Of the 162 sampled individuals who were eligible to be interviewed, interviews were actually completed with 137 people, or 85 percent (Table IV.1). All of these respondents confirmed that they had received cash assistance in LBP month six but not in month seven. Most of the interviews were completed by telephone (121 interviews) rather than in person (16 interviews). Interview completion times ranged from 25 to 117 minutes, with a mean completion time of 44 minutes. Interviews could not be completed with 25 sampled individuals for the following reasons:
- Thirteen individuals could not be located.
- Seven individuals could be located but could not be contacted.
- Three individuals were contacted but refused to participate in the survey.
- Two individuals began, but did not complete, the interview.
Data on a limited number of characteristics of all sampled individuals (both survey respondents and nonrespondents) and their associated FIP cases were available in DHS administrative files. These data are used in Table IV.1 to profile the 162 survey-eligible individuals and the subset of 137 survey respondents. Survey respondents are somewhat more likely to be male and somewhat less likely to be minorities than all survey-eligible individuals. For the other five selected individual and case characteristics (marital status, case size, cash benefit amount in LBP month six, birth year of respondent, and birth year of youngest child in case), there are no notable differences between the survey respondents and the broader group of survey eligibles.
The high survey response rate and the relatively small differences between the survey respondents and all survey-eligible individuals suggest that the LBP Survey results reliably represent the experiences of all individuals and their associated FIP cases in SDAs 9, 10, and 11 that entered LBP month seven in late 1995 and early 1996. The principal limitation of the survey data derives from the small number of persons who were eligible for the survey, implying that descriptive statistics generated from the data may have large confidence intervals.