Abstract
The extent to which response and nonresponse errors occur in the reporting of time deposits is investigated in two consumer financial surveys. The findings point not only to substantial errors in estimates of the means and size distributions of accounts in such surveys, but also suggest that confidence intervals computed from such data are badly distorted. Many different factors appear to be associated with these errors, including the questionnaire approach as well as respondent characteristics.

This publication has 0 references indexed in Scilit: