Jump to ContentJump to Main Navigation
Cheating WelfarePublic Assistance and the Criminalization of Poverty$

Kaaryn S. Gustafson

Print publication date: 2011

Print ISBN-13: 9780814732311

Published to NYU Press Scholarship Online: March 2016

DOI: 10.18574/nyu/9780814732311.001.0001

Show Summary Details
Page of

PRINTED FROM NYU Press SCHOLARSHIP ONLINE (www.nyu.universitypressscholarship.com). (c) Copyright University of NYU Press, 2022. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in NYSO for personal use. Subscriber: null; date: 20 May 2022

(p.189) Appendix A: Critical Methodology

(p.189) Appendix A: Critical Methodology

Source:
Cheating Welfare
Author(s):

Kaaryn S. Gustafson

Publisher:
NYU Press

My Approach

The core of my research involved interviews with welfare recipients, individuals who must negotiate compliance with increasingly complex welfare regulations. My data collection was limited to one county in Northern California, a county I call Bayview. Bayview County was an attractive research site because of its size, racial and ethnic diversity, and its coverage of both urban and rural areas, and because other researchers have conducted welfare research in the county, allowing me to draw upon existing studies. Studying the policies in only one county allows me to control for policy variations that may occur from county to county.

While limiting the scope of research to one county may limit some of the findings, the general nature of the questions I asked and the similarities of welfare policies and practices across both the state and the country allow reasonable generalizations about many of my findings. Rules, regulations, and their enforcement all vary from county to county. Limiting the interviews to welfare recipients in only one county is most likely to hinder the generalizability of the interviewees’ responses about rule knowledge. It is likely that welfare recipients’ access to institutional sources of information varies from locale to locale, as office practices and caseloads vary from county to county. Still, the degree to which welfare recipients obtain information from unofficial sources may or may not vary across counties or sites. More importantly, examining the degree to which welfare recipients skirt the rules and their perceptions of the welfare system’s legality offers revelations of until-now-hidden behaviors and beliefs.

I began recruiting and interviewing welfare recipients in November 1998 and conducted the final interview in April of 2002. This meant that all the interviews occurred between the time the CalWORKs program was instituted in the county and December 31, 2002, the day that the sixty-month lifetime welfare limit was reached for California families who had been (p.190) receiving welfare since the beginning of my study. The welfare recipients I interviewed were recipients of TANF and were the heads of household in an aid unit. Not all of the interviewees, however, received cash assistance themselves. Some either were having their cash grant sanctioned at the time of the interview or were excluded from the aid unit. Welfare time limits and work requirements were inapplicable to the excluded cases where only the child or children received aid. Still, all of the interviewees were responsible for compliance with the other welfare rules, including the reporting rules. Moreover, while some of the adults were excluded from the aid calculations done by the welfare office, they all shared the collective income resources with the other members of their households.

With the consent of the research participants, I audiotaped the interviews. (Only one interviewee, Carmen, wished not to be recorded. I took notes by hand during her interview.) After each interview, I either wrote or recorded additional notes or observations about the interview and the interviewee. I also discarded any contact information I had for the interviewee and changed his or her name in my notes. I transcribed the recorded interviews, a time-consuming (and for the half I had professionally transcribed, expensive) endeavor that allowed me to more thoroughly and accurately review and analyze the interviews.

I followed each interview with a debriefing, explaining to the participants that I was most interested in their knowledge and assessment of the welfare rules. I carried with me a list of legal and community service referrals and, after the interview, provided research participants with appropriate contact information based on information they shared with me during the interview.

The interviews with the welfare recipients generally lasted one to two hours. Most of the interviews occurred in the back corners of restaurants or cafes, though I interviewed four of the participants in their homes. (I interviewed Lisa and Shanelle, who were best friends, in Lisa’s house.) I interviewed one participant, Viola, in a conference room at her drug treatment program.

The interviews were semistructured and covered welfare recipients’ knowledge of the rules and of the sanctions for violating the rules and requirements; their means of support and household composition (both reported and unreported); their socioeconomic, educational, and work histories; their experiences with the welfare office and with its fraud and error control measures; and their attitudes about the welfare rules, the welfare system, and other welfare recipients (see the Interview Schedule, Appendix B). I asked probing questions about sources of income and ways of making ends (p.191) meet when the money got tight. I asked follow-up questions to find out if they knew that some of the things they were doing were against the rules and how they regarded their actions that fell outside the rules. I also asked what they thought about people who broke the rules; whether they thought the work requirements, time limits, and reporting requirements were fair; and what changes, if any, they would make to the welfare system if they could.

Much of the information the participants shared with me would be impossible to obtain through administrative data and difficult to obtain in a survey. I asked participants what they might have considered sensitive, personal questions. In addition, I often asked them about hidden practices that they may not have wanted others, especially welfare officials, to know. Certainly, the strength of my data and conclusions relied upon my ability to obtain revealing information from the participants in my research. I describe some of my experiences with the interviewing process below.

While some of the interview questions involve theory testing—for example, testing the assumption that welfare recipients possess accurate knowledge of the rules and the consequences for breaking them—many of the questions and the study as a whole are designed to peek into dark, unexplored crevices, to generate rather than test theories.

Gaining Access

This description of my methodology differs in one significant way from many of the social scientific and legal texts written about welfare in that much of this section is written in the first person. My choosing to write in this voice is no error or oversight but rather a deliberate decision to reveal the research process. That process involves a continual back-and-forth between honing the research questions, developing the appropriate methods, and, in response to everyday practicalities and new revelations, readjusting both. The criticism and danger I face in rendering the research process transparent is exposure of the problems I encountered and therefore the weaknesses in the research. However, every researcher faces similar problems, whether she or he chooses to reveal them or not. I hope that by revealing the problems I can make the efforts of other researchers easier.

I had originally hoped to solicit a random sample of welfare recipients in my research county by letter. I thought this would allow me to reach more reliable general conclusions about welfare recipients in my research county. I drafted a proposal to the county welfare department in 1997, requesting minimal contact information for three large samples (one thousand households) (p.192) of welfare recipients, with the samples drawn in different quarters of the year. I requested that administrators destroy the samples after providing them to me. From these samples, I hoped to subsample one hundred cases, making it virtually impossible for the administrators to trace who my sampled research participants would be.

Shortly after I submitted the proposal, the county and the state both drafted new policies about providing identifying information about welfare recipients to researchers. These county and state policies were ambiguous and, in some ways, contradictory. As a result, officials were hesitant to release the samples, though they admitted that doing so would have been simple. After more than seven months of discussions with county officials, I realized that this sampling technique was not going to occur. (Admittedly, I had anticipated a very low response rate by writing solicitation letters to welfare recipients.) I ultimately tried another strategy.

The strategy had two parts. First, I contacted and left recruiting leaflets at legal aid and welfare rights organizations in the county in an effort to contact welfare recipients who might be more “rights conscious” than the average welfare recipient. Through this method I located only two participants—Jane and Tanya. Second, I recruited participants by approaching individuals who were exiting the primary county welfare office. I parked myself on the sidewalk outside the offices and did my best to approach every adult who left the office. To protect participants from possible embarrassment, I did not approach individuals if others were within earshot. After catching individuals’ attention, I stated that I was a researcher from the nearby university and was conducting a study of people’s experiences with the welfare system. I explained that participation involved an interview of approximately an hour and a half and that the interview would be scheduled at a convenient time and place.

When recruits expressed interest in participating, I assured them that I was not employed by or otherwise affiliated with the welfare office. I explained that there were two requirements to participation—current receipt of welfare benefits and an age of eighteen or older. Because my interview questions concerned familiarity with welfare rules, and because I did not want to contaminate the responses I received, I did not tell participants before the interviews began that my research particularly focused on welfare rule knowledge and compliance.

Participants usually set up a meeting time on the spot, though a couple asked me to call them at home so that they could coordinate their schedules or check their calendars. I told participants that my questions addressed sensitive personal information, specifically questions about money and questions (p.193) about problems they might have experienced with the welfare system. I also told them that I was interested in facts that the welfare office might not know and that I was taking measures to minimize the risk of the welfare office finding out any of this information. I informed participants that their participation was completely voluntary and that their consent could be withdrawn at any time.

The Interviews, Trust, and the Presentation of Self

At the 1998 American Sociological Association Meeting, I discussed my early thoughts on this research project with a social scientist who had related research interests. She warned me that I would have a lot of trouble gaining trust and getting people to reveal honest information. My experience working with welfare recipients was quite the contrary. Not only did people tell me things about themselves that were not particularly admirable, they also commonly shared much more information than I wanted to know.

I have come to realize that I was not only studying rule breaking but also engaging in a sort of rule breaking in the course of research. In doing the type of qualitative research that I did, I violated social scientific norms about who does research, guidelines about researcher neutrality, social expectations of fixed social status, and everyday norms that caution people against asking personal questions. Failing to acknowledge these violations would, I believe, give an air of dishonesty to my data presentation and analysis.

The very fact that I was a graduate student surprised many of my research participants. For the record, I am an African American woman who uses a wheelchair. At the time I began recruiting research participants, I was visibly pregnant. For the first six months after my daughter’s birth, I toted her along with me both when recruiting research participants and when conducting interviews. As a researcher, I was engaged in an unusual and ambiguous status relationship with my research subjects.

The renowned sociologist Erving Goffman described status as “a pattern of appropriate conduct, coherent, embellished, and well articulated” (1959, 75). The status I held as a researcher was sometimes at odds with the intersecting axes of privilege and subordination that construct my identity, as well as with the image I projected to strangers. No matter how coherent my conduct as a researcher, my appearance to some degree undermined my researcher status. The result of my various positions: status ambiguity.

Textbook training for aspiring social scientists is geared toward those individuals who have traditionally been aspiring social scientists: white, (p.194) socially and economically privileged students and researchers who lack visible disabilities. Advice on interviewer comportment and interaction is based on the assumption that researchers already project a certain image. Practical training for qualitative research is often hard to come by. Instructors offer the advice that new studies should look like existing studies and that methods and approaches to research should not stray toward innovation for risk of compromising study comparability and data integrity. For those of us who do not look like most other American social scientists, however, our bodies may become influential media in the inquiry process.

When I began this research project, I was not prepared for the ways that the image, or images, I project would affect my interactions with the research participants. The fact that identity plays a role in research would come as no shock to most social scientists. Identity affects both research interests and comfort in pursuing certain research agendas. Social researchers who conduct interviews or engage in direct observation tacitly take into account their social identities when gauging their ability to gain access and trust. The interplay between researcher identity and research subjects’ identity is rarely discussed in methodological literature, except as something to be overcome. (For a notable exception to this general rule, see a collection of research stories by Francis Winddance Twine [2002].) A researcher’s identity, though, is a variable in the research. While this variable is sometimes brought to light by the researcher (Brunskell 1998), this occurs far too infrequently.

My unusual social location affected basic issues of methodology, including recruitment; interviewer neutrality; openness and trust; critical distance; and interviewer self-disclosure. In fact, my initial contact with research participants was unusual. Many of the individuals I recruited approached me before I approached them. Apparently concerned to see a disabled woman sitting outside the welfare office, a number of people walked right up and asked whether I needed help. Sometimes when my infant daughter was with me in the early days of recruiting interviewees, people came up to make baby talk to my daughter. A couple times, as I was making my way from the car to the sidewalk, women approached me and asked, “Are you going to the welfare meeting?” In these situations, my response that I was a university researcher and my solicitation of their participation came as unexpected. Though I did my best to look the role of the researcher—dressed-up but not dressy; glasses rather than contacts; folio or clipboard in hand as a prop—my race, disability, motherhood status, and perhaps even gender seemed to undermine my “performance” of the researcher (Goffman 1959). Thus my interactions with research participants began with the experience of status ambiguity.

(p.195) Many of the qualitative studies of the poor and their experiences with the welfare system that have been published in the last two decades (e.g., Berrick 1995; Edin and Lein 1997; Sarat 1990; Seccombe 1999) have been conducted—like most academic studies in the United States—by white, financially stable, able-bodied researchers. For these researchers, their social status both put them at a distance from their research participants and served as a known barrier going into the research. For example, John Gilliom, in his study of welfare mothers in Appalachia (2001), quickly realized that his age, gender, economic status, and regional accent were all barriers to data collection and hired recent welfare recipients from the area to conduct the research interviews. Gilliom and the other researchers knew they would have difficulty gaining the trust of individuals who might consider themselves (or be considered by the researchers) as social subordinates. My ambiguous social status, however, meant that there were no established interaction rituals, no clear indications of which of us should be playing the subordinate and therefore deferential role.

My status ambiguity also fogged the issue of researcher neutrality in my study. The traditional literature on qualitative research methods stresses the need for the appearance of neutrality on the part of the researcher. Interviewers, in particular, are instructed not to give off any signs of political stances, provide clues about their socioeconomic background, or provide hints about their feelings about interviewees’ responses (Singleton and Straits 1999, 262, 269-70). But this traditional advice seems specifically aimed at traditional-looking interviewers. There is nothing neutral about the image I gave off; it called out to have stereotypes projected upon it. Moreover, there was only so much of my image that I could control. The people I interviewed commonly assumed from my race, gender, disability, and motherhood status that I had direct experience with government aid programs. This affected our interactions, leading interviewees to assume that we shared experiences and perspectives that we might not have actually shared. Anne-Marie Fortier writes that the images we researchers project are unstable and are commonly negotiated and renegotiated in the course of our interactions with research subjects (1998, 49). There may be, however, a point where unstable images become so unstable that they destabilize traditional modes of social inquiry.

Qualitative researchers do not universally agree upon appropriate research methods or even on the epistemological assumptions underlying the research. Departing from classical social science methods, some feminist methodologists not only take a less rigid stance on neutrality but even encourage researchers to reveal more about their agendas and themselves (p.196) in efforts to gain trust and obtain personal information from interviewees (Acker, Berry, and Esseveld 1991; Oakley 1981). These researchers, aware of gender biases in the substance and methods of traditional research, are constantly engaged in continual “epistemological debate about the nature of legitimate inquiry” (Brunskell 1998, 43). Feminist methodologists argue that only by identifying with the research subject can the researcher find “truth.” If truth includes interviewees’ admissions of deception, then my nonclassical approach to methodology was effective in eliciting the truth about rule breaking. For the most part, however, the “truth” I was searching for in the interviews had less to do with interviewees’ descriptions of their behaviors than with their beliefs about the welfare rules and the welfare system—the view from the inside.

My breaches of convention extended further. In the course of my research, I violated social norms by asking strangers personal questions about finances and family, as well as invasive questions about lying and cheating. Before conducting my research, I believed that eliciting this information from research participants would take some time and effort. I found, to my great surprise, that welfare recipients were quite forthright in disclosing rule-breaking behavior. Often the women volunteered such information before I even asked. My guess is that my apparent subordinate social location as defined by my gender, race, and disability made me nonthreatening.

“Nonthreatening,” though, may not completely describe the interaction. I sense that, despite my role as researcher, the participants I interviewed considered me neither a social superior nor an equal. My identity was, and is, so marginal that the possibility that I might develop negative judgments of them did not seem to concern them. I did not fall within their social peerage. As a result, they offered deeply personal information about their finances, families, and social failings.

I also found that the interviewees had no hesitation in turning the tables, making me the object of their questions. The magnifying glass became a two-way lens. While self-disclosure by interviewers is typically discouraged (Weiss 1994, 79), I found it unavoidable. Being asked personal questions seemed to be the price I paid for doing the same. While I bristled at these questions and tried to give short, dampening responses, the counterques-tions were a persistent aspect of my interviews. Again, my status ambiguity seemed to inspire curiosity. In some interviews, I felt as if I were asked to ante up with some degree of self-disclosure for each sensitive question I put forth. These were experiences for which my graduate school research methods courses had not prepared me.

(p.197) I noticed two effects of my violating the conventions of social interaction and social scientific research. First and foremost, I was able to obtain rich data. The people I interviewed described their frustrations with not understanding complex welfare rules; their pain at having little education and poor reading skills; their mundane and extraordinary strategies for coping economically; their fears of being caught by “the system,” a term that seemed to envelop welfare fraud investigations, child protective services, and the criminal justice system; and their admissions of mental illness, drug use, prostitution, and victimization by abuse. There is, I believe, no way that I could have captured this information through less interactive methods.

Moreover, the interviews tended to go long, and the interviewees seemed quite comfortable participating in the research. Classical methodologists view an interviewer’s response effects as a systematic error in research, a source of bias (Singleton and Straits 1999, 267). My experience suggests, however, that response effects may be neutral or even positive aspects of an interview if the effect is merely the interviewee’s divulging of honest information.

The second effect of my approach, however, was my frequent discomfort, even prompting my hesitation to pursue similar research in the future. Both delving into the lives of others and revealing so much of myself to my research participants were emotionally draining. Sometimes I found myself dreading to ask interviewees follow-up questions about personal matters, anticipating that they would feel free to ask equally touchy questions of me. I believe that my marginalized social location also raised the psychic costs of interview research.

Qualitative research is particularly well suited to revealing actors’ understandings and misunderstandings, motivations, and hidden behaviors. The types of data gathered in the course of social research are affected not only by the types of questions that are asked but also by how the questions are asked and by whom they are asked. While reflexive social inquiry has been gaining ground, few social scientists acknowledge that the researcher herself (or himself) is a variable in the research process, both as a data gatherer and as a data analyst. For much of U.S. academic history, there has been little variation in the social status of those researchers pursuing sociolegal or public policy questions. Social scientists are wedded to the idea that for data to reflect truth it most be possible for all inquirers to be able to gather and observe the same data.

Patricia Hill Collins writes that African American women have often held an “outsider within” status because they have been nonparticipant observers in mainstream, white society (1990, 11-13; 1991). For African American (p.198) women intellectuals, she argues, marginality provides a distinctive perspective on social life and theory. Collins suggests that this status affects analysis and interpretation of data, as well as engagement with and evaluation of social theory. In developing my research agendas, my research questions, and my methods of research, I commonly find myself an outsider within. As a social scientist studying traditional theory and methodology, I find that my minority identities are not represented in those theories, those methods, or the society composed of my disciplinary colleagues.

I am additionally an “outsider within” as a researcher doing fieldwork. My multiply subordinate social location offered me some entrée into the circle of welfare recipients I was interviewing. At the same time, my ambiguous social status and my status as a social scientist maintained my place as an outsider. This status created unusual researcher-subject interactions, allowing me to explore data that have been until now gone unexplored. (My sense is that Sudhir Alladi Venkatesh, a sociologist studying life in low-income, African American neighborhoods in Chicago, has been able to gain unusual access because of his “outsider” status as a man of South Asian descent [Venkatesh 2000, 2006].)

My experiences as an interviewer suggest that qualitative researchers—of all races, ethnicities, ages, genders, abilities, and sexual orientations, not just minorities—should reveal their social locations in presenting their findings. These statuses are significant variables in the ability to observe, to ask. Until researchers reveal these variables, they will not be given the attention in methodology that it deserves. Second, more diversity in the pool of qualitative researchers might reveal layers of social phenomena that are yet unexplored. Bringing more outsiders into the discipline of sociolegal studies might just open up new areas of understanding.

The Research Sample

The response rate for the research was admittedly low. I approached more than two hundred potential interviewees to obtain the thirty-four interviews. Fewer than half of the individuals who initially agreed to be interviewed actually followed through. In addition, in many of the cases of successful interviews, it took multiple attempts at meeting to finally complete the interview.

Most of the interviews lasted approximately an hour and a half. During the interviews, I asked questions about what had led the interviewees to apply for welfare, about their familiarity with the rules and requirement of (p.199) the welfare system, and about their knowledge of the consequences of violating welfare’s rules or failing to meet the requirements. I asked the interviewees to explain how they made ends meet financially. I also asked whether they thought specific rules were fair and necessary, whether they had particular rules, regulations, or aspects of the welfare system they wanted to change.

This study departs from most qualitative studies of welfare recipients in that I tried to interview a cross section of welfare recipients. I did this by recruiting near the county welfare office, approaching every person who left the building. My lack of fluency in any language other than English certainly prevented me from sampling from a broad ethnic and linguistic spectrum that Bayview enjoys. Still, I found great variation in levels of rule knowledge, in things respondents did to make ends meet, and in perceptions of the law to convince me that I was gathering rich and valuable information.

In addition, some of the sociolegal studies with the poor tend to sample individuals who may be distinct—those who have mobilized the law through legal services. For example, Austin Sarat (1990) interviewed welfare recipients in a legal services office in his article on the welfare poor. Welfare recipients seeking legal services, however, are likely to have a certain degree of knowledge of the rules, or at least of the procedural rules, and are arguably in the office with the belief that they have some access to law that they can use in their favor. In short, they are using rules strategically. They also have a certain orientation toward the legal system in general. They must invest some legitimacy in the rules and the procedures for calling upon or challenging those rules, even if they view the system as biased.

While Sarat attends to rights and due process as legal symbols at play in the legal consciousness of welfare recipients, he does not attend to other social and cultural symbols that may also be relevant to welfare recipients. In the context of the American welfare system, gender, race, and motherhood are not merely axes of individual identity but also axes of cultural experience and markers of social status. To discuss rights as salient symbols without discussing “welfare queens,” “deadbeat dads,” or “vagrants” as salient cultural symbols limits the analysis of the complex cultural milieu, where various symbols and stereotypes inform individual action, individual identity, and individuals’ relationships to the law.

Qualitative studies of welfare recipients that use snowball sampling techniques (e.g., Edin and Lein 1997; Gilliom 2001; Seccombe 1999) also select a particular group of welfare recipients: those who have social networks that include other welfare recipients. While some of the individuals (p.200) I interviewed did know other welfare recipients, others did not. In some cases, in fact, the respondents were quite socially isolated, with limited or nonexistent social networks. One of the biases that may exist where interviewees are part of a network of welfare recipients is that they are likely to share knowledge, or at least understandings, of the welfare system. Since part of my goal was to assess welfare recipients’ knowledge of welfare rules, I wanted to avoid conducting interviews that followed established circuits of shared knowledge.

In some ways my recipients reflect the local population of welfare recipients; in some ways they do not. Admittedly, my sample does not reflect the racial and ethnic diversity of the county, in large part because of my limited language ability. In 1999, 31 percent of the welfare recipients in the county used a language other than English as their primary language. My research therefore excluded a sizable population of Vietnamese, Chinese, Laotian, Cambodian, and Spanish-speaking welfare recipients. It also excluded small but significant populations of Afghan and Ethiopian welfare recipients.

While a disproportionate percentage of my interviewees were African American, I do not consider this weighty oversampling a liability in the research design. The political scientist Martin Gilens (2003) has conducted research finding that many Americans hold a racialized view of welfare, associating welfare with African Americans, especially urban African Americans. And at the time of the data collection, slightly more welfare recipients—58 percent (Coulton 2003, 95)—lived in urban areas rather than rural areas. But race and racism are particularly salient to studies of welfare recipients. In fact, focusing on how African Americans respond to the legal and moral regulation that occurs through welfare policy is compelling because they have become the focus of welfare policies. Kenneth Neubeck and Noel Cazenave write that the stigmatization of African American women in discourses about welfare has been fostered by the media, academics, and the general public (2001, 31). Studies by Ange-Marie Hancock (2004) and Martin Gilens (1999, 2003) reveal that negative public attitudes about welfare correlate with, and perhaps rise from, negative stereotypes about African Americans. While focusing on this population poses the risk of perpetuating the myth that poverty and welfare use are limited to city-dwelling African Americans, research about the very group most associated with welfare in the public consciousness will provide a clearer picture of the lives these families lead. In short, holding interviews with individuals who fit not just the stereotype but the very archetype of welfare recipients is of some value in bringing a lens of reality to the myth.

(p.201) The Data Analysis

This study was guided by Barney Glaser and Anselm Strauss’s notion of grounded theory (Glaser and Strauss 1967). In a grounded-theory approach, a researcher generates theory from data rather than testing a priori theories against found data. To do this, according to Glaser and Strauss, “one generates conceptual categories or their properties from evidence; then the evidence from which the category emerged is used to illustrate the concept” (23).

My interview schedule and my interviews thus included numerous questions about how and where welfare recipients received information about welfare rules, what they knew about the rules, how they navigated the rules, what their financial needs were each month, and how they made ends meet. I reviewed the interview schedule after each interview to assess which questions and topics were effectively generating information and which ones were not. I then adjusted the interview schedule to better obtain useful information. For example, after a handful of early interviews where I asked very detailed questions about monthly income and expenses, I discovered that often the interviewees simply could not recall even approximate dollar amounts well enough for specific time-consuming questions to be particularly useful assessments of need. More general questions such as “How do you make ends meet?” or “How do you feed your family?” were somehow better able to capture the economic dynamics in the family—especially if those questions were repeated at different points in the interview.

I developed conceptual categories during the course of data collection and well afterwards. I periodically reviewed the interview tapes and transcripts to refresh my familiarity with the material and wrote analytical notes based on the recurrent issues I found. Common issues and themes emerged from the interviews. For the most part, these themes became and headings and subheadings of data chapters.

I began analyzing the data by turning the conceptual categories into a detailed coding scheme. I coded a number of the transcripts and tried a qualitative software program to analyze the coded data. Once I had developed the categories, however, I found that most of the analytic work had been done and that taking the time to code the remaining transcripts offered little value to the project. With a sample of only thirty-four recipient interviewees, much of the data could be organized in Excel spreadsheets without the use of software specifically designed for qualitative analyses. Some of the basic characteristics of the interviewees—age, education, marital status, number of minor children—I compiled in distribution tables.

(p.202) One of the frustrations of collecting rich data in the course of lengthy qualitative interviews is that so many of the details of individual experience are necessarily lost in the effort to confine the discussion to the topics of the research. That loss seems particularly poignant in this study because the men and women who were the subjects of this research lead lives that are often invisible to middle-class Americans. My hope is that in the future other types of projects will allow those who are economically privileged to gain more complete and complex impressions of the lives of those who are economically disadvantaged. More importantly, I hope that others—scholars, members of the press and popular media, and policy makers—will better allow poor families the opportunities to express and share their experiences with everyone else.

As the author, I have no way that I can write without my voice being the dominant one in this manuscript. My framing of the issues and my choices in selecting quotes particularly filter the view of the recipient interviewees. Still, the discussion borrows from the lives and views of many other individuals. In an attempt to make their voices come through, I include a lot of unpolished quotes that are in many cases longer than the scholarly norm. I encourage the reader to engage with these quotes rather than treat them as textual distractions.