Reforming the rules for statistics producers: our views
Most people don't know that statistics producers - the people in government who collect the numbers we all rely on, from unemployment to inflation - have an independent regulator, just like solicitors or dentists.
Well they do, called the Office for Statistics Regulation, or OSR for short. Their job is to ensure our statistics are accurate, unbiased, and answer the important questions.
The OSR publishes a Code of Practice, which sets out the rules that producers have to follow, and it has opened a call for evidence about how well the Code is working. This might look like an obscure consultation by a minor government body, but it's far more important than that.
It matters because when official statistics break or go missing, the stakes are high. We make bad decisions. We can't hold government to account. And when important questions can't be answered, conspiracy theories and distrust fill the gaps.
We have submitted evidence that warns that the Code urgently needs updating. Most importantly, the current Code does not have an effective way to handle data gaps - the places where reliable statistics do not exist.
What’s wrong with the Code?
The Code aims to ensure that official statistics answer the most pressing questions. But there’s often a significant gap between those questions, and the statistics that actually get published.
There are plenty of high-profile examples - for example, reliable, trusted statistics on migration to the UK were lacking for many years.
In our own research, in 2022 we examined the UK’s criminal justice system and found many examples where questions that experts and MPs care deeply about cannot be answered from official statistics. Two examples are:
How many people appear in magistrates’ courts without a lawyer to represent them. This data is a basic measure of access to justice that has been repeatedly (and unsuccessfully) requested by MPs, campaigners, and journalists.
How long people are held on remand, i.e. held in custody before being convicted of a crime. Worryingly, FOI requests by campaigners have shown many people are being held beyond the legal time limit, but there are no official regular statistics on the topic.
The Code should ensure that when stakeholders ask for better statistics, producers help them. But at the moment, it doesn’t. It has very clear requirements that statistics are impartial and accurate, but not that they should be relevant. It doesn’t require producers to understand statistics users’ needs, engage with them, or tackle major gaps.
No stakeholder we spoke to during our research had ever been approached by a producer, or knew about the official channels for engagement. And our attempts since to flag data gaps with producers have been of limited success.
How should the Code be strengthened?
It’s really important that statistics answer the questions that people care about. We propose the following:
Define ‘statistics users’ explicitly in the Code, and set the definition to include the full range of users outside government: Currently, producers tend to focus on the needs of ministers and civil servants than on users outside government. The Code doesn’t help this, because while it makes many references to ‘users’, it doesn’t actually define who those users might be. There should be a formal definition that should include MPs, researchers, journalists and groups representing citizens and service users.
Require producers to do modern, active user research: Again, the Code doesn’t require users to do active research - only to listen to the views of the groups that approach them. But few groups have the time, knowledge and connections to approach statistics producers themselves. Inevitably, this means the groups that get heard are self-selecting and far from diverse. Instead, the Code should require producers to track signals of need for better statistics (such as disputes in the media), and to recruit and interview a diverse set of users from the groups above.
Require published work programmes to address these needs: The Code encourages producers to publish programmes of their future work, but this rarely happens in practice. Publishing work programmes should be compulsory, should allow time for consultation, and should require producers to describe their user research findings and how they are being addressed.
Introduce penalties for producers: Finally, the Code should provide some stick as well as carrot - it should penalise producers who can’t show that they are meeting the needs of a wide range of users.
With limited resources, it’s understandable that statistics producers focus on the needs of ministers, ahead of more distant users like MPs and journalists. But the result is that questions of significant public interest go unanswered, leaving space for misinformation to flourish.
We call on the OSR to update the Code to help statistics serve the public good. For more detail, see our full response.