Role of statistics in a democratic society Thank you very much for inviting me to speak to you this morning. I am delighted that you have extended this invitation to me to speak on today’s important topic.

I would like to set out my thoughts about the relationship between statistics and the mechanisms and processes of a modern democratic society. The processes of democracy do not run on political rhetoric; punditry and anecdote. Or at least they do not run very well on those fuels, no matter that they are in seemingly unlimited supply. What, in fact, democratic processes need is information on which the public can rely in deciding who to vote for, information on which elected representatives can rely in holding public institutions to account, and information on which public institutions can rely in devising and carrying out public policies. That is the democratic model – an informed public driving the machinery of public administration through elections, through accountability to the legislature and through direct feedback to public institutions.

Official statistics are a prime source of public information of this kind. General elections can and do turn on, for example, unemployment figures, balance of payments numbers, retail price indices, and GDP growth or contraction. Local elections can turn on the performance of schools, on immigration, and on local jobs. Social services and police forces should be judged by their overall performance not by scandal and witch-hunt. And all that, or most of it, means statistics. There is no other way. Statistics are not the only driving force but they are a specially important one.

When I addressed this gathering two years ago I remember acknowledging that the safeguarding of democracy required an improvement in public confidence in official statistics ; and acknowledging that that was a huge challenge. I would like to make two points about that today. The first is that an action aimed at improving public trust in the longer term may well in the short term diminish it. Each time the Statistics Authority criticises a statistical series, or the use a Minister or a Government Department makes of an official statistic, we are told that we are damaging trust in that Department’s statistics. I do not accept that. It may be necessary to make the criticism to try to change the Department’s behaviour, in order to make its statistics more trustworthy; and to show the public that the Statistics Authority is there to be a guarantor of quality and trustworthiness. To achieve greater trust in the long run we may need to damage trust today or next week.

My second point is more familiar, and more depressing. Public trust is a fragile commodity and no respecter of boundaries and categories. Recent events on the political level in terms of public confidence in Parliament and its Members will certainly provide a challenging backcloth to the Statistics Authority’s efforts to improve trust in official statistics. There is nothing we can do about this contextual difficulty.

Naturally, over the past year or so a very great deal of attention has been given to the economic indicators. The scale of the financial crisis, and of the recession, has been such as to create anxiety and fear about how rapidly, and how deeply, British and world GDP was contracting, how long the contraction lasted, whether it has ended, and if so how clearly and definitely. This situation lends urgency to a set of further questions: – what is the effect on individuals and on households, on inequality, between rich and poor, between different regions, and different social groups, and so on.

The Report of the Sarkozy Commission deals eloquently with these issues, and with much more besides – notably on sustainability, on the need to improve our ability to measure the state of our environment; and also on the gap between our ability to measure economic outputs, and our ability to measure individual and society’s well-being. You have today many speakers who are better qualified than I to discuss the recommendations in the Commission’s report. But I would like to add a personal angle to the debate.

I have a PhD in philosophy, and one philosopher I wrote about – many years ago – was Aristotle, who was well aware that well-being, or happiness, was wholly distinct from wealth, income and other measures of material well-being. The conceptual work which underpins a notion of measurable well-being is important, and difficult. It should be possible to devise agreed and internationally accepted measures of individuals’ health and education, their personal activities, including work, their political voice and governance, their social connections and relationships, security and environment; but it is more difficult to assign weightings to these characteristics and even more difficult to arrive at agreed measurements of such things as self-respect, self-esteem, and other components of what we ordinarily regard as happiness.

What I think would be wrong, however, is at this point to give up, on the basis that these latter characteristics are subjective matters, and not measurable or comparable. The task for statisticians, as I see it, is to discover what in this area can be measured, to chart the relationships between such characteristics, than to seek agreement within nations, and internationally, so that comparisons can be made, across time and across nations.

As I see it, the concept of well-being which this process will carve out is unlikely to be identical with the concepts of happiness of philosophers, poets and novelists, because what we will discover is a statistical concept, measurable and comparable by numbers. But that non-identity should not deter us from trying to establish – as the Commission recommends that we should – better measures of well-being, both objective and subjective.

Now I return more strictly to the subject of my contribution this morning. I have said many times before that an over-riding goal for the Statistics Authority is to improve UK official statistics to ensure they meet the needs of the user community in our democracy, of which citizen-users are a vitally important part, and to ensure we have as accurate a picture of our society and our economy as it is possible to paint. However improving statistics means more than ensuring that they are as reliable as they reasonably can be. It means also that the user can readily find the statistics in which he or she is interested; and having found the figures, is told all they need to know about the strengths and weaknesses of the statistics, so that the ‘end-use’ of the statistics is as well-informed and beneficial as possible. This is a big agenda. It embraces finding ways dramatically to improve online access as well as developing explanatory text that is itself easy to find and easy to understand. I understand that there will be a presentation later today about developments in this area in the United States. I can assure you that the Statistics Authority is paying close attention to these and is keen to work with Andrew Dilnot and the whole user community to explore the best way forward in this country.

I, Jil Matheson, the National Statistician and all our colleagues on the Statistics Authority Board are very much aware that the statistical service does not produce official statistics for their own sake. They are not trying to archive the 21st century in figures – as interesting as that might be. The value of their work, of all official statistics, is in the policy discussions they facilitate or contribute to, in the empirical evidence they provide to guide action, and in the intrinsic value they are to those who use them and rely on them. If official statistics are genuinely to be used to drive democratic debate, then the user must not be seen merely as a passive participant – the user community must represent more than just a final resting place at the end of the production process. It is not just statisticians who must show how statistics can add value to the processes of democracy; there is a vital role for users, academics, the media and other commentators to demonstrate that their use of official statistics supports the public good. By demonstrating the value of their use of statistics, users will be a much stronger influence over the shape and form of the statistical product; and thus help us to ensure it continues to benefit society and democracy.

The powerful role of official statistics mean that there need to be effective governance structures in place to oversee it all, and this is very much at the heart of what the Statistics Authority is all about. But, while governance is important, there also needs to be more emphasis on identifying, documenting and meeting the current and future statistical needs of all sectors of society. The Authority will be publishing a report about that in the New Year. A well-informed society pushes and encourages, and sometimes constrains, the institutions of the state to take the right decisions, for the right reasons, and on the back of the right evidence. Our role in the Statistics Authority is to support users, academics and commentators to engage with, and have influence over, the statistical product and how it is used for the public good and the wider public interest.

Before I close and hand over to Enrico and Paul, I just wanted to pause and say how pleased I am that the Statistics Authority has been able to assist again this year with the funding for this Conference. We see the work of the Users Forum and, indeed, this conference, to be well worth supporting in whatever way we can. I am also pleased that the Authority has seconded one of our staff – Liam Murray – to work with Andrew Dilnot in providing administrative support to the Forum. Please keep the Statistics Authority informed of your views and thoughts, either through Liam or with us directly.

Thank you very much indeed again for inviting me to speak to you this morning.