Abstract
The Survey of Consumer Finances (SCF) collects dollar amounts for a wide variety of assets, liabilities, payments, incomes, and other items. Non-negligible problems with item nonresponse for these variables in the past has been mitigated by the collection of range information. However, reports from interviewers suggested that probing for ranges was sometimes awkward. The redesign of the 1995 SCF for computer-assisted personal interviewing (CAPI) allowed the collection of range data to be incorporated more formally into the interview process than was feasible with a paper questionnaire and to capture more detailed information about the nature of probing. The results from the use of the new approach suggest that many responses that would previously have been recorded as ``don't know'' were instead collected as range information, and that ranges may have displaced some point estimates made by the respondent. In addition, the data suggest there are complex interactions effects that determine the types of ranges that interviewers and respondents negotiate. One justification for imposing this additional burden on interviewers and respondents is the potential for addressing nonignorable item nonresponse more fully. Although the results presented here tend to support the collection of range data as an efficiency-improving measure, there is not strong evidence that imputations would otherwise be biased.