The Accuracy of Interview Paradata: Results From a Field Investigation
Wang, K. H., & Biemer, P. P. (2010, May). The Accuracy of Interview Paradata: Results From a Field Investigation. Presented at AAPOR 2010, .
Survey statisticians and methodologists are inventing new ways of using interview paradata. Novel uses range from monitoring and controlling errors in the field to post-survey adjustments. As an example, the authors are experimenting with using number of callback attempts and call outcomes for unit nonresponse adjustment. However, some problems with the quality of the paradata have limited the effectiveness of these uses. This paper reports on an investigation of paradata quality in a large-scale field survey. The study included both focus groups conducted with field supervisors and an informal survey of 601 interviewers regarding their practices in recording call attempts and call outcomes. A particular focus of this study is how interviewers record situations that are not be straightforward such as drive-by sightings, calls to neighboring units, encounters with gate keepers, and so on. The investigation identified a number of situations where call attempts and outcomes were recorded in error and where the collection of paradata was done in a manner inconsistent with the concept of “level of effort” that nonresponse adjustment models rely on. We speculate on the causes of the problems and discuss what steps could be taken to reduce the errors.