Using auxiliary data to model nonresponse bias The challenge of knowing too much about nonrespondents rather than too little?
Butt, S. & Lahtinen, K. Using auxiliary data to model nonresponse bias The challenge of knowing too much about nonrespondents rather than too little?. Paper presented at the International Workshop on Household Nonresponse 2015, 02 Sep 2015 - 04 Sep 2015, Leuven, Belgium.
Abstract
The ADDResponse project (www.addresponse.org) explores the potential for using auxiliary data from multiple sources to understand and correct for nonresponse bias in general social surveys in the UK. Data from the census and other administrative sources together with consumer profiling data and geographic information about local neighbourhoods have been matched to data from Round 6 of the European Social Survey in the UK.1 Preliminary bivariate analysis suggests that a large number of these variables may be associated with response propensity and worthy of further investigation. Here we discuss some of the preliminary steps we have taken to try and identify the most likely candidates for nonresponse adjustment and compare the results from propensity models employing theory-driven vs. automated variable selection.
| Publication Type: | Conference or Workshop Item (Paper) |
|---|---|
| Subjects: | H Social Sciences > HM Sociology |
| Departments: | School of Policy & Global Affairs > Sociology & Criminology |
Download (265kB) | Preview
Export
Downloads
Downloads per month over past year
Metadata
MetadataActions (login required)
Admin Login