The Office for National Statistics has just published a report examining public attitudes to personal data being utilised for official statistics (PDF). Drawing from a number of different studies, it explores what the public knows about the use of personal data for statistical purposes as well as attitudes towards it. The report also follows the recent recommendation by Jil Matheson that the next census should be supplemented by the use of administrative data.
A ONS survey conducted in February 2012 found that three quarters of the public trusted the ONS to protect confidentiality of personal data, with 59 per cent supporting the retention of personal data in a central database for statistical purposes. A second survey (conducted in 2013) also found 60 per cent public support for personal data being stored in a central database if it were anonymised (significantly higher than if data were not anonymised).
However, just one half of the public supported it sharing their personal data with government departments. For commercial organisations, it was even less, at just a third. A number of focus groups (conducted in 2010 by the ONS, and then in 2013 by Ipsos MORI) found that participants expected a direct and public benefit to sharing their data. Linkage for ‘socially beneficial purposes’, for example, had increased support.
When it comes to understanding the issues, a series of cognitive tests conducted by the ONS of a small sample of 45 people indicated that caution should be applied in using terms such as ‘administrative data’, ‘data linking’ and ‘anonymisation’. Only limited understanding of the terms ‘address register’, ‘data sharing’ and ‘demographic’ were evident from these tests.
The focus groups mentioned above also found ‘a clear lack of understanding of the distinction between the use of data for statistical purposes and for operational purposes.’ The distinction between the two is quite significant, as explained in the report. Statistical use of data is defined as where 'no information about an identifiable individual will be disclosed (the data are published in aggregated form) and no actions or decisions will be made which could have a direct individual impact. Operational use, on the other hand, 'requires identifiable data about an individual for the provision of public services, and may impact directly on that individual – for example the use of information in administering the tax or benefit systems.' The 2013 workshops conducted by Ipsos MORI also showed little prior familiarity with the purpose, processes and use of social research and statistics.
Detailed findings from each of the unpublished studies mentioned in the report are due to be released in April 2014. The ESRC and Ipsos MORI’s Dialogue on data focus group research was published on 14 March 2014, and which David Walker covers in his recent StatsLife opinion piece.