When working with survey data, a bad representation of the target population or measurement errors may considerably bias results. Survey methodologists identify such sources of error drawing on the ‘Total Survey Error’ (TSE) framework. In this talk, Oliver Lipps (Head of Survey Methodology at FORS) will present different components of the TSE using recent large-scale social science surveys.
In the first part of his presentation, Oliver will cover the analysis of representation bias from selective non-observation, such as selective undercoverage or selective non-response in telephone surveys. He will further give examples of recent research on how to increase observation rates and reduce selection bias, for example by adding new survey modes. In addition, he will show that conventional ‘solutions’ such as weighting or more effort put into fieldwork may increase bias.
In the second part of his presentation, Oliver will introduce interviewer effects, a major source of error. Specifically he will elucidate the role of the interviewer in making contact and obtaining cooperation, discuss how to optimize interviewer-household assignments in telephone surveys, and show, which substantive topics may suffer from gender of interviewer effects.
Oliver Lipps studied Mathematics at the University of Freiburg (Breisgau) and holds a PhD in civil engineering from the University of Karlsruhe and a habilitation degree from the University of Basel. He worked for the German Mobility Panel (MoP), the Survey of Health, Ageing, and Retirement in Europe (SHARE), and the Swiss Household Panel (SHP) before he became head of the FORS methodological research unit in Lausanne and private docent at the Institute of Sociology at the University of Berne.
Oliver Lipps published more than 25 papers about survey data quality issues with a focus on nonresponse and attrition in panel surveys. Together with colleagues from NCCR LIVES he wrote articles on social inequality and representativity of minorities in surveys. In addition, he is lecturer in survey methodology and survey research at the Institute of Sociology at the University of Berne and at the Swiss Summer School in Lugano, focussing on methods of panel data analysis.