The effect of events between waves on panel attrition

  • Mark Trappmann Institute for Employment Research (IAB) Otto-Friedrich-Universität Bamberg
  • Tobias Gramlich
  • Alexander Mosthaf
Keywords: panel survey, attrition, nonresponse, weighting

Abstract

Panel surveys suffer from attrition. Most panel studies use propensity models or weighting class approaches to correct for non-random dropout. These models draw on variables measured in a previous wave or from paradata of the study. While it is plausible that they affect contactability and cooperativeness, panel studies usually cannot assess the impact of events between waves on attrition. The amount of change in the population could be seriously underestimated if such events had an effect on participation in subsequent waves. The panel study PASS is a novel dataset for labour market and poverty research. In PASS, survey data on (un)employment histories, income and education of participants are linked to corresponding data from respondents' administrative records. Thus, change can be observed for attritors as well as for continued participants. These data are used to show that change in household composition, employment status or receipt of benefits has an influence on contact and cooperation rates in the following wave. A large part of the effect is due to lower contactability of households who moved. Nevertheless, this effect can lead to biased estimates for the amount of change. After applying the survey’s longitudinal weights this bias is reduced, but not entirely eliminated.

Author Biography

Alexander Mosthaf
Universität Mainz
Published
2015-04-10
How to Cite
Trappmann, M., Gramlich, T., & Mosthaf, A. (2015). The effect of events between waves on panel attrition. Survey Research Methods, 9(1), 31-43. https://doi.org/10.18148/srm/2015.v9i1.5849
Section
Articles