Survey data is subject to a wide range of threats to its quality, including non-response and measurement error. While a number of these threats are increasing, a positive development is that several sources of information about quality are also increasing, in particular as a by-product of the evolving technologies used in survey data collection.
There has been significant research on ways of using such information to inform quality control and management of survey data collection, but there has been an absence of rigorous methodological research on how to make the most effective statistical use of such information when analysing the data. Chris Skinner has been awarded a three-year ESRC Professorial Fellowship from February 2010 to address this gap.
The focus of the research will be on the use of indicators of quality recorded at the level of a survey unit, such as an individual. Although the availability of such information is growing, particularly in the form of paradata, its incorporation into survey data analysis is not straightforward since, in particular, there is a risk that attempts to use such indicators to improve the precision of survey estimates may be offset by the creation of unintended biases.
The primary theme of the research will be to develop and investigate methods for using indicators of measurement error. Research on additional themes, including nonresponse, data linkage and statistical disclosure control, will enable the fellowship to develop a systematic basis for using quality indicators across a diverse range of settings. Applications of the methodology will be explored across a range of survey datasets.
The research will focus on the needs of survey data analysts, but will also seek to provide insights to data collection agencies which wish to make effective use of information on quality in the management and design of data collection processes.