Changes in Labor Market Behavior Due to Panel Conditioning in a German Panel Study

• Activity Number: 77
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Survey Research Methods Section
• Author(s): Ruben Bach* and Stephanie Eckman
• Companies: Institute for Employment Research and Institute for Employment Research
• Keywords: administrative data ; longitudinal surveys ; panel conditioning ; propensity score

Abstract

Panel conditioning can refer to either changes in respondent behavior due to participation in repeated interviews or changes in their reporting of behavior. Difficulties arise when distinguishing between the two forms: Do changes over time reflect true change or do they reflect change in reporting (or both)? Using administrative data linked to a German panel survey on labor market outcomes, changes in the take-up of labor market programs are analyzed. These linked data allow us to distinguish between the two types of panel conditioning. We use propensity score weighting to estimate the average treatment effect of participation in several waves of the panel survey. Results show that respondents participate in more labor market programs than a control group of unselected (but eligible) persons. We interpret these results as evidence that the survey induces changes in behavior among respondents. These results will be of interest to all researchers who collect or analyze panel survey data.