communicating statistics: not just true but also fair...communicating statistics: not just true but...

HC 190 [incorporating HC 662-i, Session 2012-13] Published on 29 May 2013

by authority of the House of Commons London: The Stationery Office Limited

House of Commons

Public Administration Select Committee

Communicating statistics: Not just true but also fair

First Report of Session 2013–14

Report, together with formal minutes, oral and written evidence

Ordered by the House of Commons to be printed 21 May 2013

£14.50

The Public Administration Select Committee (PASC)

The Public Administration Select Committee is appointed by the House of Commons to examine the reports of the Parliamentary Commissioner for Administration and the Health Service Commissioner for England, which are laid before this House, and matters in connection therewith, and to consider matters relating to the quality and standards of administration provided by civil service departments, and other matters relating to the civil service.

Current membership

Mr Bernard Jenkin MP (Conservative, Harwich and North Essex) (Chair) Alun Cairns MP (Conservative, Vale of Glamorgan) Charlie Elphicke MP (Conservative, Dover) Paul Flynn MP (Labour, Newport West) Robert Halfon MP (Conservative, Harlow) David Heyes MP (Labour, Ashton under Lyne) Kelvin Hopkins MP (Labour, Luton North) Greg Mulholland MP (Liberal Democrat, Leeds North West) Priti Patel MP (Conservative, Witham) Mr Steve Reed MP (Labour, Croydon North) Lindsay Roy MP (Labour, Glenrothes)

Powers

The powers of the Committee are set out in House of Commons Standing Orders, principally in SO No 146. These are available on the Internet via www.parliament.uk

Publications

The Reports and evidence of the Committee are published by The Stationery Office by Order of the House. All publications of the Committee (including press notices) are on the Internet at http://www.parliament.uk/pasc

Committee staff

The current staff of the Committee are Emily Commander and Catherine Tyack (Joint Clerks), Rebecca Short (Second Clerk), Alexandra Meakin (Committee Specialist), Paul Simpkin (Senior Committee Assistant) and Su Panchanathan (Committee Assistant).

Contacts

All correspondence should be addressed to the Clerk of the Public Administration Select Committee, Committee Office, First Floor, 7 Millbank, House of Commons, London SW1P 3JA. The telephone number for general enquiries is 020 7219 5730; the Committee’s e-mail address is [email protected].

Communicating statistics: Not just true but also fair 1

Contents

Report Page

Summary 3

1 Introduction 5

2 Finding official statistics 6

3 Presenting and explaining statistics 9 Presenting statistics 9 Explaining statistics 12

4 Statistics on demand 15

5 Misuse of official statistics 17

Conclusions and recommendations 19

Formal Minutes 22

Witnesses 23

List of printed written evidence 23

List of Reports from the Committee during the current Parliament 24


Summary

A good evidence base should underpin all public policy. Across Government, robust statistics are essential to drawing up that evidence base. Public trust in the integrity of government policy will be more likely if the public understand the evidence base and the statistics used. Communicating statistics effectively is therefore an important way for the Government to uphold its accountability to the public and to ensure transparency in what it does.

However, finding government statistics is not easy. Both expert users and occasional users struggle to navigate their way through the multiple places in which statistics are published. In particular, the website of the Office for National Statistics must be improved. As well as being hard to find, statistics are often presented in a confusing way, for example, in formats which are not easily understandable. Government statisticians should work much more closely with different kinds of users in order to present statistics in ways which meet their different needs.

Government statisticians have an important role to play in explaining statistics as clearly and helpfully as possible. In some cases, the story behind the statistics is reduced in its presentation to such an extent that the picture is no longer true and fair. Government statisticians could do a lot more to explain statistics clearly.

In addition to the many routinely-produced statistics, government statisticians produce thousands of pieces of data on demand, known as “ad hoc statistics”. Whilst we welcome this openness, more of this kind of data should be published proactively, rather than simply in reaction to requests, and greater transparency around the process for ad hoc requests is needed.

An important part of the role of the UK Statistics Authority is to monitor the use and abuse of official statistics. Where the Chair of the Statistics Authority judges that there has been misuse of official statistics, we support his independence and his right to intervene.

We welcome efforts being made by the UK Statistics Authority, the Office for National Statistics and government statisticians to communicate statistics better, for example, through more media appearances, bringing together statistics on the ONS’s website on key themes like “population” and creating user-friendly ways to present statistics such as interactive guides. However, wider and deeper improvements are still needed to the presentation and explanation of government statistics if public trust in them, and therefore in public policy, is to be earned and kept.


1 Introduction 1. A good evidence base should underpin all public policy. Across Government, robust statistics are essential to drawing up that evidence base. Public trust in the integrity of Government policy will be more likely if the public understand the evidence base and the statistics used. Communicating statistics effectively is therefore an important way for the Government to uphold its accountability to the public and to ensure transparency in what it does.

2. This study is part of a wider programme of work we have announced on statistics and their use in Government. We are undertaking a series of short studies looking at particular ways in which statistics are used in Government; their accuracy and relevance; and their availability, accessibility and intelligibility to the public. A full description of the studies is set out under the heading “Statistics” in the “inquiries” section of our website, which can be found at www.parliament.uk/pasc.

3. We have made progress on several of the studies. In January 2013, we conducted post-legislative scrutiny of the Statistics and Registration Service Act 2007: our report, “Public Trust in Government Statistics”, stated that we found that “the Act has indeed helped to improve the operation of the statistical system. However, if the Act is to achieve its intentions, there needs to be greater clarity and transparency in the way it operates and, indeed, in the functioning of its primary creation, the UK Statistics Authority”.1 We have also looked at the work of the Office for National Statistics (ONS), publishing correspondence with the UK Statistics Authority (the Statistics Authority), taken oral and written evidence on the issue of migration statistics, and will be working through the other statistics topics over the coming months.

4. Our study into communicating and publishing statistics looked at existing processes for communicating statistics, including the format of statistics releases, the needs of users of statistics and the role and effectiveness of the Statistics Authority and the ONS. We called for written evidence, and took oral evidence from the Cabinet Office minister responsible for statistics, Nick Hurd MP, Full Fact, an independent fact-checking organisation, two journalists, the Chair of the Statistics Authority and the National Statistician. We are grateful to our Specialist Adviser on statistics, Simon Briscoe, for his help with this inquiry2.

1 Public Administration Select Committee, Ninth Report of Session 2012-13, Public Trust in Government Statistics, HC

406

2 Simon Briscoe declared interests in relation to his work as Specialist Adviser to the Committee. These can be found in full at: www.parliament.uk/pasc but of specific relevance here: he is a member of the Royal Statistical Society and trustee of Full Fact.

6

2 Finding official statistics 5. In its written evidence to us, the Statistics Authority expressed a view common to most evidence submitted to this inquiry:

The Statistics Authority starts from the perspective that official statistics are collected and managed at public expense and must justify that expenditure by contributing as much as possible to decision-making in all parts of society and the economy. However, official statistics can only do that if those who need to use them know that they exist, can find them when they need them, and can understand their relevance and utility.3

6. Producers of official statistics, including the ONS and government departments, will generally publish their statistics on their own websites. Two sites bring together official statistics: the National Statistics publication hub brings together first releases of accredited National Statistics, while data.gov.uk is a more general site for public data releases.

7. The Statistics Commission, the predecessor of the Statistics Authority, undertook two research projects into the ease of access to public statistics. In its detailed report resulting from these projects, published in June 2007, it set out eight principles of statistical dissemination, as listed in the box below.4 Many of the findings of that report have been echoed in evidence to this inquiry.

Principles of statistical dissemination, Statistics Commission, June 2007

1 Statistics are collected to be used and as wide a use of them as is possible should be encouraged.

2 UK government statisticians should adopt an exploratory and experimental approach to dissemination and access to statistical data through the Internet.

3 Government departments that publish official statistics should seek the full involvement of other web professionals in the presentation of statistical data on their websites.

4 Government departments that publish official statistics should recognise that web design and web culture are still developing and should set up an appropriate mechanism to keep accessibility issues under review.

5 User needs, interests and capabilities should determine the design and operation of statistical dissemination over the Internet.

6 Statistical products should be specifically designed for the Web.

7 Data should be presented in a layered or hierarchical way to allow users to drill down to the level of detail they desire.

3 Ev 50

4 Statistics Commission, Report No.34 Data on Demand – Access to Official Statistics, June 2007


8 There should be one point of entry – a government statistics portal – giving access to official statistics across the UK government and those of the devolved authorities.

8. The ONS website has long been a subject of complaint by users. It was redeveloped in 2011, but many users of statistics still report problems in finding the official statistics they need. This point was frequently made in the written evidence to this inquiry. The Market Research Society wrote “It is not easy to find data on the ONS website or publication hub. Making it easier would be the single greatest contribution to better access and communication”.5 The Statistics Users Forum told us “All too often it is extremely difficult even for the expert user to find the statistics they need from the ONS and departmental websites. Search engines leave much to be desired – most users rely on Google”.6 Chris Giles, Economics Editor at the Financial Times, talked us through the laborious process through which he – even as an expert user – had gone through to find the answer to the question “is unemployment now higher or lower than it was in the mid-1990s?”7

9. Jil Matheson, the National Statistician, and Andrew Dilnot, the Chair of the UK Statistics Authority, told us that they recognised the picture painted by our witnesses, and that the website was poor and had got worse following redevelopment in 2011. Andrew Dilnot said that the “relaunch of the ONS website [...] was not one of our greatest moments, and at that time the website became difficult to use, difficult to navigate, difficult to search”.8 Jil Matheson described the website relaunch in 2011 as “really disappointing”, and stated “I am a user of the [ONS] website as well as responsible for it and I share that frustration”.9 She added “There have been improvements, but the improvements that are there now are only part of a process. There is more to come [...] this is an ongoing development programme”.10

10. Although the issue of effective dissemination of statistics has been thoroughly explored, including six years ago by the Statistics Commission, progress has been slow. The ONS website and its relaunch in 2011 is a disappointment but we welcome the acknowledgment by the ONS of the problem. We note that they have undertaken to make substantial improvements to the website. Further improvements should be made as soon as possible to make the website accessible to ordinary users. We recommend that the ONS report progress to us and publish the report on the ONS website. We recommend the ONS also publish its plans for future improvements on its website. We recommend the ONS systematically seek and publish the views of users in order to inform further improvements to the functionality and presentation of official statistics on the ONS website.

11. There are many places in which official statistics are published; this is confusing to both the regular and the occasional user. The relationship between data.gov.uk,

5 Ev 33

6 Ev 44

7 Q 48 [Mr Giles]

8 Q 90

9 Q 93

10 Q 93

8

departmental websites and the ONS website is not clear. We recommend that the National Statistician review, update and adopt the principles set out by the Statistics Commission in 2007 and urgently take a greater role in sign-posting users to different groups of statistics. The Statistics Authority should publish information on how data.gov.uk relates to other websites showing government statistics.


3 Presenting and explaining statistics

Presenting statistics

12. Users of government statistics are a very diverse group —from expert analysts, journalists and civil servants, to Members of Parliament and members of the public. Evidence to this inquiry suggested that current modes of communication assume too much homogeneity within the user group and that focus is on some stakeholders to the detriment of others.11 The Social Research Association stated that:

More needs to be done to make official statistics accessible to a wide range of audiences. The user community is wide and diverse, ranging from analysts requiring the latest economic data for modelling work to a member of the public wanting to know the size of population of their town or village so a range of access routes are needed which assume different levels of knowledge and expertise.12

13. We were told that the form in which government statistics are currently presented is often quite limited. Witnesses suggested that statistics should be presented in a range of forms. The Statistics Users Forum told us that they were working with the National Statistician’s Office “to develop best practice guidelines for user engagement” and that:

Access arrangements should accommodate the requirements of the full range of levels of expertise among users: open data formats for those whose main aim is to use official statistics for secondary analysis, modelling, etc, through to simple tables and charts with informative commentary for the lay audience whose main aim is to be informed about trends in society and the economy.13

Full Fact suggested in their written evidence that:

The goal of publishing official statistics should be to ensure that users can get the information they need, in its full context, in the most convenient way. In particular, it should strive to present a coherent statistical picture in important or contentious areas of public debate. This entails different things for different users, which might be met in some of these ways:

· Graphically-led presentations such as graphs and maps. · Web pages using links to bring the full picture into view (suited to the

increasing proportion of mobile internet users). · PDFs, easily downloaded and printed, with full background and

interpretation in one place. · Spreadsheets for those who want raw data. · In due course, interactive tools.

11 For example, evidence from the Market Research Society (Ev 1), Social Research Association (Ev 3), Statistics Users

Forum (Ev 4) and Full Fact (Ev 7)

12 Ev 38

13 Ev 39

10

At the moment, official statistics are primarily presented in PDFs and spreadsheets, which seem geared toward more technical users.14

14. Both the Statistics Authority and ONS were clear that they saw effective communication of statistics as being central to their roles. Andrew Dilnot told us that communication was “a high priority” for the Statistics Authority, and explained in written evidence that the Authority was starting a programme of work to help to improve the communication of statistics and related advice to users:

Earlier this year, I convened a workshop to review how a series of ONS statistical releases could be better communicated to users. In light of that, the Authority has established, on a pilot basis, a ‘good practice team’ which will assist the statistical service more generally in developing and implementing improvements to current communication practices.15

15. Jil Matheson was aware of shortcomings and said “we have to improve in lots of ways”.16 She also told us of a number of planned improvements and that there was a new division in ONS called the public policy division, “whose aim is explicitly to be sensitive to the wider public debate and what the issues are and how statistics can be presented to inform that debate”.17 The Statistics Authority’s statement of strategy, published in February 2013, states “in support of the National Statistician’s vision for the GSS, the Authority will support and champion the role of departmental statisticians in effectively communicating statistics and in providing statistical advice for users”.18

16. We welcome the Statistics Authority’s programme of work to improve the communication of statistics across government. In particular, we welcome the creation of a public policy division in ONS. We recommend that the Statistics Authority publish information on the work of this team.

17. We are pleased to note that the Statistics Users Forum has been working with the National Statistician’s Office to develop best practice guidelines for user engagement, although these guidelines are not yet published and so we cannot comment on their content. We recommend that ONS disseminate and promote the best practice guidelines, as soon as possible, throughout Government. We recommend that the Statistics Authority and ONS, together with government departments, work much more closely with different kinds of users of statistics in order to present statistics in ways which meet their different needs.

18. As well as suggesting improvements to the presentation of statistics, written evidence suggested that more needs to be done to bring statistics together. Hard copy annual and monthly compendia, on topics such as “Social Trends”, “Financial Statistics” and “Economic Trends” have been discontinued, a decision which Andrew Dilnot told us was

14 Ev 46

15 Q102, Ev 50

16 Q100

17 Q117

18 UK Statistics Authority, Statement of Strategy, February 2013


taken “in the face of cuts [...] after some consultation with users”.19 He added “The serendipity that used to come from these compendia is important, and I am pretty sure that the National Statistician also shares the Chairman of the Authority’s slight discomfort at this decision”.20 Some Members of this Committee also share this discomfort.

19. There is therefore less attempt than in the past to coordinate the presentation of statistics where they pertain to a wider topic, such as inflation or health, or an event, such as those statistics relating to Scottish independence or the proposed referendum on the UK’s membership of the EU. The Royal Statistical Society wrote that:

A deeper communications challenge for the official statistics service is to present a coherent statistical picture of what is going on in areas where debate needs to concentrate on the issues rather than on explaining particular statistics. The debate on Scottish independence is an example where statistics need to be brought together and well communicated in order to foster good debate.21

The National Statistician suggested to us that government statisticians successfully did just this following the riots which started in Tottenham in 2011.22 Although we welcome in principle all efforts to communicate statistics more coherently, it is unclear why this topic was chosen for special treatment.

20. The ways in which statistics are presented sometimes present a challenge even for expert users. The lay user is left confused and disengaged. We recommend that the Statistics Authority work proactively to bring together and clearly present key statistics, from various sources, with associated commentary and in printable format, around common themes or events, such as elections and referendums, as well as broader topics such as the labour market, economic trends and so on. This is especially important given the ending of hard copy compendia on such topics.

21. We recommend that the Statistics Authority continue to explore more creative ways of communicating statistics, for example, through interactive guides. This should be in addition to the publication of more raw data in machine-readable format for experts who want the full results, not just the edited highlights presented in releases for the mass audience.

22. Evidence also suggested there was a disconnect between producers of statistics, interpreters of statistics and audiences for statistics. Witnesses told us that, given limited resources available, frequent users of statistics would prefer fewer, but better presented statistics. In their written evidence Full Fact suggested that:

To standardise and improve ONS releases, many of which are poorly written, all should be sent through a desk of sub-editors and the communications office before publication. The Norwegian equivalent of the ONS employs journalists full-time. They work closely with statisticians to produce public-friendly press releases. While

19 Q 94

20 Q 93

21 Ev 44

22 Q 117 [Ms Matheson]

12

this system might not translate perfectly into the UK, the Committee could certainly look to this model as inspiration for what can be achieved when communication is made a first-class part of a statistical office’s task.23

23. The Statistics Users Forum stated in their written evidence that:

Many users obtain official statistics through the press, broadcasters, social media, and other secondary sources. The Government Statistical Service (GSS) could do more to help these mediators to disseminate statistics, through presenting them in simple formats with informative explanations to which links can be made. This would not only widen the use of official statistics but would also improve the accuracy with which they are reported.24

24. Whilst cautioning against seeking “eye-catching headlines” or seeking “maximum coverage”, Andrew Dilnot told that “we need to employ journalistic skills” although not necessarily employ journalists. The National Statistician suggested to us that in future, very well-explained statistical releases could form the news release.25

Explaining statistics

25. There is a difficult balance to be struck by those responsible for communicating statistics between, on the one hand, explaining statistics in a way which is useful to the different user groups, and on the other, appearing to jeopardise their impartiality by providing the wrong sort of commentary. The Statistics Authority’s overview report on its own statutory assessment of Official Statistics between 2009-2012 found that more could be done across the statistical service to communicate statistics and their limitations to users:

One common pattern we found was a degree of inhibition among those who write the commentary in statistical releases (which are in effect statistical press notices) that accompanies the publication of official statistics. Government statisticians are acutely aware of the political implications of their work and are concerned to maintain a hard-won reputation for impartiality. The pressure from the Authority and others to include in statistical releases advice about the main messages from the statistics, and advice about the uses of the statistics and their strengths and limitations, may seem to some statisticians to risk exposing them to the charge of making politically loaded comment. However, saying nothing about the strengths and limitations of the statistics is not necessarily politically neutral either; it may also lead to misinterpretation by the news media and users. The Code of Practice thus takes a clear line that strengths and limitations of the statistics must always be explained clearly.26

23 Ev 46

24 Ev 39

25 Q 109 [Ms Matheson]

26 UK Statistics Authority, Monitoring Report, The Assessment of UK Official Statistics 2009-2012, August 2012


26. Witnesses suggested to us that the ONS did not properly explain the implications of statistics; sometimes the press releases were misleading and gave a “true but not fair” picture.27 Chris Giles gave us an example of an ONS press release about trade statistics: the headline of the press release highlighted the most recent increase in the UK deficit in trade in goods and services, and did not explain that, taking into account the past trend in the trade deficit, the real story behind the figures was in fact “Britain’s trade deficit [...] remained broadly stable”.28 Because the ONS did not properly explain the figures, the media coverage suggested that the story was that the trade deficit had massively increased, which was very misleading.29

27. Producers of government statistics do not always present their figures in the clearest way, sometimes going too far to create a newsworthy headline, when the true story is more nuanced. Government statistics press releases do not always give a true and fair picture of the story behind the statistics. We recommend that press officers and statistics producers work together much more closely to ensure that press releases give an accurate and meaningful picture.

28. Will Moy, Director of Full Fact, thought that the National Statistician and Heads of the Statistical Profession in individual government departments should have a higher public profile in explaining important statistics.30 In written evidence, Full Fact stated:

When it comes to government departments, it can be hard to get hold of statisticians, and conversations mediated through either press officers or freedom of information officers are less likely to be fruitful. Government Statistical Service personnel should be exempted from the requirement that officials do not speak to the media. Their names and telephone numbers should appear on statistical releases.31

29. The Statistics Authority told us that “ONS policy is to provide direct contact details for lead statisticians on all published Statistical Bulletins”.32 The number of broadcast media interviews conducted by ONS statisticians has also risen in recent years, from around 20 in 2008 and in 2009, to around 300 in 2012, with about a further 100 in that year on the census.33 Other ways in which government statisticians communicate statistics include:

Quarterly GDP 'live' broadcast briefings to accompany the publication of the preliminary release of data;

Advanced media training of statisticians who are experts in topics including the labour market, retail sales and public sector finances;

Growth in 'short story' formats and articles which are designed to be of particular use (and re-use) by journalists;

27 Q 82 [Mr Giles]

28 Q 62

29 Q 62

30 Q 60 [Mr Moy]

31 Ev 46

32 Ev 58

33 Ev 58

14

Use of BBC Radio’s General News Service to deliver multiple local radio interviews across the UK; and

Pro-active engagement with the BBC Economics and Business Unit, Sky News, ITN, C4 News and others.34

30. The improvements to be made to the presentation and explanation of statistics reach beyond the written document. A “public face” to statistics would help enhance trust in the figures and encourage their use. A fear of appearing politically biased sometimes means producers of statistics are reluctant to explain them properly. We welcome the fact that more staff in ONS were presenting their figures to the media, and by extension, the world beyond. Producers of statistics – in both the ONS and across government departments – should be bolder in ensuring that statistics are presented with a factually-accurate, but helpful explanation.

31. We recommend that the Statistics Authority take the lead across Government in coordinating the effective presentation of regularly- and occasionally-produced and key statistics in relation to high profile topics or events. The National Statistician should raise her public profile to promote statistics and their value without fear of appearing politically compromised, and go further to encourage other government statisticians to do the same.

32. We recommend that all Government Statistical Service press releases and statistical statements have named contact points of people with an in-depth understanding of the statistics in that release.

34 Ev 58


4 Statistics on demand 33. Producers of Official Statistics will generally produce simple statistical tables on request, as long as these do not cost too much time or money to produce; these are known as “ad hoc statistics”. More complex analysis or data may be provided and sometimes at a fee. ONS alone responds to around 13,000 requests each year for additional analysis or tables.35

34. There is some suggestion from users that the provision of data and tailored analysis can be problematic. In their written evidence Full Fact wrote:

Ad-hoc requests for data are dealt with inconsistently by the ONS. Sometimes staff will respond with data immediately, sometimes a request will be diverted into the FOI [Freedom of Information] process and at other times we must submit an FOI request to get any headway. From the outside, there appears to be little consistency in, or explanation of, the reasons for these actions.

...Of particular concern is that we have had topical requests turned into FOI requests, which are slow to process. It seems to us that requests which would illuminate a current topic of public debate should be prioritised.36

The Market Research Society wrote:

Ad hoc output has not been very significant for commercial users. The technical process of producing such output has had limitations, timetables can be long and uncertain, definitions used in official statistics may differ from those used elsewhere, statistical disclosure control tends to restrict detail, and poor value for money may discourage commissions.37

35. The release of statistics and data in this way is part of a wider Government agenda to be more open with the information it holds. We were told by Nick Hurd MP that there was now “more transparency and accountability around the process [of applying for and generating ad hoc statistics]” and that the Government “takes some pride [...] in trying to establish a reputation for being the most transparent Government ever”.38 In June 2012, the Government published its Open Data Command Paper, stating “we are determined that all of us can reap the benefits of transparency and data sharing in the future”.39 We have undertaken to look at the issue of open data in more detail as part of our programme of work on statistics and their use in government.40

35 Letter from Andrew Dilnot, Chair, UK Statistics Authority, to Will Moy, Director, Full Fact, 18 April 2012, published

on www.statisticsauthority.gov.uk, Reports and Correspondence

36 Ev 46

37 Ev 33

38 Q25, Q28

39 HM Government, Open Data White Paper Unleashing the Potential, June 2012

40 See www.parliament.uk/pasc under Inquiries – statistics

16 Communicating statistics: Not just true but also fair

36. Andrew Dilnot told us that “of course people want more access and more responses. The best thing would be for our website and our data release to be such that the number of ad hoc requests fell because people had direct access themselves, and more and more that is the case. More and more of our data are being published in an accessible form, in an electronically accessible form, and we want to go further down that road [...] one of the very first acts that I took was to agree that we would publish much more of the outcomes of these ad hoc requests than we used to”.41

37. Finding information which has been released under an ad hoc request is also not straightforward. The ONS website lists such information in a number of different locations: under Freedom of Information, published ad hoc data, data available on request and the publication scheme. These are not easily searchable. Only a small proportion of the 13,000 requests annually appear to be published.

38. The fact that so much unpublished data sits in Government naturally leads to a steady flow of requests for information which in turn puts the statisticians under avoidable resource pressure. Government statisticians should aim for the best practice that has been adopted in other parts of the public sector, namely to publish anything that can be put in the public domain. It would be expected that requests for additional data would fall sharply as, over time, less and less would rest hidden waiting to be requested.

39. We welcome the efforts made by the Statistics Authority to be transparent with ad hoc requests for data. However, the way in which such requests are processed and presented are still a cause for concern by users. We recommend that the Statistics Authority review the ways in which statistics and underlying data are drawn up and communicated, and that it draw up expected standards for ad hoc data which government departments should be expected to comply with when publishing statistical information, which should be in formats which meet users’ needs.

41 Q124, Q125


5 Misuse of official statistics 40. The Statistics Authority has a role in monitoring the use and abuse of official statistics, and intervening where judged necessary. Its statement of strategy has a strategy priority of “trustworthiness” and states that it

[...] will speak on matters of statistical controversy, reporting publicly to Parliament, especially where there are concerns about political involvement in the production or publication of official statistics, or about damage to the integrity of official statistics through misrepresentation. The Authority will continue to give priority to investigating any significant concerns in these areas and reporting those findings publicly to Parliament.42

Jil Matheson told us

The intervention of the Statistics Authority is very important in sending the signal within Departments that this is something that will be taken seriously and that they have a role in alerting Ministers, special advisers and senior officials that there is potentially a Code compliance issue.43

41. The Statistics Authority has not been shy about addressing some issues of significant public policy importance. For example, the Shadow Secretary of State for Health raised concerns with the Statistics Authority in November 2012 about government claims about a real-terms increase in health spending. The Statistics Authority wrote to the Secretary of State for Health in December 2012, stating that real-terms expenditure on the NHS had “changed little” and requesting that the Government clarify statements made on the matter.44

42. On a different issue, youth unemployment, the Parliamentary Private Secretary to the Minister for Employment wrote to the Statistics Authority about a claim made by the Leader of the Opposition that “only Spain has higher numbers of young unemployed than the UK”. The Statistics Authority replied stating that while this claim was factually true, “accepted statistical practice is that comparing unemployment rates between countries is preferable to comparing absolute numbers” and included a table of figures which showed that in terms of rates the UK was not high relatively.45 This and other similar correspondence is published on the Statistics Authority’s website.46

43. The evidence received for this inquiry was broadly supportive of the extent to which the Chair of the Statistics Authority had intervened in public to raise issues relating to the misuse of statistics.47 The Royal Statistical Society commented “[the Statistics Authority’s

42 UK Statistics Authority, Statement of Strategy, February 2013

43 Q137

44 Letter from Andy Burnham MP to Andrew Dilnot, 1 November 2012, and letter from Andrew Dilnot to Jeremy Hunt MP, 4 December 2012, accessible at www.statisticsauthority.gov.uk

45 Letter from Harriett Baldwin MP to Andrew Dilnot, 31 January 2013 and letter from Andrew Dilnot to Harriett Baldwin MP, 26 February 2013

46 At www.statisticsauthority.gov.uk, Reports and Correspondence

47 See for example, Demographics User Group (Ev 42), Royal Statistical Society (Ev 44) and Full Fact (Ev 46)


former Chair’s] interventions were generally effective, established the Authority as a force to be reckoned with and were clearly uncomfortable at times for those concerned. Further, the Authority appeared even handed between government, opposition and other politicians”.48

44. Where the Chair of the Statistics Authority has judged that there has been misuse of official statistics, we support his independence and his right to intervene. We are grateful to both the current and former Chairs for their role in upholding the integrity of government statistics and in therefore striving towards achieving higher levels of public trust in government statistics. It would be prudent, given the controversy of the areas in which the Authority intervenes, to reduce the scope for future misunderstandings, if the Authority set out why it chooses to intervene publicly on some issues and not on many others that are raised.

48 Ev 44


Conclusions and recommendations

Finding official statistics

1. Although the issue of effective dissemination of statistics has been thoroughly explored, including six years ago by the Statistics Commission, progress has been slow. The ONS website and its relaunch in 2011 is a disappointment but we welcome the acknowledgment by the ONS of the problem. We note that they have undertaken to make substantial improvements to the website. Further improvements should be made as soon as possible to make the website accessible to ordinary users. We recommend that the ONS report progress to us and publish the report on the ONS website. We recommend the ONS also publish its plans for future improvements on its website. We recommend the ONS systematically seek and publish the views of users in order to inform further improvements to the functionality and presentation of official statistics on the ONS website. (Paragraph 10)

2. There are many places in which official statistics are published; this is confusing to both the regular and the occasional user. The relationship between data.gov.uk, departmental websites and the ONS website is not clear. We recommend that the National Statistician review, update and adopt the principles set out by the Statistics Commission in 2007 and urgently take a greater role in sign-posting users to different groups of statistics. The Statistics Authority should publish information on how data.gov.uk relates to other websites showing government statistics. (Paragraph 11)

Presenting statistics

3. We welcome the Statistics Authority’s programme of work to improve the communication of statistics across government. In particular, we welcome the creation of a public policy division in ONS. We recommend that the Statistics Authority publish information on the work of this team. (Paragraph 16)

4. We are pleased to note that the Statistics Users Forum has been working with the National Statistician’s Office to develop best practice guidelines for user engagement, although these guidelines are not yet published and so we cannot comment on their content. We recommend that ONS disseminate and promote the best practice guidelines, as soon as possible, throughout Government. We recommend that the Statistics Authority and ONS, together with government departments, work much more closely with different kinds of users of statistics in order to present statistics in ways which meet their different needs. (Paragraph 17)

5. The ways in which statistics are presented sometimes present a challenge even for expert users. The lay user is left confused and disengaged. We recommend that the Statistics Authority work proactively to bring together and clearly present key statistics, from various sources, with associated commentary and in printable format, around common themes or events, such as elections and referendums, as well as broader topics such as the labour market, economic trends and so on. This is


especially important given the ending of hard copy compendia on such topics. (Paragraph 20)

6. We recommend that the Statistics Authority continue to explore more creative ways of communicating statistics, for example, through interactive guides. This should be in addition to the publication of more raw data in machine-readable format for experts who want the full results, not just the edited highlights presented in releases for the mass audience. (Paragraph 21)

Explaining statistics

7. Producers of government statistics do not always present their figures in the clearest way, sometimes going too far to create a newsworthy headline, when the true story is more nuanced. Government statistics press releases do not always give a true and fair picture of the story behind the statistics. We recommend that press officers and statistics producers work together much more closely to ensure that press releases give an accurate and meaningful picture. (Paragraph 27)

8. The improvements to be made to the presentation and explanation of statistics reach beyond the written document. A “public face” to statistics would help enhance trust in the figures and encourage their use. A fear of appearing politically biased sometimes means producers of statistics are reluctant to explain them properly. We welcome the fact that more staff in ONS were presenting their figures to the media, and by extension, the world beyond. Producers of statistics – in both the ONS and across government departments – should be bolder in ensuring that statistics are presented with a factually-accurate, but helpful explanation. (Paragraph 30)

9. We recommend that the Statistics Authority take the lead across Government in coordinating the effective presentation of regularly- and occasionally-produced and key statistics in relation to high profile topics or events. The National Statistician should raise her public profile to promote statistics and their value without fear of appearing politically compromised, and go further to encourage other government statisticians to do the same. (Paragraph 31)

10. We recommend that all Government Statistical Service press releases and statistical statements have named contact points of people with an in-depth understanding of the statistics in that release. (Paragraph 32)

Statistics on demand

11. The fact that so much unpublished data sits in Government naturally leads to a steady flow of requests for information which in turn puts the statisticians under avoidable resource pressure. Government statisticians should aim for the best practice that has been adopted in other parts of the public sector, namely to publish anything that can be put in the public domain. It would be expected that requests for additional data would fall sharply as, over time, less and less would rest hidden waiting to be requested. (Paragraph 38)


12. We welcome the efforts made by the Statistics Authority to be transparent with ad hoc requests for data. However, the way in which such requests are processed and presented are still a cause for concern by users. We recommend that the Statistics Authority review the ways in which statistics and underlying data are drawn up and communicated, and that it draw up expected standards for ad hoc data which government departments should be expected to comply with when publishing statistical information, which should be in formats which meet users’ needs. (Paragraph 39)

Misuse of official statistics

13. Where the Chair of the Statistics Authority has judged that there has been misuse of official statistics, we support his independence and his right to intervene. We are grateful to both the current and former Chairs for their role in upholding the integrity of government statistics and in therefore striving towards achieving higher levels of public trust in government statistics. It would be prudent, given the controversy of the areas in which the Authority intervenes, to reduce the scope for future misunderstandings, if the Authority set out why it chooses to intervene publicly on some issues and not on many others that are raised. (Paragraph 44)


Formal Minutes

Tuesday 21 May 2013

Members present:

Mr Bernard Jenkin, in the Chair

Paul Flynn Robert Halfon Kelvin Hopkins

Greg MulhollandMr Steve Reed Lindsay Roy

Draft Report (Communicating statistics: not only true but also fair), proposed by the Chair, brought up and read.

Ordered, That the draft Report be read a second time, paragraph by paragraph.

Paragraphs 1 to 44 read and agreed to.

Summary agreed to.

Resolved, That the Report be the First Report of the Committee to the House.

Ordered, That the Chair make the Report to the House.

Ordered, That embargoed copies of the Report be made available, in accordance with the provisions of Standing Order No. 134.

Written evidence, ordered to be reported for publishing on 23 October 2012, 8 January and 24 April, was ordered to be reported to the House for printing with the Report.

[Adjourned till Tuesday 4 June at 9.15 am


Witnesses

Tuesday 11 December 2012 Page

Nick Hurd MP, Minister for Civil Society, Cabinet Office Ev 1

Will Moy, Director, Full Fact, Michael Blastland, freelance journalist and Chris Giles, Economics Editor, Financial Times Ev 7

Wednesday 12 December 2012

Andrew Dilnot CBE, Chair, UK Statistics Authority and Jil Matheson, National Statistician Ev 18

List of printed written evidence

1 Market Research Society (4STATS 01) Ev 33

2 Tom King (4STATS 02) Ev 36

3 Social Research Association (4STATS 03) Ev 38

4 Statistics User Forum (4STATS 04) Ev 39

5 Demographics User Group (4STATS 05) Ev 42

6 Royal Statistical Society (4STATS 06) Ev 44

7 Full Fact (4STATS 07) Ev 46

8 UK Statistics Authority (4STATS 08) Ev 50

9 Cabinet Office (4STATS 09) Ev 57

10 Supplementary evidence submitted by UK Statistics Authority (4STATS 10) Ev 58


List of Reports from the Committee during the current Parliament

The reference number of the Government’s response to each Report is printed in brackets after the HC printing number.

Session 2013-14

First Special Report Public Trust in Government Statistics: A review of the operation of the Statistics and Registration Service Act 2007: Government and UK Statistics Authority Responses to the Committee’s Ninth Report of Session2012-13

HC 77

Session 2012-13

First Special Report Public Appointments: regulation, recruitment and pay: Government Response to the Committee’s Fourteenth Report of Session 2010-12

HC 18

Second Special Report Leadership of change: new arrangements for the roles of the Head of the Civil Service and the Cabinet Secretary: Further Report: Government Response to the Committee’s Twenty Third Report of Session 2010-12

HC 313

Third Special Report Strategic thinking in Government: without National Strategy, can viable Government strategy emerge? Government Response to the Committee’s Twenty Fourth Report of Session 2010-12

HC 573

Fourth Special Report The Role of the Cabinet Secretary and the Resignation of the Chief Whip: Government Response to the Committee’s Eighth Report of Session 2012-13

HC 968

Fifth Special Report The Prime Minister’s Adviser on Ministers’ Interests: independent or not? Government Response to the Committee's Twenty Second Report of Session 2010-12

HC 976

First Report The Big Society: Further Report with the Government Response to the Committee’s Seventeenth Report of Session 2010-12

HC 98

Second Report The Honours System HC 19

Third Report Business Appointment Rules HC 404

Fourth Report Appointment of the Chair of the Charity Commission HC 315-I

Fifth Report End of term report: 2011-12 HC 316

Sixth Report Special advisers in the thick of it HC 134

Seventh Report The Honours System: Further Report with the Government Response to the Committee’s Second Report of Session 2012-13

HC 728

Eighth Report The Role of the Cabinet Secretary and the Resignation of the Chief Whip

HC 864 (HC 968)

Ninth Report Public Trust in Government Statistics, A review of the HC 406


operation of the Statistics and Registration Service Act 2007

Session 2010-12

First Report Who does UK National Strategy? HC 435 (HC 713)

Second Report Government Responses to the Committee’s Eighth and Ninth Reports of Session 2009-10: Goats and Tsars: Ministerial and other appointments from outside Parliament and Too Many Ministers?

HC 150

Third Report Equitable Life HC 485 (Cm 7960)

Fourth Report Pre-appointment hearing for the dual post of First Civil Service Commissioner and Commissioner for Public Appointments

HC 601

Fifth Report Smaller Government: Shrinking the Quango State HC 537 (Cm 8044)

Sixth Report Who Does UK National Strategy? Further Report with the Government Response to the Committee’s First Report of Session 2010-11

HC 713

Seventh Report Smaller Government: What do Ministers do? HC 530 (HC 1540)

Eighth Report Cabinet Manual HC 900 (HC 1127, Cm 8213)

First Special Report Cabinet Manual: Government Interim Response to the Committee’s Eighth Report of Session 2010-12

HC 1127

Ninth Report Pre-appointment hearing for the post of Parliamentary and Health Service Ombudsman

HC 1220-I

Tenth Report Remuneration of the Parliamentary and Health Service Ombudsman

HC 1350

Eleventh Report Good Governance and Civil Service Reform: ‘End of Term’ report on Whitehall plans for structural reform

HC 901 (HC 1746)

Twelfth Report Government and IT — “a recipe for rip-offs”: time for a new approach

HC 715-I (HC 1724)

Thirteenth Report Change in Government: the agenda for leadership HC 714 (HC 1746)

Fourteenth Report Public Appointments: regulation, recruitment and pay

HC 1389

Fifteenth Report Smaller Government: What do Ministers do? Further Report with the Government Response to the Committee’s Seventh Report of Session 2010-12

HC 1540 (HC 1746)

Sixteenth Report Appointment of the Chair of the UK Statistics Authority

HC 910

Seventeenth Report The Big Society HC 902

Eighteenth Report Change in Government: the agenda for leadership: Further Report, with the Government Responses to the Committee’s Eleventh, Thirteenth and Fifteenth Reports of Session 2010-12

HC 1746

Nineteenth Report Leadership of change: new arrangements for the roles of the Head of the Civil Service and the Cabinet Secretary

HC 1582

Twentieth Report Government and IT-“a recipe for rip-offs”: time for a new approach: Further Report, with the Government response to the Committee’s Twelfth Report of

HC 1724


Session 2010-12

Twenty First Report Future oversight of administrative justice: the proposed abolition of the Administrative Justice and Tribunals Council

HC 1621

Twenty Second Report The Prime Minister’s adviser on Ministers’ interests: independent or not?

HC 1761

Twenty Third Report Leadership of change: new arrangements for the roles of the Head of the Civil Service and the Cabinet Secretary, Further Report, with the Government Response to the Committee’s Nineteenth Report of Session 2010-12

HC 1914

Twenty Fourth Report Strategic thinking in Government: without National Strategy, can viable Government strategy emerge?

HC 1625

Public Administration Committee: Evidence Ev 1

Oral evidenceTaken before the Public Administration Committee

on Tuesday 11 December 2012

Members present:

Mr Bernard Jenkin (Chair)

Charlie ElphickePaul FlynnRobert Halfon

________________

Examination of Witness

Witness: Nick Hurd MP, Minister for Civil Society, gave evidence.

Q1 Chair: Welcome to this session about publicaccess to statistics and the operation of the Statisticsand Registration Service Act. Could I ask you toidentify yourself for the record, please?Nick Hurd: Good morning, Chairman. I am NickHurd, Minister for Civil Society.

Q2 Chair: Could I start by asking what effect theGovernment feels the Statistics and RegistrationService Act has had on the quality and integrity ofstatistics?Nick Hurd: My view, Chairman, is that the legislationhas helped us reach a much better place in terms ofthe integrity of the way in which statistics areproduced and communicated in this country. Weclearly had a problem; over the last five or six yearswe have reached a much better place. The Act hasbeen fundamental to that.

Q3 Chair: What do you think is the effect of limitingthe UK Statistics Authority’s work to what we callofficial statistics? It has been suggested that thedefinition of official statistics is not clear enough inthe legislation. How would you like to see the termofficial statistics defined?Nick Hurd: To be honest, Chairman, no one has cometo me with representations that we have a problem inthat respect.Chair: They will now.Nick Hurd: In which case I am all ears, but up untilnow no one has come to me, as I said, withrepresentations that we have a problem there.

Q4 Chair: The problem is that quite a lot ofpublished information is designated as administrativedata or research, and therefore is used in the publicdomain in the same way as official statistics but hasnot been certified as official statistics. When theGovernment says something in numbers, the publicdoes not make the distinction.Nick Hurd: I understand that point and I come backto my starting point, which is the importance ofimproving the integrity of the system, where weclearly had a problem. I think we are in a much betterplace. We are very clear about wanting to preservethe integrity of statistics. As you know, Chairman, theNational Statistician polices the quality of statisticsand the way in which they are published, and is

Greg MulhollandPriti PatelLindsay Roy

supported in doing so by the heads of profession ineach Government department. That is the situation. Ifthis Committee and others feel there is a problem thatwe need to revisit, the Cabinet Office is all ears.

Q5 Lindsay Roy: Good morning. Can you explainthe rationale behind the distinction between officialstatistics and categories of information described asadministrative data, management information andresearch? Maybe specifically, why was one onclaiming benefits released as a research note?Nick Hurd: We do have a hierarchy of statistics.National statistics is a subset of official statistics. Asyou know, Lindsay, the designation “nationalstatistics” indicates the statistics have been certifiedby the Statistics Authority. We have three types ofnational statistics. I am sure the Committee is awareof that. It is our view that the Statistics Authorityshould be able to award accreditation independentlywhere it considers statistics to be robust and reliable.In terms of the statistics that are published bydepartments as administrative, management orresearch, I come back to the point I made to yourchairman: it is for the National Statistician to policethe quality of statistics and the way in which they arepublished, and she is supported in that role by headsof profession in each Government department. As Isaid, I have not been presented with any evidence yetof a problem in that hierarchy in that system.

Q6 Lindsay Roy: What are the criteria for somethingto be designated as official statistics?Nick Hurd: That is all set out in the legislation. As Isaid, the legislation has been reviewed. I do not thinkthat process threw up a great debate about this, butthis is an evolutionary process. If we have to reviewit, we will, but I think the definitions are all set out inthe legislation.

Q7 Greg Mulholland: One problem we have beentold about by a number of those who have givenwritten evidence is that there is a lack ofunderstanding of what “national statistics”, asdesignated, means. One view from a Governmentstatistician was, “The label of ‘national statistics’ isan inappropriate choice of words to describe officialstatistics that have been assessed as compliant withthe Code. Few, if any, members of the general public

Ev 2 Public Administration Committee: Evidence

11 December 2012 Nick Hurd MP

are aware of the designation. Few, if any, members ofthe media ever use the expression. The label is alsomisleading insofar as it is often interpreted to meanonly those statistics that have a national coverage. The‘national statistics’ label should be withdrawn.”Do you accept that there is a lack of understanding?Would you be open to a change in the way that theseclearly key statistics are described?Nick Hurd: Yes. We are open to discussion, if thereis evidence of confusion and that it is underminingpublic trust in the integrity of the system. Since thatis our starting point, of course I would be more thanhappy to revisit that. I am very happy to support theStatistics Authority’s efforts to broaden understandingof its processes and what national statistics and theCode of Practice for Official Statistics mean. It isquite clear that we have a long way to go in terms ofwhat you might call outreach, so that more people areaware of the definitions and the importance of theCode.

Q8 Greg Mulholland: As someone who used towork in marketing, what awareness is there of thenational statistics kitemark? Does that work? Is it akitemark or brand that people are familiar with? If itis not, obviously it is not working.Nick Hurd: I do not honestly know. It is notsomething my constituents are talking about a greatdeal, but maybe I would not necessarily expect that. Iwould expect the Statistics Authority to have a viewon that.

Q9 Greg Mulholland: With respect, that is not thepoint, is it? The point is that taxpayers’ money hasbeen spent on developing this with the belief that it isimportant in the job it does. Has there been anassessment of whether our taxpayers’ money waswell spent?Nick Hurd: Not that I am aware of. That does notmean that it has not happened.Greg Mulholland: Do you think it might be sensibleto see if this was a worthwhile exercise?Nick Hurd: It is certainly something I will be askingabout after this session. I am not aware of anyassessment of the value for taxpayers’ money in termsof that investment, but it is certainly I will ask aboutafter this session.

Q10 Robert Halfon: It has been suggested that thedata-sharing powers in the Act are too restrictive.Would you like to see changes?Nick Hurd: As the Minister who has taken through alot of the SIs that enable this, I am well aware that wehave a rather cumbersome process for doing that. Ihave to say that there are no current plans to changethe system, but it is one that will probably have tocome under review—not least as there are clearlygrowing demands for data sharing. It will becomerelevant in the context of the future of the census aswell.

Q11 Robert Halfon: The Royal Statistical Societybelieves there should be wider powers in order toobtain official data. What is your view on that?

Nick Hurd: My view is what I just said: there isclearly growing pressure for more data sharing. It is asensitive area. It is something we have to look atcarefully. We have a system at the moment that isperhaps not perfect but seems adequate. At themoment, that is what we are sticking with.

Q12 Robert Halfon: In the recent spat between theGovernment and the Statistics Authority about theNHS statistics, the Government said one thing and theStatistics Authority said another. Do the StatisticsAuthority come to you first and say, “We will releasethese statistics questioning the Government’sargument,” or do they just release it and say, “TheGovernment have got it wrong”? Do they give you achance to rebut what they have said before theypublish?Nick Hurd: They certainly in that case did not cometo me personally in that respect.

Q13 Robert Halfon: Do you think that if theStatistics Authority does publish statistics questioningGovernment figures, they should come to theGovernment first and give the Government a chanceto rebut it?Chair: They do that.Nick Hurd: We are getting into the pre-releaseaccess situation.Robert Halfon: But you are saying that for the NHSthing they did not.Chair: They did, actually.Nick Hurd: They did not come to me personally; thatis what I said.

Q14 Robert Halfon: I see. How much notice didthey give you? When did it happen?Nick Hurd: I imagine it is subject to the pre-releaseaccess requirements.Chair: To correct you, it is nothing to do withpre-release access. Before Andrew Dilnot writes anofficial letter to the Government, there is an exchangebetween the Minister and the UK Statistics Authorityto check there has been no misunderstanding. In thecase of the NHS statistics, the Government actuallysaw to it that the necessary information was altered inline with Andrew Dilnot’s request.Nick Hurd: I do stand corrected, Chairman. I havenow got the correspondence in front of me, and thatis exactly how it works.

Q15 Robert Halfon: Are you still content with thecurrent arrangements for pre-release access tostatistics? Do you think they provide the right balanceof informing debate and also ensuring that Ministersare properly briefed?Nick Hurd: Yes, you are right. There must be the rightbalance. We are in an infinitely better place than wewere five or six years ago. We have come from anunsustainable place to a situation that continues tofrustrate some people but that the Government feelsstrikes the right balance, in terms of the publicinterest, between seeking not to undermine thepublic’s trust in the process and the need to satisfy thepublic’s reasonable desire for proper accountability,proper explanation and proper response. A system of



up to 24 hours with much tighter control of who hasaccess to data and much greater transparency aroundthat process is clearly a much better place than wewere in five or six years ago.

Q16 Robert Halfon: Is it right that the argumentfrom the Treasury, for example, is that they need tosee a pre-release with further advance notice becauseof problems with the markets? If the information hasan effect on the markets, the Government needs tohave time to react accordingly.Nick Hurd: Some data is more sensitive than others,but I think the general principle is that whenparticularly sensitive data is released, Parliament, themedia and the public have a quite justifiable demandfor proper accountability, proper explanation and,where appropriate, a proper response. There willalways be noise expressing frustration, but I thinkthere is a broad consensus that it is sensible to have ashort window of opportunity for the peopleresponsible for those statistics to shape a response thatmeets that requirement for proper accountability.

Q17 Robert Halfon: Should that pre-release be madeequally to the opposition parties, so as to give them afair chance to prepare and examine?Nick Hurd: As it is, we have a system that determineshow much time is allowed and an access list, as itwere, that is transparent and of which the StatisticsAuthority is notified. We have quite a transparentprocess in that respect. On balance, this is reviewedquite regularly. There has been a conversation veryrecently between the Minister for the Cabinet Officeand Andrew Dilnot on this. The position of theGovernment is to stick with the current situation.

Q18 Robert Halfon: You do not think it should begiven to the opposition at the same time. If statisticsare given to the Government, should they be given tothe main opposition party at the same time?Nick Hurd: My position is that we are sticking withthe current arrangements.

Q19 Chair: Of course, other countries haveabolished pre-release altogether. That was originallythe position of the Conservative Party when we werein opposition.Nick Hurd: Chair, as you know there is quite a broadspectrum of positions. Scotland and Wales did notfollow us in terms of the big move down from up tofive days to up to 24 hours. The United States have ashorter period of time than us, but we are much closerto the United States than we are to Scotland and Walesin this context. There are balances to be struck. Onbalance, we think the current situation is sustainable.

Q20 Chair: Recently the Prime Minister adverted toforthcoming good news. There was speculation thathe already knew about positive GDP statistics as aresult of pre-release that had not been released intothe public domain. He received a letter from the UKStatistics Authority for abusing the privilege ofpre-release. Does this improve confidence in publicstatistics and how they are used by politicians?

Nick Hurd: Just to be clear, in terms of thecorrespondence between Andrew Dilnot and thePrime Minister, the letter from Andrew Dilnot wasslightly more balanced than you suggest. AndrewDilnot made that clear when he said, “Although it maynot have been your intent, your remarks were widelyinterpreted as providing an indication about the GDPfigures.”Chair: I say no more than that.Nick Hurd: As you will see from the response fromthe Prime Minister, he was extremely robust inpointing out that his comments were in a widercontext and were not intended to give any indicationabout the GDP statistics.

Q21 Chair: I appreciate that, but because he mightalready have had that information, it is open to thatmisinterpretation. If pre-release were abolished, suchmisinterpretations could not arise and publicconfidence in statistics would be stronger as a result,which is what we were saying when we were a partyin opposition.Nick Hurd: I think we are speculating, because I donot think there has been any public confirmation ofthat.

Q22 Chair: It is the Treasury that is objecting to this,is it not? We know the view of the Cabinet Office isto abolish pre-release, but there is resistance in otherquarters of Government.Nick Hurd: No, I think you are speculating there aswell, Chairman. The position of the Government is tostick with the current arrangements, and that is whathas been communicated to Mr Dilnot.

Q23 Chair: When do you think we will next makeprogress on this issue?Nick Hurd: It is kept under review, but I do not detectany political will to change the system at the moment.

Q24 Paul Flynn: I think I am the only member ofthe present Committee who took part in the debatewhen the Bill went through. There were not manyareas where there was disagreement between the twosides. In this House and in the House of Lords, therewere impassioned pleas from the Conservatives to getrid of pre-release altogether. Is it the case that changesof Government mean changes of scripts, and somethings that you believe passionately in opposition youno longer find palatable in Government?Nick Hurd: Processes and positions evolve, Mr Flynn.You know that as well as I do; you have been inpolitics longer than I have.Paul Flynn: They do not evolve. They deteriorate—they degrade. That is the word.Nick Hurd: I think evolve is the polite term.Paul Flynn: You have taken the line of a denial ofinformation, and you have taken the opportunity todistort the information to take advantage, which youthought was inappropriate when you were inopposition. What happened to you? You used to beagainst lobbying as well. What has happened?Chair: Steady, order. Stick to statistics.Nick Hurd: Mr Flynn, you are drawing me off-piste.



Paul Flynn: Is there some sort of lobotomy thatMinisters have when they get into Government, whenthey have been sensible people in opposition?Nick Hurd: I see no sign of any lobotomies. I comeback to my serious point. We had a bad system before.As a result of that legislation, for which there wascross-party support, we are in a much better place interms of pre-release access. Clearly, we will notsatisfy everybody on this issue. At the moment, I donot detect any political will to change the currentarrangements.Chair: We are probably satisfying nobody at themoment, because the abolitionists are not satisfied butthe people who want more pre-release are upset. MrFlynn, might we move on to the question of ad hocstatistics.

Q25 Paul Flynn: We have had evidence that peoplebelieve the release of ad hoc statistics is very patchyand inconsistent. Full Fact has said to us that ad hocrequests for data are dealt with inconsistently.Sometimes staff will respond with data immediately;at other times they will refer the request to a Freedomof Information department themselves or ask theapplicants to do it for them. Is there any desire in theDepartment to have a policy that is more consistentand readily understood in relation to the use of adhoc statistics?Nick Hurd: I would say two things. If the process ofapplying for and generating ad hoc statistics iscluttering the process or causing difficulty for theStatistics Authority, our door is open to a discussionon improving the process. I think a very usefuladdition and an improvement has been around thetransparency of that process, so all requests for ad hocinformation are now public. That transparency is afriend of efficiency in this respect. I think there ismore transparency and accountability around theprocess, which I hope the Committee would welcome.

Q26 Paul Flynn: Do you believe that the requestsare frivolous, unnecessary or excessive?Nick Hurd: Is that in relation to the whole generalityof the number of requests? I cannot answer that.

Q27 Paul Flynn: One of the other issues raised wasthat some of the statistics that are produced in thisway—in a fairly inadequate way—are thenrepublished by the people who applied for them.There is a feeling that this general publication mighthave results that are misleading, rather than comingwith the authority of the ONS and other bodies.Nick Hurd: I would be concerned if that were true,but nobody has come to me with any representationson that.Paul Flynn: You seem to be short of representations.Nick Hurd: I am.Paul Flynn: Are you on speaking terms with yourstaff?Nick Hurd: I am. However, people have not beenbeating a path to my door—nor, I believe, to that ofthe Minister for the Cabinet Office—saying that thissystem is particularly broken. This Committee mayunleash that, but to date, I have to be quite honestwith the Committee and say that—

Q28 Paul Flynn: It is refreshing to hear yourofficials are giving you such a quiet life and notbringing this information to your attention. Regardingyour general view on the accessibility of statistics,what has happened to social trends? This used to be aregular page-turner that we would examine as MPs,burning the midnight oil and going through everypage. There are other data that you have thatapparently you have been retentive about on youthunemployment, for instance. We are told there isinformation there that is not readily published. Is it afair criticism that you have gone backwards in theaccessibility of information?Nick Hurd: No, I think the whole direction of travelis completely the opposite, Mr Flynn. This is aGovernment that takes some pride and has certainlyraised the bar in trying to establish a reputation forbeing the most transparent Government ever.Paul Flynn: I think you have raised the bar in sayingthat you are the most transparent Government ever.Nick Hurd: No, I think that is unfair. These things donot happen overnight, but I think the public arealready beginning to get a sense that there is muchmore information out there in terms of what is beingdone in their name, particularly in the area of howtaxpayers’ money is being spent. It is a journey, but Ithink we have a signalled ambition and intent that wasnot there before under previous administrations.Government is becoming more open. In that process,will we satisfy everyone? No. Will we frustrate somepeople? Yes.

Q29 Paul Flynn: It appears to be a journeybackwards into the past. Is information on socialtrends still going to be published on a regular basis?Nick Hurd: I cannot comment on social trends.Paul Flynn: Could you start to open a dialogue withyour officials and find out what is going in yourDepartment? It might help us as a Committee. Youhave given several answers to questions where youhave said you have no knowledge on the matter. Thesewere issues we were likely to bring up.Nick Hurd: It is not a question of not havingknowledge. I am trying to suggest that the CabinetOffice and I do not have a strong sense that there isanything particularly broken in terms of the system.You mentioned social trends; I do not happen to knowthe future of that, but my main point, Mr Flynn, whichI stick by, is that this Government set its stall out tobe very open and transparent and is actuallyrecognised on the world stage in terms of open datato be world leading. It is working in partnership withother governments to completely change the cultureof government across the world in terms of opennessand transparency. Are we getting everything right?No, but that is clearly the direction of travel and thatis recognised by more objective, politically neutralfigures, to be the case. As to the fate and future ofsocial trends, I will have to come back to you if thatis of particular concern to you.

Q30 Chair: Mr Flynn, can I interject for a second?By way of a public advertisement, can I say that ifthere is anyone listening to this session who feels theyhave made representations to the Cabinet Office about



these matters, could they draw this to our attention aswell as the Minister’s? I think there is some disquietabout these matters and we are a little bit surprisedthat these representations have not reached you.Nick Hurd: Let me just, if I may, throw a little claritydown there. Of course, there has been a review of thelegislation. In terms of the pre-release access, therewas a special look at it again in 2010. There haveobviously been opportunities for people who areinterested in this to make their views known throughsome formal processes. What I was trying to say isthat in terms of direct representations to me, I havenot received any in this space. Just for the sake ofclarity, that is the point I am making—no more, noless.

Q31 Paul Flynn: You may have observed, Mr Hurd,that since you started to give evidence, threeConservative members of this Committee have left theroom, whether through embarrassment or guilt or not.Chair: Order. Come on.Paul Flynn: While we are still quorate, can I speedilypoint to your attention to the fact that the StatisticsUser Forum and the Social Research Association havemade similar pleas to us? Briefly, the Statistics UserForum said, “All too often it is extremely difficulteven for the expert user to find the statistics they needfrom the Office for National Statistics anddepartmental websites. Search engines leave much tobe desired. Most users rely on Google.” Thechangeover to the new website was generouslydescribed as a disaster—in fact it was probably worsethan that—but the journey you are taking is onebackwards into a denial-of-information policy.Nick Hurd: No, I reject that completely. If you arereferring to the Directgov replacement, I think the newwebsite is a major step forward in terms of theinformation we give the citizens we serve in terms ofwhat is being done in their name.Paul Flynn: I think you are alone in thinking that.Nick Hurd: There has been a transformation.

Q32 Chair: Minister, what dialogue have you hadwith statistics users about this yourself?Nick Hurd: Do you mean me personally, directly?Very little.Chair: Can we suggest that you have some?Nick Hurd: I am certainly happy to receive thatsuggestion. As I said to you before, there have beenchannels for people to make their views knownthrough the reviews that have taken place of thelegislation. Those have been absorbed by the system.

Q33 Chair: I see in my mind’s eye a very usefulCabinet Office seminar on the availability of statistics,their users and how happy everybody is—a bit ofmarket research. I think it really would be verypositive.Nick Hurd: Yes, but please do not underestimate whatis going on in terms of the open data agenda, whichis being driven by the Cabinet Office, such as thecreation of the Open Data Institute. There has beenwidespread engagement with people who look atstatistics and data. There has been widespreadengagement with people who want to create

businesses out of that. This is a very serious agendafor the Government, but if you ask me directly, interms of personal representations made to me, I willbe quite honest with the Committee: I have notreceived any. There is, however, a massive agenda forthe Department in terms of open data to improveaccountability and to support enterprise.Chair: Mr Flynn, are you finished?Paul Flynn: I do not want to be offensive to theMinister, so I shall not ask any further questions. Iwill just say that your performance, I believe, has beena very sad one today because you have batted backmost of the questions.Chair: You said you would not be offensive.Paul Flynn: Well, I am, but I am being nice. I thinkyour performance has been disappointing, and I canunderstand the reason why three Tory members haveleft the Committee.

Q34 Chair: Moving on, there is an awful lot of extradata coming into the public domain, which is verywelcome, but how well explained do you think thisdata is? Is there not a need to explain it as well as justpublish it? What improvements do you think need tobe made in order to achieve that?Nick Hurd: I think that, looking across the widerpiece, Chairman, at the open data agenda, there is avery strong recognition that you cannot just dump thedata. You need to put it out in a way that people canuse and share. If the Committee wants to look at thework of the Open Data Institute and what we have setup in terms of the architecture to nudge and push thesystem to get more data out there, you will see thatthere is a very real drive to make sure the data is asaccessible and usable as possible. Despite what MrFlynn says, we are at the start of a journey here. Weare learning, but there is a very open process ofengagement with the very many people that care aboutthis agenda a lot and who keep our feet to the fire.

Q35 Chair: Can I caution you about something thatParliament certainly will not want? Parliament willnot want vast quantities of information, which wecannot possibly cope with because we have verylimited resources compared with Government. Eventhe US Congress is complaining about the volume ofdata that is being published by the administration inthe United States. Even the Congress, with all of theirstaff, feels they cannot cope with it. Do youunderstand this danger of data overload with too littleinterpretation and conversation about what the dataactually means?Nick Hurd: In a way, that comes back to ourconversation about pre-release access, which is theneed for information to be presented in some context.There are real tensions here, Chairman, but I thinkwhat I would say—partly in response to the questionfrom Robert Halfon—is that the genie is coming outof the bottle here. It is not just Parliament; it is thepublic; it is civil society. They are also driving thedemand for information and data. The capacity ofParliament to absorb this is one thing, but we shouldnot underestimate the growing demand from thepublic and civil society.



Q36 Chair: The Royal Statistical Society has said,“A deeper communication challenge for the officialstatistics service is to present a coherent statisticalpicture of what is going on in areas where debateneeds to concentrate on the issues rather than onexplaining particular statistics. The debate on Scottishindependence is an example of where statistics needsto be brought together and well communicated inorder to foster good debate.”The written evidence that we are getting does suggestthat even official statistics are not being presentedclearly enough—particularly for the non-technicalreader. Do you think official statistics are presentedclearly enough? This is an indirect Governmentresponsibility, because the official statistics areproduced by the National Statistics staff. It is aquestion of resourcing this service effectively so thatit can do its job.Nick Hurd: I doubt we are in a perfect world in termsof how statistics are communicated. I come back,however, to what I said at the start. The starting pointis about wanting the public to feel trust in the systemand the integrity of the system. That is our startingpoint. The Government is committed to working withthe independent statistics authorities to makeimprovements where improvements are needed. Thedoor is open for that conversation.

Q37 Chair: Do you think we do enough to ensurethat for big political events—like referendums, forexample—there is enough preparation of the statisticalbasis for the arguments that are going to be had? It ispart of our inquiry into statistics to ask whether theScottish referendum will be properly informed. Whatis the Government doing to ensure that there is aproper statistical base? Otherwise, the referendummight descend into an argument about what thenumbers mean or whether the numbers are accurate orwhether the numbers are relevant. There could be ashared understanding of what the statistics mean anda debate about that, rather than a spat about whetherthey are lies or damned lies.Nick Hurd: I take the point in terms of the importanceof it. I do not think the Government’s premise is thatwe have an improper system of statistics at themoment.

Q38 Chair: Do you not think there is a case for youmaking special efforts to ensure that there is a properstatistical basis for a very important referendum aboutthe future of this country?Nick Hurd: I can certainly undertake on the part ofthe Cabinet Office to make sure those conversationstake place and we are satisfied. As I said, ourpresumption today is that we do not preside over animproper system.Chair: I am sure that is the case, but I am suggestingthat maybe you should be proactive in that regard. Wehave galloped through a number of topics in shortorder. I do not know whether there are any otherquestions that colleagues wish to ask the Ministerbefore he leaves.

Q39 Robert Halfon: Can I just come in onsomething very briefly? Touching on what the

Chairman just said, if you look at the recentconsultation regarding gay marriage, where there were500,000 or 600,000 responses against and—whateverit was—70,000 in favour, do you think there is a rolefor the Government or the Statistics Authority to lookat the responses that come in on major consultationslike that and assess them statistically, rather than justqualitatively?Nick Hurd: I am very happy to have a conversationwith the Statistics Authority about that. That is awider issue about the value of public consultation. Ithink that is a bigger issue than the statistical approachto them, but I am certainly happy to have aconversation with Andrew Dilnot to get his view onthat.

Q40 Lindsay Roy: To pick up on a point theChairman made about the referendum in Scotland, doyou think there is more work needed to clarify theBarnett formula?Nick Hurd: That is a longstanding issue, which willbe decided way above my pay grade.

Q41 Chair: If you are the Minister for Statistics—Iam sorry to press you on this—are you not the onewho is meant to understand this? Are there not someissues that you need to get a grip of in order for youto be satisfied that there will be a coherent debatearound a robust statistical base? That is what we verymuch hope you will do.Nick Hurd: You have made that point and I haveresponded to it.Chair: I do not think it is above your pay grade.Nick Hurd: The debate needs a strong evidence baseunderpinning it; the Government is extremelycommitted to the integrity of our statistical system.Chair: You said it was above your pay grade.Nick Hurd: As we said right at the start, we are in amuch better place than we were five or six years ago.Are we in a perfect place? No. Is that referendumextremely important? Yes. Will we be workingtogether with all those involved to make sure that thepublic have that debate on the basis of very firmevidence that they can trust? Yes.Chair: You said it was above your pay grade. Mypoint is that I think you are a much more importantperson than you may think you are.Nick Hurd: What I was trying to say to Mr Roy was,in terms of the whole question of the Barnett formulaand how it works, that is a whole debate above my paygrade. In terms of the evidence base—the statistics onwhich that debate is had—I completely take yourpoint about making sure that we are as good as wecan be when we get into that debate.

Q42 Lindsay Roy: Have you had discussions withthe Secretary of State for Scotland on the statisticalbasis for the referendum?Nick Hurd: No, not yet.

Q43 Paul Flynn: The purpose of setting up theStatistics Authority was to counter the distrust ofGovernment statistics. Do you think trust andconfidence in Government statistics has beenimproved in the last two-and-a-half years?



Nick Hurd: I do. I expect you to disagree with me,but I remember at that time there were fierce debatesin Parliament about the Government’s approach tostatistics, which was, in part, what prompted thelegislation. I do not feel—you may disabuse me ofthis—there is quite the same degree of intensity aboutthat now. My sense is that there is a feeling that thereis more integrity and transparency in the system.There is certainly a better process in terms ofpre-release access and things that were very emotiveat the time. I feel we are in a better place; if thisCommittee feels differently, I am sure the report willmake that quite clear and we will listen.

Examination of Witnesses

Witnesses: Will Moy, Director, Full Fact, Michael Blastland, freelance journalist, and Chris Giles, EconomicsEditor, Financial Times, gave evidence.

Q44 Chair: Welcome to this second session on theusability of Government statistics and the operation ofthe Statistics and Registration Services Act. Could Iask each of you to identify yourselves for the record?Will Moy: My name is Will Moy. I am the Director ofFull Fact, an independent fact-checking organisation.Michael Blastland: I am Michael Blastland; I am afreelance journalist.Chris Giles: I am Chris Giles. I am Economics Editorof the Financial Times.

Q45 Chair: Mr Blastland, I think in our parlance youwould declare an interest in that you have workedclosely with the Chair of the UK Statistics Authority,Andrew Dilnot.Michael Blastland: I know Andrew well. I suspect Iam not alone in this room in that, but I do, yes.

Q46 Chair: It is a small world. First of all, wouldyou like to give a reaction to what you have heardfrom the Minister this morning?Will Moy: I think various elements of what he saidmay come up in particular questions. I was interestedin his remarks on ad hoc statistics. I was glad to seethe Committee pushing him on the Scottishreferendum, but I think we can probably deal withthem as they come up.Michael Blastland: My sense is that there is a prettyvigorous debate about the quality of all aspects of thestatistical system. I am surprised that the Minister isnot quite as well acquainted with it as I would expect.Chris Giles: I would share Michael’s surprise.

Q47 Chair: Do you think this is a problem withofficials not briefing the Minister properly or what?He must have been prepared for this session.Michael Blastland: I have no idea.Chris Giles: I do not believe that anyone in thestatistical community or anyone who knows people ordeals with statistics officially in Britain could beunaware of the big dissatisfaction with thecommunication of statistics in Britain at the moment.The Minister is the first person in the field I am awareof who is not aware of it. That surprised me.Chair: That is quite a strong comment to make.

Paul Flynn: It is just that we are not overwhelmedwith evidence from the statistical community—oranyone else—to suggest there has been thatimprovement.Chair: Mind you, they are a demanding lot.Nick Hurd: They are. They keep our feet to the fireand they always will.Chair: I am certainly on a learning curve with regardsto statistics, as I suspect you are too, Minister. Wemay well want you in front of us again later on in ourset of inquiries, but we do regard this as a veryimportant part of our work. Thank you very much forbeing so helpful this morning.

Will Moy: I will certainly make sure, on Full Fact’sbehalf, that he receives some of the representationsthat he appears not to have to date.

Q48 Chair: I am sure he will look forward toreceiving those. One of the reasons we are conductingthis set of inquiries is to raise these issues up theministerial in-trays, because otherwise these issuestend to get pushed down. Can I just ask each of youhow easy it is to find the official statistics that youneed? Is there a difference between the officialstatistics produced by the ONS and those produced bydepartments and other bodies?Will Moy: I think the short answer is that it is verydifficult. It can occasionally be easy; usually you haveto know your way around. However, it tends to bedifficult and frustrating. It is hard to know whichproducer to go to or whether to go to the ONS. It ishard to know whether there is more than one data setdealing with the same subject, and it is hard to knowhow those different data sets relate to one another. Ofcourse, as soon as you get to the websites, you are ina magic roundabout all of its own.Much has been said about the ONS website. I wouldsimply say you can get lost very quickly. Regardingwww.data.gov.uk, although I admire the sentiment—as a developer myself, I have been keenly followingit ever since it was in closed beta—I think it hasbecome an elephants’ graveyard where good data setsgo to die. Departmental websites are variable; none ofthem, I do not think, is an outstanding example ofhow to communicate statistics.Our official statistics producers are sitting on atreasure chest of data, which can inform andilluminate our public debate. Sometimes they arevirtually sitting on top of it and daring you to get in.What they should be doing is presenting it and saying,“We have all this information; it can help you. It canhelp you as citizens; it can help you as businesses; itcan help you make decisions about the Scottishreferendum and all kinds of topics. Here is how it canhelp you.”If you look at something like Statistics Norway, thatis what they do. If you go to their website, you willsee there is a section on gender equality, for example.


11 December 2012 Will Moy, Michael Blastland and Chris Giles

You can go and find out about that topic. These arethe kinds of things people are interested in. There arealso things like, “There are 123 people in Norwaywhose surname is Moy.” I found that out because Iwent on their website and they have that engagingfeature, where you can find out something you did notpreviously know. It is engaging; it is sharable; and itis interesting. They want you to share the treasuretrove of information they have. I do not think that iswhat we see in the outlets we have at the moment.Michael Blastland: I do not disagree with anythingthere. I used to work for a programme called More orLess on Radio 4, which deals with statistical issues inthe news. Just before I left, I had an idea that I wouldget myself a couple of very bright maths graduatesand shut them in a room and say, “See if you canfind—I do not know—a long-run series of data sincethe Second World War of per capita GDP?” That isquite an interesting little number. You pop in everycouple of days with some bread and water to makesure they are not dead—from the stultifyingexperience or something else. In fact, I wouldrecommend it to members. In the unlikely event thatyou do find yourselves with a spare weekend, setyourselves a few challenges of discovering this dataand see how far you get.Chair: We are very spoiled in this House; we havethe House of Commons Library to do it for us.Michael Blastland: It is an exasperating experience. Iregard it as a moment of failure when I have to phonesomebody. It is a terribly inefficient way of doing it,when we have this web-based system, but I do findmyself having to phone people, and I fancy that Iknow my way around the system reasonably well.I did want to bring this to people’s attention, becauseit is in many ways a rather trivial book. It is calledOlympic Britain and it is published by the House ofCommons Library. However, there are a few dataseries in here, “National flourish: the causes ofpopulation change since 1922.” That is a fascinatinglittle piece of information. Then you look at thesources, and the sources are, “BR Mitchell, BritishHistorical Statistics, 1988; ONS, Annual Abstract ofStatistics, 1938–89.” That is not one volume. It is hastaken the House of Commons goodness knows howlong to compile a very simple series of data. There isanother one on participation in higher education. It isfascinating. Again, that is a long-run data series.Again, if you look at the sources, goodness knowshow much you would have to delve through—a lot ofit not even available online—in order to put togethera long-run data series.I agree absolutely with Will: these are treasures. Youcannot have a sensible argument about where we arenow without consulting this data. We cannot consultthis data easily.Chair: Well, the House of Commons Library does usa great service.Michael Blastland: You are lucky.Chris Giles: I will just make one specific point. I donot disagree with anything the other two witnesseshave said. I thought I would conduct a testyesterday—both from an expert and a lay point ofview—in relation to a question that would be areasonable question to ask in a debate about society

in the pub at the moment: is unemployment nowhigher or lower than it was in the mid-1990s? If youwere a layperson, the first place you might go is tothe ONS website. If I may, I will just go through theprocedure of what you must do to answer thatquestion.If you go to the homepage, you click onunemployment on the right, which you might thinkwould be the obvious thing to click on. It is not there;you do not get anything useful there to answer thatquestion, but on that page you do see something calledlabour market statistics. You think, “That might be agood place to go.” You can click on that and, again,you will not find the information. There is a pdfdocument of the latest unemployment statistics there,which, if you are still going, you might go into. Youwill get unemployment there, but only back to 2007. Ifyou then go back to where you were, there is anotherunemployment tab on the left. You actually get a chartthen but, again, only back to 2007. I could go throughand give you another five or six steps before youwould certainly give up. There is absolutely no wayof getting an answer to that question if you are alayperson—certainly not if you are using a tablet inthe pub.If you are an expert user—which I think I probablyam because I use the ONS website all the time—andyou actually know the four-digit codes forunemployment rate and level: the four-digit code forthe level is MGSC and for the rate it is MGSX. Youwould have thought that if you type that into thesearch box on the ONS website, you should be ableto get the series very quickly. No. It takes eight clicksbefore you actually get to the data. That, as an expertuser, is really very frustrating. It should not beanything like that.

Q49 Chair: That is a very fascinating account. Doyou share my frustration that the Budget red book,which one would expect to contain some of thislong-run data, is extraordinarily subjective in the datait contains? Should the Government not do somethingto pull together some of this key economic data atbudget time and publish it in a simple form?Chris Giles: I think the key economic data shouldall be very easily accessible in long-run form—notnecessarily back to the 19th century, but clearly backto 1948, from where we have good national statistics.You can get GDP; you should be able to getunemployment; you should be able to get inflation.You should be able to get all of these very basic thingsthat the public should need to know about in a verysimple form. The ONS should put the form that itthinks is the best form up on its website, so that if youclicked on unemployment first, that wouldimmediately be what you saw—in a long-term sense.

Q50 Chair: Why do you think the ONS does not dothis?Chris Giles: I have absolutely no idea. It seems to bethe simplest thing they could do to improve theirwebsite in very short order.Michael Blastland: It is a very simple principle in ahierarchical arrangement of data that you have asimple access point. Perhaps you could do it



graphically; you could zoom in and call in other bitsof information as your level of interest or expertiserequires. It is baffling to all of us why it has taken solong to get anywhere near that. There are a fewexamples of the use of graphics—for example,population is not bad—but by and large it has beenkicking around in the outside world now for a goodfew years and we have failed to exploit it.Will Moy: I think the answer to your broader questionis that the statistics producers—the GSS and theONS—prize technical skills above communicationskills. I think there needs to be a culture change sothat communication is esteemed as highly and is asmuch an integral part of producing statistics forsociety as the technical skills of producing them.

Q51 Chair: Finally, do you think there is anyexample in the world where this is done very well bya government?Will Moy: No.Chair: There is no government in the world that doesthis well.Will Moy: There is no government in the world thatdoes it. Google is terrible at it. www.google.com/publicdata is embarrassing for Google, I am sure.Statistics is an unsolved problem, I think, of reallyhigh quality presentation. It will be solved in the nextfew years.Chair: Are there any local authorities that are doingthis well?Will Moy: Not that I know of. There are individualgood examples. I do not mean to say that nothing isdone well, but if you are talking on the level ofpresenting a coherent picture of a country or an area,I am not aware of anything that I think fully takesadvantage of the potential.Michael Blastland: There is a key point there: it isvery difficult for outside people to do it. It is the staffin the Office for National Statistics who know thisdata better than anybody. They know the wrinkles init; they have very good conversations in the pub aboutwhat it means, but they have a tremendous reluctanceto tell anybody else sometimes.Chair: They do tell us in the House of CommonsLibrary.Michael Blastland: Perhaps that is true of the Houseof Commons Library. I am thinking of the ONS here.However, there is a reluctance to do more than justgive the numbers. It is particularly hard for externalbodies—who do not have that acquaintance with thedata—to be able to give it the sort of interpretation itneeds to provide an effective service. I agree withWill: by and large, these things do not exist.

Q52 Robert Halfon: In terms of your fact-checkingservice, how do you decide which facts you willcheck?Will Moy: At risk of a tangent—do cut me off if I goon too long—we try to check facts that our audienceis interested in. We try to aim for an audience of thegeneral public—people who are engaged with politicsbut not necessarily deep political enthusiasts. We tryto maintain a political balance over time and betweentopics. We try to focus on topics we know the publicare interested in. We track the Ipsos MORI poll of

issues facing Britain today, of which the top trendingissues are the economy, heath, crime, immigrationand education.

Q53 Robert Halfon: Basically, you decide based onsome polling and you own decisions on what thingsshould be fact checked?Will Moy: Yes, there is an element of news judgmentabout it.

Q54 Robert Halfon: Do you do the Channel 4 FactCheck?Will Moy: No. We are separate.Robert Halfon: That is nothing to do with you at all.Will Moy: No. Unlike them, we check the media aswell as political claims.

Q55 Robert Halfon: Surely there are ways toimprove statistics. We had a session on digitalengagement with a lady called Beth Noveck, who hadwritten a book called Wiki Government about howthey had crowd-sourced the US Patent Office so thatpeople could actually advise whether things werepatented or not. It has been a tremendous success;they are looking to extend it in others areas ofgovernment. Because of the way that statistics are soopen to different interpretations, surely the best wayto do this would be to crowd-source statistics. Insteadof you guys making a decision on what you look aton an opinion poll, you crowd-source people and seewhat statistics they would like and get them toparticipate in the statistical analysis. Of course, youwould have the experts going through it, but wouldthat not be a better way—it would also be trusted bythe public—to engage people digitally in thecompilation of statistics in a genuine way, rather thanthe great and the good compiling the statistics?Will Moy: I think there is an element of truth in that.Actually, one of the precepts of Full Fact over thenext few years is to involve far more other people—networks of experts—and become a centre wherepeople can come together and do exactly that kind ofwork. The fact is that statistics are not just numbers.Statistics are the product of a careful and carefullydesigned process, which reaches a conclusion that isa set of numbers. The methodology is integral tostatistics, and it is known by the people who producethem. The caveats are integral to statistics, and theyare known by the people who produce them. It is theresponsibility of producers to communicate those.

Q56 Robert Halfon: There will be people outsidewho have their own methodology and are intelligentenough to offer their input. Can I take the views ofthe other members of the panel?Michael Blastland: I have heard Tim Berners-Lee,who is an inspirational figure in many ways, gettingwhole conferences to chant, “Raw data now.” I am notsure he is that familiar with raw data. I am not sure ifhe would say that if he had seen a pile of raw data,which is a lot of zeros and ones in columns. It doestake some ability to decipher this stuff. We couldleave it to a freelance initiative.Chair: The Daily Mail, for example



Robert Halfon: That is misunderstanding what I amsaying. In the example of the US Patent Office, theystill have the US Patent Office and regulations and soon, but they allow people to crowd-source. Theyfound out there were thousands of experts on patents.It enabled the department to function much moreefficiently. Surely you could democratise statistics inthe same way. You would still have people checkingit and so on.Michael Blastland: You could. I am not sure I wouldwant to democratise statistical provenance. I havesome concerns about the volume. You would get a lotof volume, I think; I do not know whether you wouldget a lot of quality. I agree with you that there willbe experts out there, and harnessing their expertise issomething we ought to try to do. I am not quite surehow to do it. I have not seen the Wiki experiment inthe United States; I would be quite interested to lookat it. I come back to Will’s point: there is a great dealof method and expertise involved in understanding theway the data is generated. Numbers are not just outthere to be gathered; the interpretation is a skilfulthing.

Q57 Robert Halfon: What you are saying is thatonly the great and the good can design statistics, andthat it should not be open to the people to getinvolved.Michael Blastland: I am welcoming you to have a go,but I think I will go to the Office for NationalStatistics.Robert Halfon: I would not have a go; I am terribleat numbers.Chris Giles: I am very happy to let 1,000 flowersblossom, but I still want the ONS there to give mestatistics I can trust.

Q58 Lindsay Roy: Are you aware of the criteria forinclusion in the category “official statistics”? I askedthat of the Minister and he said that it is in thelegislation. How is that different from managementdata, admin information and research?Will Moy: This is a mess. Official statistics are verysimply defined in the Act as any statistics producedby a Government department. That is an officialstatistic; a national statistic is an official statistic thathas been assessed by the UK Statistics Authority anddesignated a national statistic. The big question is this:what is a statistic? At the moment, we seem to havegot into a bit of a mire, where occasionally ministersseem to feel able to designate something as not astatistic because it is convenient so to do. Things canturn out as management information or research datawhen they absolutely, in our view, should not be.There have been occasions when the Authority hashad to intervene to point this out.Lindsay Roy: You are saying these are politicaldecisions.Will Moy: No, I am saying these are absolutely notpolitical decisions. The only decision I am aware ofin that chain is the factual decision as to whethersomething is a statistic or not and therefore an officialstatistic or not. That decision, to the extent that thereis a judgment, is a professional judgment that needsto be made by the head of profession for statistics

in a department, free of political influence. It is veryimportant to trust in official statistics, but that is not aministerial decision and should not be mistaken forone.Michael Blastland: My sense is there is a squabbleabout it. The statistical authorities push to includemore of the statistics that might be official or not inthe official designation, but there is resistance and thatresistance excites suspicion. You mentioned thekitemark earlier—the little green tick on officialstatistics—and I would be amazed if there were to bebrand recognition in the public at large of that greentick higher than—I do not know—0.1%. I would besurprised if it were higher than 0.1% amongstjournalists. An understanding of the distinction doesnot really exist in the public at large. As we havediscovered, it does not exist at ministerial level. Iwould be surprised if it existed almost anywhere.It seems to me that there is a choice: either we can tryto demarcate those statistics we want to say arekosher—and all the rest is better or worse accordingto whatever politicians and others want to make ofit—or we can try to say, as I think the public believes,that if the data comes from a Government source—any kind of Government source—then it is official. Isthe Secretary of State not an official source forstatistics? Nobody makes that distinction in theoutside world. We either have to make it one or wehave to bang the drum on distinguishing it. The effortat distinction, with the green kitemark, is invisible.Chris Giles: I would agree with Michael on that.There are also spats between official bodies aboutthings that are national statistics and have thekitemark. For example, for GDP, perhaps one of ourmost important statistics, a preliminary release is anofficial national statistic and does have the kitemark,yet the Bank of England, when they produce theirquarterly inflation report, produce a back-cast of theONS’s figures, i.e. an alternative version of GDPbased on the way they think it will be revised in thefuture. If you graph the Bank of England’s view ofGDP against the ONS’s view of GDP—which we did,unfortunately, on the front page of the Financial Timesquite recently—you do not get the same number andyou can confuse people. There is a great confusionabout national versus official statistics. I totally agree.I think the most important point was the point madeby Michael: if it is a number or a factual statementcoming out of a Government office or minister, itshould be treated as kosher; if it is not kosher, thereis a real problem.

Q59 Lindsay Roy: Can you cite any examples ofofficial statistics that have been very well explained?

Q60 Chris Giles: I actually take a lot of trouble totry to find out about the explanation and methodologybehind statistics. When I do go to the ONS and askfor a private briefing on how statistics are put together,I do find they are rather well explained, but I think Iam in a very small minority of people who do that. Iam also probably in a very privileged position in thatthey are willing to do that for me.Michael Blastland: You do get moments of radiance,when you fall down in gratitude. The press



conferences can be very useful. It is interesting thatthese are spoken, rather than written. I will give youa little example. John Flatley, the ONS head of crimestatistics, said—I think it was at one of these events—that they were trying to understand the decline inmurder rates. There is a common question about thatdecline—whether it is to do with medical advances,for example. He said he was sceptical of thatexplanation. He thought more effective treatment bysurgeons of victims of stab wounds, for example, wasnot the main factor behind the fall, pointing to the factthat attempted murders were falling at a similar rate.Now, that is rather interesting. It is a good littleinsight. If you are very familiar with the data, youknow that. You have looked at it; you have wonderedabout it. You feel it is a relevant comparison and youcan offer it. Most people, I think, do not know whereto start on a question like that. They do. When theydo it well, they do it brilliantly and they do it withauthority. They just do not do it often enough.Will Moy: I agree with both of those points, and Ithink this goes to an absolutely crucial point, which isthat the explanation is out there. Our statisticsproducers have it, but they very often do not share it.Most people we know get their news from thetelevision; our statistics producers are absent from thetelevision. There is a strong case for the NationalStatistician having a similar role to Robert Chote,when he got up, on TV, at the end of the AutumnStatement to explain what was going on. I think ourNational Statistician could usefully have that role.Heads of profession in particular Governmentdepartments could usefully have that role. Explanationis the key to statistics; it is not an optional extra. Youmust go where your audience is.Michael Blastland: I have seen the NationalStatistician popping up a bit more of late.Will Moy: It should be encouraged.Michael Blastland: Yes.

Q61 Chair: Is there not a danger that a governmentstatistician will always be seen to be vulnerable toinfluence, whereas someone like Robert Chote isregarded as completely independent, because he is.Michael Blastland: I was thinking of the NationalStatistician in the Office for National Statistics. Theirstatus is similar, however. I see no reason why theONS’s National Statistician could not speak with thesame authority and independence as Robert Chote.Will Moy: With respect, it is for this Committee toprotect that status.Chair: Yes, I agree with that.

Q62 Lindsay Roy: What should the officialproducers do to explain statistics to non-expert users?Is there a key role here for the media?Chris Giles: I think there is an extremely key role forthe media. The media is mixed, but the media oftengets extremely poor information, particularly from theONS in its written communications. If you bear withme, I have some examples that I would like to sharewith you.Chair: As quickly as you can, please.Chris Giles: These are not just my pet hates. This isfrom last week. I took two releases from last week,

rather than thinking about the things that have gonewrong in the past year. Take the trade statistics from6 December. In the press release, the top line of thetrade statistics was, “Seasonally adjusted, the UKdeficit in trade in goods and services was estimated tohave been £3.6 billion in October, compared with adeficit of £2.5 billion in September.” They alsoproduced a chart, lower down in the press release,from which you can see the truth is that it bouncesaround a bit and is pretty stable. The top line of thepress release makes a big point of how much it haschanged within the month, so it is only looking at thattiny fall there.Chair: I am afraid audiovisual aids do not work inthis Committee. We can publish the chart if you tableit as written evidence—but refer to it.Chris Giles: I will now read you the headlines of thenewspapers and explain how this was reported.Journalists often look at the top line of press releases,particularly from trusted sources like the Office forNational Statistics. City A.M. said, “Yawning tradegap adds to threat of fourth-quarter decline.” TheDaily Telegraph said, “Weak trade renews triple-dipfears.” The Guardian said, “UK trade gap widens inOctober: Office for National Statistics says the tradedeficit in goods and services has jumped to£3.6 billion.” Even in the Financial Times, though wedid not have it in our top line, that statistic was in oursecond line.This really matters. The ONS has to answer thestraightforward journalistic question of, “So what?Why does this matter? What is important about tradehere?” What was important in those trade figures wasclearly not that it rose from £2.5 billion to £3.6 billion.My answer to that question would be something alongthe lines of, “Britain’s trade deficit shows little signof closing as it remained broadly stable and close to£3 billion in October, its average in the past year.”That would have been a fair representation of the data.Why the ONS chose to do a monthly change, I donot know.

Q63 Chair: That is very interesting. Why do youthink they did?Chris Giles: My fear is that they pander to the worstelements of journalism.Chair: They think they need to get coverage.Chris Giles: They know what journalists might wellwant and they give it to them.Chair: That is a very strong thing to say.Chris Giles: I say that is my fear. It is not somethingI know, but it is my fear.Chair: We will put it to them.Chris Giles: It is something to put to them. Can I giveyou an even worse example? This is family spendingfrom 4 December. This, again, is last week. This isthe result from the annual Living Costs and FoodSurvey. This is what determines the basket for theinflation indices. It tells us what people are spendingin the UK. The headline was, “Household spendingedges higher while spending patterns differ byincome.” The first half of that sentence is wrong. Itedged higher in nominal terms, not in real terms. Thatwas actually giving a misleading impression. The



second half is empty. “Spending patterns differ byincome” is a complete irrelevance to anybody.That is not the worst of it. It goes on. In the first line itsays, “The annual report from the ONS on householdexpenditure in the UK found that in 2011 average UKweekly household expenditure rose to £483.60, anincrease of £10 on the level recorded for 2010.” Thatis two pretty irrelevant numbers out of context,because we do not know what sort of households theyare. Also, £10 out of what is essentially £500 is anincrease of well under 2%, when we know inflationwas running at 5%. That is a decrease.

Q64 Chair: Is there an error margin in thesestatistics?Chris Giles: There will be an error margin, but that isnot mentioned in the press release.Chair: Would it be as much as 2%?Chris Giles: It will probably not be; it is quite a bigsurvey. It goes on to say, “The 2011 averageexpenditure is the highest recorded by familyspending.” They are going into record making.Actually, in real terms, it was the lowest in 10 years.Chair: If you were the Minister for happy families,you would be quite pleased with that headline.Chris Giles: You would be quite pleased with thatheadline, and you would be wrong to be pleased. Itthen goes on to say that transport costs were thehighest category of spending because they make anelementary mistake in assuming that people withmortgages do not pay for their housing services,because they used a particular form of NationalAccounts definition of housing expenditure, which inthis survey is not the appropriate definition. Those areactual conceptual errors in the press release.Let us look at how it was then reported. The DailyTelegraph said, “Household spending rises as fuelprices soar.” The Guardian said, “Household spendingat record high fuelled by rising petrol prices.” TheDaily Mail said, “It cost £483.80 a week to run aBritish home in 2011.” Sky was rather better; theysaid, “Household spending up but families get less.”The Financial Times—this time we did it exactlyright, I think—said, “Household spending falls to a14-year low,” which is actually what the truth of thatstatistic was.Chair: How interesting.Chris Giles: I think that is a very telling example ofhow what the ONS put out in their press releases isreally important, because the press follow it. It is notjust the written press, which is not much easier to findon Google; the broadcast media will do the same. Itis not a small point; it really matters. The ONS arefantastically bad at it.Lindsay Roy: The point is noted.Michael Blastland: I have a potentially slightly morebenign explanation for the way the ONS organises itsreleases, although I am afraid now as well. I think tosome extent they feel they are just giving the numbers.These are the latest numbers, so they give you thelatest month and we put that at the top of the page. Ithink it is a feeling that there is a kind of purity aboutthis—objectivity, if you like. However, I think thatillustration shows us pretty clearly that there is nosuch thing as a purity. They have implied something

about trends and the state of things now byemphasising the rate of change rather than the longrun. You cannot make choices that do not imply someparticular take on the data. You are always, to someextent, telling a story.

Q65 Chair: You are suggesting that they would say,“It is not our job to provide a gloss on this; we willjust provide the raw facts.”Michael Blastland: Yes, but the raw facts are neverraw. They are implying something about what is goingon. I think they believe they are giving raw facts, butthey are not. When you come back to their reluctanceto say, “We should not get involved in that kind of anargument,” they are involved—whether they like it ornot. The way they present the first line of the datainvolves them in some statement about the data.Will Moy: To take the constructive explanation onestage further, maybe they do not believe they aregiving raw facts, but they basically do the same thingwith every statistical release, which is compare thelatest with the previous. In doing so, they areeffectively implying that to some extent this is amechanical process.Chris Giles: But they do not.Will Moy: We have hired thousands of capablestatisticians to help explain our country to us; weshould be saying to them, “We expect you to explainour country to us, not just give us the results of amechanical processes.”Even more importantly than that, picking up on whatChris is saying, they have a responsibility tocommunicate effectively. That involves consideringthe impact of your communication on your audience.What the ONS should be doing is taking into accounthow their releases are covered and feeding that backinto how their releases are drafted. On occasion, wehave asked them to do this when we have picked upon misleading coverage. I have to say that the resultshave not been entirely helpful.There was one example around jobs from back in thedays of Gordon Brown’s slogan, “British jobs forBritish workers.” Those statistics were particularlycontroversial and were a regular topic of coverage.They were being misunderstood in a slightly technicalway. We went to the ONS and said, “If you put a littleexplanation in your release, people will stop makingthis mistake.” We had to go to them, I think, threetimes before sufficiently high-up people decided notto do it. We then had to go to the Authority to say,“This is obviously what you exist for; why do you notchange this to avoid this mistake?” They told the ONSto do that, and in future coverage the mistake was notrepeated. That is a very simple pattern of behaviour,which serves the public well.

Q66 Chair: Does that not suggest they are aware oftheir political masters?Will Moy: I do not want to tar all of them with thesame brush; I have great respect for the people in theONS and the GSS, but I think it suggests that toomany of them believe that their job is finished oncethey have got the product out. They see themselves asa producing organisation, rather than a communicatingorganisation. Actually, their job is to instil



understanding as much as to produce a technicalproduct.

Q67 Lindsay Roy: You have given us examples ofimbalanced interpretation.Michael Blastland: Can we give you examples ofthat?Lindsay Roy: No, you have just given them.Michael Blastland: Yes.Lindsay Roy: That is very helpful. Thank you.Michael Blastland: Yes, although it is aninterpretation on the premise that they were giving nointerpretation. Had they thought about whatinterpretation they were giving, I think they wouldhave given a better one.Chair: That is very well put, if I may say so.Chris Giles: If you ever ask the official statisticianswhat they think the most important statistic in arelease is, they will not give the top line of the pressrelease. What has happened to unemployment will notnecessarily be the change in the three-month labourforce survey measure compared with the previousthree months. They will, in words, give a long-runhistorical interpretation—yet their press releases donot do that.

Q68 Paul Flynn: Can I ask you whether you hada moment of radiance or desolation listening to ourlast witness?Michael Blastland: I do not think I learned a greatdeal—other than that statistics, by and large, is notconsidered a big political priority.

Q69 Paul Flynn: Can I take up a point made by theChairman? I am sure he did not intend to impute theintegrity of statisticians, but I treasure a letter I hadfrom Mrs Thatcher in 1988, regarding the move of thestatistics department to a department that had a vestedinterest in fiddling the figures, saying that it wasunworthy of me to suggest that politicians wouldinterfere in any way with the purity and objectivity ofthe statistics. Since that time there have been manyaccusations.The whole point of setting up the Statistics Authoritywas to increase public trust. If the many hundreds ofmy constituents who work on statistics find that thestatistics are not trusted, their work is valueless. Wehave to establish trust. Are we making progress onthat? Have things improved in terms of publicconfidence? The very valuable evidence you havegiven suggested that things are probably as bad asever, because of those, press or politicians, who seekto manipulate the conclusions of the ONS.Will Moy: I am not sure I am that negative. We haveput into place some very important building blocks oftrustworthy official statistics. That has been a long,ongoing process. Onora O’Neill’s analysis of whattrust really means is that people, rather than takingthings on blind faith, defer to the ONS, contest thingsfor ourselves and reach our own conclusions aboutthem—and thus we reach informed trust. That is avery important touchstone here. More and more thatis becoming possible, so I think we are makingprogress. One of the big tests is whether misleadingclaims using statistics continue to persist. Therefore,

the tests for this Committee are whether you will backup the UK Statistics Authority when they challengemisleading claims and whether ministers will beeffectively stopped from making misleading claims.In due course, we should be talking about theAuthority doing a similar role in terms of oppositionspokespeople. I am certainly not making a point thatis partisan, even within a time period. Ultimately, trustconsists of being able to check things for yourself andthings that are not trustworthy being stopped. Thebuck, for that, stops with the Committee.

Q70 Paul Flynn: The Committee certainly supportedMichael Scholar in his frequent criticism of the HomeOffice figures, primarily. I am sure we do the samefor National Health Service figures and so on now.Could you clean up this idea that the StatisticsAuthority are sensitive to the views of their politicalmasters? Do you really believe that? Are they likelyto react to what their political masters are saying,whether they are Labour or Conservative? There wascertainly no sign of that from Michael Scholar, whowas robust in criticising the previous Government.Will Moy: Others will obviously have a point of view.Paul Flynn: Andrew Dilnot was someone who hadthe enthusiastic support of this Committee, both as astatistician and also as a communicator.Michael Blastland: I do think they need to keep theirpowder dry. You cannot go after everything. MichaelScholar’s great achievement was to pick his targetsvery carefully, so that he was able to establish theauthority of the Statistics Authority in commenting onthe general use of statistical data without looking as ifhe would become the kind of pedant who was perhapspolitically motivated because someone out therewould be adding up the criticisms of eitherConservative or Labour or whatever. He chose hiscases carefully. I think that must continue to someextent. There is an educative purpose in that. Whenyou pick your cases, what can you demonstratethrough them? Is this a case that is simply of highpolitical salience? It is an egregious example? Can welearn something about the way this data has beenabused or misrepresented? There are a lot ofconsiderations when you say, “We will go after thisone but not that one.” I have absolute faith in Andrew.I have no reservations at all about his politicalindependence. I suspect in the past there might havebeen members of the board who felt the heat fromcertain parts of Government when they pressed to takemore statistics into official statistics, and perhaps partsof the Government machine said, “Keep away.”

Q71 Paul Flynn: The hot issues are the problems. Ican recall evidence from a man with a memorablename, Alfred Hitchcock—a policeman in London—on knife statistics, saying that knife crime had gonedown in London, yet there were blazing headlines inthe tabloids saying that it had gone up. TheGovernment reacted and introduced some instantlegislation to deal with it. The popular press—sinceyou absolved the serious press of it—have a vestedinterest in creating their own fiction by using statisticsthat are completely false.



Michael Blastland: I have some sympathy with Will’soptimism here, because an increasing numberorganisations—his is one of them, actually; he cannotsay this but I can—provide scrutiny elsewhere, andthe capacity for embarrassment is a very importantpart of the process. If you can haul people up andpester them and pester them until they do somethingabout it, which, among other things, Full Fact does,you might get somewhere. We have a lot more ofthose organisations. We have Full Fact. We haveChannel 4 Fact Check. We have Ben Goldacre in TheGuardian newspaper doing a similar sort of job. Wehave the Statistics Authority. We have the More orLess programme on Radio 4. There are far more ofthese places than there used to be, and there is a bitof a tide in this. It is almost fashionable in journalismnow and I am encouraged by that.Chris Giles: I am encouraged but my encouragementis quite limited. I still think we have a long way togo. I have no questions at all about the independenceof the UKSA, but I do worry much more about howmuch other Government departments really care ifthey get their knuckles rapped by the UKSA. Does itreally hurt them? Is the Prime Minister reallychastened from what he said in the Commons theother day, whether or not he knew about the GDPfigures under pre-release? I am not sure; I am not thatconfident. Though I think all the things that Full Factand all of these other good bodies, including our own,do are important in trying to keep people honest, Ihave not noticed that it has necessarily got a lot betterout there.Chair: I can assure you that ministers hate receivingletters from the UK Statistics Authority Chairman.Chris Giles: I am sure they say they hate it, but dothey really hate it?Chair: They really hate it.Will Moy: The test, though, Mr Chairman, is whetheror not they change their behaviour afterwards.Chair: That is something that we are certainlyconsidering as a Committee and we will be discussing.

Q72 Greg Mulholland: It has been a fascinatingdiscussion. The written evidence from Full Fact saidthe goal of publishing official statistics “should be toensure that users can get the information they need, inits full context,” and went on to say it “should striveto present a coherent statistical picture in important orcontentious areas of public debate”.Clearly, statistics are there to inform the big issues ofthe day and to lead to better and more honestinformation—but also better policy going forward. Inthe end, that is how we govern and are governedbetter. How do you think we should deal with thischallenge, precisely going back to the discussion wehave had about presenting some of the findings ofstatistics in a misleading way? How can we on onehand say that just having the raw numbers is nothelpful? If you are saying we should contextualise,how can we contextualise and put those into thecontext of the big issue of the day, when, clearly, youget into the political domain and people are trying tocommunicate different parts of that? Is that possible,or does that actually lead to the kinds of problems wehave heard about?

Chris Giles: I think it is completely possible. Youneed to use some good journalistic skills, so you haveyour data and you have certain forms that you useevery time. So on the front page, you need a long-runchart, which you keep the same the whole time. Youprobably want a near-term chart, because people, innews terms, do want to know what has changedrecently. Otherwise you get a long, straight line, whichlooks very boring. You also want a correct headline incontext. That needs to be written by someone prettysenior in ONS, who can actually defend, prettycategorically, why they have chosen that as theheadline for a chart.What I would also like is any special features that weneed to know about. This is exactly what Michael wastalking about in terms of the murder-rate example. Ifattempted murder has gone down as well, that is avery relevant feature. If in public finance statistics youknow there has been a delay in winter fuel allowancepayments, for example, that means that in that monthin particular public spending is lower than you wouldhave expected it to be. Some of those things must bebrought up.You want links to the methodology for people andthen links to the full data online for people who reallywant the detail. You do not need a lot more than that;most people do not use a lot more than that. That isnot very difficult. Instead of producing 105 pages,which is the length of the retail sales first release, youcut it by 95%. I mean 95%. If it were a release ofabout five pages, you would have time to make itrather more intelligent.

Q73 Greg Mulholland: Very simply, you would say:clear guidelines and consistency.Chris Giles: Clear guidelines and consistency, butallowing the leeway for the top line to be determinedby the statisticians who know about it. If politiciansthen want to make a big point and say there is politicalbias either way in this, it would be for them to haveto defend why they are attacking the ONS, when theyare trying to inform the public.Michael Blastland: One other quality I would like tosee in more of the statistical data is a willingness tobe relevant quickly. We have a schedule of releasesand we stick to it. There are good reasons for that.However, there are issues that come up from time totime that have no prominent place in the schedule ofreleases.I will give you one quick example: disputes andstrikes. We have had something like four years, as faras I can tell from my look in the media, of speculationabout winters, springs autumns and summers ofdiscontent. It has been relentless. Look at the data,which I have seen precious little of—either in themedia or in political argument. It is astonishingly low.I mean breathtakingly low. There was a period beforethe strikes last spring when we had a 12-month periodin which there were fewer strikes in that whole yearthan there were in every single month of the Blitz,when we were all pulling together. It has been thatlow for possibly 20 to 25 years. There has been anabsolute step change in industrial relations in thiscountry and we know very little about it. In fact,regarding your point about whether we make policy



on this, there have been discussions about whether weneed further policy to do something about the threatof Britain being brought once again to its knees bystriking and so on. We have had trade union leadersrattling the sabre talking about comparisons with thenational strike of 1926—all in an entirely data-freezone.The ONS has these numbers. They come out in thelabour market release on page 22 or something likethat—in the usual spot—to show the numbers ofstrikes or days of labour productivity lost this year. Itis rather interesting and it is missing. The ONS knowsabout it; they know its historical trends. They couldpull something together to say, “This is a very topicalissue; lots of people are talking about it. Here aresome common misconceptions about the data. Here issome information you might find useful.” Awillingness to be useful from time to time would be agreat addition to Chris’ list.Will Moy: If I may, I will pick up on some of theprinciples behind what you are asking. Chris divedinto detail—I support what he said—but yes, it ispossible to write non-partisan, informative content.Actually, the opposite of explanation is not beingnon-partisan; it is being ignorant. I think we reallyneed to be aware of the danger of ill communicatedstatistics. They are positively a problem. TheCommittee, I believe, needs to give the statisticalproducers a clear steer that we expect explanation andwe do not see that as them being too political. Weknow that they know where the line is; we know thatthey know where to draw it; we know that they canexercise very careful judgment about being politicaland not being partisan. However, we do needexplanation. That is, emphatically, their role.Michael Blastland: The IFS does it.Will Moy: The IFS manages that task extraordinarilywell.Chair: Sorry, one at a time.Will Moy: Michael is right; the Institute for FiscalStudies has a stellar reputation in this area. Full Factis busily building ours. The Financial Times do prettygood non-partisan analysis and so do other bodies. Itis not impossible.

Q74 Chair: Does the BBC do good non-partisananalysis?Michael Blastland: Variable.Will Moy: Michael works for them and I used to.Michael Blastland: It is pretty good at some times,less good at others. I do not think the motives arepolitical, on the whole. I do not think that is true.Will Moy: On the practical side, Chris is absolutelyright to identify the need for good journalistic skills.I think it would be surprising if we suddenly found alot of good journalistic skills were lurking untappedamong our statisticians. If we want to docommunication well, we actually have to hire peoplewith communication skills. We have suggested thatwhat is needed is what would be normal in any otherpublishing organisation, which is a sub-editor’s deskthrough which publications go and which ensureshouse style and standards and is focused on what theaudience needs, rather than the technical demands ofproduction.

Chair: That is a very interesting suggestion.

Q75 Paul Flynn: Will the present national census bethe final one? Is it not time to bury this expensive,misleading anachronism—yes or no?Chair: That is topical.Michael Blastland: I would rather leave that one toAndrew. Over the years it has been such a tremendousresource. Whether we have the technical ability toreplace it in a much cheaper way I am not qualifiedto judge. It would be useful if we could. It isexpensive. It is not a large cost, given the amount ofpublic distribution money that hangs on it. If wecannot get it reasonably right by other sources, I thinkwe would be in trouble. If we can, let us avoid thehalf billion or whatever it costs. It would be a simplething for the authorities to satisfy themselves of: canwe replace this data from other sources?

Q76 Paul Flynn: Wouldn’t we get the sameinformation if we asked one ten-thousandth of thesample that we ask? Would not we get more accurateinformation about the number of supporters of JediKnights, for example?Michael Blastland: You are always going to have tosurvey people to some extent and they are alwaysgoing to say stupid things.Chair: On the population levels in London boroughs,for example, the census produced some real surprises.Is that not a justification for the census, or could thoseissues be tackled in a different way?Michael Blastland: I do not know the technicalquality of the alternative sources. That is the problem.The census has been magnificent over the years.Chair: If I remember correctly, the London boroughof Newham had several tens of thousands more peopleliving there than anybody realised. These are verybig numbers.Michael Blastland: Yes, they would have a markedeffect on the distribution of grants for variousservices. It sounds valuable to me.Chair: On the other hand, shouldn’t the boroughthemselves have known?Michael Blastland: I can only repeat the sameanswer: if you can find sufficiently good alternativessources, yes, but who satisfies themselves that theyare sufficiently good? I think I will leave that up tothe Statistics Authority.

Q77 Chair: We are constantly assailed by thisunofficial data problem. The most notable examplewas Sir Michael Scholar’s parting shot to the DWP,which was about research they had published aboutthe number of claimants on out-of-work benefits whowere immigrants. What is the best way of addressingthis issue? There is a danger of gumming up themachinery completely with procedural problems ifevery number must be filtered through the sameofficial statistics framework.Will Moy: Pardon me, but the official statisticsframework only exists to make sure the numbers areactually sound and not political. These are notstringent and bureaucratic requirements; these are thebasic requirements for numbers that you can actuallytrust.



Chair: You would argue that the Department waspassing something off as research.Will Moy: It has been a while since I looked at thatparticular example, but I have certainly seen examplesof things I think were perfectly clearly officialstatistics that should have been judged against thecode and were trying to avoid being judged againstthe code by being designated as either ad hoc,management information or research. It is a decision,to some extent, of professional judgment. I think itneeds to be very clear that professional judgment mustbe insulated from ministers, from the press office andother political influences. If there is any doubt aboutthat professional judgment, the heads of professionshould consult the National Statistician or theAuthority as appropriate to try to insulate it fromdepartmental political pressures. If in doubt, the codeshould be met.Michael Blastland: I would only add that there is alegitimate place for research. The ONS itselfpublishes what it calls experimental statistics; there isno problem with that. We are not confident enough togive them the green tick, but we want to have a lookand we also want feedback from other people andexperts on how good they are. There has to be somescope and freedom to do this. It is about thepresentation. Are you claiming that this is giving youa definitive account of something that is going on outthere or are you just saying, “We have had a bit of aprod around and this is what we have come up with”?The latter seems to me to be necessary and sensible—as long as you present it in that form.

Q78 Chair: The procedure is that if a departmentdoes some research and they want to publish it, itshould go through the ONS. Is that right?Michael Blastland: Or it should be very clearlylabelled as an experimental piece of research that hasnot been validated by the ONS and for that reasonmay not be an accurate representation.Chris Giles: I do not see a problem at all with aGovernment department putting out the equivalent ofa working paper and essentially saying, “We havebeen musing about this.” The Bank of England doesit all the time. There is no problem as long as they donot then try to pass it off as official statistics. It is thatpassing off process, which ministers might get into,where it becomes problematic. That is quite difficult.If ministers do that, the response will be that it mustgo through the code and it will become a verybureaucratic process. I do not see a real difficulty ifofficials are going around and trying to do their jobproperly, investigating things in their area and puttingout working papers on that to get wider views.Will Moy: The key thing really comes in the presshandling. If you generate something in a certainformat just so that you can leak it the night beforewithout getting an argument back and for over thecode, that is really where most of the argument is.

Q79 Chair: Moving on, you are all, obviously, nothappy with the way the ONS is engaging with users.How should they improve this?Michael Blastland: It is not just the ONS.

Chair: And other official statistics producers. Howshould this be addressed?Chris Giles: I will deal with what I think shouldhappen at the ONS, because I deal with the ONS alot. I like the ONS. I have been very critical today,but that is because I think that is what is more helpfulfor them to improve. I think they produce a lot of verygood statistics, but I think they are let down on theircommunications. The two most important things theyneed to do, I think, are to sort out their first releasesand their website. The website revamp was acomplete disaster.Chair: They need to engage with users. That is thequestion I am asking. How should they engage withusers so they get feedback to improve their business?Most businesses use complaints to improve theirbusiness.Chris Giles: They have used feedback, because wehave certainly gone to filmed seminars where we havesaid similar things to what I have said today in frontof a two-way mirror for statisticians to hear whatpeople say about their releases.Chair: Very good. I am very pleased to hear that.Chris Giles: They are doing that. It is important. Inrecent weeks, statisticians have become veryhelpful—more helpful than they have previouslybeen—in trying to engage with us as a unit. I thinkthe ONS is trying very hard. The most important thingfor its users is that it must think about them in threecategories.

Q80 Chair: What is the message to the leadership?What do they need to do?Chris Giles: If they have taken a decision tocommunicate better, it is that they need to really pushit forward.Will Moy: The short answer is that they are quite goodat engaging with users in the sense of the StatisticsUser Forum, professional users of statistics,statisticians themselves. They are very good at that;that is a brilliant institution. I do not knock it at all;they deserve lots of credit that we are not spending alot of time giving them today. Where they need tofocus is on decision makers, the absolutely crucialusers of statistics—you, ministers, and permanentsecretaries, who Mr Dilnot, in a previous hearing,pointed out sometimes tend to use statistics as thegarnish rather than the meat of their decisions. Also,they need to focus on the general public, who are thebig missing audience in statistics production. They areincreasingly looking for statistics themselves online;they need to be taken more into account.

Q81 Chair: I am still not clear on what the leadershipof the statistics service needs to be doing in order toimprove this?Will Moy: They need to change the culture of theONS and the GSS so they look at communicating bystarting from where their audiences are. What dopeople want to know about? How do they want toreceive that information? Those are the questions. Ifyou applied that way of thinking aboutcommunicating, you would radically transform whatthe ONS puts out.



Michael Blastland: I agree. I think they must answerthis question. Who are your users? It is a muchbroader community than they are accustomed tothinking of. There was a recent consultation on thePESA statistics, the Public Expenditure StatisticalAnalysis, by the Treasury. It invited consultation fromanybody with an existing strong interest, i.e.Government bodies, businesses, local authorities,voluntary organisations and academics. In that list Ido not think there was any sense of the general public.There is no sense of where you might expand; thereis only a sense of where we are; Let us consult peoplewe already have and decide whether they are happy.

Q82 Chair: Mr Moy, you suggested that they shouldbuild up a relationship with the Press ComplaintsCommission or, I presume, whatever succeeds it, inorder that the regulation of the press should be muchmore concerned with how statistics are being used.Can you explain a bit more about that?Will Moy: What we are concerned about is the casewe had earlier this year, where the UKSA described aparticular piece of coverage as misleading. The PressComplaints Commission reached a contradictorydecision on it. By what authority they felt theyunderstood statistics better than the StatisticsAuthority, I do not know.Chair: What issue was this?Will Moy: This was about reporting riot statistics. Icould go into it but it is probably unnecessary.Chair: Yes, I remember it.Will Moy: It is clear, as members of the Committeehave raised, that there are issues with the waystatistics are reported in the press. Part of theresponsibility for that lies with the way they arepresented by statistics producers; part of it lies withthe press. Some people have said the Authority oughtto be going around shooting its mouth off at the presssaying, “Fix this; fix that.” I agree with Michael thatthis would be unhelpful. It needs to save itsinterventions for the relevant points.There is a body, in the case of the press, whose job itis to uphold accuracy. It is a fundamental principle ofjournalism that everybody subscribes to. If the UKSAhad a positive working relationship with whatevercomes out of Leveson, with the Advertising StandardsAuthority, where necessary, or with the Committee forStandards in Public Life, which is also currently underreview, all of these bodies, whose fundamentalconcern is trust in public life, can make sure that inthe relevant areas they are the lead organisation and,where necessary, they take advice from the otherorganisations. They can all intervene in their differentways to uphold the standard, which is common acrossall of them, of accuracy and substantiation of what issaid in public life. It seems to me to be a basic bit ofco-operation.Chris Giles: Can I just make one final point? For theleaders of the ONS and the UKSA I think there is one

very important strategic objective they need to thinkabout, which is that they in some ways need to turnthemselves into accountants. Accountants have a testcalled a true and fair test. The biggest way in whichthe public are misled in the use of statistics is bythings that are true—I read out quite a lot of things,like that family spending had gone up by £10. Thatwas true, but it was not a fair representation of whatwas going on. True but not fair is as good as beinguntrue, in my opinion. It is that difficult boundary.When we do things wrong, we say, “Yes, but that wastrue.” Ministers do this as well. They say, “That wasstrictly, definitively true.” If it gave a misleadingaccount of what is actually happening, that is as goodas not being true. That is the test the heads of theUKSA and the ONS should be going around anddoing to find out what is really egregiously wrong inthe communication of statistics.Michael Blastland: Also, they need to start again onthe website.Will Moy: It cannot be said enough. We do not needa website that is 10%, 50% or even 100% better; weneed one that is 1,000% better. We need one that isactually designed in 2012 and not an upgrade ofsomething that started in 1995.

Q83 Chair: I could invite you to make yourself clear.What do you have to say about pre-release? You heardthe Minister.Will Moy: It is wrong in principle. It violates theprinciple of equal access to official statistics. We paidfor them. We are all citizens and we deserve to getthem at the same time. It is wrong in practice: it leadsto the kinds of problems we constantly see and willcontinue to see. I do not believe there is a realadvantage to it. I believe that is the Committee’sstanding position. I think you should say it again, andsooner or later it will change.Michael Blastland: I have nothing to add except thatpossibly if there is really little to be gained, as I haveheard ministers say on occasion—that they do notachieve any great advantage from it—then why do it?Chair: It is custom and practice—it is cultural.Michael Blastland: A custom and practice that excitessuspicion is not a custom and practice you really wantto continue.Chris Giles: I just have one thing to add: I deal withquite a lot of market-sensitive statistics and I deal withthe Treasury a lot, and I do not think I have everhad a well-informed briefing at 9.31 from the Treasurybecause they had pre-release.Chair: That is something we will have to ask themabout. Gentlemen, you have given us a reallyinteresting session. Thank you very much indeed. Ithas been really exciting. I have enjoyed thisenormously. Thank you very much indeed for yourtime.


Wednesday 12 December 2012

Members present:

Mr Bernard Jenkin (Chair)

Charlie ElphickePaul Flynn

________________

Examination of Witnesses

Witnesses: Andrew Dilnot CBE, Chair, UK Statistics Authority, and Jil Matheson, National Statistician, UKStatistics Authority, gave evidence.

Q84 Chair: May I welcome our two witnesses to thissession about communication and statistics? Pleaseidentify yourselves for the record.Andrew Dilnot: I am Andrew Dilnot. I am Chairmanof the UK Statistics Authority.Jil Matheson: Jil Matheson, National Statistician.

Q85 Chair: Thank you very much for being with us.First, in reaction to yesterday’s session with theMinister, we were a little surprised: he seemed lessengaged with some of the issues that many peoplethink are very, very important in the statistics world,not just in relation to the communication andpublication of statistics. Could you say somethingabout the relationship between UKSA and the CabinetOffice? I mean, do you have regular meetings? Whatkind of engagement do you have? Do you feel youhave a good relationship with the officials briefing theMinister? Is there an issue to address here?Andrew Dilnot: We have two very different sorts ofrelationship: we have a relationship with officials,because in some senses the Cabinet Office isresponsible for some elements of the pay and rationsassociated with the UKSA; and I have a relationshipwith Mr Maude. I have not met Mr Hurd in hisministerial capacity, but I have met Mr Maude, I think,on either two or three occasions since I took office. Ihad a meeting with him two or three weeks ago, andwe are in line to have regular meetings. I suspect, Ihope, that I will meet him a couple of times a year totalk about statistics business. When I have met him,we have discussed the interventions that have beengoing on. We have discussed pre-release access. Wehave discussed data sharing, the future of the Census,and ways in which the Statistics Authority might wantto look to the Cabinet Office and Cabinet OfficeMinisters for support. I also have interactions with theCabinet Office, both at official and ministerial level,over things like appointments of new members to theboard: soon after I arrived, we had to make three newappointments because of the sad death ofRoger Jowell and the retirement of others. There is awhole process of that going forward. So, by and large,that seems to be working okay. I was not able to watchMr Hurd’s evidence session yesterday. I was in ameeting elsewhere, and all I have seen is a rather briefsummary of it.Chair: Right. Ms Matheson, do you want to sayanything about that?Jil Matheson: No, my contacts are, as Andrew said,on the pay and rations side, and also on matters ofCivil Service reform, where one of the strands is about

Kelvin HopkinsGreg Mulholland

capabilities and strengthening of the professions, ofcourse, which I take a great interest in. So there iscontact with Cabinet Office officials.

Q86 Chair: Mr Dilnot, what about with Secretariesof State and permanent secretaries?Andrew Dilnot: Permanent secretaries: I am going ona circular trip around the great Departments of stateseeing permanent secretaries, often seeing permanentsecretaries with their heads of statistical professionand sometimes an appropriate director general. I thinkI have now seen three or four, including the permanentsecretary at BIS, the permanent secretary at DWP. Iam off to CLG quite soon.

Q87 Chair: So you have a programme of meetings.Andrew Dilnot: Yes, I will not call them state visits,but that is all good. I have met the Cabinet Secretaryto talk about statistics matters twice, I think, since Ihave taken office, and the Cabinet Secretary has madeit clear that if we have issues, I am at liberty to ringhim up and say, “Sir Jeremy, we need help with this.”

Q88 Chair: And Sir Bob Kerslake?Andrew Dilnot: I bumped into Sir Bob two weeks agoon my way in to see Sir Jeremy. I have not had aformal meeting with him, but again he is certainlyopen to that.Chair: Well, we think there is scope for improvementin terms of what we heard yesterday from theMinister.

Q89 Charlie Elphicke: Can I ask who is primarilyresponsible for managing and maintaining the ONSwebsite?Andrew Dilnot: There is a range of different answersto that. I imagine I will turn to Jil in a moment fordetail on how the structure works. In the end, the UKStatistics Authority Board has to take responsibilityfor that as for all other matters, but we have a DirectorGeneral of the Office for National Statistics, GlenWatson, relatively new in post, and the NationalStatistician, and, within the ONS, we have a structurethat has changed relatively recently. I will pass to Jilfor an accurate description of that. At the beginningof November, I think, we set in place a new analysisand dissemination directorate, and it is now withinthat that the structure of the ONS website iscontrolled.Jil Matheson: Part of the restructuring of the Officewas to bring a new leadership focus to the way thatwe communicate and disseminate statistics—the way


12 December 2012 Andrew Dilnot CBE and Jil Matheson

that we engage with users. That of course includes thewebsite, so a lot of work is going on on the websiteto bring about some much needed improvements;some of them are already in place, but theresponsibility is within ONS.

Q90 Charlie Elphicke: And would you accept thatno one or virtually no one now consults hard copycompendia? Inevitably, it is the website that everyoneis going to consult to find statistics and to get thelatest information and, effectively, the ONS now is thewebsite from the point of view of anyone looking foror utilising statistics.Andrew Dilnot: Yes. I think I made clear at mypre-appointment hearing coming up to a year ago thatI thought the website was the absolutely central wayin which people access information, certainly at themoment. Who knows what the world will be like intwo years’ time? There are some signs ofdevelopments in social media, for example, that weare going to have to take seriously, but at the momentthe website and certainly electronic communication isat the absolute heart of what we do, and we must makethat as good as it possibly can be to give the kind ofaccess to the statistics and commentary that we haveto see.

Q91 Charlie Elphicke: Before you became Chair ofthe Authority, presumably you would have consultedthe website for statistics and things like that. Did youfind it easy to use, user friendly and highly accessible,or did you find it challenging?Andrew Dilnot: I have boasted that for many years theONS has been one of my home pages on my internetbrowser. Before I came to work in the Authority, itwas indeed a website that I consulted regularly. Fouror five years ago, I thought it did what then seemedto be a good job, and I have said repeatedly we havein this country marvellous data—an extraordinarilyrich array of professionally produced statistics. Therelaunch of the ONS website in August of last yearwas not one of our greatest moments, and at that timethe website became difficult to use, difficult tonavigate, difficult to search, and a lot of work hasbeen done since then. My view is that quite a lot ofprogress has now been made. There is quite a lot morestill to be made, but even in the last two weekssignificant progress has been made in an area that wasthe one I found most frustrating, which was search. Ithink Will Moy from Full Fact in his written evidencepointed out how hopeless the search function used tobe. I think the example he used was that you couldtype in “population” and you would get 3,000 resultsin no particular order. I think the weekend before last,the ONS web team quietly put through a reform to thesearch function, and now when you type in“population”, the first thing that comes up is the latestCensus results. I spent quite a lot of time yesterday,thinking that these questions might come up, trying tobreak the search engine and not succeeding very well.So the search is now much better; there are bettertheme pages, but there is a further significant series ofsteps to be taken.

Q92 Charlie Elphicke: Would you accept that,broadly, the website has been pretty terrible? It needsa lot of improvement and you are on the case. Thesearch is just awful and that needs improvement, andalso are you going to sort out the time series, as it isalso a menace to try to work out how to do those? Allthese things are shocking to all of us who are anoraksand use statistics. Quite a few around this table,including Mr Hopkins and me, find this endlesslyfrustrating. Are you going to say to us, “We get it, andwe are on the case”?Andrew Dilnot: Absolutely, and one of the things thatwe have agreed with this Committee is that we willbe bringing a version of our strategy to you in thenext little while, and one of the things you will see inthat is the significance that we give to this. I wouldencourage you to go back and have another go at thesearch engine, because I think it is now muchimproved, but we would really value input from thisCommittee. So if there are more specifics, please letus know, because it is of the highest possible priority.

Q93 Chair: Can I invite you to look at the transcriptsof Chris Giles and others yesterday? They were prettyexcoriating about the website, in particular about theinability to obtain time series. I have to say thatMembers of Parliament are very insulated from thisproblem because we just ask the Library, but mostpeople do not have the expertise of the Library at theirdisposal, and getting time series on GDP statistics orunemployment statistics is still extraordinarilydifficult.Jil Matheson: Can I just come in on this? I absolutelyrecognise the problem and, as Andrew said, I am auser of the website as well as responsible for it andshare that frustration. There have been improvements,but the improvements that are there now are only partof a process. There is more to come. One of thelessons from the really disappointing launch of thewebsite in August 2011 was the need to engage thevery wide range of users that we have. There are, asyou say, some very expert users and there are somevery casual users and trying to meet a range ofrequirements is not an easy task, but we have groups;we have been talking to journalists and others, tryingto understand what their experience is and what theyneed from the website. That is now part of what willcontinue to develop over the next not just months butyears; I think this is an ongoing developmentprogramme.

Q94 Charlie Elphicke: The other issue is I havealways found the publications like Social Trends,Economic Trends, Labour Market Statistics, FinancialStatistics, Annual Abstract, and Monthly Digestpersonally very, very useful, particularly forprompting new thought. I published a lot of think-tankpapers when I had time before getting elected here,and these were very useful publications, and yet theydo not seem to be around anymore. I do notunderstand it, and I think it is a real tragedy that theseall seem to be discontinued or something. What isgoing on?Andrew Dilnot: My understanding is that in the faceof cuts, the compendia documents were one of the



things that were discontinued, after some consultationwith users. I have to say, Mr Elphicke, that I am withyou. I have often said that the world would be a muchbetter place if every Member of Parliament had a copyof Social Trends under her or his pillow, because itcontains such a rich source of information and theputting together of information that is not necessarilyprecisely what you thought you wanted. Theserendipity that used to come from these compendiais important, and I am pretty sure that the NationalStatistician also shares the Chairman of theAuthority’s slight discomfort at this decision. Now, itmay not be that physical copies are what is required,but I think we do need to look at whether eitherphysical or electronic compendia are a way forward.Of course, there is a new electronic publication, whichI hope Jil will talk about in a moment, called Life inthe UK, which has grown out of the well-being work,which has much of the sorts of information that SocialTrends used to have.Jil Matheson: Yes, the content of all of those thingsthat you referred to is still available, and one of thereasons that the big paper publications stopped wasbecause increasingly the content was available muchmore rapidly online. There was a delay in compilingSocial Trends or whatever it was every year, comparedwith much of that data now being available quarterlyor monthly on the website.

Q95 Charlie Elphicke: So are you saying that youpublish a PDF of Social Trends on the website or thatyou are planning to do that?Jil Matheson: No. Life in the UK is one of those onthe website, but there are other publications that bringtogether data. We could do more, and again this is anarea that is evolving and, as Andrew said, I am a bitsympathetic myself, being of that age, to the notion ofhaving something that is easy to carry around and easyto look at.

Q96 Charlie Elphicke: Yes, but I would say,broadly, discontinue the hard copy stuff anyway,because no one reads it and, to be honest, even if Ihad a copy, I would suspect it was out of date. I wouldjust go to the website. So you should not botherpublishing hard copy. Put Social Trends, RegionalTrends, Economic Trends, all these great publicationsthat you used to do, on the internet as PDFs, which ishow we used to access them anyway. I mean, whenwas the last time anyone saw a hard copy? Mr Dilnot,when did you last read a hard copy of Social Trends?You would look at it on the internet, presumably, ifwe are being honest.Andrew Dilnot: Well, I do look at it endlessly on theinternet, but I have to confess I went and had a lookat the 1991 Social Trends only last week.

Q97 Charlie Elphicke: Because it is not on theinternet probably.Andrew Dilnot: Probably. I have two hoods on myanorak; I am that committed to nerdishness.

Q98 Charlie Elphicke: But you get the point that Iand indeed many others are making. As Chair of theAuthority and National Statistician, will you both look

at this issue and see if you can draw together thesecompendia in a way in which anoraks like me canaccess most easily and understand?Andrew Dilnot: My sense is that one of the tasks thatthis new, one-month-old directorate within the ONShas—the analysis and dissemination directorate—isprecisely to think about how we can help usersthrough the astonishing array of data there is. We needto find ways of putting it together into differentpackages, and my guess is it may not even necessarilybe PDFs. Even that technology is moving on and itmay be other forms of web publication, but I amcertainly happy to say that we will look at how wecan meet the needs that were met for all those decadesby things like Social Trends.

Q99 Paul Flynn: Did I understand you to say thatyou have not seen yesterday’s session we had with theMinister and with the other witnesses?Andrew Dilnot: I have not. I have seen a briefsummary of it.

Q100 Paul Flynn: Why have you not seen it? It wasbroadcast to an attentive nation.Andrew Dilnot: It was, but I am afraid that yesterdayI was focusing on my other day job, in Oxford.

Q101 Paul Flynn: Well, had you watched it, youwould have seen an astonishing performance from theMinister, who seems to have a relationship with hiscivil servants and with reality generally that is verytenuous. Every question seemed to come as acomplete surprise to him. None of his civil servantsseem to be telling him what is going on in the areasof his responsibility. It is very well worth reading, andI would urge you to contact him and communicatewith him as quickly as possible. The other threewitnesses were great value and were, surprisingly,very critical of the communications. At yourpre-appointment hearing, which you referred to,where we were all greatly enthusiastic about yourappointment, we saw that as one of your greatstrengths. One of your colleagues in broadcasting onthe More or Less programme was saying that thereis no questioning the quality of the statistics and theprofessionalism of statisticians, but thecommunications are awful generally. I see you have anote in your report saying that you agree with this,and you convened a workshop on how a series of ONSstatistical releases could be better communicated tousers. In the light of that, the Authority established,on a pilot basis, a good practice team. You have beenthere almost a year now; should it not have gonebeyond the gleam in your eye? Communications arein such a poor state.Andrew Dilnot: I have been there since 2 April, so Ihave been there eight months and I think we havemade progress on all of this. For many years I havebeen critical of statistical communication by almostthe whole statistical community, of which the ONSand the GSS are a part, but there are other parts of thestatistical community. I would also want to emphasisethat while there are some very serious shortcomingsstill, I think there are also some signs of really goodstuff. While your evidence session was going on



yesterday, a second tranche of Census data wasreleased, and I thought the way in which that wascommunicated was extremely effective. We see in thepress this morning a number of, I think, very wellwritten stories that have demonstrated understanding.We saw on the ONS website yesterday some very,very good visualisations of data, which weresyndicated widely across the Guardian and BBC, forexample, so I think there are some good signs there.We have made, I think, considerable progress. Wehave set up a good practice team. I convened ameeting in June, which was followed up in July andthen again in October, to look at three specificstatistical bulletins: the GDP statistical bulletin, theretail sales statistical bulletin and the populationestimates. We thought that we drew those at random—well, not quite at random; we thought they were veryimportant—and on those I think we have made someconsiderable progress. The population statisticalbulletin we thought was pretty good. We thought therewas significant progress needed on the GDP and retailsales bulletins, and I think significant progress hasbeen made. I have brought them with me to show thekinds of things where we are making progress.But it is important to remember that the volume ofstatistical communication is enormous and we are notgoing to transform it overnight. As Chairman of theAuthority, I have two sorts of levers: I have theNational Statistician as the leader of the ONS and ofthe GSS and the work she can do to encourage them.I also have an assessment function. Richard Aldrittis the Head of Assessment, regularly assessing andproviding monitoring. So I think we are makingprogress. The good practice team that we set up inOctober is now working with, I think, ninedepartments. The ONS and many other departmentsare working with them on their releases.I think we are making progress, but there is muchmore progress to be made. It will not happenovernight, but I think we are moving forward. In thewritten evidence that I sent in to you I described sevenareas where I think we need particularly to focus.Some of those clearly came up, from the summary Isaw yesterday, and we need to be focusing much moreon the uncertainty surrounding estimates. We need tobe focusing much more on time series—on givingpeople the long context. We need to think aboutcontext. We need to think about using appropriatelanguage and not attributing meaning and causationwhen there is none. There is a long way to go, but Ido not feel as though there is no momentum. I thinkthat we have lots of signs of progress.Jil Matheson: I did hear and welcomed the commentsfrom the three colleagues that you heard yesterday,because this is really important and it is becomingmore and more important. I was also pleased to hearabout the quality of statistics, because that underpinseverything that we do and we must not sacrifice that.But it is undoubtedly true that the ability tocommunicate and explain what we do is even moreimportant than it has been in the past, and we have toimprove in lots of ways. We have talked about thewebsite. We have talked about the kinds of statisticalreleases and written communication, which I think isreally important, but it is interesting for me to have

heard that yesterday, because one of the things thatwe also often hear is: “Just give us the data. We donot want any of that.” So to have a strong voice sayingthat we do want clear understanding—because peopledo not have time and because they need to be able toreally understand the key messages and the trends inthe data, and that is what a professional statistician’sjob is—is really important. It is about writing, andAndrew has talked about what the good practice teamis doing, working with nine Departments. There is anappetite. There is something about skills andconfidence, and the confidence to talk as well as towrite—so being able to talk to the media, as theCensus team did very effectively yesterday.

Q102 Paul Flynn: Full Fact said yesterday that manyof the communications from you are poorly written,and one of the witnesses gave an example of Norway,where, in the equivalent of ONS there, they employjournalists. While the Census yesterday was wellcovered, virtually every paper played on their ownprejudices, whether they were going on immigrationor religion or whatever it was going on; it fitted withthe lines taken previously by the papers themselves. Ido not think we would have any criticism of the paperyou put in, but I am rather depressed by the fact thatwhen the website was reorganised and was launchedagain with trumpets and trombones, it turned out tobe a flop and more impenetrable than the previousone. One of the witnesses yesterday said it took himseven clicks to get the information he required. Onehopes that between the dream that you have, which issplendid, and the reality, the shadow does not fall. Areyou confident of that?Andrew Dilnot: This is a big task. The volume ofstatistical output is enormous, absolutely enormous,and this is an institution that is relatively stretchedfinancially. We spend about £3 per year per person inthis country on the Office for National Statistics.Chair: That is far more than the monarchy.Paul Flynn: Yes, but the Statistics Authority is auseful body.Andrew Dilnot: That is a debate into which I shall nottravel, but the GDP per head is between £25,000 and£30,000, so we are spending about one ten-thousandthof the resource we generate on collecting data that wethen use to allocate most of that. So we are relativelystretched. There is a huge amount to do. We areproducing many, many hundreds of bulletins all of thetime. We have to have a world where we make itdesirable for everybody doing this production to do itin the right way. Simply mandating it, I think, willfail. What we are finding is that there is realenthusiasm among statistical producers to do thismore effectively. They want to be well understood,they want to get their stories out, but I would befoolish to say to you that we could be sitting herethis time next year and it would not be possible forChris Giles or Michael Blastland or Will Moy to pickout a statistical news release and say, “Well, I do notlike that very much.” What I think we absolutely cancommit to you is that the proportion of all of theoutput that is good and effective, which is alreadygrowing, will have grown very substantially more bythen and that we will gradually drive up standards



across the board. As I say, just taking retail sales, ifwe look at the July release, we had statements of theform “the annual implied deflator”, which by Octoberhad become “the estimated price of goods sold in theretail sector”. In July, we talked about value andvolume; by October, we were talking about amountspent and quantity bought. We have moved fromstatements that meant nothing to the public tostatements that did. In July, we had charts that,frankly, when we sat down to look at them in a groupat the ONS, we could not work out the meaning of.By October, we had a chart that shows what hashappened to retail sales since the year 2000, so itshowed its growth and then, at the beginning of therecession in 2008, it became flat and stays flat.So we are making progress. There is more to do, butmy sense is that our colleagues in the ONS and theGSS do want to do this. We need to keep the pressureup, and it helps us and is encouraging for them thatthis Committee, representing Parliament and thewhole of the nation, is saying, “Come on, you need todo better. You have to do this. We care enough aboutthe data for you to need to communicate themeffectively.” I cannot promise that it will changeovernight, but we are making progress and itabsolutely is a high priority for us.

Q103 Paul Flynn: I will let you come back to thislater, but can I give you just one opportunity to answersomething that was suggested yesterday in a question?It was that those who work in the ONS might betempted to trim the figures in some way to please theirpolitical masters. Would you give me an emphaticdenial of that or an astonished confirmation?Andrew Dilnot: I have absolutely no evidence of that,and we have been through an assessment of all thestatistics—all national statistics. I do not think anyevidence of that came through. I think it particularlyunlikely that it would happen in the ONS, becausethere is no Minister responsible. I am responsible; Jilis responsible. There is not a Minister responsible.Francis Maude is not the Minister in charge of theONS. In the GSS you might think that there was morerisk, but every statistician in the GSS knows that ifthey were to be subjected to such pressure, they wouldgo immediately to Jil, and Jil could come to me, andI would get out the very large stick with rusty nails init and we would go all out on the attack, because themarvellous existence of the legislation that set theStatistics Authority up in 2008 makes it clear that suchbehaviour would be unacceptable.Jil Matheson: And the culture and the values of theONS absolutely prioritise integrity and professionalconduct and transparency about our methods. So Iwould be amazed, absolutely amazed, if there wereany foundation in that observation.Andrew Dilnot: But if there were any such evidenceand it was brought to your attention, we would act.We would also bring it formally to this Committee’sattention.Paul Flynn: Of course. Thank you very much.

Q104 Charlie Elphicke: I think you have brieflytouched on the issue of revisions. Taking the secondquarter GDP figures, how many revisions have there

been? They started at was it -0.7 or something andended up lower. Can you just explain that?Andrew Dilnot: I think off the top of my head itstarted at -0.7, closed at -0.4 or -0.5. I think there weretwo revisions. My recollection is that at the time ofinitial publication the uncertainty surrounding themwas emphasised because of the additional BankHolidays for the Jubilee, and so there was uncertaintyand a revision of 0.3 of 1%, so three one-thousandths.The process is that we publish the first release 25 daysafter the end of the period to which it relates, whichmeans that we have, I think, between 30% and 40%of the data that we will finally have when we makethe final estimate. Now, of course, it would be possibleto say we do not want an estimate as soon as 25 daysafter the end of the period, but our experience of users,including the users in the Bank and the Treasury, isthat there is an enormous demand for data as quicklyas possible. It is important for us to recognise theuncertainty around it, which is why the first of theseven points I make is that we must be clear aboutuncertainty, and you will note that the first line ofthe preliminary estimate first released in October was:“GDP in Q3 is estimated to have grown by 1%,”whereas in July we did not say “is estimated to” andinstead said “has”. We need to be clear about that.

Q105 Charlie Elphicke: Understood, but then wealso have the situation where the OBR come alongand say the depth of the recession in 2008 was worsethan we thought. That looks to me like a five-year-agorevision. What is going on there?Jil Matheson: There are two parts of this. One is thequarterly estimates, where there are two revisions: weput out the very first estimate each quarter, and thenrevise it a month later and again at the end of thequarter. That is a perfectly normal part of the processand, as Andrew says, the thing that is important, andit is part of communicating statistics, is to be clearabout what the levels of uncertainty are at each point.There is then another stage where we think about whatwe are trying to do when we measure something likeGDP, where we are taking data in a long time afterthe reference period. For example, corporation tax orwhatever it is that we are getting from HMRC andother sources comes a long time after the year, so werevise then.The other part of it is where there are methodologicalor structural changes. For example, that may be a newindustrial classification, which means that differentparts of the economy have different weights, and thoseget introduced often under EU or UN standards. Theback theories can be revised again when there arethose big methodological changes. All of that is anormal part of the process.The important point is that we explain very carefullyand clearly what the status is at any one point andthere is a revisions policy on the ONS website.Charlie Elphicke: If you can find it.Jil Matheson: I hope you can find it. It explains whatthat process is and, in particular, what the revisionsare at each stage in the process.Chair: We need to make faster progress.



Q106 Kelvin Hopkins: I have to say that I amsomething of a hard copy dinosaur. I at leastmetaphorically sleep with a copy of the House ofCommons Library’s Economic Statistics under mypillow. I use it all the time, and if I want anythingspecial I go and ask the Library to sort it out for me,so it is nice to have them there. But I am concernedabout the non-expert users. We have known for sometime now that more than 50% of the population do notunderstand what “50%” means. The understanding ofpercentages is particularly poor, and when I write mylocal newspaper articles, if it is “11%” I tend to put“one in nine people” in brackets afterwards, so that itis understood. I walk around with this—I am verymuch a nerd. That is national debt as a proportion ofGDP, which is just over a quarter of what it was in1945; it is not quite the crisis that some people makeout, but there we are.Chair: Audio-visual aids do not work in thisCommittee.Kelvin Hopkins: Well, I have used that frequentlyand people say, “Oh my gosh, how interesting.”Anyway, I am always concerned about the non-expertusers and the way that particularly the media and thepress distort things: they cut off vertical bar charts atthe knees so it exaggerates the vertical scale and soon. How are you working on that kind of visualsimplification, which is not unfair and not untrue butgives a good picture?Andrew Dilnot: Well, we think you are absolutelyright. The first principle of our Code of Practice isuser needs, and so in thinking about these statisticalbulletins, we have comprehensibility high up the list.One of the sets of things we have been talking aboutwith producers is making the numbers meaningful: ifit is sensible to describe the number per person, thendescribe it as per person, rather than using hundredsof millions or billions of pounds; if it is sensible toput some international comparisons in, do that. Weare thinking very hard about trying to make thingscomprehensible and avoid the use of jargon, so youwill also notice that the first line of the GDP releaseused to be: “The chained volume estimate has doneX.” We now say “GDP is estimated”. So we are tryingto think all the time now about avoiding jargon,making things come alive, making them relevant topeople. It is not easy. It is not at all easy. I have spentmuch of the last 35 years trying to do that, but I dothink we have some very good examples of that, andin some of the data visualisation that we see oninternal migration or some of the Census stuff, I thinkwe are finding good ways of using maps and figuresto make it come alive.Jil Matheson: Visualisation is an important part ofthat, and there is a fantastic team in ONS working onvisualisations, so you will see more and more of thatbecause that is one way that helps the non-expert user.Explanation is part of it. There is also another strand,which is about talking to the media, and so briefingsincreasingly play a part as does making statisticiansavailable to the media. We are also working withorganisations like the Royal Statistical Society; theirGetstats campaign is trying more broadly than wecould do to improve what they call the statisticalliteracy of the population—it is really important—so

that people have a sense of the kinds of questions tohave in their mind when somebody presents them withstatistical information. There is a broad range of effortand activity to try to improve both what we produceand the understanding of what we produce.

Q107 Kelvin Hopkins: Even the term GDP we usecasually, but for most people it does not mean verymuch, so if we talk about the total value of the outputof the country, it might mean something. Anyway,there is another question: do you think that moreneeds to be done to improve the reporting of theuncertainty associated with statistics, for examplethose from surveys? Some things are uncertain.Andrew Dilnot: Yes. I said earlier that is at the verytop of my list of things we need to do, because thereis uncertainty, which does not mean it is wrong. Thestatistic will be uncertain but it will still be muchbetter than not having any data at all, but if we helppeople to understand the uncertainty, that makes a bigdifference. To pick up one of the issues that I knowChris Giles raised yesterday—the trade statisticsrelease—if you look at what has happened to tradeover the last year, the monthly figures have just goneup and down. So there was not really very muchinformation in the fact that last month it went from2.6 to 3.5 or vice versa, and elsewhere in the releasethat is said, but it is the kind of thing we should sayright up at the front. So rather than saying, “There hasbeen this shift,” we should say, “Over the last year,there has been oscillation around what looks like atrend.” So yes, getting good ways of describinguncertainty is a crucial part of getting the meaning outof statistics and into people’s understanding.Jil Matheson: That is alongside the more technicalquality reports, which again are available, althoughoften not up front in the statistical releases that wewant to communicate with. But there are technicalreports associated with the main series, which talkabout things like the uncertainty from surveys andsampling and demonstrate what that is.Kelvin Hopkins: Harold Wilson always claimed, Ithink, that he lost the 1970 election because of fourjumbo jets in one month’s trade figures.Jil Matheson: Explaining that is exactly what Imean, yes.

Q108 Kelvin Hopkins: And moving averages ratherthan monthly figures—those sorts of things. Anotherpoint here: unfortunately, I was away yesterday, andit sounded like an interesting session the Committeehad, which I missed—great fun. We talked about pressreleases being misleading and concentrating, as yousay, on the immediate and short-term data changes;they may be true but not fair, and it is trying to getfairness into the picture.Andrew Dilnot: Again, I think I have talked in thisCommittee about how, when we publish GDPreleases, GDP numbers, the important thing is to setit in the context of the last 10 or even 60 years. Ofcourse, what the City will often be interested in issimply the news—the extra bit of information.

Q109 Chair: This is a very important point, becauseChris Giles was very clear that if the press release



goes out saying that the trade figures changed againstlast month, most of the press will report it as such,but it does not mean anything. True but not fair is avery, very powerful point, I think. He made anotherpoint: pandering to the worst elements ofjournalism—that the headline writers for the ONSpress releases are simply thinking about what willcatch the eye of the media rather than thinking, “Whatis the real headline of this set of numbers we are justabout to publish?” As Mr Flynn suggested, should younot be employing sub-editors to make sure that yourpress releases are, in fact, the news and not just aneye-catching headline?Andrew Dilnot: Well, we certainly should not belooking for eye-catching headlines. My strong view isthat it is not the task of the ONS to seek to maximisecoverage. The task of the ONS is to seek to informthe whole public with the right story about what thestatistics tell us. I am not myself absolutely sure aboutemploying journalists. I am more persuaded, and thisis certainly the view I always took at the IFS, that weshould make sure that our analysts, our statisticians,can write properly. So we need to employjournalistic skills.Chair: Well, train them to be sub-editors.Andrew Dilnot: I think there are often risks inemploying journalists, and I think sometimes whenthings have gone wrong it has been because we havebroken the connection between the statisticians andthe people writing the releases.Chair: Ms Matheson, you are smiling.Jil Matheson: Well, as Andrew knows, I have a visionof a world where ONS does not have separate newsreleases but very clear, well written, well explained,visual statistical releases, which is, in itself, the newsrelease, so you do not have this kind of disjuncturebetween the two. Now, it is going to take a time toget there, but the point about not having misleadingheadlines is absolutely right. We must not do that, butthe way to achieve that I think is to get the statisticalbulletin, the statistical release, where we want it to be.

Q110 Kelvin Hopkins: It is important to have a veryprominent health warning with those sorts ofimmediate statistics. Otherwise you get the situationwhere Paxman has a go at some poor politician andthen they come back and say, “Yes, but you arelooking at one month’s statistics. They are uncertain;it is unfair to look at them.” It looks like an excuse.Andrew Dilnot: It is important to have a healthwarning, but I think we can be more positive. One ofthe things that I know came up yesterday, which wehave been running for the last few months, is thatthere should always be a time series chart in the pressrelease. In the statistical releases that we have workedon you will see that they now all have a reasonablylong time series, which allows you to see what isgoing on—to see whether the latest bit of data tellsyou something new. So, yes, there needs to be a healthwarning, but I think we can be more positive aboutwhat we have got to say at the same time as gettingthe health warning in.

Q111 Chair: Moving on, it has been suggested thatthe National Statistician should be more ready to be a

public figure and answer questions on an importantset of numbers. You should be up there on Sky andBBC News 24 explaining what these statistics mean,rather in the way Robert Chote does for the OBR.What do you think about that, Ms Matheson?Jil Matheson: I have a couple of thoughts on that.One is my job is very different from Robert Chote’s.I do not want to talk about what his role is. I think itis important that we do have people on Sky or on BBCNews 24 or wherever explaining the statistics that weproduce. That does not have to be me, given the rangeof statistical outputs from across the whole of theGovernment Statistical Service.

Q112 Chair: So when you issue the quarterly tradestatistics, you say, “So-and-so and so-and-so areavailable for media interviews.”Jil Matheson: Absolutely—absolutely we do.

Q113 Chair: Why do you think they are not taken upvery often?Jil Matheson: They often are. One of the things thatis noticeable is the increase in the number of peoplefrom ONS, in particular, who are talking about theirstatistics. The GDP numbers regularly are broadcastand then interviews occur. The Census yesterday wasa perfect example, where we had a team of peoplewho were doing lots and lots of media interviews. Imyself did so when we released the Census resultsearlier in the summer. So it is not just about oneperson doing it. It is about having, increasingly, a teamof skilled and confident people who are able to talkabout their statistics, and I think there has been a bigchange in that. Much more of that happens than everdid in the past, and it is good to see.Andrew Dilnot: Perhaps we might get some data onthat. Since we are statisticians, why do we not getsome data on that and we will send in to theCommittee?Jil Matheson: Yes.

Q114 Chair: I am just looking at your press releasetoday on labour market statistics. You do not have aheadline on it.Andrew Dilnot: That may well be a sensible thing. Ido not know. I never know any numbers until theycome out.

Q115 Greg Mulholland: We had a number of piecesof written evidence and also an interesting discussionyesterday about the format of statistics—how they arepresented—not in terms of what we have just beendiscussing in terms of headlines, etc, but the overallpicture. Some of the evidence suggested that is notquite good enough in painting a coherent overallpicture that would then help in guiding policy. So howwell do you think the formats currently used forreleasing official statistics are meeting the needs ofusers?Andrew Dilnot: I think it varies. I think we have somepieces that are doing a very good job. I think oftenthe bits of output that come out once a year do a goodjob, because there has been lots of time to think. Ithink we have been less effective in the area of someof the data that come out once a month, and that is



where we are putting most effort in. One canunderstand why it is tricky, because the statisticiansthemselves are under enormous pressure to producethis stuff every month and then they do not have verylong to craft the output at the end, and that can tendto make it very formulaic. It can tend to grow year byyear as new things get added in. What we are trying,and I think having some success in, is absolutelyseeing the statistical bulletin as the core of what wedo and then being able to spin off shorter bits of it onto landing pages on the website, social media. So Ithink we are making progress there. We are putting agreat deal of effort into ensuring that the statisticiansunderstand that they need to tell the story. The leadstatistician needs to say to herself, “What is it thatthese data most powerfully say?” There may not be aheadline, but there needs to be a story, a context, thatwill help people to understand.So, I think there is still a considerable way to go, avery considerable way to go, but I think we aremaking progress. As I say, I think it tends to be themore frequent releases where there is most work tobe done.Jil Matheson: I agree with that, and you will gatherwe are both passionate about improving this.However, one thing I would say about even themonthly releases is that there are users out there whoreally like them, because they are familiar. When wechange things we also get people saying, “Please donot change it, because we know exactly where to look.We know exactly what the format is and what the styleis.” Even in making change there is a process, andoften a process of consultation or seeking feedback. Itis very much a process. It is taking a whole range oftypes of users with us as we develop these things.

Q116 Greg Mulholland: Just to give an example, theRoyal Statistical Society said, “A deepercommunications challenge for the official statisticsservice is to present a coherent statistical picture ofwhat is going on in areas where debate needs toconcentrate on the issues rather than on explainingparticular statistics. The debate on Scottishindependence is an example where statistics needs tobe brought together and well communicated in orderto foster good debate,” and not only debate, of course,but also good policymaking, which is, in the end, whatstatistics primarily should be there for. Have you anyideas on further publications that you might produceto inform major events or big public policy decisionsor a debate like the referendum on Scottishindependence?Andrew Dilnot: These are interesting questions. Weare all here today because we love statistics, and thereason we love statistics is not just for their own sakebut because we believe passionately that good policy,whether it is private policy of individuals or publicpolicy of governments, can only be made if we havethe right data. If there are gaps, we need to find anew way of making that work. I think some of thosecompendia used to fulfil that role, which was one ofthe points Mr Elphicke was making about compendiapublication.The question of data for the Scottish referendumdebate is an interesting one. It raises all the issues

about comparability of data across the four nationswhere we have an explicit role set out in the statute.It also raises some quite tricky questions of analysis.I think it is perfectly reasonable for us, as anAuthority, to think about the needs for that debate andwhether they are currently being met appropriately.But you were using the Scottish independence debateas an example.

Q117 Greg Mulholland: It was the Royal StatisticalSociety who used it.Andrew Dilnot: Yes. There is a wider question abouthow, when there are big policy questions, we assesswhether the statistical material is appropriate. It seemsto me that is a task for Jil and her colleagues—but itis also a task for my absent colleague Richard Alldritt,as the Head of Assessment, and for David Rhind, theDeputy Chairman responsible for official statistics—with their monitoring hats on to have periodicexaminations of whether we have the right kind ofdata. Recently, that side of our work produced amonitoring review on health statistics, askingquestions about whether the array of health statisticswas coherent and really answered the kinds of policyquestions that need to be answered. I think that issomething we need to go on and possibly do more of.Jil Matheson: I have a couple of points to add. Thefirst is this is about relevance, and making sure thatthe statistics we produce are not only of goodtechnical quality but are relevant is again fundamentalto what we are trying to do. There are some examplesof where the statistical service has been able to bevery responsive to a policy or political debate. Theexample that comes to my mind is the riots last year,where the statisticians in the Ministry of Justice,working with other colleagues, were able to veryrapidly produce some excellent statistical analysis andreports in the public domain that helped shape thedebate. That was a really good example, which I holdup because that is the kind of thing that I would liketo see more of.Within ONS, I think earlier I was talking about someof the restructuring that we have done, and part of thenew analysis and dissemination directorate in ONS isa division called the public policy division, whose aimis explicitly to be sensitive to the wider public debateand what the issues are and how statistics can bepresented to inform that debate. There are of course,as Andrew referred to earlier, going to be questions ofpriority and capacity. There are lots and lots of thingsthat we would like to do and we cannot do them all,so there will be a process of discussing with theAuthority and others what the priorities are.On the comparability point, I think our role is to makesure that we have a bedrock of comparable statisticsthat can be used for a variety of purposes. Forexample, coherent statistics across the UK is one ofthe issues that certainly the Statistics Authority hasraised with the Government Statistical Service. We arelooking at it. I have to say it is not something that wehear about very often from users. It is not somethingthat I get representation on or that I know mycolleagues in Scotland, Wales and Northern Irelandget much representation on. It is part of our remit toexplain and particularly to understand where there are



differences between Scotland and England and Wales,for example, and we have a group working on that atthe moment so that that is more helpful to users.

Q118 Chair: Can I just interject at that point? Thisis obviously a huge existential debate for our countryabout whether Scotland will leave theUnited Kingdom or not, and the statistical analysisaround that debate will be hugely important. Whatexactly are you planning to do in order to make surethis debate is properly informed? Should there not bea compendium of statistics that are producedespecially to support this debate, where you canhighlight what is indisputable, what is disputable orwhat might be a matter for negotiation of a separationsettlement, so that we can engage in the debate on acommon understanding of what the statistical basis is?Andrew Dilnot: I think my response to that is that itis extremely helpful for us to have that user need soclearly spelt out and, in the light of that, I think it isperfectly reasonable for me, as the Chairman, and Jil,as the National Statistician, to go away and thinkabout exactly what it is that we might seek to do thatwould fulfil that need.

Q119 Chair: You might do the same for the EUquestion as well, though I hesitate to start thatargument in this meeting.Jil Matheson: This is why getting the bedrock rightand getting some clarity about the comparability isreally important, so that we are not just trying torespond but we have the infrastructure right.

Q120 Chair: Are you free to do this? Nobody canstop you doing it?Andrew Dilnot: Let us leave the EU question to oneside.

Q121 Chair: Can you do this on your own initiative?Andrew Dilnot: If we have the resources, then I thinkin principle we can.Jil Matheson: I do not think there is anything thatwould stop us doing it impartially and bringing ittogether based on the statistics that we already haveor the data that we already have.

Q122 Chair: Do you anticipate Ministers leaning onyou not to publish controversial series of statistics?Andrew Dilnot: Jil has been doing this much longerthan I have. I have only been here for eight months.No Minister has ever attempted to lean on me in anyway.Chair: Right, well, we look forward to it.

Q123 Kelvin Hopkins: The Chairman touched onresources, and I am concerned that you are trying towork within fairly difficult financial constraints. Ipersonally believe that we should spend more, giveyou more resources, and help to reflate the economya bit maybe as well in the process. It is a worry thatyou are constrained in doing more. Is that a worry?Am I right?Andrew Dilnot: Yes. We face limited resources.Everybody faces limited resources, but this is an areawhere people very rarely want to stop doing

something. One of the challenges that we absolutelyface—and it is a strategic challenge for us—iswhether, with the resources that are going to beavailable to us, we can go on doing everything that wehave done in the past and have some slack available todo the kind of work that the Chairman has justencouraged us to do.

Q124 Chair: If there is to be a referendum onScotland leaving the UK, surely you should say, “Partof the cost of this referendum is going to be producingthe statistics, and you need to reckon that that is partof the cost of what you have decided to do,Prime Minister,” and get the extra money for it.Andrew Dilnot: Well, it is certainly the case that weneed to have a range of discussions about our fundingin the medium term that reflect the way in which thethings that we are doing are very important. It is notjust this; there are discussions about the financialcrisis and whether we need additional sorts of data tohelp us understand financial flows, for example. Ithink there may be a strong case for that, but that isalso not cheap, and if that is important for the Bankand the Treasury, then we may well need some helpwith that.Kelvin Hopkins: I can see you can go into politicaldeep waters. It might be helping the Prime Minister,but not the First Minister in Scotland, and that mightbe a real controversy. On the European thing,Ministers time and again talk about our exports to therest of Europe. They very rarely talk about theimports, which are far larger.Chair: Kelvin, it pains me to stop you, but we are notgoing there. Shall we move on?

Q125 Paul Flynn: Ad hoc requests: I was surprisedto see that you have 13,000 of these requests.Yesterday, we heard some complaints, understandably,from journalists, who want the informationimmediately for their deadlines. There wasunhappiness about the fact that they can never speakto the person involved who produced the figures.There was no contact number often on the pressreleases and they were irritated by that. They werealso suggesting that many of the ad hoc requests wereturned down for reasons that did not appear to berational and consistent. Many of them were turnedinto Freedom of Information requests, which took along time to do. Are there improvements you thinkare necessary on this?Andrew Dilnot: I will turn to Jil for confirmation onthis, but I am pretty sure there is a telephone numberon all press releases and all statistical releases.Paul Flynn: The press releases are not from theperson involved in producing the figures. I think theythought that the statisticians who were mainlyresponsible were too remote.Andrew Dilnot: I think one of the very first acts thatI took was to agree that we would publish much moreof the outcomes of these ad hoc requests than we usedto. I think we published 700 or so. That seems to mevery good, and every couple of weeks I go and havea look at what is there and there is often somethingvery interesting to me. Of course people want moreaccess and more responses. The best thing would be



for our website and our data release to be such thatthe number of ad hoc requests fell because people haddirect access themselves, and more and more that isthe case. More and more of our data are beingpublished in an accessible form, in an electronicallyaccessible form, and we want to go further downthat road.Jil Matheson: There may be an ONS versus otherdepartments issue here, but I am proud of the fact thatour statisticians who produce statistics are available,and we do give names on statistical releases. It doesnot mean that they are necessarily there waiting for aphone call if somebody has a very short deadline, andso it is simply not going to always be possible to turnround requests very quickly. That is the reality, but wewould try to meet requests and we would try, asAndrew said, to make it known what requests havealready been made, because it might save us doingsome work twice as well as being helpful for users.

Q126 Paul Flynn: If you had an ad hoc request froman organisation like Republic, for instance, on thesubject raised by the Chairman—the preposterousclaim put out by the propaganda machine in the palaceabout the cost of the royal family—do you think youshould include in it the cost of the Jubilee last year,with this wet procession down the Thames and thecost of security in that? I know minor royals come tomy constituency and my constituents mass nil deepon the pavement, but the cost in security can be£20,000 to £30,000. Do you think that you have a jobwith your stick with a rusty nail in it to make sure,when you get some figure that is clearly dodgy,propagandist and where someone has an interest inminimising the cost, to come out and tell Republicwhat the total true cost is, particularly of the BankHoliday for the Jubilee last year?Andrew Dilnot: I have no desire to get into thatparticular matter, but on the wider question of whenwe should intervene, we should intervene when amatter of serious public interest is going on, and thatis a judgment that we need to take. But we cannotpossibly get involved in all such debates.

Q127 Paul Flynn: If you had an ad hoc request fromRepublic to do that, what would you do?Andrew Dilnot: The ad hoc request would go to theONS, and they would work out whether it seemed tobe something that they could do or that had a highpriority.Jil Matheson: An ad hoc request would be a requestfor analysis of data that we already had. If we do nothave the data, then we cannot do it.

Q128 Paul Flynn: This might be an appropriate timeto raise this: under Michael Scholar and thepioneering work he did, he was very severe, I think,at times—entirely justified—on the Home Office fora number of very dodgy claims that they made, andthere were half a dozen times that he wrote to them. Ithink you have written two letters, as far as I know,both to the Prime Minister, about two matters. One inparticular was the cost of the Health Service. Wouldyou like to tell us about that and what your approachis going to be on this in making sure that we do not

have self-serving distortions of statistics by thePrime Minister and other Ministers?Andrew Dilnot: I have written many more than twoletters, and only one of them has been to thePrime Minister. I think I have intervened on six orseven matters so far. Three have been reasonably highprofile. One was the discussion over what wasinterpreted in the press as being pre-release ofstatistics by the Prime Minister over GDP inPrime Minister’s Questions. One was in response to acomplaint from Andy Burnham about descriptions ofhealth statistics. Another was a discussion abouteducation rankings and what could be read into thedifferences in number from 2000 to 2009.The way in which we tackle this is that, if matters areeither drawn to our attention or we notice what wethink are serious questions where the public interest ismaterially involved, we will investigate them, and ifwe think that things have not been as they should be,then I will say so and say so very loudly. I am seekingin that to follow the policy and practice thatSir Michael developed and I think in the first seven oreight months that is exactly what we have been doing.

Q129 Paul Flynn: Are you happy with the situationas far as the Health Service statistics are concerned?Are they being fairly presented?Andrew Dilnot: Well, my letter, which was a verycareful letter, laid out our analysis of the data, and myunderstanding is that that description of the data hasnot been challenged; those figures drew on nationalstatistics published by the Treasury. So I am contentthat we have put into the public domain an analysisof the figures for the real levels of public spending inthe years 2009–10, 2010–11, and 2011–12. So, yes.

Q130 Paul Flynn: Did the Prime Minister accept inhis reply that he left a misleading impression?Andrew Dilnot: I did not write to the Prime Ministerabout health statistics. I wrote to the Secretary of Statefor Health, and towards the end of his reply, which bythat stage I think I already knew, he accepted therewere some questions around definition and that“consequently it would be helpful if we wereconsistent in the definitions we used. TheConservative Party website previously stated, ‘Wehave increased the NHS budget in real terms in eachof the last two years.’ While this again is correct—the NHS budget has increased in both 2011–12 and2012–13—it would be better if it were consistent withwhat I and the Prime Minister previously stated.Consequently, I have asked the Conservative Party toamend its website accordingly.”

Q131 Chair: Are you happy?Andrew Dilnot: I am content that the ConservativeParty website has been changed in response to theletter that we sent.Chair: Because if at any stage you were unhappy withthe responses you were getting from the Government,obviously we would like to know about that.

Q132 Paul Flynn: There is a book advising MPs onhow to ask questions in the House, and there is a bitafter the bit on how to ask questions, on how to



question the answers that follow, because most of theparliamentary answers we have are as nebulous aswhat you just read to us, I think. The whole point,which you wrote about very eloquently while the Billwas going through the House, of setting up thestatistics body was to increase public trust in statistics,and we should be in a position now where there ismore confidence in statistics than there was 10 yearsago and 20 years ago. Are we heading in that directionnow? Does your relationship with the presentGovernment encourage you to think that there isprogress being made?Andrew Dilnot: We should be very clear that it is notsimply Government political spokesmen who are atrisk; opposition spokesmen are at exactly the samerisk. My sense is that both Government and oppositionpoliticians are aware that, if they step out of linestatistically, there is the risk that it will be drawn toour attention and that we will be critical. Then, I hope,indeed, that if we were critical, there is a chance thatthey would also face a discussion with thisCommittee. So I think there is an awareness that thereare risks associated with statistical sleight of hand thatperhaps were not there in the past, and my sense isthat Ministers, senior civil servants and oppositionpoliticians are aware of that and so are looking totread carefully. Of course, no Minister or senior civilservant likes to be in a position where there is implicitcriticism, but I have not felt that this is deemed to bein some sense unfair in my discussions with Ministers.

Q133 Chair: Can I just press you on the DWP issueand their ad hoc release, which they maintained wasresearch and therefore was not covered by the Code?How was that resolved? Has that been resolved toyour satisfaction?Andrew Dilnot: I think this was one that was at thevery end of Sir Michael Scholar’s time as chairman.My understanding is that it has been pretty muchresolved to our satisfaction. There is an acceptance inthe Department that they will think very carefullyabout whether this should be thought of asmanagement information or official statistics, andhave moved in that direction. They have nowpublished, last month, some data that have beenpublished as an official statistic and therefore issubject to the Code of Practice and is legitimate forus to assess. So that does seem to have been areasonable outcome.

Q134 Chair: Ms Matheson, you said earlier thatstatisticians in Government Departments have thisindependence. Were there any conversations betweenyou or the GSS and the Department about that releasebefore it went out, or did anybody in the Departmentexpress any anxiety about that original release goingout? How does it work?Jil Matheson: Sorry, which release?Chair: The DWP statistics about the number ofpeople on out-of-work benefits who were immigrants.Jil Matheson: How it works is that if there is aconcern from the head of profession in a Department,then he or she would contact either me or my office.

Q135 Chair: So was there a complaint in that case?

Jil Matheson: No, no, I do not remember—

Q136 Chair: Should there have been?Jil Matheson: I am struggling with the detail on thisparticular one.

Q137 Chair: My question is: how does this happen?For the health statistics one, the website is not underthe direct control of the Secretary of State. This wasall in Government. There clearly has been a failure inthe Statistical Service not to have flagged that inadvance.Jil Matheson: To have flagged what?Chair: To flag to a Minister that this ad hoc releaseof research information might fall foul of the Code. Iam just asking, technically, how does this work? Howdo we make the people, the statisticians inGovernment, stronger and more confident and able tospeak truth unto power, so that we do not finish upwith the Chair of the Statistics Authority having towrite a letter to the Secretary of State?Jil Matheson: The intervention of the StatisticsAuthority is very important in sending the signalwithin Departments that this is something that will betaken seriously and that they have a role in alertingMinisters, special advisers and senior officials thatthere is potentially a Code compliance issue. That iswhat normally happens, and the fact of theinterventions builds up this case law of how thathappens.

Q138 Chair: If you do not have the detail to hand,could you send us a note of what the chronology ofevents was? Did this go out unchallenged from withinthe Department or was somebody not consulted withinthe Department who should have been consulted?Because clearly there was some failure, so that whenthe letter arrived from the UKSA Chair, it came as agreat surprise to the Minister. It seems to me that it isabout strengthening the confidence of the Service sothese accidents do not happen. Agreed?Jil Matheson: I will do a note, because I do not recallthe detail.

Q139 Chair: Thank you very much. Can I ask aboutuser engagement, because there is a very, very strongsense we got from yesterday’s panel that there is amountain to climb for users outside the Service, thatthey are not engaged, they are not listened to.Listening to Jil Matheson today, I hear a very strongwill to make changes, but is this a new initiative tomake changes or have you been trying to makechanges since you were appointed in 2009? There isa strong sense amongst users that, however muchchange you say you want, it is taking a very long timein coming.Jil Matheson: It is taking time.Chair: What are the impediments?Jil Matheson: Well, let me just say what we havebeen doing.Chair: Well, I would rather know what theimpediments are. By all means give us the good news.Jil Matheson: I think it would be helpful if Iexplained what we are doing and why we are doing



it, because that is aimed at addressing some of theimpediments.Chair: As briefly as you can, but we would like tounderstand the difficulties you are facing.Jil Matheson: Okay. Part of the difficulty is inidentifying who those users are. There are some very,very strong relationships with some users that havebeen built up over years. It is very difficult to reachout to new users, which is what we need to do, orpotential users even.

Q140 Chair: That suggests to me that you are ratherfed up with some of the sort of terrorists who keepturning up on your doorstep and being ultra criticaland complaining that things are not working, and youwould like to engage with some new people who aregoing to be a bit nicer to you. Is that unfair?Jil Matheson: I think that is unfair, yes.

Q141 Chair: Statistics is a very small world and Iam getting to understand the very, very bigfrustrations from very well informed users. It seemsto me that, if you cannot engage those very wellinformed users and get them to feel that they areinvolved with improving your business, engaging witha wider set of users is not going to resolve theproblem.Jil Matheson: Again, I think that is not entirely fair,from my perspective, in that there are some very wellengaged users who I am sure do feel as though theyhave a say in what happens. They may not always likethe outcome, of course, and that is partly becausethere is such a wide range of users, and one of thechallenges is to understand that wide range of use andthe kinds of decisions that people are making basedon statistics. One of the challenges is to hear thatbreadth of perspective, not just from, I willparaphrase, the usual suspects.Part of the way of doing that is, of course, talking, butthere is a limited number of conversations that youcan have. Part of it is through a new web-basedsystem called StatsUserNet, where there arecommunities of interest developing, so that we canagain try to get the breadth of views. Some of it isabout established user groups, where there areestablished user groups, and some of it is about moretraditional kinds of consultation. Certainly one of thethings that I hear from users is that there is much moreconsultation—in fact, some of them say too much—going on than there ever has been in the past, and sopart of the problem for some users is trying to respondto the plethora that comes from across the StatisticalService on users. At the moment, my office is workingwith the Statistics Users Forum on a set of guidanceon user engagement. We are doing that together to tryto improve the way in which we all engage with thebreadth of user communities. But even they wouldadmit in the discussions that I personally have hadwith them that they find it difficult sometimes toengage with their communities, because there is nosingle community. The former chair of the StatisticsUsers Forum is sitting next to me and he might liketo add to this.Chair: Well, great things are expected of him.

Andrew Dilnot: Being the chairman of the StatisticsUsers Forum was difficult. We found it quite difficultto get people to engage, for all the reasons that Jil hasdescribed. But I think, turning it back on ourselves atUKSA and the ONS, the most effective way ofencouraging people to engage with us is showing thatit makes a difference. What we have been engagedwith for probably longer than I have been around istrying to make it clear that it does make a difference.If we can show we are responding to the needs ofusers, particularly on the communications side, then Ithink it is more likely that users will want to engagewith us, because I think otherwise there is a tendencyfor people to think, “What is the point?” I hope thatsome of the steps that we are taking will succeed inmaking that work and we have to take it seriously. Itis, after all, the first principle in our Code of Practice.

Q142 Chair: Right. Now can you answer the otherbit of my question: what are the impediments that arepreventing you from changing more quickly, moreresponsively?Jil Matheson: The impediments are some of thosevery well established relationships, particularly withinGovernment. This notion of thinking about statisticsfor a public good is a relatively recent one. It onlyreally was enshrined in the legislation that we weretalking about.

Q143 Chair: So what is the recommendation that weneed to make in our report that will relieve you ofthese impediments?Jil Matheson: Well, the crucial thing is the one I wasgiving earlier: a very strong message that producingand communicating statistics for a wide range ofusers, not just for your traditional policy colleague inyour Department or the traditional relationships thatwe have had, is an intrinsic and important part of theStatistical Service.

Q144 Chair: So you need to spend more money onengaging your users and communicating yourstatistics to the wider public.Jil Matheson: There is certainly a resource on thecommunications side. There are certainly, as I wassaying earlier, skills, training and culture changes aswell as the technology changes that are needed, andthat takes resource and it takes time.

Q145 Chair: So, technological restraints, skillsrestraints, resource constraints—these are allpreventing you from changing as quickly as you wish.Jil Matheson: It will take time.Chair: I see Mr Dilnot nodding.Andrew Dilnot: Yes, because at the moment the usersthat it is not at all difficult to engage with are the bigGovernment Departments.

Q146 Chair: Indeed, we get representations thatprivate sector users of statistics are given much lowerpriority than the public sector.Andrew Dilnot: Yes, although there I think we have arelatively recent real success story. We had a businessusers engagement event two or three weeks ago that Ispoke to, and 100 people came along; it was huge. I



do not know what it was about doing it this time thatmeant that it worked, but we had a huge array ofpeople come along who were really actively using theONS and GSS output in their businesses. We had areally exciting day and lots came out of it. I wish Icould bottle whatever it was that made that sosuccessful.Chair: I suggest that people are beginning to feel thatit is worth coming and making representations to you,which is a jolly good thing.

Q147 Kelvin Hopkins: Just a quick question onrecruitment. Do you have a problem recruiting staffwith the sorts of skills that are required, and is it thecase, yet again, that the City is sucking up masses ofpeople with mathematical ability, which makesengineering suffer as well: we have to importengineers because we cannot produce enough, as theyall go and work in the City.Jil Matheson: There are recruitment difficulties,interestingly; although, of course, for ONS, it is notso much the City, because we have so few staff inLondon now. But because of the recruitment ban,there is now a backlog of vacancies that we are tryingto fill across the GSS, not just in ONS. We have hada couple of recruitment campaigns; they have notnecessarily been as successful as we would like, andwe are talking about how we can do better.

Q148 Chair: What is your analysis of why they havenot been successful?Jil Matheson: I think there are several things. Theban on advertising has not helped.

Q149 Chair: But surely statistics people all work inquite a small world. There must be a website forstatisticians looking for jobs.Jil Matheson: Not if you are recruiting directly fromuniversities, which we are. These are not necessarilypeople we know. These are new graduates and noteven statisticians: numerate social scientists,mathematicians; there is a whole range of skills thatwe are trying to recruit to, so it has been difficult. Wehave upped the recruitment effort. There are stillvacancies.

Q150 Chair: So you need to improve what we usedto call the “milk round”.Jil Matheson: Exactly. The milk round is part of it.Getting people out to the milk round, and getting thesejobs known among careers advisers.

Q151 Chair: Is this not something that all yoursenior management should be involved in?Jil Matheson: Yes, and it is not just the GSS. I havebeen having these conversations with the otheranalytical professions within Government, who arefacing the same kinds of problems.

Q152 Chair: Are the salaries an issue?Jil Matheson: I am not aware that salaries per se arethe issue.

Q153 Chair: Have you done some analysis andresearch into that?

Jil Matheson: Yes. Certainly for ONS, I am not awarethat salaries are an issue.

Q154 Chair: Do people like going and working inNewport? Is that a constraint?Jil Matheson: The staff engagement scores for ONSwere improved this year, which suggests that peopledo like working in Newport.

Q155 Chair: But is it easy to recruit top graduates togo and work in Newport?Jil Matheson: I do not think the site is the issue. I donot think location is the issue. It is getting to the rightpeople and recruiting them.I think the other part of it is that I do have, along withothers, some concerns about the supply of numerategraduates and whether the universities are training anddeveloping enough of the kinds of people to fill theroles of the future. It is not just the City that we willbe competing with. It is people like Google and theinformation sector as a whole.

Q156 Chair: Come to the University of Essex. Weproduce lots of social scientists and statisticians.Paul Flynn: But they have an inferior standard ofparliamentary representative.Andrew Dilnot: The only other thing I would want tosay on that is, seriously, people need to realise howimportant and glamorous it is. The whole statisticalcommunity has a problem of being a little bit too shyand not explaining to the world quite how importantit is. We need young people to realise that it isincredibly exciting and astonishingly important, andthat is a big and long-term task.

Q157 Chair: But that is a big PR task, is it not, forthe Statistical Service and for UKSA?Andrew Dilnot: Yes, and a great way of starting thatwould be that people got used to thinking that theONS website was a fabulous place to go.

Q158 Kelvin Hopkins: Going back to the theme Iwas talking about earlier, about appreciation by peoplewho are not necessarily statisticians themselves—andthis goes for quite a lot of graduates and people whostudied non-numerate subjects and so on—do youthink that statistics producers should do more to helpeducate people about statistics? You talk about peoplewith an enthusiasm for it and I feel the same, but otherpeople may not.Andrew Dilnot: Yes, but I do not think we should doit in a dry way. I think we should do it by helpingpeople to understand why it is important. I do notthink we should try to educate people about what astandard error is; I think we should help them torealise that of course estimates vary and of coursethere is uncertainty. I think we can do the educationthrough having really good statistical bulletins, but Ido not think it needs to be obvious to people that theyare being educated statistically when we are helpingthem to understand things. We do not need to explainregression to the mean. We just need to show peoplewhat is going on. So, yes, we have a role, but I thinkour role, by and large, is through explaining clearlyand then the stuff should come alive.



Q159 Kelvin Hopkins: Even in primary schools, justplaying games with statistics, and there is secondaryeducation, obviously. I am involved with NationalNumeracy, one of these new organisations, whichrecently had a parliamentary reception. Sadly, theyhave found that numeracy has actually declined overthe last 30 years or so—they did a comparison test—which is very worrying indeed. Numeracy is a realproblem for Britain, I think. Is that your impression?Andrew Dilnot: I do not have a professionalunderstanding of whether it has got better or worse,but what I do have a professional understanding of iswhether it is high or low. It is also worth saying thatthe media have to work alongside us in this. Themedia will be the medium in general through whichmuch of our work gets out, and we need to make surethat what we are producing helps them and recognisesthat many people in the media feel very anxious andinsecure about all of this and so are looking for us togive them material that they can work with. I thinkthat is a crucial part of what we do.

Q160 Kelvin Hopkins: Many people in general feelnervous and insecure because they do not feel at easewith numbers and statistics.Jil Matheson: As a small example of one other thingthat we can do, during the Census we provided lotsand lots of material for schools to help theCensusAtSchool, which was really a way of gettingpeople to think about their area, about how data arecollected, how you present those data, what they looklike. Again it was not teaching the mathematics of itbut teaching the understanding, and that was very,very widely taken up.

Q161 Paul Flynn: On your relationship with outsidebodies, I complained to the Advertising StandardsAuthority in 1988 about what I thought was a lyingadvertisement for personal pensions. They rejected mycomplaint, but a few years later 6 million people werecompensated because they were mis-sold personalpensions partly as a result of those adverts. The otherorganisation we have in mind and there werecomplaints about is the Press ComplaintsCommission, this remarkable body that I think waschaired possibly at the time by the intellectual giantPaul Dacre. You, I believe, approached the Authorityto complain about the Daily Mail’s coverage of theprosecution of offences relating to the 2011 riots. ThePress Complaints Commission decided that you werewrong, their statistics were better and they rejectedyour complaint. Clearly, this body is going to beeuthanised and buried in a deep hole with a concreteslab on top soon, but do you think it is worthwhilewhen the new reformed body does emergepost-Leveson that you engage with them, possibly atan early stage, to make sure that they do not get awaywith this? If you have decided that their statistics arewrong, they really have very little right to questionyou on the basis of the statistics alone and purely outof self-interest. I would associate the AdvertisingStandards Authority in the same way, who haveacted outrageously.Andrew Dilnot: This happened just before my time,but my understanding is that indeed a concern was

raised with the Authority about a particular piece ofcoverage and that the then chairman wrote about thatcoverage. It is not my understanding that the PressComplaints Commission in any way dissented fromthe view that the Authority had taken.Paul Flynn: They said the Daily Mail had notbreached the Code of Practice, so it was all right to lie.Andrew Dilnot: No, not our Code of Practice. Theyhave no jurisdiction over the statistical Code ofPractice; it was their own Code of Practice. Now, myown view is that independent regulators, which in thiscontext is what the Authority is, should always actsimply on the basis of their own judgment. I think itwould make sense when this new institution, whateverit is, is set up for me to have a meeting with itschairman to explain what our work is and the basis onwhich we do it. But I think I would always fight very,very shy of ever appearing to collude or co-operateover judgments. The judgments, I think, shouldalways be the judgment of the Authority. Of course, Iwould then be very disappointed if any other bodywere to try to second guess that judgment, but thatagain, it would seem to me, would be up to them. Iwould be happy to meet with the chairman andexplain our work, but I think I would probably stop atthat point.

Q162 Paul Flynn: As I understand it, both these twobodies have codes of practice that allowed them todissemble, to cheat, to lie, and you, as the keeper ofthe flame of truth, honesty and objectivity, have a roleto get them to behave in a reasonable way.Andrew Dilnot: If somebody were ever to complainabout the statistical action, these bodies are within ourdomain, so if a concern reached us that any body hadacted statistically inappropriately, that is somethingthat we could take on. But my sense is that we standalone. We are an independent Authority and weshould not seek to collude or co-operate in coming toany decisions with any other body.Chair: I think we have reached a conclusion.

Q163 Paul Flynn: Just one comment, if I may,seeing as Newport came up. One of the pieces of workcarried out at Newport is on a sense of well-being. Aswe referred to the parliamentary representatives, thereis no question but of the sense of well-being that myconstituents have in Newport, for various reasons,partly because the Statistics Office is located there.But this is an area that seems to be very promising inproviding us with a measure of success ofGovernments and others, not just in GDP but inmeasuring this area that is very difficult to measure—the question of well-being. Is this work continuing?Andrew Dilnot: It is. I will pass to Jil in a moment,who is leading the work, but it is funded for a periodand it is important to be clear about what it is. It is anattempt, as I understand it, to bring together a widerange of data, some of it economic, some of it aboutthe physical environment; some of it, but only a verysmall part of it, is answers to questions aboutsubjective states of mind. It is an attempt really to bea kind of modern Social Trends—a modern way ofdescribing the whole of what it is to live in the UK.



Jil Matheson: Exactly so and it is developing. It ispart of an international agenda. There are lots ofcountries that are doing this, so it will evolve. TheOECD is very active, but it is looking at the threepillars: the economy, the environment and quality oflife for individuals and households, and putting thosetogether into a single place. There is a lot of interest.

Q164 Chair: How much is the well-being workcosting?Paul Flynn: Not enough.Jil Matheson: The funding is for £2 million a year,which includes data collection, survey costs and thecost of the team in putting the data together,processing it, analysing it, producing it,disseminating it.Andrew Dilnot: We were given a specific fundingstream for it.

Q165 Chair: How will you judge the effectivenessof this expenditure?Jil Matheson: The same way as with any other: isthis expenditure doing what it should be doing? Is itvalue for money in the sense of: are there other waysin which we could have done this work more cheaply?Is there a user need? It takes its place alongside thekind of prioritisation that we will have to have of therange of work that ONS does.

Q166 Chair: Apart from the Government, who isreally enthusiastic?Jil Matheson: There are a huge number ofconstituents. I spoke very recently at a conference ofvoluntary sector organisations, who were intrigued,fascinated and incredibly supportive. A lot of themoperate within local authorities or particular areas, andthey found it really, really valuable in understandinghow those three dimensions play out together andwhere the pockets of poor well-being or whatever arethat the voluntary sector may have a role in helping,which may not be the same as areas of more materialdeprivation. So, it is real insight for them and there isa lot of support. There is support from parts of thebusiness community as well, so that is a very, verywide range of constituents and a lot of new users whowe have managed to engage with as part of the workthat we have been doing on well-being.

Q167 Chair: How confident are you that it is notsubject to political manipulation—that different

people could measure well-being in different ways inorder to support different political outcomes?Andrew Dilnot: Well, one thing that we are not goingto do, or at least not while I am the Chairman and Jil isthe National Statistician, is produce a happiness index.What we are doing here is bringing together lots ofdifferent streams of data: some about the economy,not just GDP but employment, job tenure, housingexperience; some about life, experience of health care,education; some about the subjective.

Q168 Chair: Lord O’Donnell will be sodisappointed.Andrew Dilnot: I have had a discussion with LordO’Donnell about happiness indices, yes. This range ofdata is extraordinarily interesting and valuable. Icannot imagine how it could be collapsed in astatistically coherent way into a single number to say,“Look, this shows you what has happened to thewell-being of this country.” So, yes, the range of datawill be open to a variety of interpretations, and peoplewho value the performance of the Health Service orthe labour market will come up with differentsummaries, but in a way that is precisely what is soexciting about it: that we are presenting a rich diet ofdata for people to look at and make their ownjudgments.

Q169 Paul Flynn: And it is challenging the view thatonly material wealth determines happiness andcontentment and so on.Andrew Dilnot: Yes, but it is not saying that materialwealth does not matter.Paul Flynn: Oh no, indeed, no, no. It is a factor, butso is fairness.Chair: There wouldn’t be many people sitting aroundthis table if we thought material wealth was the onlyroute to personal happiness.Andrew Dilnot: Quite so.Paul Flynn: Indeed no; it wouldn’t be on for thisCommittee.Chair: May I thank you very much indeed for yourevidence and can you convey the Committee’s thanksto all your staff for the work they do for the publicgood? We appreciate that the cost pressures onGovernment make life difficult at the moment,nevertheless we thoroughly value the work that youall do. Thank you very much indeed.Andrew Dilnot: Thank you very much.


Written evidence

Written evidence submitted by Market Research Society (4STATS 01)

I am writing on behalf of The Market Research Society (MRS) in response to the PASC fourth study oncommunicating and publishing statistics. No part of the MRS submission needs to be treated as confidential.

With members in more than 70 countries, MRS is the world’s largest association representing providers andusers of market, social, and opinion research, and business intelligence. All individual MRS members andMRS Company Partners agree to self-regulatory compliance with the MRS Code of Conduct. Full detailsregarding MRS and its activities can be found via: www.mrs.org.uk

MRS has an advisory board, the MRS Census and Geodemographics Group (CGG), which recommendsMRS policy in relation to geographic, demographic and census information and which has advised MRS in theformulation of this response. CGG has been operating continuously for the last twenty years and has been aleading voice representing business users. It includes members with experience in the creation and use ofpopulation statistics, census data and geographic information. MRS and CGG have worked with the UK CensusOffices, via the Office for National Statistics, representing the needs of the research sector in consultations onthe 1991, 2001 and 2011 censuses.

MRS and CGG welcome the commitment of Government to the principles of Open Data and transparency.We believe the commitment to improved accessibility and use of government data will provide long termbenefits to the wider economy.

We have attached an Annex that contains our responses to the questions raised by PASC. The issues thatMRS and CGG consider particularly important are:

— Official statistics are widely used in marketing and research.

— But private sector needs have low priority, which runs counter to government policy to help growthe economy by use of its data in business.

— MRS’s top priority is better access to data, essential to good communication and to realising fullvalue from the investment in official statistics.

— The supply of official statistics as open data is a major advance by the government which is warmlywelcomed by MRS and its members.

— There are many opportunities to get more value from official statistics, such as a self-service facilityallowing for ad hoc output within defined parameters and disclosure control rules.

— It could be counter-productive to try to put all official statistics on one website, better search servicesare a more feasible option.

— Insufficient resources are devoted to communicating data and messages in the data. This issue shouldbe a priority for the Inquiry.

Annex

MARKET RESEARCH SOCIETY RESPONSE TO THE PASC PROGRAMME OF WORK ONSTATISTICS

Fourth Study: Communicating and Publishing Statistics

Summary of the Market Research Society (MRS) Response

— Official statistics are widely used in marketing and research.

— But private sector needs have low priority, which runs counter to government policy to help grow theeconomy by use of its data in business.

— MRS’s top priority is better access to data, essential to good communication and to realising full valuefrom the investment in official statistics.

— The supply of official statistics as open data is a major advance by the government which is warmlywelcomed by MRS and its members.

— There are many opportunities to get more value from official statistics, such as a self-service facilityallowing for ad hoc output within defined parameters and disclosure control rules.

— It could be counter-productive to try to put all official statistics on one website, better search services area more feasible option.

— Insufficient resources are devoted to communicating data and messages in the data. This issue should bea priority for the Inquiry.


The Market Research Society’s Interest in Statistics

Importance of official statistics

Official statistics are widely used in marketing and market, social and opinion research both directly and tohelp structure and analyse other data. Figures for very local areas, mainly from the population census, appliedin geodemography, help in understanding patterns of customer preferences and in effective location of services.

Priorities for marketing and research

The broad scope of the Inquiry is welcome and necessary, as effective communication through all media isessential to realise the full value of the investment in official statistics. There are many opportunities forimprovement, with a top priority of better access to data, and deployment of relatively modest additionalresources can bring a high return from greater use.

Open data, a major advance in communication

The supply of statistics as open data with minimal restrictions on re-use is major advance especiallyappreciated in marketing and research. It facilitates adding value though further products and services,ultimately adding to effective communication of the statistical data. However, restrictive terms of use and costof key supporting data such as Ordnance Survey address gazetteers incorporating data from the Royal MailPostcode Address File discourage use, for example in the coding of data to match with statistical geography,and diminishes potential benefit from open statistics.

There is a strong case for extending the scope of open data, and the Government’s transparency and opendata initiatives are welcome as is inclusion of this important area later in this Inquiry.

The Committee’s Questions

Q1. How well are the practices for the release of official statistics, and pre-publication access to thosestatistics (“pre-release access”) working?

1.1 The publication of schedules for the release of official statistics, as under current practice, is importantwhen planning use in commercial activities where timetable slippage is costly, and particularly where servicesto third parties are involved. The arrangements are working, although lead times once the actual dates forrelease are announced seem short.

1.2 Pre-release access is generally less important, provided that there is no commercial gain from anyprivileged access.

Q2. (a) Has the right balance been struck between the resources devoted to producing statistics and thosedevoted to communicating them?

2.1 Insufficient resources are devoted to communicating data and messages in the data.

2.2 This is illustrated by the Census, and although there is not yet a publicly available breakdown ofexpenditure for the 2011 Census, the indications are that, given for example the lengthy timetable byinternational standards for publication of results and a lack of innovation in access via the Web, little haschanged since the 2001 Census when only 3% of the budget was devoted to preparing and disseminatingresults. The Committee could helpfully ask for more information on the current position.

Q2. (b) What should be the future of the National Statistics publication hub, particularly in the context of theproposed single Government domain for communicating data?

2.3 The publication hub is convenient but not essential for commercial users who generally are familiar withrelevant sources, and there are better ways of providing access.

2.4 It would not be the best way forward for statistics to become part of a monolithic Government domain,and indeed could be counter-productive, on the evidence of the long standing shortfalls of the smaller scaleONS website intended to integrate all output. A less restrictive “umbrella” would be more appropriate to thevery varied nature of statistical output—from single key indicators to millions of Census figures.

Q3. In what circumstances should the UK Statistics Authority comment on the use or misuse of statistics?

3.1 UK Statistics Authority should act to ensure supply of “full and frank” commentary to help pre-emptcriticism, for example of the discontinuation of a series, as well as ensuring compliance with the Code ofPractice for Official Statistics and responding to perceived abuse.


Q4. (a) To what extent are the requirements of users taken into account in decisions on the communicationand publication of official statistics?

4.1 Experience shows that private sector needs tend to have low priority, unless shared with the public sector,with communication and publication being no exception, which runs counter to government policy to helpgrow the economy by use of its data in business.

Q4. (b) How easy is it for users to find the official statistics they need, whether via the Office for NationalStatistics website, other websites, the National Statistics publication hub or elsewhere?

4.2 It is not easy to find data on the ONS website or publication hub. Making it easier would be the singlegreatest contribution to better access and communication.

4.3 Search facilities are poor, often giving long, undifferentiated lists. First and foremost the need is for asearch service which covers all relevant sites—rather than trying to put all data on one site. The objectiveshould be to reach a point where all official statistics sites have identical search engines so that results for aspecific search are identical whichever site is used.

4.4 Further measurable objectives should be set to help improve access and meet user needs:

— Making all data easy to download, and developing ways such as Application Programming Interfaces(APIs) to download updated data automatically into users’ systems.

— Providing clear and consistent guidance and metadata, targeted at a range of user communities andexpertise from schools to businesses.

— Providing clear and consistent explanation of methodology.

Q4. (c) How well do the formats used for releasing official statistics meet user needs, both in terms ofpublication type (compendia, reports, press releases) and data format (pdf, excel, open data standards), andwhat changes, if any, should be made?

4.5 In one key area—the supply of large Census datasets in CSV format—needs have only been acceptedafter repeated requests.

Q4. (d) How effectively does the UK Statistics Authority engage with the user community to understand itsrequirements, and what, if anything, should it do differently?

4.6 The activities of UK Statistics Authority (UKSA) tend not to come to the notice of users outsidegovernment who may also not be clear about the distinction between the Authority and statistics producers—an area for improved communication and promotion. UKSA should put more emphasis on strategic, crossgovernment issues such as the publication of comparable output across all parts of the UK—a majorrequirement for businesses operating nationally.

Q5. How well are trends in, and limitations of, official statistics (produced by the Office for NationalStatistics and elsewhere) explained to users? Do these explanations have the right level of detail and howbalanced are they?

5.1 There is need for better explanations, issued concurrently with statistics, and at a more detailedgeographical level.

5.2 Users in marketing and research, as well as elsewhere, need quality measures to indicate the degree ofconfidence which can be put on results. Otherwise there is a risk that figures will be taken as perfectly correct.Quality measure should be published concurrently and be well publicised to help minimise this risk. As qualityis not consistent in a set of output, measures should also be published for all levels of area aggregation.

Q6. How well do producers of official statistics respond to ad hoc requests for data, and what should be thepublication policy be in respect of responses to these requests?

6.1 Ad hoc output has not been very significant for commercial users. The technical process of producingsuch output has had limitations, timetables can be long and uncertain, definitions used in official statistics maydiffer from those used elsewhere, statistical disclosure control tends to restrict detail, and poor value for moneymay discourage commissions.

6.2 Policy should be to actively exploit the many opportunities to get greater value from existing datasets,avoiding the cost of collecting new data, with better and more flexible production systems, co-operativeproducer-user partnerships, and generally higher priority for ad hoc output. Self-service systems should be usedmore widely to allow users to produce their own ad hoc output within defined parameters and disclosurecontrol rules.

6.3 An example of the way datasets can be opened up to yield more value is the coding of data in majorofficial surveys by type of small area—the classification itself developed in a producer-user partnership—afterencouragement by geodemographers, thus adding a completely new geographical dimension to the analysis ofthe survey data.


Q7. How well are the opportunities afforded by the internet being exploited in the publications strategy forofficial statistics?

7.1 In the context of the websites of large organisations, the Office for National Statistics (ONS) and otherproducers have embraced internet technology relatively well in principle, with examples of innovation such asthe work of the small ONS team producing interactive Web content, but there is a need for significantimprovement in structure and function as commented in response to Question 4.

7.2 However, it is not easy for outsiders to get a full picture of what is realistically possible, and theCommittee would help greatly by focussing on this highly relevant area.

October 2012

Written evidence submitted by Tom King (4TATS 02)

Summary

Statistics can of necessity be complex in their formulation; where this is the case the assumptions behindthem need to be clearly explained. Such explanations need not be written anew for new releases but could bedesigned to support a particular class of release and simply linked to each new iteration.

Central user engagement is dominated by large, fixed uses; a shift from users to uses should be the next stepin the user engagement strategy. It should be possible to engage individuals in research and evaluationenvironments for this purpose.

UKSA is known for criticising statistical use but should balance this by commending good practice anddeveloping good communication. Statistics without good metadata invite problems of interpretation socommunicating statistics without linking to appropriate sources of information should be controlled.

Finding statistics and what information may be held is harder than it should be, maintaining a substantialbarrier to use. As other users suggest developing resources including some derived from ONS data, there maybe partnership opportunities but certainly insights into user needs.

Public understanding of statistics has wide variations but for unfamiliar applications it will need support.The value of statistics in providing accountability should be researched not presumed. Consideration could begiven to those who do not use statistics but may benefit.

The text roughly follows the ordering of the questions paper until paragraph 18 when more general issuesare discussed.

1. The sheer number of breaches in relation to release and pre-release access and the lack of evidence ofproblems caused by these demonstrate that the purpose of pre-release access and a publication timetable arenot well understood by anyone.

2. There may be a more indirect effect of appreciation of the unaccountable nature of unequal access toinformation. An example comes from a recent Public Accounts hearing on diabetes: A person giving evidencewished to cite the latest performance statistics which were to be published later that week in the belief thatthese would be more useful. It was readily understood that the NAO had not had access to these data and soreferring to them was inappropriate. Equal access to data facilitates accountability by agreeing certain factswithout privileging the producer of the data.

3. Not enough is understood of the use made of statistics to judge whether their communication is particularlylimiting. However, it is certainly true that statistics would find more uses were they communicated better andthat many users would request further statistics if they were better communicated.

4. The UKSA is well known for criticising political misrepresentation of statistics and little known foranything else. It would do better to promote serious use and understanding of statistics by instigating productionof explanations of technical aspects where they are being misunderstood. (Where commentary exists, it focuseson release-to-release trends, not underlying issues.) Criticism can foster a lack of trust if other issues are nottaken in hand.

5. Two particular issues where interventions might be made are misrepresentation of the precision of officialstatistics, and to dispute “zombie statistics” which have assumed the status of official fact. These issues mightrequire its position on statistics which were published before the Act was passed to be clarified. An examplefrom inquiries into Mid Staffordshire Foundation Hospitals Trust illustrates all three issues.

6. A leaked inquiry draft report included an observation on mortality of roughly 800–1200 above the averagefor similar hospitals in that period, which might statistically be attributed to poor standards of care. This valuewas never published and an estimate of around 400–600 over a three year period was the official confidenceinterval for the excess mortality. Media reporting subsequently took on the interval 400–1200, or “as many as1200”. These misrepresent the precision and vivify the value of 1200. The origin of the 1200 value wasconfusing enough that the independent inquiry struggled to identify its origin.


7. There is an underlying problem that the needs of users are not well understood. More details are givenfrom paragraph 18 but publication and communication rarely facilitate information integration in the sense ofcomposite statistic construction.

8. Finding simple statistics is such an ordeal for novice users that there are repeated suggestions that onlineresources be independently developed for teaching in schools and universities (eg http://rsscse.org.uk/images/stories/ts_pedagogy_full_report.pdf). The premise is that it is too difficult to expect teachers to do this tosupport topical lessons.

9. Recurring needs of large users are easier to address and engage with and various user groups exist tofacilitate this. However, they have not been successful in coalescing concerns into changes in known underlyingproblems in methodology (eg RPI/CPI).

10. Much more metadata to support releases and more proximate links to other, related (in the minds ofusers) datasets would be very helpful.

11. UKSA would do better to change its focus from the demands of large users to establish the uses ofstatistics in end users which are often not evidenced.

12. Initiatives such as the MAC are excellent but there is no option for users to contribute to the agenda.Thus the integrity of the statistical system as a whole has no forum for user consultation as opposed to specificsurveys and datasets which have consultations.

13. There needs to be further consideration of public understandings of statistics: misconceptions ofassumptions, resistance to quantification and unilateral acceptance/rejection of findings. A presentation by theGCSA to the ESRC RMF 2012 showed that statistical understanding is essential when considering futurepolicy. Statistics will be fundamental to citizenship and therefore governance but UKSA has not looked to seewhether these uses are being served by present dissemination of official statistics.

14. Trends and limitations can be difficult to understand and can be hard to explain (PISA is a recentexample). Information is usually available but may not be directly associated with the release but this resultsin the context being lost in translation. Such contextual caveats are rarely passed on in use even though it canbe important to the interpretation. However, trends are usually misinterpreted because of limitations of thedata, not because trends themselves are difficult—concentration should be on limitations not trends.

15. An exemplar of what can be achieved is “Trout, Catfish and Roach: the beginner’s guide to censuspopulation estimates”. This publication was balanced enough to highlight the independence assumptionnecessary to identify the DSE as well as giving an accessible explanation of stratification. More of thesestandalone guides and promotion of them would be worthwhile as statistical series are unchanged for a longtime by design.

16. Responses to requests for statistics and metadata are poor but much could be done to improve the processfor requests. It is difficult to find data and therefore easy to request something which is already publiclyaccessible. It takes a lot of work to establish what might exist in order to make a request as other data held byproducers are rarely publicised.

17. Requests for metadata (ie details of sample, geographical detail, covariates) can be misunderstood as(unreasonable) requests for entire (and disclosive) datasets and paradata is scarce. Although GSS memberscomment that one “should ask to speak to a statistician”, the process for achieving this is opaque enough tobe offputting.

18. Statistics are produced with the premise of realising value through use—they do not have intrinsic value(evidence of the positive impact of their collection is variable). Statistics are routinely used in resourceallocation, project evaluation, policy formation, research and performance monitoring. There is also apresumption that they can be used in processes of accountability. It is worth noting that they can be combined,in a process of “information integration” rather than being used directly, and some of this integration is impliciteg weighting, using poststratification, to census proportions.

19. Statistics still can be interpreted as very powerful indications of success and failure. This is amisinterpretation on either side of the accountability but historically tends to produce perverse consequences.The intentions, conclusions and limitations of statistics should be clear to all; from this it might be useful todevelop case studies of the value added by statistics in accountability forums.

20. Some larger users are synonymous with the use they make of the same statistics every time they arereleased and other frequent users eg media may benefit from commentary on the nuance of the release. Otherresearch, evaluation or small users may be regularly looking at data which is new to them due to continuallychanging needs, sporadic updating and synthesis of several sources. Such users will benefit much more fromweb infrastructure and visual summaries to help identify the data they want before considering commentary. Itshould be possible to establish a link with this type of users to monitor the accessibility of relevant data whichshould result in fewer dubiously sourced statistics being presented or left unchallenged.

21. An inquiry by your Science and Technology Committee was frustrated by the lack of submissions frombusiness of the social science uses of census data. While it may be that business may not conceptualise its use


as social science, there is considerable intellectual property involved in the uses data is put to, which may beconsidered valuable enough to keep confidential. Thus any engagement with users needs to take into accountthe stake they have in the methods they use to analyse data which may not be public knowledge.

22. A bigger problem involves making data releases interpretable to small, sporadic users—parastatisticians,perhaps. Evidence is that experienced statisticians can misinterpret unfamiliar graphs and tables on firstreading—the area of application is a key component of statistical practice. The focus of UKSA has been onmisuse rather than misunderstanding but statistical concepts can be very involved.

23. Research on understanding of statistics is generally very limited—it focuses on toy examples (eg theprobability of two coin tosses both landing heads) when statisticians do it at all. This omits the area ofapplication which we know is fundamental to the practice of statistics. Taking statistics out of context is not agood way of eliciting understanding.

24. Other work on understanding comes from those trying to communicate on specific issues such as shareddecision making in medicine and environmental risks (ranging from local flooding to climate change). Thereare also some collaborations on very specific decision tools and modes of calculating specific kinds ofuncertainty (eg economic and weather forecasting).

25. What evidence there is shows that there are a wide range of misconceptions about statistics on matterssuch as sampling, inference, uncertainty, independence, seasonal adjustment and change over time.

26. Trust in statistics implies a belief that they are an accurate representation of something real. A populationprojection is often mistaken for a forecast in the sense of a weather forecast. A rational response on seeingsuch projections are wrong would be to say that the statistics are not trustworthy. Statistics cannot be trustedunless their purpose is understood and clearly communicated. It is the understanding of new and sporadic userswhich will benefit and through them larger swathes of the public but at present they are not being engaged.

27. For some statistics, expected uses are part of citizenship, through democratic policy and accountability(crime statistics are an oft cited example). It is not clear that sufficient efforts are made to support publicunderstanding of such. This is not helped by much use of statistics being simply as colour eg “X% of peoplehave attribute N; this is too low; we will do this to improve it” where the value of X has no bearing on theremainder of the statement or solution.

28. The work by the statistics authority on “strengthening user engagement” referred to public use but thiswas only evidenced in crime statistics. Tracing this to the evidence, this was a comment by one police officerstating that the statistics helped the public to hold them to account. This is easily explained as an example ofreactivity rather than accountability with an internal response to unexpected poor numbers not an externalchallenge.

October 2012

Written evidence submitted by Social Research Association (4STATS 03)

Background

The SRA comprises a wide range of practising applied social scientists from across the sectors of academia,central and local government and their associated agencies, the voluntary sector and the commercial socialresearch agencies and not for profit organisations. Many of our members are occasional rather than regular andfrequent users of statistics. As such they are often less competent to find their way through the maze thatofficial statistics sometimes appears to infrequent users. It is in this context that we have looked at yourquestions and thoroughly endorse the response of the Statistics User Forum.

Some of the points particularly pertinent to the SRA are set out in the general introductory paragraph of theSUF’s response. This highlights issues that while general are especially important for the applied socialscientists we represent.

— The balance between resource expended on producing statistics and communicating/providing accessto them is not yet right—more effort should go into the latter in order to maximise utility.

— Access arrangements should accommodate the requirements of the full range of levels of expertiseamong users.

— Users need a “one-stop shop” to find out what official statistics are available on a particular topic.

— Effective user engagement should be a continuous dialogue, not just a series of one-off consultations.

— Explaining the strengths and weaknesses of particular statistics is an essential part of theircommunication to users, as is providing impartial interpretation of what underlies trends andcomparisons.

— The publication strategy for official statistics is responding rather slowly and quite patchily acrossdepartments to the opportunities offered by the internet.


Responses to Questions

We strongly endorse that:

Q2. National Statistics Hub should be a “one-stop shop” for users to find out what official statistics areavailable on a particular topic. Ideally this would then provide them with links through to thestatistics themselves, irrespective of the government department/agency website on which they arestored.

Q3. UK Statistics Authority should comment on any misuse of statistics of national importance, forexample where misuse affects Parliamentary debate or if commentators outside government aremaking sustained attempts to mislead the public on the meaning of a particular set of officialstatistics.

Q4. More needs to be done to make official statistics accessible to a wide range of audiences. The usercommunity is wide and diverse, ranging from analysts requiring the latest economic data formodelling work to a member of the public wanting to know the size of population of their town orvillage so a range of access routes are needed which assume different levels of knowledge andexpertise.

Q6. Our members are more likely to experience difficulties than more frequent and experienced users.The picture is mixed between different departments. We are concerned however that the movetowards greater transparency will put statistics producers under increased pressure as there mightwell be an increase in requests. This might well mean the urgent more straightforward request isdealt with at the expense of the more complex but more important request.

Q7. We welcome moves to greater use of the internet and more data visualisation but feel more resourcesneed to be expended on this if the less experienced user is to have their needs met.

October 2012

Written evidence submitted by Statistics User Forum (4STATS 04)

A. Introduction and summary

Statistics cannot be used unless they are made available and communicated in a clear and open manner.Thus championing the need for excellent communication and publication of statistics lies at the heart of theconcerns of the Statistics User Forum (SUF). SUF has welcomed the fact that these requirements are nowreflected in the Code of Practice for Statistics. However, there is still a long way to go before users’ needs inthis respect are met. SUF views on each of the questions raised in the study are given in detail below but wehighlight the following points as being of particular importance:

— The balance between resource expended on producing statistics and communicating/providing accessto them is not yet right—more effort should go into the latter in order to maximise utility.

— Access arrangements should accommodate the requirements of the full range of levels of expertiseamong users.

— Users need a “one-stop shop” to find out what official statistics are available on a particular topic.

— Effective user engagement should be a continuous dialogue, not just a series of one-off consultations.

— Explaining the strengths and weaknesses of particular statistics is an essential part of theircommunication to users, as is providing impartial interpretation of what underlies trends andcomparisons.

— The publication strategy for official statistics is responding rather slowly and quite patchily acrossdepartments to the opportunities offered by the internet.

SUF is an umbrella organisation for a growing number of groups and networks of users of official statisticsand represents the shared views of this very diverse community. It was set up in 2004, succeeding the long-established Statistics User Council, and operates under the aegis of the Royal Statistical Society.

B. Response to Specific Questions

1. How well are the practices for the release of official statistics, and pre-publication access to thosestatistics (“pre-release access”) working?

1.1 The mechanics of the processes for the release of “headline” official statistics work reasonably well.Future publication dates are published and updated and it is possible in many cases to sign up for alerts fornewly published data. However, it is in the range of what is made available and the formats in which detailedinformation is released which pose much greater problems for users (see our response to question 4).

1.2 Users of official statistics rarely find pre-release access an issue of concern in their day-to-day work.Apart from users in the media, most are probably unaware of pre-release access, and would be surprised if theyknew that it existed. Where there is that knowledge, it acts to reduce public confidence because it reinforces the


perception that the data may have been subject to political manipulation, even when there is no evidence thatthis has happened.

2. Has the right balance been struck between the resources devoted to producing statistics and those devotedto communicating them? What should be the future of the National Statistics Hub, particularly in the contextof the proposed single Government domain for communicating data?

2.1 In response to the first question, the short answer is no. Priorities are skewed too much towards producingstatistics rather than communicating and providing access to them. Ideally we would prefer more resource tobe devoted to communication without reducing that devoted to production. However, if resource constraintsmake choices necessary, then we would argue that resource should be diverted to make existing statistics easilyaccessible and excellently communicated. Improving access to statistics increases their use, and thus maximisestheir value: this can be described as “going the last 100 yards”. Simple extracts with explanatory commentary—for example, opening up the richness of data available from sample surveys such as the Living Costs and FoodSurvey—can engage non-specialist users, whereas very large and detailed datasets can be daunting. Also,investing in the ONS and departmental websites to improve access, leading to greater use, must be a toppriority.

2.2 Many users obtain official statistics through the press, broadcasters, social media, and other secondarysources. The Government Statistical Service (GSS) could do more to help these mediators to disseminatestatistics, through presenting them in simple formats with informative explanations to which links can be made.This would not only widen the use of official statistics but would also improve the accuracy with which theyare reported.

2.3 Our vision for the National Statistics Hub is that it should be a “one-stop shop” for users to find outwhat official statistics are available on a particular topic. Ideally this would then provide them with linksthrough to the statistics themselves, irrespective of the government department/agency website on which theyare stored. Unfortunately at present the Hub falls far short of this ideal.

2.4 It is not clear to us how a single government domain for communicating data would improve thissituation. Clearly the Hub should be integrated with any such initiatives, but it is important that users shouldbe able to distinguish between national statistics, which have been validated and assessed against the Code ofPractice, and other information.

3. In what circumstances should the UK Statistics Authority comment on the use or misuse of statistics?

3.1 To be effective, the UK Statistics Authority has to be selective in the instances of misuse on which itchooses to comment. Our feeling is that such interventions should focus on issues of national importance, forexample where misuse affects public or Parliamentary debate or if commentators outside government aremaking sustained attempts to mislead the public on the meaning of a particular set of official statistics. Weconsider that this is the approach which was taken by Sir Michael Scholar in his term of office at the Authorityand we hope that Andrew Dilnot will continue to exercise this sort of judgment.

4. To what extent are the requirements of users taken into account in decisions on the communication andpublication of official statistics? How easy is it for users to find the official statistics they need, whether viathe Office for National Statistics website, other websites, the National Statistics Publication Hub orelsewhere? How well do the formats used for releasing official statistics meet user needs? How effectivelydoes the UK Statistics Authority engage with the user community to understand its requirements and what, ifanything, should it do differently?

4.1 There are some examples of good practice in taking into account users’ requirements in decisions on thecommunication and publication of statistics, notably the extensive consultation in the lead-up to the 2011Census, but they are few and far between. The user community is wide and diverse, ranging from analystsrequiring the latest economic data for modelling work to a member of the public wanting to know thepopulation of their town or village. The GSS needs to do more to embrace this diversity by finding out whatthe differing needs of users are, and then by providing a range of access routes which provide for differentlevels of knowledge and expertise. One result of such an approach would be to involve more users, which inturn would help the producers to develop and improve the statistics for which they are responsible.

4.2 All too often it is extremely difficult even for the expert user to find the statistics they need from theONS and departmental websites. Search engines leave much to be desired—most users rely on Google. Therelaunch of the ONS website in late summer 2011, if anything, made things worse, as many links were lost.The Publication Hub does not provide a user-friendly interface for the non-expert (see 2.2).

4.3 Access arrangements should accommodate the requirements of the full range of levels of expertise amongusers: open data formats for those whose main aim is to use official statistics for secondary analysis, modelling,etc, through to simple tables and charts with informative commentary for the lay audience whose main aim isto be informed about trends in society and the economy. The general move towards HTML based publications,with user friendly spreadsheets, is to be welcomed.


4.4 The UK Statistics Authority, through the assessment process and its championing of the need for userengagement, has made considerable efforts to understand the needs of the user community. Its report“Strengthening User Engagement” (2010) was an important articulation of the importance of user engagementand how it could best be achieved. The Authority has followed this through with the secondment of a UserEngagement Programme Manager (UEPM) to support the work of SUF within the RSS and this has enabledreal progress to be made towards much more productive user engagement through the internet using thepotential of social media. The UEPM was largely responsible for the successful launch in late February 2012of StatsUserNet, a new website which facilitates user-user and user-producer engagement. There are alreadyover 1,200 registrations on the site, and it has provided an online forum for discussing important issues suchas those surrounding the Retail Prices Index and the Consumer Prices Index.

4.5 However, it is primarily within the ONS and other government departments rather than the Authoritythat the culture shift advocated in the “Strengthening User Engagement” report has to be achieved. Effectiveuser engagement should be a continuous dialogue, not just a series of one-off consultations, particularly whenusers’ perception is often that consultations are held when decisions have in effect already been made. Someprogress is being made in this area: SUF is working with the National Statistician’s Office (NSO) to developbest practice guidelines for user engagement which we would like to see implemented throughout the GSS andthe user community.

5. How well are trends in, and limitations of, official statistics (produced by the Office for National Statisticsand elsewhere) explained to users? Do the explanations have the right level of detail and how balanced arethey?

5.1 Explaining the strengths and weaknesses of particular statistics is an essential part of their communicationto users, as is providing impartial interpretation of what underlies trends and comparisons. This sort ofcommentary is particularly important for non-expert users. It requires culture change in the GSS, where manystatisticians still see their task as just providing the numbers. We welcome best practice guidelines on statisticalcommentary recently developed by the GSS, and the formation of the “Helpers” team by the UK StatisticsAuthority and the NSO to embed these and other best practices across departments. However, thesedevelopments need to be maintained and developed.

5.2 Even when this sort of information is provided, the casual user may find it difficult to find and interpret.Often, commentary concentrates exclusively on the latest published figures rather than providing a longer termcontext. Producers need to develop the ability to put themselves in the place of users: what story might theydraw from the data and why are the data of interest/importance to them?

5.3 There is no “right” level of detail in explanations of trends and limitations: different users requiredifferent levels of detail and this should be recognised in the way that the GSS structures their communications,particularly on the web. Producers need to be proactively diagnosing where users (expert and non-expert alike)are evidently not understanding the commentary and explanations they provide and adjusting them accordingly.

6. How well do producers of official statistics respond to ad hoc requests for data, and what should be thepublication policy in respect of responses to these requests?

6.1 This is a very mixed picture. Some government departments respond well, others not so well. Requeststhat are not too time-consuming are responded to fairly readily. However, often users face seeminglyinsurmountable barriers in gaining access to official data and it is often only in response to initiatives by users(rather than producers) that progress is made—for example, the development of the Samples of AnonymisedRecords from the Census.

6.2 The data provided in response to ad hoc requests should routinely be made available to other users, andthe evidence from ad hoc/FOI requests about what users’ needs are should be used to extend the range ofinformation made routinely available. This will then help to avoid the need for future ad hoc requests

7. How well are the opportunities afforded by the internet being exploited in the publication strategy forofficial statistics?

7.1 The publication strategy for official statistics is responding rather slowly and quite patchily acrossdepartments to the opportunities offered by the internet. There has been some progress—less reliance on largePDFs, and a better range of more user-friendly formats for download of data. Most GSS outputs are availablesomewhere on the web. However, it is still very difficult for many users—particularly the non-expert users—to find what they need, in a format that is suitable to their needs.

7.2 The ONS website and the Publication Hub are still far from satisfactory in many respects—for exampletheir search facilities. For example, the first publication appearing in a search for National Accounts in theEconomy theme in the Publication Hub is the Annual Abstract of Statistics and the Blue Book does not appearuntil number 32.

7.3 Data visualisation, harnessing the potential of technology to provide interactive content, is very importantto engage a lay audience. There are some good examples of its effective use on the ONS website, such as theinteractive presentation of data from the 2011 Census. However, such examples are not routine throughout the


GSS or indeed in all areas of ONS output. Similarly, the potential for hierarchical presentation of informationhas not been widely exploited—ie top level very user-friendly and simple to engage the lay audience, withability to drill down for the expert user to reach the level of detail that they require.

October 2012

Written evidence submitted by Demographics User Group (4STATS 05)

“THE LAST 100 YARDS”

Executive Summary

The Demographics User Group (DUG)1 represents the views of 15 major commercial companies includingBoots, Co-operative Group, Everything Everywhere, John Lewis, Marks & Spencer, Nationwide, Sainsbury’s,and Whitbread. The members make extensive use of government statistics and geographical data to understandlocal markets and consumers, and make decisions about large investments in delivering better services. Theseare the tip of the iceberg of 2.3 million businesses in the UK, many of which can increase their efficiency, andgrow, by using data gathered by government, which has the great advantage of consistent collection across thewhole of the country.

The key themes of this note are the need for Government suppliers of data to:

— Put more emphasis on “the last 100 yards”, making it easier for new users, particularly smallbusinesses, to find and use data.

— Shift some resources from statistical production to marketing (identifying needs and meeting them)within the various parts of the GSS.

PASC’s Questions


1.1 Analysts in commercial companies rarely find this an issue of concern in their day-to-day work. Mostare probably unaware of pre-release access, and would be surprised that it occurs.

2. Has the right balance been struck between the resources devoted to producing statistics and those devotedto communicating them?

2.1 The short answer is “No”. As regular users of government statistics, we believe that Governmentsuppliers of data need to put more emphasis on “the last 100 yards”, making it easier for occasional users,particularly small businesses, to find and use data. The Government Statistical Service has traditionallyemphasised the importance of its methodology, and the quality of its statistics, but sometimes this has been atthe expense of putting effort into making access to statistics easier. We make some proposals below. There isa consequent need to shift some resources from statistical production to marketing (identifying needs andmeeting them) within the various parts of the GSS.

What should be the future of the National Statistics publication hub, particularly in the context of theproposed single Government domain for communicating data?

2.2 Commercial users usually view all statistics from government simply as “government data”, rather thanNational Statistics/Official Statistics/etc. Members of DUG typically look at the ONS website first and,increasingly, www.data.gov.uk The real need is to develop this as a portal that enables easy searching, andlinks to various other government websites.

3. In what circumstances should the UK Statistics Authority comment on the use and misuse of statistics?

3.1 We feel that the occasional public interventions by Sir Michael Scholar hit the right balance whenseeking to restrain the misuse of statistics by politicians and the media.

4. To what extent are the requirements of users taken into account in decisions on the communication andpublication of official statistics?

4.1 This is good in some areas (such as the Census), but why have others moved so slowly, or not at all?Looking for possible explanations, we have:

— A lack of understanding of users’ needs? No: whilst there is a need for more & better userengagement, some of the major needs of users are well known—although they may be given littleweight when determining priorities.

1 http://www.demographic.co.uk/dug.html


— The costs of users’ demands? Not generally: looking at the areas of little progress, only theredevelopment of websites might cost a significant sum. Much progress with the others could bemade with marginal effort. Users’ demands for less onerous Statistical Disclosure Control wouldactually reduce costs of policing, and greatly increase the value of the statistics released.

— GSS & government priorities and culture? Yes: for a variety of reasons, the GSS can find it difficultto respond to external users’ needs. The Census (both 2001 and 2011) stands out as an exception:its size, and interest to so many user sectors are obvious factors; also, its fixed date imposes atimescale. The progress with www.data.gov.uk is a second exception: here, prolonged userdissatisfaction with trading funds such as Ordnance Survey found a catalyst for change on the arrivalof a personality, Sir Tim Berners Lee. Other areas, however, lack the imperative of a timescale, or aforceful political focus, and are liable to drift, unless they coincide with departments’ own priorities.

How easy is it for users to find the official statistics they need, whether via the ONS website, other websites,the National Statistics publication hub or elsewhere

4.2 This is patchy. The broken links that remain in the ONS’s website after its redevelopment are a continuingsource of frustration.

How well do the formats used for releasing official statistics meet user needs, both in terms of publicationtype (compendia, reports, press releases) and data format (pdf, excel, open data standards) and what changesif any should be made?

4.3 This breaks down into three elements, each of which needs improvement:

— Packaging of data into forms that suit users/customers. This varies from the need for simple extractsof large and complex datasets (such as the Living Costs and Food Survey), to UK-wide files ofCensus Key Statistics (rather than having to stitch together files region by region).

— Providing files in popular formats: only recently has ONS agreed to provide 2011 Census statisticsin csv files, which had long been requested by users.

— Alerting users by email that new datasets have been published, rather than simply putting them onthe website and awaiting their discovery.

How effectively does the UK Statistics Authority engage with the user community to understand itsrequirements, and, what if anything, should it do differently?

4.4 There can be a disconnection between consultation to understand users’ needs, and carrying this through.The steps that need to be taken include:

— More & better user engagement (aka market research/customer insight).

— A proactive role for the UKSA’s Monitoring function. User’s views need to be captured andformalised into documents that can be used to summarise and evaluate their cause. Neither individualgroups nor the Statistics User Forum have the resources to swiftly write authoritative reports ontopics such as “Adding Value by packaging statistics”; “Priorities for improving UK-wide statistics”;“The Business Case for a National Address Register”; etc. The Monitoring function should emulatethe former Statistics Commission in acting as a swift and independent lightening conductor for users’concerns and priorities.

— Cultural change: setting the tone that customer service is the top priority. Most users of statistics arevery understanding of statisticians’ professional pride in doing the best possible technical job,avoiding errors, and warning of potential misuse of statistics. But this emphasis can be to thedetriment of maximising their use by both existing and new users. The importance of topmanagement setting a new tone—finally laying to rest risk aversion, which took hold in 2003—cannot be overstated. Such a change of tone could also be reinforced by UKSA/ONS appointing aDirector of Customer Services, and having small teams dedicated to customer knowledge (userengagement) and improving access. We believe that it would be helpful to draw on the experienceof commercial companies: for example, O2, speaking at the 2009 DUG Conference, emphasised its“absolute focus on customers & customer experience”, with the rallying cry of “turning customersinto fans”.

5. How well are trends in, and limitations of, official statistics (produced by ONS and elsewhere) explained tousers? Do these explanations have the right level of detail and how balanced are they?

5.1 Datasets have different value according to the project in hand, and we are very doubtful aboutdesignation, or attributing kite marks. All statistics do, however, require good metadata, explaining thestrengths & weaknesses of the particular dataset. Members of DUG (and, we are sure, most commercialcompanies) would rather have reasonable data now, rather than wait months for marginal improvements. Thecorollary is that such datasets should be accompanied by informative metadata.


6. How well do producers of official statistics respond to ad hoc requests for data, and what should be thepublication policy be in respect of responses to these requests?

6.1 Commercial users usually seek data very quickly, and members of DUG would usually assume that it’snot worth asking.

7. How well are the opportunities afforded by the internet being exploited in the publications strategy forofficial statistics?

7.1 The greater use of visualisation techniques and mapping would attract more interest in governmentstatistics, and greatly expand the market/user base, and hence the value.

October 2012

Written evidence submitted by Royal Statistical Society (4STATS 06)

A. Introduction

i. This submission by the Royal Statistical Society (RSS) focuses on questions 1 to 3 of the Study. TheStatistics User Forum is providing a more extensive submission, which the RSS endorses and supports.

ii. Communication has always been a weak area of UK official statistics. Since the introduction of the Codeof Practice following the 2007 Act, and the first Assessment round of the UK Statistics Authority, moreattention has been paid to it. Improvements have resulted but much more effort is needed. We are aware thatthe UKSA is undertaking some new initiatives in this area; the success of these and other measures need to bemonitored carefully and further measures undertaken if required.

iii. In this submission we focus on communications and presentations around statistical releases and otherstatistics outputs that already exist. A deeper communications challenge for the official statistics service is topresent a coherent statistical picture of what is going on in areas where debate needs to concentrate on theissues rather than on explaining particular statistics. The debate on Scottish independence is an example wherestatistics needs to be brought together and well communicated in order to foster good debate. The old styleSocial Trends was a good example of when this was done.

iv. The RSS is the UK’s only professional and learned society devoted to the interests of statistics andstatisticians. Founded in 1834, it is one of the world’s most influential and prestigious statistical societies. Itaims to promote public understanding of statistics and provide professional support to users of statistics andto statisticians.

B. Specific Questions


1.1 As far as we are aware the actual mechanics for the release of official statistics are working well, barsome technical problems in ensuring that Office for National Statistics’ (ONS) statistics are released andavailable on the website at 9.30 am prompt. The issues around statistical releases concern essentially contentand presentation.

1.2 As indicated in our response to Study 1, the RSS believes there is no justification for pre-release accessother than in exceptional circumstances; at most this should be just two—three hours.

1.3 The arguments for pre-release access (as a reading of the Cabinet Office review of the arrangementsmakes clear) rest on the alleged necessity for ministerial comment or a departmental press release being issuedat or around the same time as the statistics. In our view the practice of issuing such commentary to coincidewith statistical releases is pernicious. It skews any debate over the figures and perpetuates the impression thatministers control the data. There is strong evidence from surveys of public confidence in official statistics tosuggest that lack of confidence in statistics is in large part due to perceptions of political control ormisrepresentation. This practice perpetuates that perception so is a major obstacle to improving confidence, inour view.

1.4 We accept that straightforward abuse of the system is rare but even 24 hours allows a degree of mediamanagement. The situation in Scotland and Wales is worse in that up to five days pre-release access is allowed.

2. Is the right balance struck between the resources devoted to producing statistics and those devoted tocommunicating them? What should be the future of the National Statistics Publication Hub, particularly inthe context of the proposed single Government domain for communicating data?

2.1 Statistics are of value only if they are used; it is a waste of money to collect data and compile figures ifthey do not reach potential audiences. More resources need to be devoted to communicating statistics, includingappropriate training sessions, even if this has to come at the expense of reducing expenditure on statistical


collection. The Assessment process carried out by the UK Statistics Authority has confirmed the need toimprove communication which regular users have long known.

2.2 Improving communication will not be cost free but need not always be very expensive. In some casessimple improvements such as improving the writing and presentation in statistical releases is all that is required.The ONS communications unit does some good work in finding innovative ways to present statistics; theremay be a case for strengthening it and for increasing efforts to disseminate its expertise across the rest of theGovernment Statistical Service. However, in many cases what is essentially required is an improved ability ofstatisticians to put themselves in the shoes of potential users, especially the non-expert user, and to tailorcommunication appropriately, including, importantly, communication via websites.

2.3 Good communication of statistics includes the provision of appropriate information on the forms, notes,definitions, and classifications used. The ONS website and the statistical sections of Departmental websitesshould include these as standard practice.

2.4 Good communication of any technical subject also educates the recipient. Government statisticians mustkeep the “education” element in mind when preparing web pages, statistical releases and other outputs. Moregenerally, however, government statisticians in general and the ONS in particular should be conscious of theirbroader responsibility to help develop a more statistically literate society. Websites could, for example, explainin plain language issues such as the sampling and survey design, seasonal adjustment and, importantly, thelimitations of and margins of error, which lie behind particular statistics. Educating journalists, reporters andothers in the media is of particular importance.

2.5 The National Statistics Publication Hub clearly needs to be integrated appropriately with othergovernment communication initiatives but we do not see that a single government domain for communicationdata would necessarily mean that the Hub has no purpose. There can be more than one route to finding data.Improvements in the Hub are, however, needed. Many of its aims—for example giving an overview ofparticular statistics with links to more detailed explanations are good. But the quality of execution is patchyand it is not as easy to use or as user friendly as it could be. For example, while clicking on a “theme” orbroad subject group brings up a fairly short list of more detailed subjects, clicking subsequently on one of thedetailed subjects often brings up a very long list of publications in alphabetical order which take a long timeto scroll down and where it is not always easy to guess which items are relevant to what the user is searchingfor. Another example is that the section on National Accounts lacks consistency in nomenclature making itconfusing to a non-expert.


3.1 It is clearly impossible for the UK Statistics Authority to comment on every instance of statistical misuseby any person or organisation in the public domain. Further, instances of misuse are not equally serious. Inpractice the Authority has concentrated on misuse by politicians and has selected instances where it feels thatits intervention is appropriate and will be effective. We feel that the Authority under Sir Michael Scholarjudged this well; Sir Michael’s interventions were generally effective, established the Authority as a force tobe reckoned with and were clearly uncomfortable at times for those concerned. Further, the Authority appearedeven handed between government, opposition and other politicians.

3.2 It is perhaps more difficult to assess whether and when the Authority should intervene in other cases,notably misuse in the media. In practice the Authority has only done this on one or two occasions. We agreethat the Authority should keep its powder dry and intervene only occasionally, when there has been significantmisinterpretation rather than a simple mistake and where the misinterpretation is serious and clearly the faultof the journalist (rather than, for example, arising from a poorly written release). Having established itscredibility and effectiveness under Sir Michael Scholar, the Authority is now well positioned to consider someextension to its interventions on these occasions issuing forceful reminders to editors and writers that they beara heavy responsibility.

3.3 Similarly, it should be prepared to reprimand advertisers and businesses that badly and deliberatelymisuse statistics. It should liaise closely with regulatory bodies in marketing and advertising and with whateverstructure emerges from Lord Leveson’s inquiry. It might also consider “strategic” and programmatic reports onhow statistics are being used by business and the media.

October 2012


Written evidence submitted by Full Fact (4STATS 07)

A. Summary

Our evidence shows inconsistent standards in communicating official statistics and suggests possible waysstandards of communication could be improved, including:

— Excellence in communication must be valued as much as technical excellence.

— Senior statisticians should have a prominent role in promoting public understanding of officialstatistics, including talking directly to the media.

— More emphasis on non-professional audiences for statistics, including policy-makers, the media andthe public.

— More emphasis on providing topical statistics, including publishing all ad hoc releases which providedata not readily available anywhere else.

— More and better use of the internet, websites and interactive online material.

We endorse the role the UK Statistical Authority has carved out for itself so long as it continues to intervenewhen intervention is necessary. We also recommend more formal coordination between the Authority and otherbodies concerned with standards in public life.

Full Fact is an independent factchecking organisation. We research claims, many statistical, made bypoliticians, journalists and pressure groups. As an organisation dealing in official statistics, we have frequentcontact with the Office for National Statistics and the Government Statistical Service.

B. Response


1.1 Normal statistical bulletins are released satisfactorily as far as we are aware; ad-hoc statistical releasesare more problematic. We address this in our response to question 6. We have expressed our concerns aboutpre-release access in our submission to the first study.

2. Has the right balance been struck between the resources devoted to producing those statistics and thosedevoted to communicating them?

2.1 Full Fact does not have a view on the current allocation of resources. We do, however, believe that thecommunication of statistics needs improvement. If more resources must be allocated to achieve this, theresources would be well spent, but communicating better should not necessarily cost more.

2.2 It is important that, as well as technical excellence, excellent communication is valued and nurturedwithin the Office for National Statistics and the Government Statistics Service. Communication is not asecondary task but crucial to how statistics serve the country.

2.3 All good communication starts from its audience. Increasingly, policy makers, journalists and the publicaccess statistics directly because they are only a click away. Statistics producers need to take this into accountin deciding how to communicate.

2.4 To that end, it is encouraging to see that there are clearly people with senior support thinking hard abouthow information is best presented not just online, but in publications and on the airwaves. The most effectiveinnovations need to be made widespread, and the baseline raised. Much of this, we suspect, is not aboutspending more money but rather a question of emphasis, culture and to some extent skills.

2.5 We would ask the Committee to consider the role of senior statisticians in this process. As data becomesever more prominent within public life, we think the National Statistician could usefully have a higher publicprofile, engaging in and informing debate. This could be done without in any way compromising her neutrality.We would extend this to Heads of Profession in government departments too.

2a. What should be the future of the National Statistics publication hub, particularly in the context of theproposed single Government domain for communicating data?

2.6 The Publication Hub is not perfect or particularly engaging, but it fulfils its basic role as a portal forfirst releases. The principle of the single portal for discovering all new releases across different producers andacross the UK is a good one, and this should not be lost in any government overhaul of statisticscommunication.

2.7 We would point out that data and statistics are sometimes treated as interchangeable terms—but statisticsare the product of professional expertise, careful methodology and review, and necessarily come with importantcontext. For example, raw crime data from police are distinct from the national statistics that are ultimatelyproduced from that data. We would be concerned, therefore, if statistics and data in its broader sense weremixed in a way that eroded the distinction: not all numbers are equal.



3.1 The UK Statistics Authority should continue to exercise broadly its current approach, monitoringstatistics used by a wide range of people including those in politics, the media and business, and reservingintervention for serious misuses of statistics only.

3.2 As Sir Michael Scholar has said, the UK Statistics Authority cannot “assume the role of a referee oneach occasion that an official statistic is misquoted or misinterpreted.”2 It would be impossible for the UKStatistics Authority to intervene every time a statistic is misused, and it is ultimately a matter of judgementwhen an intervention would be useful. The test for this judgement is whether those who might misuse statisticsare conscious of and deterred by the possibility of an intervention from the Authority.

3.3 Civil society must play its part in informing public debate and Full Fact’s work helps to fulfil this role.We research claims, many statistical, made by politicians, journalists and pressure groups. When we findmistakes, we ask for them to be corrected. Ministers, government departments, a Select Committee, MPs, theBBC, most national newspapers and a variety of pressure groups have made corrections when we have pointedout inaccuracies. Where necessary, we turn to appropriate regulatory bodies to ensure that corrections are made.

3.4 Pursuing individual factual claims sometimes exposes wider policy issues. For example, when wedirected the UK Statistics Authority’s attention to a series of opaque media reports on benefits, the Authority’sintervention led to the Department for Work and Pensions starting to publishing its ad-hoc analyses systemicallyonline. This principle is now expected to be implemented more widely across government. Similarly, a recentletter from the Chairman of the Authority about the limitations of comparative education statistics used by HMChief Inspector of Education has roots in concerns we raised with the Authority back in 2010.3

3.5 These examples illustrate that it is not necessary any more than it is possible for the Authority to pickup every concern for itself. However, this model of course only works if the UK Statistics Authority can berelied upon to bring its full weight to bear down on more serious statistical misuses. So far we have found thatit can.

3.6 The only gap we would point to at the moment is one of cooperation. There are quite a number ofauthorities broadly concerned with aspects of trust in public life, from the Advertising Standards Authority tothe Press Complaints Commission. In March, the UK Statistics Authority criticised the Daily Mail’s coverageof prosecution of offences related to the 2011 riots.4 Surprisingly, the Press Complaints Commission ruledthat the Daily Mail had not breached the Code of Practice, effectively overruling the UKSA.

We feel there would be value in the UK Statistics Authority approaching other relevant bodies, to establisha working relationship in advance of situations when they might deal with cases involving official statistics.These bodies would then be able to reply with the canonical advice of the UK Statistics Authority. That wouldprobably be a better approach to the worlds outside government and politics than attempts to intervene directlywhere the UK Statistics Authority has no clear means to make its writ run.

4. To what extent are the requirements of users taken into account in decisions on the communication andpublication of official statistics?

4a. How easy is it for users to find the official statistics they need, whether via the Office for NationalStatistics website, other websites, the National Statistics publication hub or elsewhere?

4.1 The overall experience of finding statistics is one of frustration. Unless you know your sources, it canbe very hard to access even basic information.

4.2 The ONS website is notoriously poor and needs to be redeveloped again, this time from first principles.In particular, the search function is hopeless. For example, a search for “population” yields “2008-basedSubnational Population Projections for England” as its first result. The site does not consistently make linksbetween different datasets and analyses that may illuminate each other. Since 2011, when the website wasreplaced, all links to ONS data used in Full Fact articles have broken. This is frustrating, and a breach of basiconline norms. We would be concerned if the same mistake is made when the site is overhauled again.

4.3 However, there is good work going on. The “key figures” provided on the front page and theme pagesdo often answer the kinds of questions we receive from our audience. The interactive material available onlineis another promising initiative. Eventually this should be expanded and moved to a more prominent location,but first the website’s organisation and search function should be improved.

4.4 There is no standardised or centralised way of locating statistics in the websites of governmentdepartments, so finding data is neither consistently easy nor hard. data.gov.uk is messy and has gaps. We donot think an average member of the public would find departmental websites user-friendly (we find most datawithout huge problems partly because we use statistics outlets frequently); however, it is hard to see how the2 “David Cameron could damage public trust in crime figures, says statistician”, The Guardian, 14th July 2010

http://www.guardian.co.uk/uk/2010/jul/14/david-cameron-crime-figures3 See letter from Andrew Dilnot to David Miliband, “Programme for International Student Assessment statistics”, 3rd October

2012, http://www.statisticsauthority.gov.uk/reports—correspondence/correspondence/index.html4 “Statistics watchdog criticises Mail’s claims on riot-related crime”, fullfact.org, http://fullfact.org/blog/UK_Statistics_Authority_

Daily_Mail_riots-6801


UK Statistics Authority can fulfil its remit to improve public access to statistics without raising the standardof departmental and other publishers’ sites.

4b. How well do the formats used for releasing official statistics meet user needs, both in terms ofpublication type (compendia, reports, press releases) and data format (pdf, excel, open data standards), andwhat changes should be made?

4.5 The goal of publishing official statistics should be to ensure that users can get the information they need,in its full context, in the most convenient way. In particular, it should strive to present a coherent statisticalpicture in important or contentious areas of public debate. This entails different things for different users, whichmight be met in some of these ways:

— Graphically-led presentations such as graphs and maps.

— Web pages using links to bring the full picture into view (suited to the increasing proportion ofmobile internet users).

— PDFs, easily downloaded and printed, with full background and interpretation in one place.

— Spreadsheets for those who want raw data.

— In due course, interactive tools.

4.6 At the moment, official statistics are primarily presented in PDFs and spreadsheets, which seem gearedtoward more technical users. Web-native presentation should have a much greater role, being more searchable,less intimidating and quick to use. It also allows the possibility of linking between information to give peoplea fuller picture in a way that static formats do not.

4.7 Spreadsheets sometimes seem to prioritise appearance over functionality, for example spacer columnsobstruct calculations. This should stop. Simply putting data into a format that can be used for calculations canentail hours of work. There are better ways to make spreadsheets easy to navigate and the priority should bemaking data easy to work with.

4.8 To standardise and improve ONS releases, many of which are poorly written, all should be sent througha desk of sub-editors and the communications office before publication. The Norwegian equivalent of the ONSemploys journalists full-time. They work closely with statisticians to produce public-friendly press releases.While this system might not translate perfectly into the UK, the Committee could certainly look to this model asinspiration for what can be achieved when communication is made a first-class part of a statistical office’s task.

4.9 Finally, the rarity of topical publications is glaring. For example, the UK Statistics Authority couldcoordinate the publication of statistical factsheets to inform the debate on the forthcoming independencereferendum in Scotland.

4c. How effectively does the UK Statistics Authority engage with the user community to understand itsrequirements and what, if anything, should it do differently?

4.10 The Statistics User Forum demonstrates that the UK Statistics Authority has an active programme ofprofessional user engagement. However, policy makers are, in effect, the most important users of statistics, andwe share Andrew Dilnot’s concern that senior civil servants sometimes seem to see statistics as “what you putin after you have worked out what the policy should be”. We agree that the UK Statistics Authority needs towiden its engagement and work with the Cabinet Secretary and Permanent Secretaries “to boost the sense ofconfidence of analysts and statisticians within the civil service.”5 The same argument applies to working withParliamentarians, their staff, and the staff of the two Houses.

4.11 We also endorse Andrew Dilnot’s view that the UK Statistics Authority should work with journaliststo improve the communication of statistics.6 As we argued in our submission to the first study, this meansbroadcast first and foremost. Our experience of journalism and journalism education suggests that first, thereis an increasing awareness of the importance of statistical competence and second, not enough is being doneto help trainee or existing journalists develop it.

5. How well are trends in, and limitations of, official statistics (produced by the ONS and elsewhere)explained to users? Do these explanations have the right level of detail and how balanced are they?

5.1 There are examples of ONS and GSS releases which explain statistics clearly, but there is no consistentstandard. Other releases are ill-structured and presented less coherently. We have two systemic concerns—

5.2.1 Statistics producers need to monitor how their statistics are used in public debate and adjust howthey present them accordingly. In the summer of 2011 Full Fact raised concerns with the ONS about

5 House of Commons Oral Evidence taken before the Public Administration Select Committee: Pre-Appointment Hearing for thePost of the Chair of the UK Statistics Authority, Tuesday 6th December 2011, Q38 http://www.publications.parliament.uk/pa/cm201012/cmselect/cmpubadm/uc1634-i/uc163401.htm

6 House of Commons Oral Evidence taken before the Public Administration Select Committee, Wednesday 12th September2012, Q51 http://www.publications.parliament.uk/pa/cm201213/cmselect/cmpubadm/uc406-i/uc40601.htm. “We have to workalongside those who become the medium through which most people get access to statistics to try to make sure that they reportthem in the most sensible and best way.”


media misrepresentation of foreign workers’ jobs statistics. The ONS repeatedly resisted our requestfor guidance about the statistics’ limitations to be included, and only after we raised the issue withthe UK Statistics Authority did they agree to clarify future bulletins.7 Part of the ONS’s job is toexplain statistics and when it sees statistics being misused it should alter their presentation—forexample making caveats clearer—to guard against further misrepresentation.

5.2.2 Producers rarely do enough to explain which changes are statistically significant. This reflects thefirst point, in that it is one of the most frequent areas of misinterpretation. For example, variationsin crime level numbers have been treated as major news regardless of whether those variations aresignificant or whether those numbers really mean that crime has stayed the same as near as we canmeasure. If such caveats are not prominent in the original release, they stand no chance of makingit into the media, the debate or public consciousness.

6. How well do producers of official statistics respond to ad-hoc requests for data, and what should be thepublication policy in response to these requests?

6.1 Ad-hoc requests for data are dealt with inconsistently by the ONS. Sometimes staff will respond withdata immediately, sometimes a request will be diverted into the FOI process and at other times we must submitan FOI request to get any headway. From the outside, there appears to be little consistency in, or explanationof, the reasons for these actions.

6.2 When it comes to government departments, it can be hard to get hold of statisticians, and conversationsmediated through either press officers or freedom of information officers are less likely to be fruitful.Government Statistical Service personnel should be exempted from the requirement that officials do not speakto the media. Their names and telephone numbers should appear on statistical releases. Of particular concernis that we have had topical requests turned into FOI requests, which are slow to process. It seems to us thatrequests which would illuminate a current topic of public debate should be prioritised.

6.3 Any request which generates data not readily available elsewhere should be published to give effect tothe guiding principle of the Code of Practice for Official statistics of “equal access to all”.8 Much of thepoint of the Code of Practice is about building trust in official statistics, and by extension in public life. AsOnora O’Neill argued, “if we want a society in which placing trust is feasible we need to look for ways inwhich we can actively check one another’s claims.”9 Ad-hoc releases often underpin prominent politicalclaims and need to be released for analysis. Even when they do not, it is the public’s information and it belongsin the public domain.

6.4 For example, a number of articles appeared in the press earlier this year derived from ad-hoc ONSstatistical releases, describing these releases as “exclusive”, “unpublished” and even “secret”. The impressiongiven by these newspapers of the ONS’s approach to releasing data was at odds with the Code of Practice andcannot have helped to foster public confidence in how figures are compiled and obtained. After our interventionin a similar practice by the Department for Work and Pensions in 2010, the Department started to publish allits ad-hoc statistical analyses online. This is surely the right default position. Again with the ONS, Full Factraised this with the UK Statistics Authority10 and we have since seen progress toward establishing thatprinciple throughout the ONS and government, but we would welcome the Committee’s support for thesemoves.

7. How well are opportunities afforded by the internet being exploited in the publication strategy for officialstatistics?

7.1 The opportunities afforded by the internet include the ability to reach the public directly; ease ofsearching; speed of publishing; interactivity and engagement; informative visualisation; being responsive tocurrent debates and concerns; easy cross-referencing between information; personalisation; provision ofdetailed explanations at miniscule marginal costs and the ability to make large datasets easy to interrogate. Sofar few of these opportunities seem to have been explored, alone exploited.

7 “ONS moves to end misreporting of foreign worker job statistics”, fullfact.org, 12 July 2011 http://fullfact.org/blog/foreign_worker_statistics_ons_clarification-2815

8 Code of Practice for Official Statistics, January 2009, p.149 Onora O’Neill, Reith Lectures 2003, “A Question of Trust: Trust and Transparency” http://www.bbc.co.uk/radio4/reith2002/

lecture4.shtml10 See letter from Will Moy to Andrew Dilnot and reply, “Handling of ad-hoc requests for statistics”, 2nd and 18th April 2012,

http://www.statisticsauthority.gov.uk/reports—correspondence/correspondence/index.html


7.2 In a sense, that is understandable. The state of the art in publishing statistical information online is stillfairly rudimentary and government can be expected to be behind the curve. What is concerning is that muchof the approach to publishing official statistics online seems to be at odds with web culture, where breakingother people’s links is profoundly anti-social, not linking well between your own information is perverse andnot being part of the conversation is simply odd. The ONS has a “twitter policy”, which includes the phrase:“We encourage users to follow us.”11 At the least, this is not an organisation which is comfortable online.

October 2012

Written evidence submitted by UK Statistics Authority (4STATS 08)

Further to the Committee’s recent call for written evidence on communicating and publishing statistics, theUK Statistics Authority’s response is set out below.

I would like to make a few general observations about the importance of effective statistical communication.As I have mentioned in a number of public settings over the past few months, further developing thecommunication of official statistics across the statistical service is a particular priority for the StatisticsAuthority and for me as Chair of the Authority. We see improving statistical communication as playing a vitalrole in securing the maximum value from official statistics.

In October 2010, the Statistics Authority published guidance for government statisticians on thecharacteristics of a good statistical report, based on the requirements in the Code of Practice for OfficialStatistics. Building on that Statement, we have identified seven themes to focus further work with the statisticalservice to improve communication. These will be reflected in revised guidance which the Authority willpublish shortly:

— uncertainty—emphasising that official statistics are estimates rather than absolutes; and theimportance of guidance for the user about relevant quality considerations in relation to potential usesof the statistics;

— trends—highlighting the rich value of time-series, rather than over-emphasis on month-to-month orpoint-to-point comparisons of estimates;

— accessible explanation—introduced, defined, and clearly explained concepts, so that the user canunderstand them;

— international comparisons—providing comparisons with other countries to frame the UK’s situationwhere appropriate;

— context—explaining how statistics relate to our wider economy and society;

— attribution and causation—should not be asserted without sufficient statistical evidence, or otherwisein a misleading manner; and

— written outputs should use plain English and should meet expected standards for publication.

We understand the direction of the Committee’s questions to be about whether official statistics are beingcommunicated well enough to those who will want to use them. This is one of the most important andchallenging aspects of the work of official statisticians. The Statistics Authority starts from the perspective thatofficial statistics are collected and managed at public expense and must justify that expenditure by contributingas much as possible to decision-making in all parts of society and the economy. However, official statistics canonly do that if those who need to use them know that they exist, can find them when they need them, and canunderstand their relevance and utility.

The Code of Practice for Official Statistics contains several specific requirements to ensure these matters areaddressed. In practice, a good proportion of the many recommendations made in the Authority’s formalAssessment Reports relate to communicating statistics and supporting users more generally.

The fact that the Authority continues to press for improvements in this area is itself indicative that currentstandards of communication and accessibility are not as good as we think they need to be. In effect, officialstatistics are not yet being enabled to do the maximum amount of public good that we would wish to see. Ourrecent report reviewing the first cycle of the statutory Assessment programme, Assessment of UK OfficialStatistics 2009–2012, discusses the priorities for the future and how the Authority intends to address them. Acopy of this report was provided to the Committee recently.

At the most general level, the Authority is actively supporting planning to shift the balance of resourcesslightly away from data collection and management towards increasing accessibility and communication. Thiswill not be easy to achieve in practice. It is likely to require difficult decisions to reduce data collectionactivity—at least at the margins—and use the resources saved to make improvements in communication. Usersmay be reluctant to accept changes which reduce the range of figures available but nonetheless this may wellprove, somewhat paradoxically, to be a key step towards ensuring that the users get as much good as possiblefrom the statistical product.11 ONS Twitter policy, http://www.ons.gov.uk/ons/about-ons/what-we-do/social-media/twitter-policy/twitter-policy.html


In respect of the Committee’s detailed questions, the Authority’s response is below. Further operationalinformation in respect of the Office for National Statistics (ONS) and the wider Government Statistical Service(GSS) is provided in the attached Annex.

1. Practices for the Release of Official Statistics

While there are many high quality statistical reports published every week, we recognise that there areweaknesses in terms of the consistency, coherence and accessibility in the practices for the release of officialstatistics. A vast amount of statistical material is published by a large number of official bodies—includingsome 40 producers of “National Statistics”. These bodies publish statistics largely autonomously. There is somemeasure of central co-ordination, shared practice and control, enshrined in advice to government departmentsfrom the National Statistician and in the Code of Practice; and in centralised arrangements such as the NationalStatistics Publication Hub. But from the perspective of the user of official statistics, the impression is still oneof differing practice; problems with accessing what is there in the form the user wants; and inconsistency interms of the provision of explanation and advice. The statistical service itself is doing what it can to addressthese weaknesses but there are undoubtedly real obstacles. While the decentralised nature of the statisticalservice does have some real virtues, it also means that each body that produces official statistics is free tooperate largely as it wishes in respect of communicating and publishing official statistics, subject to compliancewith the Code of Practice. In addition, the volume and rapid evolution of statistical material being produced issuch as to make its integration particularly difficult. For example, a recent Authority review found some 22producers of official statistics on health topics and a further 21 bodies that were releasing health-relatedstatistical material. Integrating statistical products across that number of bodies would be a significant andwelcome undertaking but, in a climate of budgetary reductions, we do not necessarily expect it to be made ahigh priority.

Pre-release access to official statistics

Under the Statistics and Registration Service Act 2007, responsibility for determining arrangements for pre-release access to official statistics rests with Ministers and the UK Parliament in respect of official statisticsproduced by UK government departments and public bodies, and with Ministers in the respective DevolvedAdministrations and the devolved legislatures in respect of wholly-devolved official statistics. The relevantMinisters are free to draft secondary legislation on pre-release access as they choose. As a result the maximumperiod for pre-release access is 24 hours for UK departments’ statistics and for statistics produced by theNorthern Ireland Executive, and five days in Wales and Scotland.

It is the view of the Authority that current practices for pre-release access to official statistics areunsatisfactory. Large numbers of Ministers, their advisers and officials outside the statistical production chainare routinely receiving privileged pre-release access to large numbers of statistical releases. For example, areview undertaken by the National Statistician’s Office during 2011 found that around 140 officials have beengranted pre-release access to the Crime in England and Wales statistical release. Now that responsibility forthe production of this release has been transferred to ONS, the Authority intends to review this. The Departmentfor Education has also provided pre-release access to over 100 people—inside and outside government—to itsstatistics on Key Stage 2 performance for England; likewise the Home Office in respect of its statistical releaseon firearm certificates. This is an unacceptable and unjustifiable number in my view, and the Authority wouldwish to see the numbers reduced as much as possible.

The Authority believes that, under current arrangements, there is a risk that pre-release access mayinadvertently be given to non-approved officials. For example, in May 2011 HM Treasury accidentally releasedmarket sensitive consumer price index (CPI) data 17 hours prematurely to around 400 people not on theapproved pre-release access list, due to what was believed to have been a clerical error. This followed anotherbreach in February 2011 on producer price indices. In December 2011, pre-release access rules were againbreached when comments regarding the direction of CPI figures were made at an internal meeting of seniorcivil servants within the Department of Business, Innovation and Skills. These are matters about which mypredecessor wrote to the departments concerned.

As the Authority has previously said to the Committee, we believe that pre-release access has the potentialto cause real harm. Providing privileged pre-release access may be seen by others as giving Ministers and theiradvisers time to consider how best to react to, or possibly distract attention from, a statistical release. There isalso a public perception that pre-release access gives time to pressure statisticians to alter releases, althoughthe Authority has not seen any evidence that this has happened. We also believe that pre-release access canbring the statistical system as a whole into disrepute, reducing public trust in the quality and integrity ofstatistics.

Although international practices vary considerably, the Authority’s Monitoring Report on Pre-Release Accessto Official Statistics published in March 2010 noted that most advanced countries have sophisticated regimesin place to “control and minimise” pre-release access. The report also noted that the 24 hour period adoptedby the UK is “very much at the upper end of the range where specific limits exist”. The Authority will preparefor the Committee a note providing an update on current international practice in respect of pre-release access.


In light of these types of issues, the Authority recommends (as outlined in our 2010 report) that the maximumperiod for pre-release access should be reduced to no more than three hours, with less being the norm. Inaddition, we recommend that the number of people who are granted pre-release access to official statistics, andthe number of statistical releases to which pre-release access applies, both be reduced significantly. We remaindissatisfied with the present arrangements across the UK, and would wish to see the Authority givenresponsibility for determining them throughout the UK administrations.

2. Balance between Production and Communication

The Authority aims to ensure that official statistics are well explained, including their strengths andweaknesses, leading to better decision-making in the public interest. In a decentralised statistical system,decisions on the appropriate balance between data collection, management and statistical production on theone hand, and the resources devoted to the accessibility and communication of official statistics on the other,are for the department(s) concerned. However, the Code of Practice for Official Statistics requires all statisticsto be accompanied by commentary that aids communication. Specifically, it requires the main messages fromthe statistics to be summarised early in the commentary; description of the policy or operational context forthe statistics, including any relevant government targets; description of the statistics in neutral language;comparisons over time; avoiding specialist terms and jargon; and explanation of the quality and reliability ofthe statistics in relation to their potential uses.

As discussed above, the Authority’s Assessment programme has highlighted that communicating officialstatistics and explaining their strengths and weaknesses remains something of a challenge for many producersacross the statistical service; and that many producers of official statistics may not yet have achieved the optimalbalance between the resources devoted to the production of statistics and their subsequent communication tothe user. To this end, the Authority is beginning a programme of work with the statistical service to help toimprove the communication of statistics and related advice to users. Earlier this year, I convened a workshopto review how a series of ONS statistical releases could be better communicated to users. In light of that, theAuthority has established, on a pilot basis, a “good practice team” which will assist the statistical service moregenerally in developing and implementing improvements to current communication practices.

ONS is currently working to improve theme and topic pages on the ONS website to achieve a broaderaudience appeal. Interactive visualisations, engaging statistical commentary, and improved signposting,navigation and search functionality are being added to the website to seek to attract and engage more users.

Future of the National Statistics Publication Hub

The UK National Statistics Publication Hub was established in April 2008 as a central part of the launch ofthe UK Statistics Authority. During the passage of the legislation in advance of the 2007 Act coming intoforce, the concept of a Publication Hub received a good deal of parliamentary support. The Hub provides a“one-stop-shop” for all National Statistics and some official statistics releases, to maximise the accessibility ofthe statistics to users. Users do not need to know the producing department of the statistical releases they arelooking for since the Hub provides a central place from which they are all linked. Crucially, the Hub ensuresthat links to statistical releases are clearly separated from any ministerial statements, thereby ensuring that theuser accesses statistical releases without being potentially biased by political comment put alongside.

ONS and the wider GSS are working with the Government Digital Service in respect of investigating howbest also to provide access to official statistics on the new single Government domain (www.gov.uk). The Hubwill continue to be maintained until this exploratory work has concluded. At that point, a decision will bemade in respect of any further development of the Hub that might be necessary. The independence of thepreparation and uploading of statistical content, sufficiently separate from ministerial and policy comment, areimportant considerations which the Authority will require reassurance about, alongside the maintenance ofstrict controls on pre-release access in advance of publication.

3. Commenting on the Use and Misuse of Statistics

The Statistics Authority reviews concerns about the use or misuse of official statistics raised by externalparties or concerns raised from within the Authority itself. Where official statistics are leaked beforepublication, where statistics are used in such a way as to mislead public debate, or where the advice of agovernment statistician has been ignored, the Authority will investigate and publish its findings. As I notedduring my oral evidence to the Committee in September, we would also investigate any suggestion thatParliament had been misled on a statistical matter.

A range of interactions and interventions are available to the Authority, from discussions with the partiesconcerned to broader, public interventions. As an independent body with a statutory objective to serve thepublic good, the Authority operates transparently and openly. Reports and correspondence relating to ourinterventions and findings are published on the Authority’s website for all to see. Being willing to commentopenly on cases where official statistics may have been used in a potentially misleading way is an importantpart of the Authority’s monitoring work as provided for in section 8 of the 2007 Act.


4. User Requirements in Respect of Communication and Publication

The Code of Practice for Official Statistics puts particular emphasis on the need for producers of officialstatistics to take the requirements of users into account. The Code also requires producers to “publicise officialstatistics in ways that enable users to identify and access information relevant to their needs [and] make accessto official statistics as straightforward as possible” and for official statistics to be accompanied by “commentaryand analysis that aid interpretation”.

The Authority’s Assessment programme has systematically required evidence from bodies that produceofficial statistics that they are researching and documenting user needs and compiling evidence that user needsare being met.

The National Statistician regularly meets statistical Heads of Profession and representatives of the StatisticsUser Forum to discuss issues of common interest, and these discussions include departmental statisticalcommunication and publication strategies.

The Authority fully recognises the importance of producer departments releasing official statistics in a rangeof publication types and data formats that reflect current user needs, and accommodating future userrequirements. Further information is provided in the attached Annex.

Statistics Authority engagement with users

The Statistics Authority is committed to engaging with the user community to understand its requirements.The Code of Practice for Official Statistics identifies meeting user needs as its first principle, noting “theproduction, management and dissemination of official statistics should meet the requirements of informeddecision-making by government, public services, business, researchers and the public”. In this area, the Codeof Practice is both consistent with, and places added emphasis on, the relevant principles set out in the UnitedNations Fundamental Principles of Official Statistics and the European Statistics Code of Practice.

The Authority’s Monitoring Report on Strengthening User Engagement, published in June 2010, noted thatthe statistical service had taken important steps in strengthening its engagement with users, including dialoguewith various user groups and in improving online access to official statistics. We also highlighted several areasfor improvement, including the navigability of official statistics on the internet, consultation with experts toexploit digitised data more effectively, and to enhance the accessibility of official statistics and related advice;and encouraging further communication between statisticians and journalists, including opportunities fortraining and other engagement opportunities.

The Authority continues to support ongoing initiatives for improving user engagement, including the RoyalStatistical Society’s (RSS) StatsUserNet user engagement portal, active support for Statistics User Forumconferences and other user-focused meetings, and through the funding of a secondment to the RSS to managethe Society’s user engagement work.

5. Trends in and Limitations of Official Statistics

As set out above, the Authority’s Assessment programme found that more could be done across the statisticalservice to communicate statistics and their limitations to users. The Authority is continuing to work withrepresentatives from across departments to ensure that official statistics are well explained to users.

Communicating significant trends is an important part of this process. With many statistics, currently thereappears to be a tendency to place too much of a commentary focus on the latest figure released. Focusing onthe latest figure in this way may unnecessarily confuse or distract the user from a more important longer-termtrend and the wider statistical message. The Authority is, therefore, working alongside the statistical service,the broader analytic community, and the media to improve understanding of the importance of communicatinglonger-term statistical trends. Further information is provided in the attached Annex.

6. Responding to Ad-hoc Requests for Data

The Authority believes that all ad-hoc requests for data should be dealt with in an open and transparentmanner. We support the Government’s Open Data initiative and want to see all statistical material (that is notdisclosive of confidential information) readily available in the public domain. We are aware that practice variesacross the GSS, where some departments (eg Department for Work and Pensions and the Ministry of Justice)publish a range of their ad-hoc statistical analyses, whereas others do not do so routinely. ONS practice inhandling ad-hoc requests for unpublished analyses was the subject of discussion by the Board of the Authorityearlier this year. The Board instructed ONS to ensure that responses to ad-hoc requests from the media andother bodies engaged in public debate should be made available on the ONS website at the time the data andanalysis is released. While the Board asked ONS to make the publishing of responses to ad-hoc requests fromthe media a priority, alongside responses where the data might be of high-profile interest to others, the Boardalso asked ONS to consider how all responses to ad-hoc requests could best be made available in a cost-effective and accessible way for the benefit of all users. ONS has undertaken to maintain a published list of allanalyses provided on an ad-hoc basis and to make those freely available on request. The National Statisticianis also preparing related guidance for the wider GSS. Further information is provided in the attached Annex.


7. The Internet and the Publication Strategy for Official Statistics

The Authority recognises the importance of embracing opportunities afforded by the internet in disseminatingofficial statistics. As we set out in our Monitoring Report on Strengthening User Engagement, improvements inanalytical tools, visualisation software and database design, supported by increasingly fast internet connections,facilitate the presentation of statistics in ways that were once unimaginable and that offer the potential forusers to extract additional value from them. For example, interactive population pyramids on the ONS websiteenable users to understand readily the changing population structure of the UK; and the NeighbourhoodStatistics website pulls together a range of detailed statistics on a variety of topics within specific geographicalareas, adding to the value of each of these sources when considered in isolation. We would not wish to seethese kinds of developments as the exception rather than the norm. Producers of official statistics should striveto present statistics in ways that capture users’ interest and improve their understanding. Further informationis provided in the attached Annex.

I and my colleagues would be pleased to elaborate further on these or any other matters as the Committeewould like.

Annex

FURTHER OPERATIONAL INFORMATION

This Annex provides operational information in respect of the Office for National Statistics (ONS) and widerGovernment Statistical Service (GSS).

4. User Requirements in Respect of Communication and Publication

ONS staff regularly hold press briefings for major ONS releases, and undertake a significant number ofbroadcast and radio interviews. There are also a number of education and awareness events which are held,designed to meet user requirements. For example, ONS routinely arranges public seminars on aspects of labourmarket and economic statistics to appeal both to specialist and non-specialist audiences, and ONS will shortlyhold a workshop to discuss the relationship between economic productivity and labour market data in thecontext of the current economic climate.

The GSS Presentation and Dissemination Committee is continuing to support departments in improving thecommunication and publication of their data, and related advice and commentary. The Committee, supportedby the National Statistician’s Office, has established a network of “commentary champions” in a number ofdepartments and will soon, alongside the “good practice team”, further develop the champions’ network andproduce a “resource kit” to ensure they are well supported. The “good practice team” is also producing guidancewith examples of good practice for wider dissemination across the GSS. The National Statistician’s Office willcontinue to review departmental statistical releases and will hold workshops with departments to help continueto raise standards of communication in statistical releases, and to share findings and examples of good practice.

Accessibility of websites

Official statistics are disseminated via a range of digital platforms, including the UK National StatisticsPublication Hub, the ONS website, the Neighbourhood Statistics website, the local-area labour market statisticssite Nomis, departmental websites, and www.data.gov.uk.

The Publication Hub provides free access to all National Statistics produced across the statistical service,with 2.25 million visits in the last year. A survey of users, carried out in Spring 2011, identified that 66% ofrespondents were able to find what they were looking for on the Hub. Anecdotal evidence suggests that those34% who did not find what they were looking for on the Hub may have been searching for datasets currentlyheld on departmental websites, looking for statistics that are not currently collected, or for data not otherwisepublished. Since the 2011 survey, enhancements have been carried out on the Hub to improve the userexperience, including improved search and sorting functionality, updated contact information for enquirers, andadditional statistical material on relevant European statistics.

ONS launched a new website in August 2011. This programme of work involved the re-engineering of thewebsite’s underlying architecture. Initial feedback from users was disappointing, due to difficulties withsearching and navigating the new site. ONS took action by launching a website recovery project, expandingONS’s publishing support services and identifying a series of enhancements that would be required to tacklethe outstanding issues. These have resulted in a reduction in negative user feedback. The increasing use of datavisualisation tools on the website has lead to an increase in syndication, significantly expanding the reach ofONS data. This involves other websites picking up and reusing what ONS publishes. For example, a snapshotof one week in July 2012 showed that while 10% of views of the “100 years” animation of Census data werevia the ONS website; the remaining 90% were via the BBC (mainly) and the Daily Telegraph websites. Furtherfunctional enhancements to the ONS website are in progress, including the development of an ApplicationProgramming Interface (API) to allow machine-to-machine access to data, and a new “Data Explorer” facilityto allow users to customise large datasets and to download them according to their individual requirements.


Formats for releasing official statistics

ONS continues to publish its official statistics in a wide variety of formats. These range from statisticalreleases; longer articles and discussion papers on, for example, statistical methodology or setting officialstatistics in a wider policy or historical context; media releases; shorter website theme and topic pages; topodcasts and social media alerts designed to supplement statistical releases published on the ONS website. Inthe year to September 2012, ONS published 303 statistical bulletins, 268 supporting datasets, 177 compendia(annual, quarterly, or monthly) publications, 99 longer articles and discussion papers, 81 news releases and 56podcasts, alongside over 800 other guidance notes.

All ONS statistical releases are published in .html as well as .pdf format to ensure accessibility and toencourage re-use. ONS undertakes periodic reviews of the ways in which its official statistics are disseminatedand released, to seek to reflect the changing needs of users. In a recent audit exercise, around 55% of statisticalreleases on the ONS website met the 3-star open data re-usability standard, and ONS is seeking to maximise theamount of data released to the higher four—and five—star advanced data linkage formats as soon as possible

ONS regularly contributes to various international publications comparing, for example, the situation of theUK with other EU member states, produced by Eurostat, the statistical agency of the European Commission;and likewise publications produced by other international organisations such as the Organisation for EconomicCooperation and Development.

ONS is currently in the process of establishing a new Analysis and Dissemination directorate within whichdigital publishing and stakeholder/external communications activity will be located. The digital publishingteam is developing ONS’s long-term dissemination strategy to review all ONS statistical products and toinvestigate the possibilities for partnership working to develop further ONS’s portfolio of digital content. Thisteam will bring together designers, developers, analysts and communications expertise in a multi-disciplinarysetting.

Engagement with users

ONS recognises the importance of improving its own user engagement activity, and is currently undertakinga programme of work to disseminate ONS statistics and analysis in more user-friendly and accessible ways. Ameeting of ONS staff with a special interest in user engagement is held quarterly to share good practice andto take user engagement forward within ONS. Best practice guidelines have recently been produced andcirculated to all ONS directorates. ONS obtains annual formal feedback from key government departments andagencies to assess how well ONS is meeting their needs as users of ONS data. ONS publishes a summary ofthese assessments in its annual report. Work on user engagement is also underway across the GSS, includingworking with the RSS and other parties to produce case studies of good user engagement and a good practiceguide, alongside seeking to maximise additional departmental engagement with users, particularly via the RSS’sStatsUserNet forum.

5. Trends in and Limitations of Official Statistics

ONS continues to make progress in this area. For example, in the most recent ONS statistical release onRetail Sales, a longer-term time-series was added to provide further analysis of the underlying pattern withinretail sales. Similarly, further context around international migration data has been added to the MigrationStatistics Quarterly Report published by ONS, alongside a “migration timeline” as an interactive product todemonstrate recent migration trends within the context of the last five decades. The timeline also shows other,relevant key statistics such as the unemployment rate, percentage change in gross domestic product, the mid-year population estimate and any other historical events that may have impacted on migration patterns.Migration data in published spreadsheets have also been enhanced during 2012. New functionality was addedto charts to enable users to quickly identify migration trends relevant to specific groups of migrants.

Across the GSS there is an ongoing programme of work to make improvements to the explanation of patternsin time-series. Some recent examples of further development include, the Health and Social Care InformationCentre’s release on prescriptions dispensed in the community which includes much wider explanation of thecontext in which prescribing has increased; commentary about livestock patterns and trends in agricultural landuse in the Scottish Government’s agricultural census; the Department for Business, Innovation and Skillsrelease on trade union membership which provides a long time series back to 1892 and analysis of variousfactors involved in the change in membership rates over time; and the Forestry Commission’s release onwoodlands which includes a longer time-series and explains the significant changes in new woodland plantingin relation to various initiatives such as changes in tax benefits and the introduction of Rural DevelopmentContracts.

6. Responding to Ad-hoc Requests for Data

ONS routinely receives a large number of requests for data. Since June 2012, ONS received over 500 ad-hoc requests for data, alongside a number of requests for data made under the Freedom of Information (FOI)Act. These requests are made by a broad range of users, including government, the media, researchers,businesses, and the general public.


In June 2012, ONS developed a new policy on responding to ad-hoc requests for data. Under this policy, allad-hoc requests are published on the ONS website, providing equal access to all in line with the principles ofthe Code of Practice for Official Statistics. Responses to ad-hoc requests are either published in full on theONS website, or listed in summary form where interested parties may contact ONS to be sent the data direct.Responses published in full on the website include requests from the media or other users that are likely tolead to onward publication of the data or otherwise where the data might be seen as being of high-profileinterest, alongside responses to FOI requests for data and requests for large datasets. Other responses to ad-hoc requests continue to be listed on the ONS website and are made available to users on request.

The GSS is undertaking similar work designed to improve accessibility to responses to ad-hoc requests fordata. Effective engagement between producers and users of official statistics enables producers to be madeaware of the needs of users in respect of access to data which are not routinely made available. Theseopportunities can allow producers to plan more effectively and to consider the benefits of revising their portfolioof regular statistical outputs. The National Statistician is also currently updating her guidance on identifyingofficial statistics to include advice to producers on this issue. The updated guidance will be published shortly.

Department for Work and Pensions statisticians have established a page on the Department’s website toprovide links to additional statistical analyses that are not included in their portfolio of regular publications.This includes previously unpublished statistics, management information, or statistical analyses used inministerial statements or in briefings to the media and others which could be potentially high profile orotherwise attract media interest. Ministry of Justice (MoJ) statisticians have also developed a similar page tohost the publication of statistical information released by MoJ in response to ad-hoc requests. In line with theprinciples of the Code of Practice for Official Statistics, the Authority expects all other departments notcurrently doing so to do likewise, and where this does not happen the Authority would regard the matter as abreach of the Code.

ONS continues to make available unit record information to approved researchers under the Statistics andRegistration Service Act, either in anonymised form or under controlled conditions through:

— the Virtual Micro data Laboratory (VML), an on-site research access facility. Data available includesbusiness datasets from ONS surveys on e-Commerce, Foreign Direct Investment, Products of theEuropean Community (PRODCOM), monthly output data, the Annual Survey of Hours and Earnings,and Consumer and Producer Prices; additionally, social survey datasets on Labour Force Survey,Annual Population Survey, Living Costs and Food Survey, and the Longitudinal Study.

— the UK Data Archive, hosted by the University of Essex. This makes anonymised social survey unitrecord data available including data from the Annual Population Survey, English Housing Survey,General Lifestyle Survey, and Labour Force Survey, as well as a disclosure controlled samples ofdata from the 2001 Census. These data are restricted via user registration and access agreementschemes; and,

— the Secure Data Service (SDS), an academic facility hosted at the UK Data Archive. ONS continuesto work closely with the SDS and deposits some of the data from the VML for use within the SDSby approved researchers under controlled arrangements.

These services enable accredited academics and researchers to use advanced statistical techniques to extractfurther value from the data and to apply them to relevant research and public policy questions. There arecurrently around 2,000 approved researchers who have been accredited formally against strict criteria.

7. The Internet and the Publication Strategy for Official Statistics

The GSS, led by the National Statistician, is currently preparing a Presentation and Dissemination Strategy,to further develop departmental dissemination of official statistics through the internet, social media, and mobiledevices. The strategy will draw on a range of dissemination initiatives from across the statistical service,including better use of email notification and other alerting tools tailored to individual users’ interests; furtherdevelopment of user-driven tabulation in place of pre-prepared, standardised reports; and greater availabilityof interactive data visualisation and other innovative dissemination tools.

ONS is continuing to develop new theme and topic pages on the ONS website, featuring headline storiesand “info-graphics”, as well as optimising the ONS website for mobile devices and developing more digitalcontent for promotion via both social media and the ONS website itself. In addition, ONS is developing morecustomisation features to enable users to access only the content they are interested in, through the use of “RSSfeeds” and online alert services to enable users to choose which subjects they want to receive alerts about atthe point at which data are published.

In 2007, ONS was the first National Statistical Institute to establish a dedicated data visualisation team witha clear remit to exploit emerging web technology to improve the way official statistics are disseminated on theinternet. This has become a centre of excellence, recognised both within the UK and abroad. The team hasdeveloped a portfolio of over 30 interactive and animated products. All data visualisations are made availablein a single location on the ONS website, grouped by theme. They cover a wide range of visualisations from“storytelling”, where ONS analysts provide visual narratives (eg the history of the Census), through to dataexploration, where users can explore patterns in big datasets visually (eg income estimates).


The data visualisation team has also worked proactively to support a number of UK government departmentsand statistical offices around the world in allowing them to re-use graphical data visualisation technologydeveloped at, and by, ONS. The ONS data visualisation team won the Royal Statistical Society’s 2010 awardfor statistical excellence, and was also shortlisted for the Civil Service awards in 2010. The team is currentlyworking with emerging technologies (eg HTML5) to support the dissemination of statistical content on thenext generation of mobile devices. The first output from this work was an interactive map of internal migration,designed for use on a range of mobile devices and standard technology, alongside new functionality that enablesusers to share statistical content and data visualisations on social media.

October 2012

Supplementary written evidence submitted by Cabinet Office (4STATS 09)

I wanted to extend my thanks to you and the Public Administration Select Committee for the opportunity togive evidence last week as part of your Inquiry into Communicating and Publishing Statistics. The Committeehas an important role to play in interrogating the Government’s approach to the publication and communicationof statistics: although much progress has been made in recent years, as I said to the Committee last week Iappreciate that there is much still to do. The evidence you are taking from key stakeholders will, I am sure,inform a robust, challenging and helpful final report which the Government will consider in depth.

The publication of robust, accessible data is critical to the Government’s commitment to improvetransparency in Government, and an important contribution to its wider efficiency and reform agenda. TheStatistics and Registration Service Act 2007 was a radical and essential overhaul of the framework for handlingand communicating statistical information. The Cabinet Office led a process of post-legislative scrutiny of theAct in 2011, consulting key stakeholders on the impact of the legislation and drawing on survey data from theUK Statistics Authority’s 2010 monitoring report Strengthening User Engagement.

Through post-legislative scrutiny, we identified a number of areas where further reforms may be needed,including potentially widening data-sharing powers (which I discussed with the Committee last week) anddoing more to distinguish statistics which are official from those that are not. As I said to the Committee, mydoor is open to discussions about ways in which the Government could help to clarify this distinction in orderto make it more meaningful to users of statistics.

One of the most important elements of the Statistics and Registration Act was the creation of the UKStatistics Authority as an independent arbiter of quality and robustness, a role which Andrew Dilnot is takingforward with vigour and determination. I look forward to meeting Andrew Dilnot and the National Statisticianshortly to discuss some of the particular issues raised during the hearing, including how to ensure continuedimprovement in the transparency and robustness of ad-hoc statistics. I would also like to reaffirm thecommitment I made to the Committee last week that I would welcome any representations from parties whohave concerns regarding the handling of statistics.

The Committee also questioned the Government’s role in providing the evidence base to inform the Scottishindependence referendum. As I’m sure you will be aware, the Secretary of State for Scotland announced earlierthis year that the UK Government would undertake a programme of work to inform and support this debate.There is a public appetite for greater information on these issues, and it is right that the Government providesthe facts ahead of the referendum. The programme will therefore set out the evidence of how Scotlandcontributes to and benefits from being part of the United Kingdom, building on national and regional statisticsthat are already publically available such as HM Treasury’s Country and Regional Analysis.

Finally, the Committee asked about the Barnett Formula, which is used to determine spending allocationsfor the devolved administrations. The formula is fully detailed in “Funding the Scottish Parliament, NationalAssembly for Wales and Northern Ireland Assembly: Statement of Funding Policy”. As the CoalitionAgreement makes clear, the Government recognises the concerns that have been expressed on the system ofdevolution funding, but any reform of the current system must await the stabilisation of the public finances.

I look forward to the Select Committee’s report following the conclusion of its Inquiry and stand ready totake part in future hearings at the Committee’s request.

December 2012


Supplementary written evidence submitted by UK Statistics Authority (4STATS 10)

At an oral evidence session on 12 December 2012, we offered to send a note to the Committee on thenumber of broadcast media interviews conducted each year by Office for National Statistics (ONS) statisticiansand ONS statisticians’ engagement with the media more generally.

ONS introduced a pro-active media strategy in 2009 which led to a large increase in the number of broadcastmedia interviews conducted by ONS statisticians. In the calendar years 2008 and 2009, ONS delivered fewerthan 20 broadcast interviews per year. By comparison, in December 2012 alone, ONS statisticians deliveredover 100 radio and TV interviews, including on 2011 Census outputs, Retail Sales, Quarterly National Accountsand Labour Market statistics. The number of broadcast media interviews conducted in 2012 was around 400,which included those on Census outputs. Excluding Census-related media activity, ONS spokespeople currentlyundertake approximately 300 broadcast interviews per year.

The ONS pro-active media strategy delivered the following changes in approach to media engagement:

— Quarterly GDP “live” broadcast briefings to accompany the publication of the preliminaryrelease of data.

— Advanced media training of “expert statisticians”, who can talk confidently on camera about keyONS outputs each month—including Labour Market, Retail Sales and Public Sector Finances.

— Growth in “short story” formats and articles which are designed to be of particular use (and re-use) by journalists.

— Use of BBC Radio’s General News Service to deliver multiple local radio interviews acrossthe UK.

— Pro-active engagement with the BBC Economics and Business Unit, Sky News, ITN, C4 Newsand others.

In addition to the pre-release “lock-in” briefings, ONS often provide group briefings for print journalists andbroadcasters. ONS hold around 10 to 15 of these each year, usually on major statistical releases such as thefirst estimate of GDP or the annual crime data. These briefings can also be valuable in handling potentiallycontentious issues—for example face-to-face briefings were recently held for economics correspondents onONS’s treatment of the Government’s interventions in the financial crisis. The broadcasters usually record theirown one-to-one interviews with spokespeople after the main briefing, though recording of the briefing itself isnot prevented.

ONS policy is to provide direct contact details for lead statisticians on all published Statistical Bulletins.Statisticians and other officials dealing directly with the media are all trained to the required standard. Contactnumbers for the ONS Media Relations Office are also provided on all Statistical Bulletins and other ONS newscontent, and Media Relations officers work closely with the relevant lead ONS statisticians in the handling ofmedia enquiries. Media interest in official statistics continues to grow. Data from the ONS Customer ContactCentre shows the unit handled 5,406 general media calls in the 2012 calendar year compared with 5,138 in2011 and 4,338 in 2010.

At the oral evidence session on 12 December 2012, the Committee requested further detail on the report,Nationality at Point of National Insurance number registration of Department for Work and Pensions benefitclaimants (link: http://statistics.dwp.gov.uk/asd/asd1/adhoc_analysis/2012/nat_nino_regs.pdf), which waspublished by the Department for Work and Pensions (DWP) on 20 January 2012.

On the advice of the Head of Profession for Statistics, and on the basis that the statistics were a one-offpiece of work, the Department for Work and Pensions treated them as not being official statistics, but ratherad hoc analysis. This has since been confirmed in internal correspondence released under the Freedom ofInformation Act.

The decision taken implied that the contents of the report and its release into the public domain were notcovered by the Code of Practice for Official Statistics. However, on the advice of statisticians, the Departmentdid nevertheless choose to adopt some provisions of the Code with its release, including pre-announcing thepublication and publishing the analysis separately from any political comment on its website.

The report was published as a “research report” on 20 January and the Authority subsequently receivedrepresentations regarding the interpretation and manner of the release of these statistics. The Authority reviewedthe report and set out its conclusions in a letter from the then Chair of the Authority, Sir Michael Scholar tothe Secretary of State for Work and Pensions dated 25 January 2012 (link: http://www.statisticsauthority.gov.uk/reports—correspondence/correspondence/letter-from-sir-michael-scholar-to-rt-hon-iain-duncan-smith-mp—25012012.pdf). These conclusions are summarised below.

The Authority considered that it would be preferable for statistics of this kind to be published as officialstatistics and in line with the Code of Practice. In reaching this view, Sir Michael noted that:

— The DWP website had referred to the research as “publication of ad hoc statistics”, whichindicated that DWP itself considered them to be official statistics. Many users had treated themas official statistics and assumed they had been published in accordance with the Code ofPractice for Official Statistics.


— In view of the political and media interest in the statistics, the Authority anticipated that therewould be an ongoing demand for these statistics to be published in future as a series.

— The statistics in the report were a cross-analysis of two existing statistical releases and theAuthority expected that most commentators would view the statistics in the report as a newrelease, which augmented the two existing releases.

— As recognised in the DWP report, the statistics were highly relevant to public policy andhighly vulnerable to misinterpretation. There were important caveats that needed to be explainedcarefully and objectively which, in the view of the Authority, could best be done by officialstatisticians producing a statistical release in line with the Code of Practice.

In his reply (link: http://www.statisticsauthority.gov.uk/reports—correspondence/correspondence/letter-from-rt—hon—iain-duncan-smith-mp-to-sir-michael-scholar—25-january-2012.pdf), the Secretary of State for Workand Pensions noted that the report had been prepared by departmental statisticians to appropriate professionalstandards, and that no breach of the Code of Practice had taken place. He also noted that although there wereno plans to repeat the release, the Authority’s view that it would be appropriate for any subsequent release tobe designated as official statistics would be noted.

In addition, the National Statistician’s office contacted DWP statisticians directly, seeking an undertakingthat any further publication of such statistics would be handled as an official statistics release. Finally, theNational Statistician’s guidance on identifying official statistics was also updated, to provide additional claritywhen determining the status of similar analyses in the future.

April 2013

Printed in the United Kingdom by The Stationery Office Limited05/2013 024384 19585

PEFC/16-33-622

communicating statistics: not just true but also fair...communicating statistics: not just true but...

Documents