The Quality of Data

Mary Thompson, PhD
Professor
Statistics and Actuarial Science
University of Waterloo

January 11, 2006
12:00 Noon - 1:00 PM
Davis Centre 1304, University of Waterloo

View Video of Presentation in HI Alive Archive: Research Seminars Archive 2005-2006

Abstract
The talk will discuss the role of survey data in health research. Health care decision makers rely on data from the census, from multipurpose surveys like the Canadian Community Health Survey, and from surveys of disease prevalence, surveys of health care use, and surveys of opinion. Survey science has recently begun to focus on an all-encompassing notion of “data quality”. Examples will be given illustrating the components of “total survey error”: measurement problems, data collection/processing error, sampling error, non-response error and coverage problems. The last part of the talk will be devoted to current research on the reduction of non-sampling error.

Biosketch
Mary Thompson is a Professor in the Department of Statistics and Actuarial Science. She is Co-Director (with John Goyder of Sociology) of the Survey Research Centre (SRC) at the University of Waterloo. SRC is interested in survey methodology and survey research and has over 100 members from three universities: University of Waterloo, Wilfrid Laurier University and the University of Guelph. The Centre holds seminars on research issues, carries out surveys, develops and teaches courses in survey administration and statistical analysis, and provides training opportunities and employment to students.

Dr. Thompson is also an investigator on a series of international surveys on tobacco control (PI Geoff Fong of Psychology), and director of the associated Data Management Core. In her own survey research she is currently focusing on the construction and use of survey weights for longitudinal, multilevel and latent variable models, and resampling methods for interval estimation and testing.