The quota sampling technique was also employed for both supplemental samples (homeowners and rural debtors), with the addition of more qualifying criteria. The supplemental rural sample consists only of those debtors living in certain predetermined rural zip codes; all respondents not in those zip codes were eliminated from the supplemental sample. The Los Angeles supplemental housing sample was developed by continuing to collect questionnaires beyond the original target of 250 until a total of 100 homeowners had been collected; nonhomeowner cases were discarded from this supplemental sample. The supplemental housing sample for Philadelphia and Chicago used two filters: The debtors were homeowners and they agreed to be interviewed by telephone; nonhomeowners and those who did not wish to be interviewed were omitted from those supplemental samples.
To recap: The core sample consists of 1,250 debtors, 250 from each of the first five states. This sample was supplemented by an extended sample of 970 debtors, which includes an extended selection of homeowners from the five core sample districts and an additional sample of rural debtors from Tennessee and Iowa. None of the supplemental data is part of the core sample.
Protection of Confidentiality
Before we entered the field, the study’s procedures and instruments were reviewed by Harvard University and the University of Texas to guarantee protection of study participants. Consistent with those reviews, access to debtor identity has been limited to guarantee respondents’ confidentiality. All researchers who had access to respondents’
names and data were required to sign a statement promising to maintain confidentiality. Finally, responses are reported only in the aggregate. When individual quotations are used, the debtors’ names and any other obvious identifiers have been changed.
Instruments
Data were collected using three types of instruments: debtor questionnaires, court records, and telephone interviews.
Questionnaires.
The debtor questionnaires included the same demographic questions asked in the 1991 and 1999 research. In addition, many new questions were added in response to the data collected from the open-ended questions asked in 1991 and 1999, as well as developments in consumer law and practice. For example, the 2001 questionnaire asked debtors why they had filed for bankruptcy; the options provided on the questionnaire were based on the eight reasons most often given by debtors in 1991, plus an option to fill in a blank after “other.” In addition, because so many debtors in 1991 and 1999 had indicated that an illness or injury was the catalyst for their bankruptcy, the 2001 questionnaire asked debtors if they had lost income as a result of their own or a family member’s medical problems. The 2001 questionnaire also gathered information on health insurance, medical debt, child support, home ownership, mortgage debt, alternatives to bankruptcy, and number of dependents.
The majority of the questions were closed-ended and required the debtor only to check a box. One opened-ended question was for occupation: Debtors were asked for their occupations and, if married, the occupations of their spouses. Debtors’ answers were entered verbatim into the database. Dr. Sullivan coded each entry using the 1970 U.S. Census codes and the corresponding prestige scores developed by the National Opinion Research Center.
7
Dr. Thorne recoded each entry to ensure accuracy and consistency.
In addition to the closed-ended questions, debtors were invited to use the back of the questionnaire to tell the story of their bankruptcies in their own words. The responses were included in the database as a text field.
At the bottom of the last page of the questionnaire, a form offered debtors $50 for participating in follow-up telephone interviews, with the possibility that they might eventually complete three such interviews over the next few years. Debtors who were willing to be interviewed signed the forms and provided their telephone numbers. Only those debtors who signed this form and provided a phone number were called. Debtors whose primary language was Spanish were given a Spanish version of the questionnaire. The questionnaire was translated into Spanish and back-translated into English to verify that the translation was accurate.
With the assistance of the Bankruptcy Court in Boston and the United States Trustee for Region One, the questionnaire was pretested in January 2001. Because this questionnaire was considerably more comprehensive than the one used in 1991, there was some concern about respondents’ reactions to the amount of time it would take to complete it. The average time required to complete the questionnaire was six minutes, and no respondent expressed any concern about the length of the questionnaire. After testing, the questionnaires were refined and retested in the Boston courthouse. Once that was complete, the same questionnaire was used in all districts.
The questionnaires asked each debtor for more than thirty pieces of information, for a total of about 75,000 pieces of information gathered from the questionnaires. Copies of the questionnaire are available from the Consumer Bankruptcy Project, Harvard Law School.
Court Records.
For every questionnaire that debtors returned, a copy of the corresponding court records was collected; thus the researchers coded court record data from all 2,220 cases in the core and supplemental samples. Court records for bankruptcy are a matter of public record. A photocopy of the court records or a printout from the online court records was added to each debtor’s file.
Three people with legal training reviewed the court records, determined which information to code, and assisted in the development of a coding program for data entry. The information reported in the debtors’ schedules, which are filed under penalty of perjury, includes income, assets, debts, household expenses, and several legal events. The research team entered approximately 160 pieces of information from each case into the database, for a total of more than 350,000 pieces of information gathered from the court records.
Telephone Surveys.
The telephone survey comprised four specialized schedules: general questions presented to all respondents, a medical section presented to those who indicated that a medical problem had contributed to their bankruptcies, a small business section presented to current or former small business owners, and a homeowner section presented to those debtors who owned a home at the time of filing or who had owned a home within the previous five years but had lost it through foreclosure or other financial problems. Interviews took about twenty minutes for each section.
Telephone interviewers were all trained and supervised by Dr. Thorne. All initial interviewers were graduate students or postdoctoral students in sociology; later, other trained interviewers were added. All interview responses were entered directly into a Microsoft Access database designed specifically for this project. Approximately 226,000 pieces of data from the telephone survey were entered in the database.
Response Rates
In some districts, the Chapter 7 and Chapter 13 trustees distributed and collected the questionnaires; in other districts, we trained a local person to complete the distribution
and collection, with the cooperation of the trustees. In almost all cases, the questionnaires were distributed and completed when the debtors attended a mandatory meeting at the courthouse, although a few people took their questionnaires home and returned them by mail a few days later.
Without the cooperation of the trustees, it would have been almost impossible to gather a useful sample from among the debtors attending these meetings. The need to rely on the trustees, however, meant that we lost some degree of control over the distribution of the questionnaires. Although we can be nearly certain that the questionnaires were distributed to every debtor who attended a meeting scheduled on a particular day, on some days the trustees did not distribute questionnaires, often because they had no help, because they felt especially pressed for time, or because they forgot. In one district, the Middle District of Tennessee, the trustees went far beyond our expectations and maintained especially careful records, and we could determine that the response rate was 99 percent. In the remaining four districts, however, we could not impose on the trustees to keep similar records, so it is not possible to identify the questionnaire response rates with precision. We estimate that in Texas and Illinois, response rates were approximately 90 percent on the days the trustees distributed the questionnaires. In Illinois, California, and Pennsylvania, when someone other than the trustee distributed the questionnaires, the response rates were lower, estimated at about 60 percent, 55 percent, and 45 percent, respectively.
Because court records are public data, we collected court records for each debtor in the sample, effectively a 100 percent response rate.
Debtors from the core sample and the supplemental homeowner samples were eligible to participate in the telephone survey. Participants in the Iowa supplemental rural sample were not part of the telephone interview portion of the study because those questionnaires were gathered later.
From the core sample of 1,250, 875 debtors (70 percent) agreed to a telephone interview. Of these, it was possible to contact and complete at least one schedule of the telephone survey with 609 debtors, for a response rate of 69.6 percent of those who agreed to be interviewed, or 48.7 percent of all debtors in the core sample. In the supplemental housing sample, 521 debtors were eligible to be interviewed; of these, 328 completed telephone surveys, for a response rate of 62.9 percent. In total, 930 telephone interviews were completed between the core and the supplemental samples.
Coding Accuracy
Data from both the questionnaires and court records were entered into a Microsoft Access database designed specifically for this instrument. Training was extensive, particularly for collecting data from court records. Ten percent of all questionnaire
entries were rechecked for errors. The error rate was approximately 0.7 percent. Because the court record data coding was extremely complex, on completion of the data entry, the researchers arranged to recode 500 cases. All mistakes were corrected; therefore, the accuracy of these 500 cases approaches 100 percent. Of the remaining approximately 1,720 cases, 15 percent were randomly selected for quality control checks. On average, there was one error per 160 data points, which translates into accuracy exceeding 99 percent.
ACKNOWLEDGMENTS
In a book about debt, we now acknowledge our own debts, which are substantial. No work involving so much data, so many different areas of expertise, and so many ideas could emerge without the contributions of many different actors.
We begin our thanks with those who provided the support that made this book possible. The Ford Foundation (grant 1010-1838) and the Robert Wood Johnson Foundation (grant 042425) provided generous support for the collection of the bankruptcy data that are used in this book. The Fellowship Program at the Radcliffe Institute for Advanced Study was the first to invest in a project about middle-class mothers and children in financial trouble, generously supporting one of us (Elizabeth) for the year of research that laid the foundation for this book. The Harvard Law School and the New York University School of Law both made substantial contributions to the Consumer Bankruptcy Project as well; we are particularly grateful to the respective former deans, Robert Clark and John Sexton, for their ongoing support for empirical work so that the project could begin.
As we noted in the appendix, the 2001 Consumer Bankruptcy Project was built on the hard labor put into the 1981 and 1991 Consumer Bankruptcy Projects. Dr. Teresa Sullivan and Professor Jay Westbrook have been my (Elizabeth’s) longtime coauthors, coinvestigators, and coconspirators. When it was time to put together the 2001 Project, they brought their considerable talents and energies to yet another empirical study. No one could ask for better fellow travelers—and friends.
The 2001 project opened up new lines of inquiry, requiring participation by people with expertise in diverse specialties. We are grateful to Dr. David Himmelstein and Dr. Steffie Woolhandler, both of the Harvard Medical School, for their contributions on health care finance issues. We wish to thank Dr. Susan Wachter, of the Wharton School, University of Pennsylvania, and Professor Michael Schill, of the New York University Law School, for structuring the research on housing policy issues. We thank Professors Bruce Markell and Robert Lawless, University of Nevada-Las Vegas, for developing the interview questions
for the self-employed. Their substantial contributions transformed the Consumer Bankruptcy Project into a richly complex, multifaceted study.
In addition, three former students (now moving into academic positions in their own right) served at different times as project directors: Katherine Porter provided the energy and enthusiasm to help design and initiate the data collection and see the project take life. Professor John Pottow added expertise as we began the work on court records. Professor Deborah Thorne completed the data collection and painstakingly assured the accuracy of the data. Each enriched and strengthened the Consumer Bankruptcy Project, and we express our thanks for their hard work.
Everyone associated with the project benefited from the continuing support of Alex Warren, who designed and managed the database with great skill. He was always willing to do “just one more run” and always alert to new ways to test our hypotheses. We are grateful for his help.
The bankruptcy judges and trustees made data collection possible. We offer our lasting thanks to the Honorable Keith Lundin, of the United States Bankruptcy Court for the Middle District of Tennessee, who provided the kind of advice and encouragement that makes empirical work possible, and to his very able assistant Cindy Odle, who did much of the ground work in Tennessee. We are also grateful for the support and cooperation of the chief judges of the districts we studied: the Honorable Steven Felsenthal, the Honorable Bruce Fox, the Honorable Lee Jack-wig, the Honorable Carol Kenner, the Honorable Geraldine Mund, and the Honorable Susan Sonderby. We also appreciate the ready cooperation of the Office of the United States Trustee and the Assistant United States Trustees, the individual bankruptcy trustees, and the bankruptcy clerks of Boston, Chicago, Dallas, Des Moines, Los Angeles, Nashville, and Philadelphia. We owe them an incalculable debt. Without their generous contributions of time and energy to this project, there would have been no 2001 Consumer Bankruptcy Project.