Panel Conditioning: Change in True Values vs. Change in Self-Report

• Activity Number: 252
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Ting Yan*+ and Stephanie Eckman and Frauke Kreuter
• Companies: and Institute for Employment Research and Joint Program in Survey Methodology
• Keywords: panel conditioning ; measurement error ; longitudinal surveys ; administrative data ; misreport

Abstract

Panel conditioning is an important source of measurement error unique to panel surveys. It refers to the phenomenon where participation in repeated interviews changes respondents' true behaviour, attitudes, or knowledge, or their reporting of their true behaviour, attitudes, and knowledge. A major weakness of (and a challenge to) the existing research on panel conditioning is its inability to distinguish between true change and change in reporting behavior. Existing studies are heavily reliant on assumptions and models when studying panel conditioning because they have no external gold-standard data source. This paper examines panel conditioning effects using data from four waves of a large German panel survey on labour market outcomes (PASS). Because administrative data on employment and unemployment benefit receipt status are available for nearly all respondents, we are able to separate panel conditioning due to change in true status and panel conditioning due to change in self-report of the true status without depending on assumptions. Our results show that PASS respondents are more likely to change their true behavioural status the longer they stay in the panel. In addition, they are less likely to misreport their behavioural status the longer they stay in the panel.