Interviewers and data quality: Evidence from the 2001 survey of consumer finances

Abstract

The most obvious pressure on field interviewers during a survey is to complete interviews. Comparable efforts to enforce data quality standards are hampered because many of the most important indicators of quality are embedded in the data in ways that are typically very difficult to extract quickly enough to be useful during a survey field period. This paper examines a number of important indicators of data quality based on the data in the 2001 Survey of Consumer Finances and uses that information to assess variations over interviewers. Of particular interest is the low correlation across interviewers between some measures of the quality of the data they collected and the rate at which they completed cases. The paper argues for three things: (1) development of feedback systems to monitor and enforce data quality, (2) reexamination of the role of interviewers with particular attention to the possibility of dividing their tasks, and (3) investigation of the labor market for interviewers.