Studying Recall Error in Survey Data Using Administrative Records

• Activity Number: 397
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Survey Research Methods
• Author(s): Sigrid Behr and Johannes Eggs and Gerrit Müller and Iris Pigeot-Kübler and Walter Schill and Rainer Schnell and Mark Trappmann*+
• Companies: BIPS - Uni Bremen and BIPS - Uni Bremen and IAB and BIPS - Uni Bremen and BIPS - Uni Bremen and Uni Duisburg-Essen and IAB
• Keywords: recall error ; seam effect ; telescoping ; panel survey ; PASS

Abstract

Survey often rely on respondents' reports based on their autobiographical memory. Therefore, survey data may suffer from measurement error due to memory failure that can affect models based on these data. The German panel survey PASS collects information on work histories using repeated interviews and retrospective reporting. The PASS data have been linked to administrative data on work histories from the German employment agency. The paper focuses on recall error with respect to recipiency of unemployment benefit/insurance payments. After describing distributions of different kinds of errors found in the data we focus on (1) seam effects and their causes, (2) the effect of length of the recall period on recall error, and (3) if likelihood of a recall error can be predicted by individual attributes and which effect it has on coefficients of time-to-event models of recipiency dynamics.