What Factors Explain Variation in Monitors' Detection of Interviewing Errors in Telephone Surveys?
Douglas Currivan
RTI International
Paul p. Biemer
RTI International
Tamara Terry
RTI International
Ivan Carillo-Garcia
RTI International
To limit the potential for interviewer behavior to bias or add variance to survey estimates, monitoring telephone surveys requires accurate and consistent detection of interviewing errors, interview protocol violations and other biasing behaviors. Multiple factors can affect telephone monitors' detection of interviewer deficiencies. Because monitoring results alone are insufficient to understand how multiple factors contribute to monitors' error detection, this research combines data from (1) monitoring sessions, (2) responses to a survey of monitors, and (3) administrative records with monitor and interviewer characteristics. These data operationalize multiple sources of variation in monitors' error detection that could not be captured by the monitoring system data alone. Multi-level models are used to analyze the contribution of these sources to variation in monitors' detection of interviewing errors. This paper discusses the implications of these results for understanding sources of variation in monitors' detection of interviewing errors and for guiding decisions on designing monitoring processes in centralized telephone survey centers.