Tackling survey nonresponse: The role of geocoded auxiliary data

Thursday 26th May 2016 17:00 – 19:15

London Art House

Low response rates – and the potential this has to lead to bias – are one of the major challenges facing survey research today. One commonly suggested approach to address non-response is to append auxiliary data available for both respondents and non-respondents to the dataset. However, despite the growing array of data that can be appended to sample frames using geocodes such as postcode, doing so can often be a time-consuming and frustrating task and research to date has struggled to identify auxiliary variables which are sufficiently correlated with both response propensity and the survey variables of interest to be useful.

This workshop provided an opportunity to hear findings from the ADDResponse Project   – which investigates the scope that auxiliary data provide to understand and overcome nonresponse bias in the European Social Survey in the UK – and discuss their implications for survey research and practice. Incorporating auxiliary data from a variety of sources and at multiple levels of aggregation – including small-area government data, household level commercial data and local geographic information – the project provides a uniquely in-depth insight into the potential afforded by geocoded auxiliary data to address the problem of nonresponse in social surveys.

Full workshop programme and speaker biographies


Sarah Butt and Kaisa Lahtinen “ADDResponse: Getting to grips with different types of auxiliary data: Was it worth it?

Kathrin Thomas and Rainer Schnell “ADDResponse: Predicting nonresponse with small area auxiliary data

Panel Discussion

An expert panel of survey methodologists and practitioners led a discussion of the following questions:

  • What are the key findings emerging on using auxiliary data to study nonresponse bias, from ADDResponse and/or other research projects you are aware of?
  • What are the main challenges associated with using auxiliary data to explore survey nonresponse?
  • What are the most promising areas for further research combining auxiliary and survey data?
  • How will/should the growing availability of auxiliary data influence survey practice in the future?

Panel participants were:

  • Patrick Sturgis, University of Southampton and NCRM
  • Tom W. Smith, NORC at University of Chicago
  • Patten Smith, Ipsos MORI
  • Michael James, ONS

Read a summary of the discussion.