05 August 2005
A democracy needs to be sure that the figures on which decisions are made are accurate and do not mask a hidden agenda. The Statistics Commission chair would like to see safeguards enshrined in law
It is self-evident that the public needs to be able to trust official statistics as being sufficiently accurate and not perverted for political ends. Without this trust, neither the public nor its elected representatives can hold government to account.
Official statistics drive a multitude of everyday decisions in government, public services, and even in our individual lives. For instance, more than £80bn a year is allocated to local government and the NHS using census-based population figures.
And, as was powerfully illustrated in the run-up to the May 2005 general election, they are also ever more central to the highest level of political debate. So official statistics matter, and public confidence and trust in them is crucial for a properly functioning democracy.
A public survey carried out last year by the Office for National Statistics indicated that more than half the people interviewed thought there was political interference in the production of statistics.
The Statistics Commission's report Official statistics: perceptions and trust, which looked into the views of 'opinion-formers', showed that, while these had a lot of respect for UK official figures, many believed that there needed to be more independence in their production.
Trust in official statistics can also be damaged if the figures are used for inappropriate purposes. Even technically accurate statistics, such as school exam results, should not be trusted too far in guiding decisions for which they are not well suited, such as on the relative performance of schools.
On the other hand, statistics such as those of recorded crime — which have been at the centre of much controversy and multiple interpretations — can tell you important things if understood and analysed with care.
The origins of this lack of trust probably lie in multiple factors – falling trust in the wider political process; a few real errors made by statisticians; 'scandals' in the news media; lack of even a rudimentary understanding of statistics in the populace; and hugely contradictory interpretations of official figures by politicians have all probably contributed.
So how can things be improved? Most countries have a Statistics Act designed to ensure that statistical work is beyond inappropriate influence and carried out according to purely professional considerations.
In the UK, we do not have such over-arching legislation but rely instead on a range of other structures, conventions and agreements to achieve the same goal - such as those in the Framework for National Statistics and related Code of Practice.
The framework established the Statistics Commission and set out the principles that should be followed, explaining the roles of the national statistician, ministers and other key players. This was an important step forward.
However, over the last year - notably in our May 2004 report Legislation to build trust in statistics - the Statistics Commission has argued that a stronger set of arrangements is needed.
Our preferred model focuses on a new, more robust, unambiguous, statutory code of practice - to be developed by the national statistician - binding on all government departments, and some other bodies, which assemble and disseminate the official figures.
Such arrangements could ensure that the national statistician, along with an independent commission answering to Parliament, had effective oversight of the quality of statistical work carried out across a multitude of government departments and agencies.
At present, the real authority is limited to the Office for National Statistics - although about half of all official statistics are produced in other departments. The new commission would approve the new Code of Practice and would have powers to obtain information in pursuit of its enforcement.
Unlike opposition parties, the government has not yet responded to our proposals for new legislation on statistics. But in the absence of a grand solution, the imminent review of the framework document provides an opportunity for government to take the reform agenda forward.
Many problems arise because of misunderstandings - resources dedicated to explaining statistics would be a good investment.
We also need a transparent planning mechanism so that what is collected can be seen to reflect the needs of users in business, local government and other sectors, as well as those of central government.
We need statisticians always to publish their figures, with notes on their limitations, before or at the same time as governmental press releases.
We could not, and would not seek to, take official statistics out of political debate, but everyone must strive to keep - and be seen to keep - politics out of official statistics.
David Rhind is chair of the Statistics Commission