Conferences

The quality of survey data is of concern to social scientists and policy makers alike. It is determined by many aspects of the survey process including sampling, patterns of response and non-response as well as survey design and data collection procedures. To provide an overview of our session, we begin by comparing and contrasting cognitive interviewing, behaviour coding, and traditional pilots used to evaluate aspect of survey and question design in order to expose their underlying assumptions about the sources of survey data quality. Drawing on examples from the British Household Panel Study, we expose how each method can lead to inconsistent conclusions about data quality. We offer examples of ways the strengths of various evaluative methods can be combined to better inform survey design decisions. Our session is then comprised of four papers. The two initial papers provide new models for assessing total survey quality. The remaining two contain examinations of data quality post fieldwork.