Public trust and pre-release access to statistics

This is one of a series of notes to support our policy positions as summarised in the RSS Data Manifesto. It was first published in April 2015, and was updated in May 2017. You can download the PDF version at this link. To discuss the policies we are advocating, contact our Policy and Research Manager, Olivia Varley-Winter.

1. Summary

End the practice of pre-release access whereby some people in government see statistics before the public.

The RSS Data Manifesto calls on government to help build public trust in statistics. Statistics are one of the key building blocks to create the evidence base for science and research, for government policy making, and for public and media use. Impartial, independent and high quality statistics are therefore essential for a high-quality democratic debate. Government ministers should not be the arbiters of whether data are accepted as an official statistic: the independent UK Statistics Authority should continue to fulfil that role, and should be the prime guardian of the correct use of official statistics.

One of our key requests in this regard is for the government to end the practice of pre-release access to official statistics, whereby ministers and their officials have access to official statistics before they are released to the public. In the latest (2016) survey of public confidence in official statistics, although the majority (78%) of the public agreed that official statistics are generally accurate, only 26% agreed that the government presents official figures honestly when talking about its policies [1]. Internationally comparative surveys suggest that in the UK, official statistics are not as widely trusted as is the case in many other countries [2]. The public needs to be reassured that statistics are not being misused to suit political or media agendas, and decisive action against pre-release access is one of the actions needed to respect the impartial nature of official statistics, and that all should have equal access to them.

The UK also needs to work hard to ensure that it sets and fulfils global standards for public trust in new uses of data. Algorithms, big data, data science, machine learning, the internet of things and smart cities are changing our data, our public policy, businesses, and daily life. Legislative changes should empower the independent Information Commissioner to audit compliance with data protection law and punish bad practices. To strengthen governance of wide ranging public concerns in this area, we have also recommended the establishment of national Council for Data Ethics. Such a body would advise government, the private sector and the public on issues of data governance, transparency, and fairness, and would help industry to innovate more safely by building consensus.

2. A clear case for the abolition of the current practice of pre-release access

Legislation should be changed to empower the UK Statistics Authority to keep pre-release access to a minimum across all four UK administrations.

Currently, ministers and key officials responsible for a policy area have access to statistical releases and publications in that area 24 hours before they are made available to Parliament or the general public. This is in direct contravention of the principle of equal access which is enshrined in international good practice [3]. It also has implications for public confidence in statistics. The most recent (2016) report on Public Confidence in Official Statistics notes (at p.21) that “the proportion [of the public] who felt that specific statistics are free from political interference ranged from just 46% for employment figures to 76% for the Census.” A majority (67%) in this survey supported official statistics being made available to everyone at the same time, with no pre-release access [4].

It is sometimes argued that access to statistics at the earliest opportunity is necessary for ministers to manage the affairs of their department. It is however the case that where statistics are derived from administrative data and management information, managers and ministers can access those data directly. There is no justification, in our view, for pre-release access to the statistical outputs of government.

The arguments for pre-release access (as attested, for example, in the Cabinet Office review of the arrangements) also rest on the alleged necessity for ministerial comment or a departmental press release being issued at or around the same time as the statistics [5, 6]. In our view the practice of issuing such commentary to coincide with statistical releases is pernicious. It skews debate over the figures and perpetuates the impression that ministers control the data. Evidence from surveys of public attitudes toward official statistics suggests that lack of confidence in statistics is in large part due to perceptions of political control or misrepresentation [7]. Pre-release access is likely to strengthen that perception, so is a major obstacle to improving public confidence. Straightforward abuse of statistics to exert political control is rare, but access 24 hours in advance of release (as is the case in England and Northern Ireland) allows the possibility for media management, and the suspicion that this could be taking place. The situation in Scotland and Wales is worse in that up to five days of pre-release access is allowed.

Additional to this, we have witnessed growing concerns about the extent of pre-release access in the UK to market sensitive (potentially market-moving) data. To provide an example of the extent of this: according to the ONS, 114 people had pre-release access to labour market data in April 2017, with their job titles being published in the Pre-release access list for UK labour market statistics [8]. On this matter we note that the greater the pre-release access we have in the UK, the greater the likelihood that such data may be leaked [9].

The latest round of peer reviews of statistical standards across Europe has highlighted pre-release access as “an area of weakness” in the UK. Reviewers concluded that the current situation of pre-release in the UK is “excessive” and recommended that it ultimately be abolished [10]. The key principle underlying this position is that of equal access to statistics for all users, as defined in the United Nations Principles Governing International Statistical Activities. These principles also state that good practice involves impartial compilation and dissemination of international statistics, and requires there to be a clear distinction, in statistical publications, between statistical and analytical comments on one hand and policy-prescriptive and advocacy comments on the other [11].
To address public trust in this area of weakness, the relevant legislation should be changed at the earliest opportunity to empower the UK Statistics Authority to keep pre-release access to an absolute minimum in line with international best practice, across all four UK administrations.

3. Ministers and public officials should handle data with integrity

All too often statistics are being misrepresented in the public debate. Ministers, MPs and other public officials are guided by the Ministerial Code to follow the UK Statistics Authority’s Code of Practice [12]. The Code states that at all stages in the production, management and dissemination of official statistics, the public interest should prevail over organisational, political or personal interests, and that no statement or comment is issued to the press or published ahead of the publication of the statistics [13]. In recent years there has been a much closer scrutiny of the misuse of statistics in public debate by the UK Statistics Authority and by fact-checking organisations such as Full Fact. The Authority in 2008 highlighted the need for their Code of Practice to apply much more widely, following the production of a Home Office fact sheet on knife crime which published administrative ‘hospital episode statistics’ data that had been accessed before ONS processing and official release [14, 15]. Subsequent interventions have addressed issues such as misreporting by the former Prime Minister, David Cameron and others of new jobs figures in relation to foreign workers, which is an issue that Full Fact has also campaigned on [16, 17]. Examples such as these continue to show that good practice in official statistics must be applied more widely in future.

4. The independent UK Statistics Authority should be the prime guardian of the correct use of official statistics

Ministers should not be the arbiters of whether data are accepted as an official statistic.

Since its inception in 2007, the UK Statistics Authority has established itself as a key independent player in the UK statistical landscape. Its statutory role is to be the prime guardian of the correct use of statistics in public debate, and its interventions, we believe, have informed both politicians and the media. We would encourage the government and other policy makers to continue to seek the Authority’s advice, and to follow its guidance to prevent recurrent cases of misuse. The government should be widely committed to improving how information, as a basis for policy making, is communicated in public. Standards in the communication of statistics, adjudicated by the Authority, have a major part in this. It is also the case that a great many statistics are produced, of variable quality, some providing a better standard of evidence than others. We are of the view that ministers should not be the arbiters of whether data are accepted as an official statistic and their categorisation as such. That responsibility should be transferred in law to the UK Statistics Authority.

5. The independent Information Commissioner's Office should have greater powers, and a national Council for Data Ethics should be established

The Information Commissioner’s Office should have greater powers to audit compliance and punish bad practices.

In addition to addressing ministers’ use of statistics, one of the key ways of building trust in statistics among the general public is to have strong safeguards in place to protect individuals’ personal data. Research commissioned by the RSS and others has shown the public lack trust in the use of data by organisations across the public and private sectors. There are a range of safeguards that, when applied and discussed, help to reassure the public [18, 19, 20]. Sufficient privacy safeguards should therefore be built into any sharing of personal data at the outset. We would also like to emphasise that there is a need for independent oversight and governance to better assure the safeguarding of information and to improve trust. The Data Protection Act provides for the Information Commissioner’s Office (ICO) to have certain powers, such as monetary penalty notices and prosecutions. However we believe the ICO should have greater powers to audit compliance and punish bad practices. Data protection provisions should also be written in to the government’s contractual arrangements, and penalties for data theft should be strengthened, to encourage greater attention to the ICO’s remit. Alongside these data protection concerns, new questions have also emerged, such as how we can hold algorithms to account for automated (or semi-automated) decisions – for example, who is responsible when a driverless car crashes? In light of major dilemmas about the ethics of new data science, we have recommended that a national Council for Data Ethics should be established to advise government, the private sector and the public on issues of data governance, transparency, and fairness [21]. Such a body would help industry to innovate more safely by building consensus around standards and ethics [22].

Endnotes
[1] Simpson, I. (2017) Public Confidence in Official Statistics - 2016 [PDF]. London: NatCEN social research. Available from:https://natcen.ac.uk/media/1361381/natcen_public-confidence-in-official-statistics_web_v2.pdf
[2] P. 13 in European Commission (2007) Special Eurobarometer: Europeans knowledge on economical indicators [PDF]. Available from: https://www.oecd.org/site/worldforum06/39562127.pdf
[3] United Nations Statistics Division (2006) Principles governing international statistical activities [html]. Available at: https://unstats.un.org/unsd/methods/statorg/Principles_stat_activities/principles_stat_activities.htm
[4] Simpson, I. (2017) Public Confidence in Official Statistics - 2016, ibid. as above.
[5] UK Statistics Authority (2010) Pre-release access to official statistics: a review of the statutory arrangements, March 2010. London: Crown Copyright
[6] Cabinet Office (2010) Pre-release access to official statistics policy [PDF]. London: Cabinet Office. Available at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/62084/pre-release-stats.pdf
[7] Bailey, R. Rofique, J. & Humphrey, A. (2010) Public Confidence in Official Statistics 2009 [PDF]. London: NatCen Social Research. Available at: http://www.natcen.ac.uk/media/25384/public-confidence-official-statistics-2009.pdf
[8] ONS (2017) ‘UK labour market statistics: Apr 2017’ > ‘pre-release access list’. Available at: https://www.ons.gov.uk/releases/uklabourmarketstatisticsapr2017#prereleaseaccesslist
[9] Bird, M. ‘New Data Suggest U.K. Government Figures are Getting Released Early’, The Wall Street Journal, 13 March 2017. Available at: https://www.wsj.com/articles/market-drift-suggests-some-investors-may-be-trading-on-u-k-economic-data-ahead-of-release-1489397403
[10] Snorrason, H. Byfuglien, J. & Vihavainen, H. (2015) Peer review report on compliance with the Code of Practice and the coordination role of the National Statistical Institute: United Kingdom [PDF]. Available from: http://ec.europa.eu/eurostat/documents/64157/4372828/2015-UK-report/d44f7d3f-64c1-4450-8a37-bfadb8542607
[11] United Nations Statistics Division (2006) Principles governing international statistical activities [html]. Available at: http://unstats.un.org/unsd/methods/statorg/Principles_stat_activities/principles_stat_activities.htm
[12] Cabinet Office (2016) Ministerial Code [pdf]. London: Cabinet Office. Available at: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/579752/ministerial_code_december_2016.pdf
[13] UK Statistics Authority (2009) Code of Practice for Official Statistics: Edition 1.0 [PDF]. London: UK Statistics Authority, Available at: http://www.statisticsauthority.gov.uk/assessment/code-of-practice/code-of-practice-for-official-statistics.pdf
[14] House of Commons (2009) ‘Minutes of evidence taken before public administration committee: Official Statistics; publication of statistics relating to knife crime’ [html], 5 February 2009. Available at: http://www.publications.parliament.uk/pa/cm200809/cmselect/cmpubadm/c229-i/c22902.htm
[15] ‘Publication of statistics related to knife crime’ [PDF], Letter from Sir Gus O’Donnell to Tony Wright MP, 16 January 2009. Available from: http://www.statisticsauthority.gov.uk/reports---correspondence/correspondence/publication-of-statistics-relating-to-knife-crime.pdf
[16] ‘Statistics on new jobs’ [PDF], Letter from Sir Andrew Dilnot to Jonathan Portes, 18 August 2014. Available from: http://www.statisticsauthority.gov.uk/reports---correspondence/correspondence/letter-from-sir-andrew-dilnot-to-jonathan-portes-18082014.pdf
[17] ‘ONS moves to end misreporting of foreign worker job statistics’ [online article], Full Fact, 12 July 2011. Available at: https://fullfact.org/blog/foreign_worker_statistics_ons_clarification-2815
[18] Cameron, D. Pope, S. & Clemence, M. (2014) Dialogue on Data: Exploring the public’s views on using linked administrative data for research purposes [PDF]. London: Ipsos MORI. Available from: https://www.ipsosmori.com/researchpublications/publications/1652/Dialogue-on-Data.aspx
[19] RSS (2014). Research on trust in data and attitudes toward data use / data sharing [PDF]. London: Royal Statistical Society. Available from: http://www.statslife.org.uk/images/pdf/rss-data-trust-data-sharing-attitudes-research-note.pdf
[20] Sciencewise (2014) Big Data: Public views on the collection, sharing and use of personal data by government and companies [PDF]. Didcot: Sciencewise. Available at: http://www.sciencewise-erc.org.uk/cms/assets/Uploads/SocialIntelligenceBigData.pdf
[21] RSS (2015) The Opportunities and Ethics of Big Data: Workshop report [PDF]. Available from: http://www.rss.org.uk/Images/PDF/influencing-change/2016/rss-report-opps-and-ethics-of-big-data-feb-2016.pdf
[22] ‘Science and Technology Committee (Commons)’ > ‘Algorithms in decision-making inquiry – publications’ > Written evidence submitted by The Royal Statistical Society (ALG0071) [PDF] available from: http://data.parliament.uk/writtenevidence/committeeevidence.svc/evidencedocument/science-and-technology-committee/algorithms-in-decisionmaking/written/69144.pdf