Affiliations: Board of Governors of the Federal Reserve System (Ret.), Main Stop 153, Washington, DC 20551, USA. E-mail: [email protected]
Abstract: Typically, the end of a survey field period is a point when a survey
struggles to achieve what is seen as an acceptable response rate. Response
is shaped by the decisions of field staff to continue applying effort and
respondents to be open to persuasion. The situation is an uncomfortable one
in that it is quite difficult to apply measurable standards to the process
in a way that can be mapped with any precision into response probabilities.
The Survey of Consumer Finances (SCF) has long gone to the bitter end of the
field period where the remaining possibilities of completed interviews are
very slim, and in doing so has held response rates approximately constant
over time. The hope is that this operational approach will allow as much
constancy across time as possible in any response biases. Yet this is a
costly approach that could not be justified easily if the last cases are not
sufficiently different from cases collected earlier. This paper examines the
set of the last cases collected for the 2007 SCF and describes the ways in
which they differ from ones collected earlier in the field period. Some
limited comparisons are also made with the nonrespondents.
Keywords: Survey of consumer finances, unit nonresponse, response bias, response rates, late cases, field work