Somewhat difficult 8.1 Very difficult 15.6 Don't know 39.5 Refused/missing 0.4 How many of the three food stamp recipients you know best sold food stamps for cash in the past year? None 34.0 One 8.2 Two 4.2 Three 5.1 Don't know three food stamp recipients 26.4 Don't know how many sold food stamps 21.9 Missing 0.3 How often did these friends sell benefits in the past year? Every month 4.9 Nearly every month 2.5 Once in a while 4.3 Question not applicable 82.3 Missing 1.0 SOURCE: Food Stamp Trafficking Survey, weighted data. NOTE: N indicates sample size. 21 neighborhoods. On the other hand, it is plausible that, despite their answers, the "don't know" respondents lived in neighborhoods that were quite similar in terms of trafficking to those of the people who did answer the question. If any significant number ofthe "don't know" respondents live in neighborhoods where substantial amounts oftrafficking do occur, this would suggest a prevalence rate higher than that implied by the small number (3.4 percent) of respondents who reported having themselves sold coupons for cash trafficked in the previous month. A related survey question asked how easy it would be in the respondent's neighborhood for someone who had cash and wanted to buy food stamps to find a seller. As shown in Table III.2,36.4 percent of respondents replied that finding a seller would be either "very easy" or "somewhat easy," compared to 23.7 percent who indicated that it would be "somewhat difficult" or "very difficult." (About 40 percent indicated that they didn't know how easy it would be.) A third relevant questioning sequence asked respondents to think about their three best friends on food stamps and estimate how many trafficked during a 12-month period. Approximately SO percent of respondents said that they knew three other food stamp recipients and had an opinion as to whether the people they knew trafficked. Approximately 17.5 ofrespondents (continuing to use the overall sample as the base) thought that at least one of the people they knew trafficked. Furthermore, of those who thought their friends trafficked, most believed that the friends did so "every month" or "nearly every month." (In these two categories, 7.4 percent responded, compared to only 4.3 percent responding "once in a while.") While not conclusive, these findings suggest that the food stamp recipients who responded to the survey perceive that trafficking occurs frequently in their communities. This is in contrast to their direct self-reports of trafficking. 22 2. Pre-Survey Focus Groups As described in Chapter II, the design work for the study included conducting focus groups of FSP participants in each of the planned survey sites. While it is not possible to quantify the focus group responses precisely, the general impression gained from both the sessions and the transcripts was that several of the participants in each session had engaged in trafficking activity and that virtually all participants were aware of trafficking taking place among their acquaintances. Respondents were able to speak with some confidence about such details as the cash price of coupons, the locations in their communities where trafficking often occurred, and what items could and could not be easily bought with food stamps. As with the written survey, these focus group sessions provide evidence that trafficking is a much more common phenomenon than suggested by the self-reports discussed in Section III.A. 3. Post-Survey Focus Groups To further explore the likely accuracy of the survey responses and to learn respondents' perceptions about the questions and our interviewing approach, we held focus groups in two ofthe survey sites several months after the survey. During this second round of focus groups, samples of the respondents to the earlier survey were assembled and asked to describe their reactions to the survey, particularly the self-administered taped interview. In addition, the focus group members were shown the survey results concerning the incidence of trafficking and asked whether they thought them believable. The responses strongly suggest that some of the respondents did not provide accurate self-reports concerning their trafficking activities. Several indicated that they had distrusted the survey and suspected that it was a "sting" operation to detect food stamp fraud. Furthermore, several focus group participants explicitly said that the use of the tape recorder did not significantly increase their 23 confidence that their replies would be kept confidential. These focus group responses provide significant evidence that the incidence oftrafficking is significantly underreported in the survey data. 4. An External Estimate of Trafficking Incidence A USDA study (Macaluso, 1995) used evidence drawn from federal trafficking investigations to estimate the approximate level oftrafficking in the United States in 1993. The data include only trafficked food stamps that were sold directly to stores and therefore were not ultimately spent for food; they thus exclude sales to people who then used the coupons to obtain food in a store. The USDA study estimated that food stamp trafficking in 1993 amounted to about $800 million, approximately 3.8 percent ofall food stamps issued during the year. As noted earlier, the tabulations from the current survey imply that about two percent ofdollar benefits are trafficked. This is further evidence that trafficking is significantly underreported in the survey. In addition, the questions in the current survey asked about all transactions involving selling food stamps (including selling them to a person who was planning to use them to buy food), whereas the USDA study, because of the methodology it used, was able to assess sales of food stamps only to buyers who then "laundered" them through stores. Thus, if the numbers could be put on a comparable basis, the gap between the survey numbers and the USDA estimates would further widen. For instance, if, as the survey data suggest, considerable selling takes place to buyers who buy for their personal use, then adjusting the survey data to exclude mis portion oftrafficking would reduce the survey-based estimates ofthe value ofcoupons trafficked to well below 2 percent, further increasing the discrepancy with the USDA estimate of 3.8 percent. 24 IV. OTHER HYPOTHESES SUGGESTED BY STUDY RESULTS As discussed earlier, the overall project was an exploratory study designed both to assess an apparently promising technique for surveying food stamp recipients about their trafficking activities and to identify which other hypotheses about trafficking behavior might warrant further study. As it turned out, the "headphone" interviewing approach we tested appears not to have been successful in eliciting reliable reports of trafficking activities. However, the various research activities conducted for the project, including the survey, the focus groups, and the ethnographic research, have yielded a number of interesting hypotheses that may warrant further study. Several of these hypotheses are discussed below, together with their supporting evidence. 1. Many Buyers in Food Stamp Trafficking Transactions May Buy the Coupons for Their Own Use; Sometimes, They Themselves May Be Food Stamp Program Participants Participants in the pre-survey focus groups repeatedly voiced their beliefthat the buyer in a food stamp trafficking transaction is often a food stamp recipient, and that many buyers use the stamps themselves at the grocery store. Focus group members reported that, as a way of stretching their food budgets, they and their friends often bought food stamps from people who wanted to sell them, with several viewing this practice as simply prudent household management. They also reported that it was common for a recipient to be both a buyer and a seller over the course of a month, perhaps selling coupons initially to obtain cash for some high-priority use and then, if cash became available later, using it to buy coupons at a discount and gain access to food more cheaply. This theme of food stamp recipients buying food stamps for their own use but also selling them came up in the focus groups in all three cities in which the groups were held. 25 The ethnographic research conducted for the study also found evidence that many buyers of food stamps are purchasing them for their own use, rather than intending to resell them to stores that can "launder" them through the regular banking system. In both cities where these interviews were conducted, respondents indicated that most purchasers they were aware ofwere buying the coupons because they represented a cheap way ofobtaining food. In Baltimore, for instance, the ethnographic researchers interviewed a respondent who lived in the suburbs but regularly came into the city to purchase food stamps at a discount so that she could use them to obtain food. These transactions were made easier because she had previously lived in the city and retained ties to her old neighborhood. In Denver, one ofthe respondents in the ethnographic study was a crack dealer wl occasionally traded crack for food stamps (about S65 ofcoupons per month) for her own use in buying food but would not accept more food stamps than she needed for food. In another Denver interview, the household (a couple) were drug addicts. Usually, they bought food stamps for use only in buying food, though occasionally they used the stamps to buy drugs. 2. There Is No Clear "Profile" of Characteristics of People Who Traffic During preliminary analysis ofthe survey data, we examined whether self-reports of trafficking were correlated with demographic, social, or economic characteristics of respondents. While some weak possible correlations were identified, no single characteristic or set of characteristics stood out as being highly predictive oftrafficking behavior. To be sure, as noted earlier, there is evidence that trafficking was substantially underreported in the survey, and this weakens our ability to identify significant correlations, if in fact they exist. (Indeed, the apparent underreporting oftrafficking is the reason that we have not formally analyzed the survey data in greater detail.) However, the fact 26 that no strong correlations were present in the survey data is at least consistent with the conclusion that traffickers do not fali into readily identifiable groups. We explored this issue further during the post-survey focus groups, because the preliminary analysis of the survey data had shown a surprising lack of regularity or patterns in trafficker characteristics. Focus group participants were asked to discuss among themselves what characteristics they would expect a person selling food stamps to have and were given several specific sets of choices (for example, men/women, younger people/older people; people with kids/people without kids). In general, no strong patterns emerged from these discussions. While there was some tendency for the groups to focus on certain characteristics, such as being young and being a woman, it appeared that the discussion group members were really focusing more on the profile of the food stamp recipients they knew rather than on the characteristics of traffickers from within that group. Overall, no clear consensus about a pattern of characteristics emerged. This would be consistent with a hypothesis that many different types of people traffic for many different reasons. 3. "Middlemen" May Not Be a Major Factor in Trafficking One issue of considerable interest in the current study was the role of "middlemen" who buy coupons and then sell them at a profit, either to people who plan to use them for food or to stores who can "launder" them. It was believed at the outset of the study that such middlemen might represent a significant share of buyers. However, as discussed below, no evidence emerged from the research to suggest this. During the pre-survey focus groups, group members were asked to talk about what groups of people bought food stamps. The group participants tended to focus on two types of buyers: (1) retail stores who cashed the coupons in at banks, and (2) low-income people who intended to use the 27 coupons to purchase food (as discussed in Section 1, above). The existence of "middlemen," who bought the coupons to make money by reselling mem, was seldom mentioned by the focus group participants. Furthermore, even after direct probing, most of them said that they did not know of such people. In addition, in one of the questions on the participant survey (QE7), respondents were asked what type ofpeople might be likely to buy food stamps. The answer categories, which were all read by the interviewer, included such responses as "people you live with," friends," "storekeepers," and, most important for the current discussion, "a middleman who buys food stamps and sells them to someone else." (Multiple responses were allowed.) Fewer than five percent of the respondents chose this "middleman" response as one of their replies. Similar findings were obtained during the ethnographic research. The ethnographers probed extensively about middlemen and found virtually no evidence that they exist. None of the 10 respondents interviewed, all of whom had themselves been buyers of food stamps for various reasons, had ever bought coupons and resold them for cash. Furthermore, they were not aware of people (other than retailers who cashed the coupons in at banks) who did so. Indeed, several ofthese respondents indicated that they were not able to figure out how a middleman could make money. In their experience, the discount price of coupons was fixed in any given neighborhood at one rate (usually SO percent); they felt that, if somebody bought coupons at SO percent, it would be hard to convince another person or a store to repurchase them at a higher rate. In their experience, coupons could be bought at the SO percent rate easily, and they didn't see why anybody would pay a middleman a higher rate. The fact that these respondents, who were located through street contacts, were not themselves major middlemen is not surprising, since such people, if they exist, would probably be harder to 28 interview. However, these respondents were being interviewed not only about their own experiences but also for their perspective on their local social scenes, and the fact that they had not heard of middlemen is reasonably telling. 4. Electronic Benefit Transfer (EBT) Does Not Completely Prevent Recipient Trafficking bat Does Change the Dynamics of the Transaction One of the three study sites had been issuing food benefits under an EBT system for several months prior to the study. A number of respondents at that site reported trafficking, and the focus group discussions yielded insights as to how this trafficking took place. The dynamics of the trafficking under EBT appeared to be quite different from trafficking in food coupons. Focus group participants indicated that, under EBT, the benefit is usually sold to someone the seller knows and trusts, since if the transaction is with a stranger, it is riskier and more time-consuming for both buyer and seller. In general, sellers were said to "spend down" the amount of the food benefits they needed for themselves before turning t!se EBT card over to a buyer. If the seller knows the buyer, the buyer may pay in advance and the seller may give out an identification (PIN) number with the card. If the buyer and seller do not trust each other, the buyer often accompanies the seller to the store. In this situation, the seller shows his or her own identification and pays for the groceries as if he or she is shopping; payment is made after the pair leave the store. The buyer and seller were reported to go to the store together often because the buyer wants to avoid (1) prepaying for an EBT card on which benefits have been depleted, and (2) prepaying for an EBT card that has been reported as lost or stolen. The general feeling about EBT as it relates to trafficking was that "where there is a will there is a way." Focus group respondents reported that the cycle with which benefits were posted to the EBT cards contributed to selling benefits. In the EBT site, the AFDC benefit was credited early in the 29 month. At this time, recipients needed food and spent the welfare benefit on groceries. The food benefit was credited four or five days later. By the time the food benefit was credited, food had been purchased and recipients needed cash. 30 REFERENCES Camburn, Donald, Marcie Cynamon, and Yossi Hard. "The Use of Audio Tapes and Wrinen Questionnaires to Ask Sensitive Questions During Household Interviews." Paper presented at the National Field Directors Conference and National Field Technologies Conference, San Diego, May 1991. Gfroerer, Joseph C, and Arthur L. Hughes. "Collecting Data on Illicit Drug Use by Phone." Survey Measurement ofDrug Use: Methodological Studies. Rockville, MD: National Institute on Drug Abuse, 1992. Hay, David A. "Does the Method Matter on Sensitive Survey Topics?" Survey Methodology, vol. 16, no. 1, June 1990, pp. 131-136. Macaluso, Theodore F. "The Extent of Trafficking in the Food Stamp Program." Washington, DC: U.S. Department ofAgriculture, Food and Consumer Service, August 1995. Turner, Charles F., Judith T. Lester, Barbara J. George, Michael L. Hubbard, and Michael B. Witt. "Effects of Mode of Administration and Wording on Data Quality." Survey Measurement of Drug Use: Methodological Studies. Rockville, MD: National Institute on Drug Abuse, 1992. 5 AU 31 M uro mm APPENDIX A SURVEY METHODOLOGY 3>3 di This appendix presents the methods used to select the sample, conduct the Food Stamp Recipient Trafficking Survey, and process the data. A. METHODS FOR SELECTING AM) LOCATING RESPONDENTS 1. Sample Design A combination ofpurposive and probability sampling was used to select three states from which the survey sample would be drawn. FCS's specifications for the work were that each state be "stratified by level of urbanization to include one local area food stamp office considered to be central city, one in a metropolitan area not considered to be central city, one in a suburban area, and one in a sparsely populated (rural) area." Random probability methods were used to select respondents from each local office area. Mathematica Policy Research, Inc. (MPR) used a three-stage sample design to implement this requirement Stage One involved selecting three metropolitan areas in different parts ofthe nation. Stage Two involved selecting a range ofurban to rural strata within and near each metropolitan area. Stage Three involved selecting Food Stamp Program (FSP) participants within strata. a. Stage One, Selecting Metropolitan Areas • The sample frame consisted of all 134 Metropolitan Statistical Areas for the 48 contiguous states and Washington, DC. The areas were stratified into three geographic strata based on FCS regions (Mid-Atlantic, Northeast, and Southeast; Midwest, Mountain Plains, and Southwest; and West) and three population size groups (more than 3 million, 1 to 3 million, and 250,000 to 1 million). Areas also were classified as issuing benefits using or not using Electronic Benefit Transfer (EBT) technology. • A random number between 0 and 1 was assigned to each area. • We used the random number to pick die EBT site first Baltimore, Maryland, had tin. lowest random number and was selected. Baltimore, in the Eastern region, was the largest stratum. 35 • Next we randomly selected L*,* third region as the second site. Using the lowest random number, we selected Spokane, Washington, among all the areas in the third region that were not in the largest size stratum. Spokane was in the smallest size stratum. • Denver, Colorado, represented the midsized population stratum in the Central region. b. Stage Two, Selecting Strata Within Metropolitan Areas • We listed every zip code in each of the three selected metropolitan areas, along with its total population, population in poverty, and population living within die urban area. • After excluding zip codes that were less than 80 percent urban, we sorted the remaining zip codes by the percentage of their households living below the federal poverty level. • We formed four strata in each area, based on the percentage living in poverty. Thestrata ranged from the very highest poverty density to no poverty. • After excluding zip codes in the lowest poverty density strata, we randomly selectedone zip code from each stratum. The selected zip codes were: Baltimore Denver Spokane 20 percent or more poverty 10 to 20 percent poverty 5 to 10 percent poverty 20 percent or more poverty 10 to 20 percent poverty 5 to 10 percent poverty 20 percent or more poverty 10 to 20 percent poverty 5 to 10 percent poverty 21231 21237 21210 80223 80207 80228 99202 99205 99016 To select a rural area, we listed every rural county contiguous to the metropolitan area and selected one at random. The rural counties selected were: Baltimore Kent County Denver Cleark Creek County Spokane Pend Oreille County • The three metropolitan sites and four subsites formed 12 strata. 36 c Stage Three, Selecting Respondents • MPR asked states for lists ofFSP participants in each stratum (that is. zip code or rural county). The states sent machine-readable data sets, which an MPR programmer edited. • The programmer sorted cases in each stratum by size of food stamp benefit. She then used interval sampling to select a sample proportionate to the size of the monthly benefit. She selected 150 cases in the Baltimore and Denver strata and 200 cases in the Spokane strata.' Cases that would have entered the sample with certainty were selected before applying the interval method. • In the "less than 10 percent" stratum in Baltimore and Denver, there were not enough FSP participants to yield 60 completed interviews. For these strata, we supplemented by adding randomly selected zip codes from the same stratum. We added zip codes 21206 and 21220 to 21210, and zip codes 99223 to 99016. Samples were drawn on the combinations of zip codes. 2. Obtaining Contact Information for Respondents Maryland and Colorado sent names and addresses of FSP participants to MPR. Washington State mailed the sample frame to MPR without names and addresses. MPR selected the survey sample and sent the results back to the state for the consent mailing. Washington sent a letter to potential sample members, offering them a two-week period in which to decline to participate. Only 2.3 percent of those to whom letters were mailed objected. 3. Screening Criteria To be eligible to participate in the study, individuals had to be active food stamp participants and be living in the stratum from which they were selected. To determine eligibility, interviewers administered a short screener before conducting the interview. Eligibility rates are discussed in Section E. 'More cases were needed in Spokane because their Human Research Review Board required a passive consent procedure before releasing names to MPR. r 37 B. SELECTION AND TRAINING OF FIELD STAFF 1. Hiring MPR hired and trained 26 interviewers for this project; 11 worked in Baltimore, 9 in Denver, and 6 in Spokane. To the extent possible, we hired interviewers who lived in the communities in which they interviewed. All but two interviewers had prior interviewing experience. The number of interviewers hired per site depended on the hours per week each could work. In Spokane, where each interviewer could work more hours per week, we hired fewer interviewers. In Baltimore, where experienced interviewers had other project commitments, we needed more. The goal was to hire enough interviewers to have at least 120 hours of labor available each week. 2. Training Interviewer training had three components. The first was a two-hour self-study module to prepare for the main training session. We expected interviewers to spend two hours reviewing the questionnaire and related materials a few days prior to the main training session. We required them to complete written training exercises to prove that they had mastered the materials. (New inter, iewers had additional time to view MPR's video, "General Interviewer Training," and to review lessons with their supervisor.) The main training session lasted eight hours. We held sessions at a local hotel. Trainers explained the background anc purpose of the study, reviewed the screener, went over the interviewer- and self-administered questionnaires, provided instructions for asking each question, discussed methods for contacting respondents, and addressed sensitivities about criminal behaviors associated with interviewing. Interviewers had ample time for role-playing, practice interviewing, and administrative procedures. After the main session, interviewers finished their training by completing mock interviews with a supervisor. 38 MPR trained the Baltimore interviewers on December 2. The Baltimore training session served as "Trainers* Training." Anne Ciemnecki, the survey director, was the lead trainer. Sharron Christofar, the Contracting Officer's Technical Representative, discussed the importance of the project to FCS. Linda Mendenko, MPR's field coordinator, covered administrative materials. Lynne MacKenzie, MPR's field manager, led the role-playing and mock interviewing. Denver and Spokane main trainings were held on December 9 and 16. Lynne MacKenzie was the lead trainer for the Denver and Spokane sessions. C. METHODS FOR COLLECTING THE DATA 1. Timing of the Field Period Interviewing began immediately after training in Baltimore and Denver. The Spokane sample was not available until December 29. We completed interviewing by April 6. Following are the numbers of interviews completed weekly at each site: Baltimore Denver Spokiine Total Week Ending Number Percent Number Percent Number Percent Number Percent December 9 23 9.4 0 0.0 0 0.0 23 32 December 16 35 14.2 13 5.9 0 0.0 48 6.7 December 23 21 1.5 3 1.4 0 0.0 24 3.3 December 30 22 9.0 7 32 1 0.4 30 42 January 6 22 9.0 12 5.4 19 7.5 53 7.4 January 13 9 3.7 16 7.2 38 15.0 63 8.8 January 20 20 L2 23 10.4 26 10.2 69 9.6 January 27 17 6.9 16 12 22 8.7 55 7.6 February 3 11 4J 23 10.4 19 7.5 53 73 February 10 13 53 18 8.1 32 12.6 63 8.8 February 17 18 7.3 20 9.0 23 9.0 61 8.5 39 Baltimore Denver Spokane Total Week Ending Number Percent Number Percent Number Percent Number Percent February 24 9 3.7 14 6.3 22 8.7 45 63 March 2 11 4.5 9 4.0 13 5.1 33 4.6 March 9 0 0.0 10 4.5 5 2.0 15 2.1 March 16 4 1.6 4 1.8 3 \2 11 IJ March 23 5 2.0 5 22 6 2.4 16 22 March 30 4 1.6 17 7.7 15 5.9 36 5.0 April 6 1 0.4 11 5.0 10 3.9 22 3.0 Total 245 100.0 221 100.0 254 100.0 720 1M.0 To encourage interviewer productivity at the end ofdie project, MPR offered field interviewers a bonus of$20 for every interview and $5 for every screener completed after March 22. This kept interviewers' enthusiasm high when sample was sparse. It also kept interviewers motivated to finish their assignment rather than to move to easier, newer projects. 2. Ensuring Respondent Confidentiality Because trafficking in food stamps is illegal, MPR took two important steps to ensure respondents' confidentiality. First was documenting contacts with the respondents. Usually, we document information about the respondent (case identification number, name, address, telephone number, and directions to the home) on a contact sheet detailing the interviewer's attempts to reach the sample member. For this study, we separated the contact information onto two forms, a Record of Contacts Form and a Contact Notes Form. The Contact Notes Form contained identifying information used to locate the sample member and schedule the interview. The Contact Notes Form was left with the respondent at the end of the interview to assure the respondent that his or her name would not be linked with data. The Record of Contacts Form contained the case-tracking number 40 and a record ofattempts to reach the respondent (that is, number and time ofcalls, interim case status codes). Interviewers mailed the form to MPR along with the completed interview. They wrote no names, addresses, or other information that could identify the sample member in any way on the Record of Contacts Form. The second step in ensuring confidentiality was to use an audiotaped interview and self-administered answer sheet to ask about the sample member's direct involvement in food stamp trafficking. The respondent used audio headphones to listen to 14 questions about personal buying or selling of food stamps. He or she recorded answers on a self-administered answer sheet and sealed them in an envelope. The interviewer signed his or her name across the back ofthe envelope over the seal. An unbroken signature implied that the envelope had not been opened before receipt at MPR. The audiotaped interview provided all the advantages ofa self-administered, pencil-paper questionnaire in increasing reports of illegal or socially disapproved behavior. It was intended to overcome the suspected barrier of poor reading skills in the food stamp population. Ninety-six percent ofthe survey respondents used the self-administered audiotape and answer sheet. Those who did not fell into three categories: 1. People with print impairments (visual disorders, innumeracy, and attention disorders) who could not focus on the answer sheet well enough to fill it out 2. People whose primary language was not English or Spanish who needed an interpreter for the interview 3. Those who refused MPR produced a Spanish language version ofthe audiotape. 41 3. Field Management and Reporting Procedures Field interviewers received most of their assignments at the beginning of the field period. Distribution of assignments early allowed interviewers to plan their time and travel efficiently. Adjustments to assignments were made as data collection progressed. Field interviewers reported progress to their field supervisor weekly by telephone at prearranged times: hours worked, expenses, and case-by-case progress. During the reporting session, the supervisor provided interviewers with feedback from MPR's quality review process, handled administrative needs such as supply orders, and answered nonurgent questions. Interviewers were encouraged to call supervisors immediately for urgent matters. The supervisor reported summaries of field progress and expenses to MPR's field coordinator weekly to provide management with information that guided staffing and sample-release decisions. 4. Problems Faced During the Field Period Two serious problems arose during data collection. Six weeks after interviewing began, an interviewer was robbed at gunpoint at the Highlander Ridge Apartment complex in Baltimore. Highlander Ridge is a gated public housing project. Because of this incident, MPR mandated that escorts accompany interviewers working in the Highlander Ridge complex. Escorts may be other interviewers but are often friends or relatives of the interviewers who travel with interviewers. Escorts added to the cost of the data collection, but no new incidents were reported after we established the policy. The second problem was in the Denver stratum with 10 to 20 percent of the households living below the federal poverty level. Here, a child was killed in a drive-by shooting during the third week of interviewing. We advised interviewers to stay out of the area for a few weeks until emotions cooled and interviewers felt more comfortable working there. 42 D. QUALITY ASSURANCE AND DATA PROCESSING PROCEDURES Interviewers mailed completed work to MPR weekly. As quality control clerks received field interviewers' packages, they logged completed work into a questionnaire tracking system. Each completed interview had: • A Record of Contacts Form • The interviewer-administered main questionnaire • The self-administered answer sheet from the audiotaped interview The clerks entered the date they received each ofthe three components of a completed interview. Weekly reports listed cases reported by the interviewer as complete but not received at MPR, and cases where one or more components were missing. MPR used these reports to track missing documents. Most of the time, when paper was missing, the work was completed but not mailed. Two interviewers reported work completed that was not, in fact, done; MPR terminated those interviewers and reassigned their cases to other staff members. Usually, MPR validates field interviewer work by calling respondents or mailing a thank-you note with a prepaid return postcard for respondent to acknowledge that the interview was completed. To protect respondent confidentiality, we were unable to link contact information with interview data for this study. Therefore, quality control callbacks and the usual means of verification were not possible. Instead, we verified interview completion by comparing the sample member's birth date (and age) and monthly benefit amount from the administrative files provided by the states with the data collected during the interview. The questionnaire tracking file contained these two variables. As the clerk logged completed work into the tracking file, she compared the age and monthly benefit amounts 43 recorded in the interview with data on the administrative record. (Record data were not available to the interviewer.) She assigned codes to indicate the closeness ofthe match. For example: Birth Date: Codel Code 2 Code 3 Code 4 Very strong match Probable match Weak match No match Exact date Month and year correct; day may be incorrect; or day and year correct; month off by one month or two months; or month and day correct, year offby up to three years Two out ofthree elements do not match, but not too far off None ofthe three elements match Age: Codel Code 2 Code 3 Code 4 Very strong match Probable match Weak match No match Exact age Off by 1 to 2 years Offby3to4years Offby 5 or more years Benefit Amount: Codel Code 2 Code 3 Code 4 Very strong match Probable match Weak match No match Respondent amount within S percent of amount on file Respondent amount within 10 percent of amount on file Respondent amount within 25 percent of amount on file Difference greater than 25 percent We produced weekly reports of verification codes by interviewer and looked for patterns of Codes 3 and 4 in both age/birth date and benefit amount fields. We uncovered no fraudulent interviews using this procedure. 44 Besides operating the tracking system, quality control staff reviewed each questionnaire prior to data entry. They determined whether problems, such as strings of missing data or incorrect recording of responses, would make data entry impossible. The quality control staff followed a set ofspecifications set up by the survey director to ensure that recorded values were within a reasonable range, response codes had been assigned correctly, and the questionnaire had internally consistent responses. Any errors, missing data, or inconsistencies could not be double-checked with the respondent because of anonymity requirements, but they were reported to the interviewer. This feedback made the interviewer aware of his or her errors and helped prevent them from recurring. After the questionnaires were manually edited and open-ended items were coded, they were batched and sent to data entry clerks who used programmable key-to-disk data-entry equipment. An editing program to perform checks on value ranges, questionnaire skip logic, and internal consistency for the full sample was prepared. Errors were flagged automatically during data entry and corrected by a senior member of the quality control staff. She reviewed the file and located error codes flagged by the quality control program. The data editor reviewed the questionnaire and corrected the error. A second operator rekeyed all the data. As the data were reentered, they were compared automatically with the first data file. Discrepancies were brought to the operator's attention. Finally, the survey data were transmitted to the research database. E. RESPONSE RATES AND OTHER FINAL STATUSES MPR interviewers completed a total of720 interviews. One was lost in the mail. The remaining 719 interviews were processed for analysis. The overall response rate was 76.7 percent. 45 TABLE A I RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE AND ELIGIBILITY RATES. BY SITE All Sites Baltimore Denver Spokane Number rcfccntamgc Number Pcfcentaft Number IVTCftitty Number FtfccnUnfc ENgibtWy Smnty Status Completed Interviews Cannot Locate 720 65 76.7 6.9 245 33 78.3 10.5 221 20 75.7 6.8 254 12 76.0 3.6 Rcfusal/Breakofrs 56 6.0 10 3.2 17 SJ 29 8.7 Multiple Unsuccessful Attempts Language Barriers Health Problem Other S3 20 19 6 56 2.1 2.0 0.6 21 0 3 6.7 0.0 19 0.3 IS 9 10 0 51 3.1 3.4 0.0 17 II 6 5 5.1 3.3 1.8 1.5 Total Eligible Saapk 939 199.0 313 100J 292 109J 334 100.0 Eligibility Rates Eligible Sample Ineligible Sample 939 375 71.5 28.5 313 87 21.8 292 182 61.6 384 334 106 75.9 2-t.l Total Saaipk Released 1.314 100.9 400 109.0 474 1MJ 440 1M.0 Reason for laebgiMHty Moved from Stratum 154 11.7 63 15.8 S3 17.5 68 15.4 No Longer Participating 214 7 16.3 0.5 22 2 5.5 0.5 96 3 20J 06 36 2 U 0.5 Total Ineligible Sample 373 28J •7 21J 182 ■at 196 24.1 NOTE: Due to rounding, percentages may not equal 100 46 1. Eligibility Rates and Reasons for Ineligibility MPR released 1,314 cases to field interviewers (see Table A.1). Of these, 375 cases, or 28.5 percent, were ineligible to participate in the survey. Cases were deemed ineligible for one of three reasons. The most common reason was that the sample member had movedfrom the stratum in which he or she was selected. Moves accounted for 214 cases, or 16.3 percent of the sample. The next most common reason for ineligibility was that the sample member no longer participated in the FoodStamp Program; 154 cases, or 11.7 percent ofthe sample, were not participating in the FSP when contacted for an interview. (Depending upon the site, contact occurred between 2 and 10 months after the survey sample was drawn from lists of FSP participants.) The third reason for a case being considered ineligible was death. Seven program participants died before we could interview them. 2. Eligibility by Site and Strata Although the overall ineligible rate was 28.5 percent, ineligibility across sites was 21.8 percent in Baltimore, 38.4 percent in Denver, and 24.1 percent in Spokane. The mobility rate was remarkably stable across the sites (15.8 percent in Baltimore, 17.5 percent in Denver, and 15.5 percent in Spokane). Differences in eligibility were related to the percentage ofprogram participants still receiving food stamps when interviewed. In Denver, where the frame consisted of all active cases in July 1995,20.3 percent of the sample were no longer receiving food stamps. In Spokane and Baltimore, where the sample was drawn from lists of cases active in October 1995, the percentages of the sample no longer receiving food stamps were 8.2 and 5.5, respectively. In the Baltimore site, ineligible rates were remarkably similar across three ofthe four strata (see Table A.2). The ineligibility rate was 19 percent in Stratum 11 (where more than 20 percent of households were under the federal poverty level), 21 percent in Stratum 12 (where 10 to 20 percent of households lived under the federal poverty level), and 19 percent in the rural stratum (Kent 47 TABLE A.2 RECIPIENT FOOD STAMP TRAFFICKING STUDY RESPONSE AND ELIGIBILITY RATES. BY STRATUM Stratum II More than 20 Percent of Poverty Stratum 12 10 to 20 Percent of Poverty Stratum 13 Less than 10 Percent of Poverty Stratum 14 Kent County (Rural) Baltimore Total Number Percentage Number Percentage Number Percentage Number Percentage Number Percentage ':■:■: -." . ■ Eligibility Survey Statas Completed Interviews 61 75J 61 77.2 54 75.0 69 85.2 245 78.3 Cannot Locale 10 113 10 12.7 11 15.3 2J 33 10.5 Refusal/Breakoffs 2 2.5 4 5.1 3 42 1-2 10 3.2 Multiple Unsuccessful Attempts 8 9.9 4 5.1 3 42 7.4 21 6.7 Language Barriers 0 0.0 0 0.0 0 0.0 0.0 0 0.0 Hearth Problem 0 0.0 0 0.0 0 0.0 3.7 3 1.0 Other 0 0.0 0 0.0 1 1.4 0.0 1 0.3 Total Eligible Sample SI 100.0 n ItM 72 100.0 SI 100.1 313 IM.t Eligibility Rates Eligible Sample Ineligible Sample 81 19 81.0 19.0 79 21 79.0 21.0 72 28 72.0 280 SI 19 81.0 190 313 87 78.2 21.8 Total Saaiple Released 1M MM 100 MM 100 MM 1M It* m ito.t Reasons for • ■eligibility Moved from Stratum 12 No Longer Participating 5 Deceased 2 12.0 16 16.0 26 26.0 9 9.0 63 15.8 5.0 5 5.0 2 2.0 10 10.0 22 5.5 2.0 0 0.0 0 0.0 0 0.0 2 0.5 Total Ineligible Sample 19 19.0 21 21.0 28 28.0 19 I9J V 2IJ 48 TABLE A2 (commutd) Stranm2l Stratum 22 Stratum 23 Stratum 24 More than 20 Percent 10 to 20 Percent of Less than 10 Percent of Clear Creek Countv Denver ofPoverty Poverty Poverty (Rural) Total Number Percentage Number Percentage Number Percentage Number Percentage Number Perec nuge Ptaim Eligibility Survey Statuses Completed Interviews 63 Cannot Locate 9 Refusal/Breakofrs S Muhipie Unsuccessful Attempts S Language Barriers 5 Health Problem 3 Other 0 656 94 8.3 8.3 5.2 3.1 0.0 62 81.6 5 66 6 7.9 2 2.6 0 0.0 1 IJ 0 0.0 55 833 0 00 1 15 5 7.6 3 4.5 2 3.0 0 0.0 41 75.9 221 6 111 20 2 37 17 0 00 15 1 19 9 4 74 10 0 00 0 75.7 68 5.7 5.1 31 34 00 Total Eligible Saatptt IN.0 76 100.0 66 100.0 54 IMA 292 Eligibility Rates Eligible Sample Ineligible Sample 96 41 70.1 299 76 49 60.8 39.2 66 52 559 44 1 54 40 574 426 292 182 616 38 4 Total Sample Released 137 100.0 125 100.0 US 180)4 94 100.0 474 ISM Moved from Stratum No Longer Participating Deceased 33 8 0 24.1 58 0.0 25 24 0 20.0 19.2 0.0 13 39 0 11.0 33.1 0.0 12 25 3 12.8 266 3.2 S3 96 3 17.5 203 06 Tctal ladigiMe Saapie 41 29.9 49 39.2 52 44.1 40 424 IS2 38.4 49 TABLE A 2 (ctmliimtd) Stratum3l More than 20 Percent of Poverty Stratum 32 10 to 20 Percent of Poverty Stratum 33 Less than 10 Percent of Poverty Stratum 34 Pend Oreille Counts (Rural) Spokane Total Number Percentage Number Percentage Number Percentage Numbe r Percentage Number rtrccnttfc aMm Eligibility Sarvty Statues Completed Interviews Cannot Locale Rcfusal/Breakoffs Multiple Unsuccessful Attempts Language Barriers Health Problem Other 61 1 7 2 2 1 0 824 1.4 9.5 2.7 2.7 1.4 0.0 62 3 9 5 4 4 0 71.3 3.4 I0J 5.7 46 46 0.0 58 13 70.7 2.4 159 37 6.1 IJ 0.0 73 6 0 7 0 0 5 80.2 66 0.0 7.7 0.0 0.0 5.5 254 12 29 17 II 6 5 76.0 3.6 1.7 5 1 3.3 1.8 0.5 Total Eligible Sample 74 IMJ S7 IMJ 82 IMJ 91 IMJ 334 IMJ Eiigibihty Rales Eligible Sample Ineligible Sample 74 29 71.8 28 2 •7 31 73.7 263 12 32 719 28 1 91 14 86.7 133 334 106 76.0 240 Total Sample Released 163 ion.* US IMJ 114 IMJ 1#5 IMJ 4M IMJ Reasons for Ueligibilir) Moved from Stratum No Longer Participating Deceased 19 10 0 114 9.7 00 It 12 1 15.3 10 2 08 22 10 0 193 8.8 00 9 4 1 86 3.8 10 68 36 2 15.5 u 0.5 Total Ineligible Sample 29 28.1 31 26J 32 28.1 14 13.4 IM 24.1 NOTE: Percentages may not equal 100 due to rounding 50 County). In Stratum 13 (where less than 10 percent of the households lived below the federal poverty level), 28 percent of the sample were ineligible (26 percent had moved from the stratum between the time the sample was drawn and the interview was attempted). At the Denver site, rates of and reasons for ineligibility differed across the strata. The highest ineligibility rate was 44 percent in Stratum 23 (where fewer than 10 percent ofthe households lived below the federal poverty level). One-t rd ofthe sample in this stratum was ineligible because they were no longer receiving food stamps on the interview day. In Denver's rural stratum (Clear Creek County), 43 percent were ineligible. One-quarter were no longer participating in the FSP when interviewed. The county food stamp fraud investigator told MPR that Clear Creek was a highly mobile county; thus, many nonparticipants may have moved. In Stratum 22 (where 10 to 20 percent of the households lived under the federal poverty level), the ineligibility rate was 39 percent, split almost evenly between those who left the FSP (19 percent) and those who moved (20 percent). The ineligibility rate for Stratum 21 (the poorest of the Denver strata) was 30 percent; one-quarter had moved out of the stratum. In the Spokane site, Strata 31 and 33 (more than 20 and fewer than 10 percent of households living under the federal poverty level, respectively), ineligibility accounted for 28 percent of the sample released. In both strata, the number of movers was almost double the number ofthose who no longer participated in the FSP. Ineligibility rates were lower, 26 percent, in Stratum 33 (10 to 20 percent living under the federal poverty ' vel) and much lower. 13 percent, in rural Pend Oreille County. Despite the lower frequency than in other Spokane strata, the ratio of movers to nonparticipants in Pend Oreille County was the same-cases were screened out because of moving twice as often as because of nonparticipation. 51 3. Response Rates and Reasons for Nonresponse Overall, 77 percent of sample members eligible to participate in the survey completed interviews. There were three important reasons for nonresponse. First the sample member could not be located (seven percent). Second, the sample member refused to be interviewed (six percent). For another six percent of the sample, interviewers could not complete the case despite multiple telephone and in-person contacts. Other reasons for nonresponse were language barriers (two percent) and respondents' poor health (two percent). 4. Response Rates and Nonresponse by Site and Strata Response rates were quite even across the three sites. Baltimore's response rate was 78 percent, whereas Denver's and Spokane's were 76 percent Baltimore had the highest nonlocatable rate (10 percent) and the lowest refusal rate (3 percent). In Spokane, the nonlocatable rate was six percent, but the refusal rate was nine percent This pattern of refusal rates increasing with the proportion of sample located may imply that sample members not easily located for this survey were, in reality, reluctant respondents. FSP offices in Denver and Spokane helped MPR with locating by checking administrative records for address updates during the field period. In the Baltimore site, however, only Kent County updated addresses. In Baltimore County, where intensive trafficking investigations were under way independently ofthis study, we were concerned that asking for address updates might hurt sample members by focusing attention on them; therefore, we did not request address updates in Baltimore County. In the Baltimore site, the average response rate in the urban strata was 75.8. The response rate in the rural county was 85.2 percent. Both refusal rates and the percentage of nonlocatable sample were lower in the rural area. 52 In Denver, where the overall response rate was 75.7 percent, two ofthe urban strata were well bove average and one was well below. The response rates for Stratum 22 (10 to 20 percent living under the federal poverty level) and Stratum 23 (less than 10 percent living under the federal poverty level) were 81.6 and 83.3 percent, respectively. The lowest response rate in the Denver site. 65.6 percent, occurred in the poorest stratum. In that area, all the components of nonresponse were higher than average. In the rural stratum, the response rate was 75.9 percent A remarkably high proportion of rural stratum sample members were either nonlocatable (11.1 percent) or unable to complete the interview because ofhealth problems (7.4 percent). In Spokane, the response rates were highest in the most urban and rural strata (82.4 and 80.2 percent, respectively) and lower in the strata with 10 to 20 percent and less than 10 percent of the households living under the federal poverty level (71.3 and 70.7 percent, respectively). In the three urban strata, refusal rates increased as the percentage of households living under the federal poverty level decreased. The rural strata had no refusals, however. a f1 OUBBBMH APPENDIX B FOOD STAMP SURVEY / 4 LU! CASE ID: I—I—l—l_!—!_! INTERVIEWER ID: I I I I I I • FOOD STAMP STUDY A. INTROOUCnOl: Hello, My name Is and I aa from Mathematica Policy Research, Inc., a research company in Princeton, New Jersey. We are conducting a research study for the United States Department of Agriculture. The study is about (food stamps/Independence food stamp benefits). You nay have recently received a letter about this study. The purpose of the study 1s to see how and where people use their (food stamps/Independence food stamp benefits). We will ask questions about you and your household, shopping for food with food stamps, and other uses of food stamps. You have been selected to be part of this very Important study. He would l?ke you to help us by participating 1n a 45-m1nute In-person interview. I can conduct the interview in your home or we can meet at any other place that is convenient for you. B. INFORMATION]: What is this study about? The purpose of the study is to see how and where people use their (food stamps/Independence food stamp benefits). He will ask questions about you and your household, shopping for food with food stamps, and other uses of Who is sponsoring this study? Who is conducting this study? The United States Department of Agriculture Food and Consumer Service is sponsoring this study. Hathematica Policy Research, Inc., an independent research and survey company located in Princeton, New Jersey is conducting the study for them. Why Is Hathematica calling me now? Hathematica would like you to take part in a 45-minute in-person interview. The interviewer can come to your home or will meet you any place that is convenient and comfortable for you. By discussing how and where you use your (food stamps/Independence food stamp benefits), you will help the researchers better understand how to improve the Food Stamp Program. 57 How was I selected? You havt boon randomly selected from (food stamp/Independence food stamp benef\tj clients In the (Denver/Baltimore/Spokane) area. ■t Do I ttmmtto participate? What If I decide not to? AlthoutpB-your participation is vary Important, it is voluntary. Your participation will not affect any benefits you get now or in the future. If I do participate, will the information be confidential? Yes. All information collected will be kept strictly confidential and used only for research purposes with no names attached. The results of the study are for research purposes only. C. ELIGIBILITY SCREEN: 1. Before I tell you more about this study, I need to confirm that I am speaking to the person I'm supposed to speak to. Are you SPEAKING WITH CORRECT SAMPLE MEMBER....01 NOT SPEAKING WITH CORRECT SAMPLE MEMBER - BtD. THANK RESTONOOrr 2. Did ydfi receive food stamps during the month of THIS».MOMW? IYES 01 - GQ mjt%4 NO 00 3. Did you receive food stamps during the month of LAST MONTH? YES 01 NO 00 - I'm sony. This aunray to aaJy for paopto «me raeaivad food thto month w toot. Thai* you for your Unto. 58 4. Oo you live in the READ ZIP CODE OR COUNTY FROM CONTACT SHEET (zip code/county)? YES 01 - NO - I'M SORRY. Wo aro wiry interviewing poopia who two in «M READ (ELKBSLE CODE/COUNTY) FROM CONTACT SHEET. Thank you for your ttano. RECORD SR CODE ORCOUNTY: !_!_:_i_;_; OR KENT 01 CLEAR CREEK 02 RENO d'ORERlE 03 5. IS THIS A TELEPHONE OR IN-PERSON CONTACT? TELEPHONE - SCHEDULE TINE TO CONDUCT IN-PERSON INTERVIEW IN-PERSON - GO TO SECTION D D. AGREEMENT TO PARTICIPATE: MARYLAND AND COLORADO: Before we begin, I would like to tell you how this interview will work. For the next 45 minutes, I will be is ing questions about you and how you use your (food stamps/Independence food stamp benefits). The information you share will be used to improve the Food Stamp Program. All the answers you give me will be kept completely confidential and will never be associated with your name. That means no one at the Food Stamp Office or any other program or agency will know the way you answered these questions. Some questions will be private. When it is time to answer the private questions, I will give you a tape in a Walkman and an answer sheet to fill out by yourself. No one, not ever. I, will know how you answered the private questions. When we are all finished, I will ask you to seal your questionnaire in an envelope. Then, I will leave this sheet [INTERVIEWER: SHOW CONTACT NOTE FORM] with your name and address with you so that you know hat your name and answers are separate. WASHINGTON: HAND STUDY DESCRIPTION TO SAMPLE MEMBER. ASK SAMPLE MEMBER TO FOLLOW ALONG AS YOU READ ALOUD. We are asking you to be in a study that Mathematica Policy Research, an independent research company, is doing for the United States Department of Agriculture (USDA). The USDA is the agency that runs the Food Stamp Program. The study is about how people use food stamps, as well as how they misuse them, for example, by selling them. The USDA will use the study findings to improve the Food Stamp Program. 59 If you agree to be in the study, we will ask questions a bout you and your household, and about how you use you food stamps. Some of the questions are about illegal uses of food staops. You do not have to answer personal questions. The interview will take about 45 minutes to complete, and can be done in your own home or someplace else that you choose. All of your answers will be completely confidential and will never be linked with your name. That means no one at the Food Stamp Office or at any other program or agency will know the way you answered the questions. Your participation in this study is completely voluntary. You can decide not to be interviewed, or you can stop the interview at any time, without any penalty or loss of any benefits you receive now or in the future. Please understand the following: • Nothing you say in the interview can or will be used against you by the State of Washington, the U.S. Department of Agriculture, or any other legal authority. The only exception is that if you give us information about the abuse or neglect of a child, the interviewer must report this to the Department of Social and Health Services. • We will not connect your name, address, or telephone number with your answers. The interviewer will not keep a record of your name, address, or telephone number. • The interviewer has signed a confidentiality pledge that prevents her from linking your answers with you as an individual. There will be a report of all the interviews conducted for this study (about 720), but it will be a summary with no names mentioned. Some que is time tape in not even question envelope name and separate stions about your own use of food stamps will be private. When it to answer the private questions, the interviewer will give you a a Walkman and an answer sheet to fill out by yourself. No one, the interviewer, will know how you answered the private s. When you finish, you should seal your answer sheet in an Then, the interviewer will leave the Contact Sheet with your address with you so that you know that your name and answers are You may also keep a copy of this Study Description. We hope you will agree to be interviewed. If you have any questions about the study, now or in the future, you may call the researchers listed on this form at 1-800-777-0085. Hay we begin the interview now? YES NO. .01 ,00 - RESCHEDULE 60 DATE: | | | - ! | | TBifli "Bay START TINE; | | |:| | | AM 01 PM....02 Al. I'd like to begin this interview by finding out about you and your household. First, when is your birthday? When were you born? "WJNTTT -HAT- -9 '^H A2. Are you now . Married, 01 Living with someone as married, 02 Widowed 03 Divorced, 04 Separated, 05 Or have you never been married? 06 A3. Which of the following best describes where you live now . . . A pi ace you own, 01 A place you rent, 02 A place where you live rent free, 03 You are homeless, 04—] You live in a shelter, group home, U SKIP TO A13, PAGE 7 or treatment facility, or 05J You live in some other type of place? (SPECIFY) 06 61 A4. Counting yourrelf, how many ptoplt live In your housthold. INTERVIEWER: COUNT PEOPLE WHO LIVE IN THE HOUSEHOLD ON A REGULAR BASIS FOR SIX NORTHS OF THE YEAR OF NORE. LIVE ALONE 01 - SKIP TO AlO | | | TOTAL PEOPLE IN HOUSEHOLD A5. How many of these people are children under the age of 16? I | | CHILDREN UNDER 16 INTERVIEWER: AS SOON AS YOU BELIEVE ALL HOUSEHOLD NOSERS ARE ACCOUNTED FOR, YOU NAY CONFIRM A6 AND/OR A7 WITH RESPONDENT. A6. How many of the people in your household are people age 16 and older who are related to you? INTERVIEWER: DO NOT COUNT RESPONDENT. I I | RELATED PEOPLE AGE 16 AND OLDER A7. How many of the people in your household are people age 16 and older who are net related to you? I | | UNRELATED PEOPLE AGE 16 AND OLDER A8. INTERVIEWER CHECK: DOES A5 ♦ A6 ♦ A7 * 1 • A4? YES 01 NO - FIX A4- - A7 A9. Including yourself, how many of the FILL A4 people in your household are covered by your (food stamps/Independence food stamp benefits)? I | | COVERED BY FOOD STAHP BENEFIT 62 A10. Dots the place where you live have a working refrigerator or freezer (that you are allowed to use)? V" 01 W 00 - SKIP TO A12 All. Is the refrigerator or freezer large enough to meet your needs? YK 01 • 00 A12. Does the place where you live have a working stove, toaster oven, or microwave oven (that you are allowed to use)? «■ 01 m 00 A13. On what date did you receive your most recent (food stamps/ Independence Food Stamp benefit)? PROBE: When was your most recent food benefit credited to your Independence Card? '^Tr1"1^1--19 hdk-1 AM. How much did you receive in (food stamps/Independence Food Stamp benefit) on FILL DATE FROM A13. ^ A15' fuftStt SSi J?]1!" 0f lour !Ind PILL DATE FROM A13 do you have left?«P«nd«"«) food stamp benefits from PROBE: That is, what is the value of your food stamp benefits that you have not yet spent? A16. During the past year, for how many months did you receive (food stamps/Independence Food Stamp benefits)? I | | MONTHS OUT OF 12 63 A17. Who takes responsibility for your food stamp benefits each month? That is, who keeps the (food steeps/Independence card) and decides how and where the (food stamps/benefits) will be spent? CIRCLE/CODE ALL THAT APPLY SAMPLE MEMBER 01 SPOUSE/PARTNER 02 CHIL0/STEPCHILD/6RAM0CHIL0 03 PARENT OR STEPPARENT 04 BROTHER OR SISTER 05 GRANDPARENT 06 OTHER RELATIVE 07 - NONRELATIVE 08 J ADMINISTRATOR AT A SHELTER, GROUP HOME, OR TREATMENT FACILITY 09 A18. INTERVIEWER: IS ONLY THE SAMPLE NOSER COOED IN A17? YES, SAMPLE MEMBER ONLY 01 - SKIP TO A20 NO, SAMPLE MEMBER ANO/OR OTHERS 00 SAMPLE MEMBER NOT COOED AT ALL -4 - SKIP TO A20 A19. Do you and your FILL RELATIONSHIP(S) FROM A17 agree about how and where to use the (food stamps/Independence food benefits) . . . All of the time 01 Most of the time 02 Some of the time, or 03 Hardly ever? 04 64 A20. Within how »any days after you rtctlvt your (food stains/Independence food stamp benefits) do you begin to spend them? SAME DAY 01 WITHIN | | | DAYS Oft BETWEEN I I I AND I I I DAYS A21. Do you usually spend the entire amount in one or two days or do you save some for later in the month? SPEND ALL WITHIN ONE OR TWO DAYS 01 - «or TO SECTION • OW MM 10. SAVE SOME FOR LATER IN THE MONTH 02 VARIES/DEPENDS 03 A22. About how much do you usually save for later in the month? PROBE: On average, what is the value of the (food stamps/Independence food stamp benefits) that you save for later in the month? A23. Typically, how long do your (food stamps/Independence food stamp benefits) last each month? IF RESPONDENT SAYS BETWEEN X AND Y, CODE Y. ONE WEEK 01 TWO WEEKS 02 THREE WEEKS 03 ENTIRE MONTH 04 65 Bl. The next questions are about your income and expenses. Let's begin with housing. What was your monthly (rent/mortgage payment) last month? B2. My next questions are about how often your household shopped for food in the last 7 days at supermarkets, neighborhood grocers, convenience stores, and specialty stores. I would like you to include all trips-trips to pick up a few items, as well as major shopping trips by all household members. SHOW CARD A How often in the last 7 days, did your household go to: B3. B3a. B4. a. Supermarkets? b. Neighborhood stores? c. Convenience stores such as 7-11 or stores that sell gas anc groceries? d. Specialty stores such as bakeries, vegetable stands, farme s markets, dairy stores, meat markets, health food stores, c- othc similar places. ASK B2a-d BEFORE GOING TO B3. 00 NOT ASK B3-B4 FOR ANY TYPES • STORES WHERE THE NUMBER OF TRIPS EQUALS ZERO. During the last 7 days, about how much did your household spenc t (TYPE OF STORE)? Include all purchases you made, whether you pe z f: them by check, cash, or (food stamps/your Independence card). K ,'NC TO NEAREST DOLLAR. Does the AMOUNT FROM B3 include the dollar value of food purchasec with (food stamps/your Independence card)? About how much of this AMOUNT FROM B3 was for non-food items such as cleaning or paper products, commercially prepared pet food, or tobacco products? INTERVIEWER NOTE: ALCOHOL IS A FOOD ITEM. Suoenaarkets b. Metghoornood norn e. Convenient* storw a. Specialty stern of Tripe LAST 7MTJ !_U is Tetet LAST 7MTS » U_l_l « l_l_LJ « U_I_J * l-l-t l IfKtl TIS....01 ■0 00 - Tft 01 m oo - m—oi ■0 00 - rts 01 »0 00 - u •J u 11 • l-J_ _LJ l_l 66 B5. Would you say that the amount your household spent on food in the last 7 days was more, less, or about the same as usual? MORE 01 LESS 02 ABOUT THE SAME AS USUAL 03 B6. During the last 7 days, what was the dollar value of purchases made (with food stamps/with your Independence food benefits) by your household? Consider all (food stamp purchases/Independence food benefit purchases) by your household, even for just a few items. B7. Now I'd like to ask about month!v expenditures. First, I am going to ask about the amount your household spent on food in LAST MONTH. During LAST MONTH, how much did your household spend in total at supermarkets, neighborhood grocery stores and convenience stores? Include all purchases you made, whether you paid for them by cash, check, or (food stamps/Independence card). Do not include any expenses for meals eaten away from home or for home-delivered or carry-out meals. B7a. About how much of this AMOUNT FROM B7 was for non-food items such as cleaning and paper products, commercially prepared pet food, or tobacco products? B8. During LAST MONTH, how much did your household spend on food at specialty stores such as bakeries, delicatessens, vegetable stands, farmers* markets, dairy stores, meat markets, health food stores, and other similar places? $11.1111 67 B9. During LAST MONTH, did anyone in your household buy and cat at home ready-to-eat meals like Chinese food, pizza or "fast food" from delivery services, carry out places or pick-up windows? YES. NO.. .01 .00 - SKIP TO 112 BIO. How much did these cost? Include food, beverages, any delivery charges, tax and tips. PROBE FOR ALL INSTANCES OF TAKE OUT « HOME DELIVERED FOOO IN THE PAST MONTH. Bll, B12 B13 Would you say that the amount your household spent on take out and delivered foods in LAST MONTH was more, less, or about the same as usual? MORE 01 LESS 02 ABOUT THE SAME AS USUAL 03 Are you currently working at a job for pay? Include any self-employment. YES 01 NO 00 - SKIP TO R15 How many hours do you usually work per week (at all jobs)? | | | HOURS PER WEEK B14. How much do you earn (at all jobs), before taxes and other deductions? RECORD HOURLY MARE UNLESS RESPONDENT CAN ONLY PROVIDE SALARY FOR A DIFFERENT PAY PERIOD. IF HOURLY WAGE IS NOT RECORDED, RECORD BOTH SALARY AJD PAY PERIOO. PER HOUR OR PER MEEK 01 DAY 02 EVERY TWO WEEKS....03 TWICE A MONTH 04 MONTHLY 05 YEARLY 06 B15. INTERVIEWER: REFER TO A4 ON PACE 6. DOES SAMPLE MEMBER LIVE ALONE? YES. NO.. HOMELESS OR IN GROUP HOME. .01 - SKIP TO BIB .00 -4 - SKIP TO BIB B16. Is anyone else in your household currently working at a job for pay? YES 01 NO 00 - SKIP TO BIB B17. How much (do they/does (he/she)) earn? Please tell me (the combined) wages earned before taxes and deductions by all of the others in your household. PER HOUR OR BVLAM on SHOW CALCULATIONS PER WEEK 01 OAY 02 EVERY TWO WEEKS....03 TWICE A MONTH 04 MONTHLY 05 YEARLY 06 69 818. During last Month, did you receive AFDC, that is Aid to Families With Deptndtnt Childrtn/FILL LOCAL NAMES FOR DENVER/SPOKANE/BALTIMORE. YES 01 NO 00 - SKIP TO B20 B19. How much did you receive from AFDC/LOCAL NAME last month? B20. How much other income did people in your household (including yourself) receive last month. Tell me about income from all other sources such as General Assistance, Unemployment, Social Security, SSI, retirement benefits or any other income you have. Oo not include the amount of the food stamp benefit. PROBE: Please tell me your other income before taxes and other deductions. NO OTHER INCOME 00 B21. Generally, over the past six months, did your income change from month to month? YES 01 NO 00 - SKIP TO B22 70 B21a. What caused your income to change? EARNINGS FLUCTUATED 01 CHILD SUPPORT PAYMENTS FLUCTUATED...02 GIFTS 03 OTHER (SPECIFY) 04 B21b. Was your Income last month more, less, or about the sane as usual? MORE ,...01 MS 02 ABOUT THE SAME 00 B22. Now I am going to ask about various events and conditions that happen to people. I'm interested in those that happened to you during the last 12 months, that is since NEXT MONTH, 1995. As I ask about the specific events, please think carefully so I can record things accurately. First, think about financial matters. Did any of the following happen to you since NEXT MONTH, 1995? a. Car, household appliance, or fumrture inniiiiiid 01 00 b. Pawned or sold off valuables to make ends meet 01 00 c Pressured to pay bills by stores, creditors, or Ml collectors 01 00 d. Maior worsening of your financial condroon Q1 00 Now I am going to ask about specific hardships. Did any of the following happen to you since NEXT MONTH, 1995. e. Fell behind in paying your rent or mortgage 01 00 f. Evicted from your aoartment/house 01 00 a- Had your utilities (water, heat, or electncrrvl shut off 01 00 h. Unable to purchase needed food 01 00 • Unable to afford needed medical care oi 00 j. Had to temporarily live with others or m a shelter or on the street* g^ QQ 71 B23. These next questions are about the foods eaten 1n your household. Which of the following statements best describes the amount of food eaten in your household—enough food to eat, sometimes not enough to eat, or often not enough to eat? ENOUGH FOOO TO EAT 01 SOMETIMES NOT ENOUGH TO EAT... 02—i OFTEN NOT ENOUGH TO EAT 03_T SKIP T0 B25 B24. Do you have enough of the kinds of food you want to eat, or do you have enough but no*, always the kinds of food you want to eat? ENOUGH OF THE KINDS YOU WANT 01 ENOUGH BUT NOT ALWAYS THE KINDS YOU WANT 02 B25. In the last 12 months, did you ever eat less than you felt you should because there wasn't enough money to buy food? YES 01 NO 00-, DOr T KNOW -ij-* *■ T° B28 B26. Did this happen in the last 30 days? YES 01 NO 00-, DON'T KNOW .lJ-* SMP T0 BM B27. In the last 30 days, how many days did you eat less than you felt you should because there wasn't enough money to buy food? I I I NUMBER OF DAYS 72 B28. People do different things when they ere running out of money for food In order to nuke their food money go further. In the last 30 days, did you or anyone 1n your household . . . a. get food or borrow money for food from friends or relatives? b. put off paying a bill so you would have money to buy food? c. get emergency food from a church, food pantry, or food bank? d. eat meals at a soup kitchen? IK NO 01 00 01 00 01 00 01 00 73 raa Cl. These next questions are about your beliefs. Please tell me if you strongly agree, agree, disagree, or strongly disagree with the following statements. SHOW CARD B a. I worry that I won't be abta to do the kind of work that I want to do because I don't have enough education b. I have had as much opportunity to succeed cs people from other neighborhoods c. If a parson works hard, s/hs get ahead STRONGLY d. As I get older, things will get better e. rt is okay to lie ;f it keeps your friends out of trouble f. It is okay to break the law if it helps put food on your family's table g. I like to take chances h. I get upset when I have to wart for something i. I act without stooping to think I l get bored easily 01 01 01 01 01 UNDEODEP DISAGREE STRONGLY DISAGREE 02 02 02 02 02 03 03 03 03 03 04 04 04 04 04 05 05 05 05 05 01 02 03 04 OS 01 02 03 04 05 01 02 03 04 05 01 02 03 04 05 01 02 03 04 05 74 C2. Now I'd like to get your opinions about a variety of issues. In the next set of questions I will read a statement and ask you to tell me if you strongly agree with the statement, agree with the statement, disagree with the statement, or strongly disagree with the statement. SHOW CARD B a. Public assistance workers (or AFOC workers) try to give everybody an even break b. it is sometimes all right to get around the r -lies if vou can get away with it. c. To get ahead, you sometimes have to do something which may be aoainst the law d. Most successful people probably used illegal means to become successful e. People who leave things lying around outside their house should expect that some of their things might be taken or stolen f. It's okav to steal from someon who is rich and can easily replace it B. It's okay to steal from the government since tne government has so much money that it won't hurt them miONQLY DISAGREE STRONGLY DISAGREE 01 01 01 01 01 01 01 02 02 02 02 02 02 02 03 03 03 03 04 04 04 04 05 05 05 05 03 04 05 03 04 05 03 04 05 C3. Now I'd like to ask you how important some things are tell me if they are very important, pretty important, important, not too important, or not important at all, is it . . . to you. Please somewhat How important SHOW CARD C a. ... for you to have a lot VERY IMPORTANT RRETTV rMRORTANT SOMEWHAT NWOftTANT NOT TOO MPORTAMT NOT BMPORTAMT AT AU m n? oa na. n« b. ... for you to have a good reputation in the community? c. ... to plan ahead? 01 01 02 02 03 03 04 04 05 05 75 C4. Please tell me how much you agree with the following statements Would you sty definitely ys, probably yes, probably no, or definitely no. a. If you lit to your public assistance (or AFDC) worker, you w« wind up IOW>Q your benefits b. If you Ml to your public assistance (or AFDC) worker, you will wind up being charged by the police c. If you lie to your public assistance (« AFDC) worker, you win wind uji being tern to court SHOW CARD D vis 01 01 01 02 02 02 UNDeCDCD MMMAMLV NO M—W NO 03 04 05 03 04 05 03 04 05 C5. Now imagine that you get caught selling food stamps. How big a problem would that cause for you? Would it cause 1 . . . SHOW CARD E Very big problem, 01 Big problem, 02 Small problem 03 Very small problem, or 04 No problem at all? 05 C6. Now tell me how much of a problem would it be to have *he following incidents occur in your life. ... If your friends find out that you had done something that was against the law? Would it be a . . . SHOW CARD E Very big problem 01 Big problem 02 Smal 1 probl em 03 Very small problem, or 04 No pro1-1 em at all? 05 76 C7. ... If your relatives find out that you had done something that was against the law? Would it be a. . . SHOW CARD E Very big problem, 01 Big problem 02 Smal 1 problem, 03 Very small problem, or 04 No problem at all? 05 NOT APPLICABLE (NO RELATIVES OR NO CONTACT WITH RELATIVES) -4 C8. ... If your children find out that you had done something that was against the law? Would it be a. . . SHOW CARD E Very big problem 01 Big problem, 02 Smal 1 probl em 03 Very small problem, or 04 No problem at all? 05 NOT APPLICABLE (NO CHILDREN) -4 77 01. My next questions ire about getting together with friends end relatives. Please think about the past year. In a typical week during the past year, how often did you talk on the telephone with friends or relatives? Did you talk on the telephone . . . IF NO REGULAR ACCESS TO A PHONE, CODE LESS THAN THAT. Every day, 07 Five or six times a week, 05 Three or four times a week, 03 Once or twice a week, or , 01 Less than that? 00 D2. In a typical week during the past year, how often did you get together in person with friends, neighbors, or relatives? By get together I mean going out together or visiting in each other's homes? Did you get together ... iv%ry day 07 Five or six times a week, 05 Three or four times a week, 03 Once or twice a week, or 01 Less than that? 00 D3. How often do you attend church or religious services? Do you go Once a week or more, 01 At least once a month, 02 Once every two or three months 03 Three or four times a year 04 Once or twice a year, or 05 Less than that? 00 D4. Do you belong to any clubs or organizations such as church groups, unions, tenant associations, athletic grojps or school groups? ^S 01 m 00 - SKIP TO 06 78 05. How often do you attend the meetings of the clubs or organizations to whi-h you belong? Do you attend meetings. . . Once a week or more, 01 At least once a month, 02 Once tvry two or three months, 03 Three or four times a year, 04 Once or twice a year, or 05 Less than that? 00 06. How long have you lived in this (city/town)? RECORD TO NEAREST YEAR. IF LESS THAU CUE YEAR, RECORD MONTHS. I I I YEARS OR | | | MONTHS D7. How long have you lived at your current address? RECORD TO NEAREST YEAR. IF LESS THAN ONE YEAR, RECORD MONTHS. |_|_| YEARS OR |_|_| MONTHS Please think about the (city/town) where you have lived the longest. D8. How long did you live there? RECORD TO NEAREST YEAR. IF LESS THAN ONE YEAR, RECORD MONTHS. I l__l YEARS OR | I I MONTHS D9. In the past year, has soreone used a weapon, force, or strong arm methods to get money, food stamps or something else from you? YES. .01 m 00 - SKIP TO 012 010. How many times in the past year did this happen? I I | TIMES IN PAST YEAR 79 Oil. In the most recent event, what was the person trying to get from you? CIRCLE ALL THAT APPLY MONEY 01 DRUGS 02 FOOD STAMPS/OTHER BENEFITS 03 S" 04 OTHER (SPECIFY) 05 012 (In addition to that,) in the past year, have you had your pocket picked or your purse or wallet sn |
OCLC number | 888048044 |
|
|
|
A |
|
C |
|
G |
|
H |
|
I |
|
N |
|
P |
|
U |
|
W |
|
|
|