Consent Error, Nonresponse Error, and Measurement Error: Assessing the Overall Quality of Linked Survey and Administrative DataISER Internal Seminars

Survey records are increasingly being linked to administrative databases to enhance the survey data and increase research opportunities for data users. A necessary prerequisite to performing exact record linkage is obtaining informed consent from respondents. Obtaining consent from all respondents is a difficult challenge and one that faces significant resistance due to societal concerns regarding privacy and data confidentiality. As a result, data linkage consent rates vary substantially from study-to-study and sometimes fall below response rates (Woolf et al., 2000; McCarthy et al., 1999). Several studies have found significant differences between consenters and non-consenters on socio-demographic characteristics, leading to concerns that inferences obtained from linked data sources may be biased (Kho et al., 2009; Dunn et al., 2004). However, these studies are limited because they rely solely on survey data to assess the consent biases. No study has attempted to determine whether systematic differences exist between consenters and non-consenters on key administrative variables. Estimating consent biases for administrative variables is complicated by the fact that administrative records are generally not available for the non-consenting portion of the sample. We analyse consent biases in the 2006/2007 German Labour Market and Social Security Survey (PASS). The PASS survey obtained an 80% consent rate to link survey records with federal employment and benefit records. With permission from the German Institute for Employment Research (IAB), we obtained access to administrative records for both consenters and non-consenters and estimated consent biases for several administrative variables. This is the first study to estimate consent biases for linked administrative outcomes. In addition, we use call record data and variables common to both the survey and administrative data to estimate nonresponse and measurement error biases. This paper compares the relative contribution of each error source (consent, nonresponse, and measurement) to the overall error observed in the linked dataset.

Presented by:

Frauke Kreuter (University of Maryland)

Date & time:

26 Jan 2011 13:00 pm - 26 Jan 2011 14:00 pm


Internal seminars home

News

Latest findings, new research

Publications search

Search all research by subject and author

Podcasts

Researchers discuss their findings and what they mean for society

Projects

Background and context, methods and data, aims and outputs

Events

Conferences, seminars and workshops

Survey methodology

Specialist research, practice and study

Taking the long view

ISER's annual report

Themes

Key research themes and areas of interest