Online Program Home
My Program

Abstract Details

Activity Number: 389 - Improving Survey Data Quality with Machine Learning Techniques
Type: Invited
Date/Time: Tuesday, July 31, 2018 : 2:00 PM to 3:50 PM
Sponsor: Survey Research Methods Section
Abstract #326535 Presentation
Title: Predicting Panel Drop-Outs with Machine Learning
Author(s): Christoph Kern*
Companies: University of Mannheim
Keywords: panel attrition; nonresponse; machine learning
Abstract:

Panel attrition due to nonresponse can lead to a substantial loss in data quality and is therefore studied extensively in survey research. A large part of this work focuses on developing or refining methods with which systematic dropout patterns can be corrected after the data has been collected (e.g. through weighting). Against this background, this study investigates the potential of moving from post- to pre-correction of nonresponse in panel surveys by predicting dropouts using information from previous waves and machine learning methods. The usage of data-driven classifiers thereby allows to model complex non-linear and non-additive relationships while focusing specifically on prediction accuracy. By comparing conditional inference trees, random forests, boosting and standard logistic regression using the German Socio-Economic Panel Study (GSOEP) it can be shown that particularly tree-based ensemble methods offer a promising avenue for predicting panel dropouts in advance.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program