Making Sense of the Code

We at OSR are often focused on all things government – how and why departments produce and publish official statistics and then use them in the public domain. It’s why OSR exists and why there is that one lovely book to rule us all – the Code of Practice for Statistics.

But what does the public think about official statistics? And the Code? It matters because the value of official statistics is that they provide everyone with the information needed to understand and improve our lives and society. Thus, we partnered with Sense About Science, an independent campaigning charity that challenges the misrepresentation of science and evidence in public life, to meet with nine members of the public to hear their thoughts.

The focus group participants ranged from a PhD student and a food blogger to a retired GP and a Comms officer at a charity. They were chatty, engaged, curious and challenging – in other words, fantastic company with which to discuss statistics.

We asked them if they trusted statistics. ‘It depends’ was the common response – depends on where they read the statistical claim, the agenda, what’s at stake, has it been peer reviewed, and what’s the source.

We asked if they had seen bad examples of statistics and all nodded vigorously.

We moved on to the topic of the Code, asking what it is and why they think it exists. Participants consider it a mechanism for quality control and best practice. They think it exists ‘for public confidence and trust,’ ‘to hold people to account,’ and ‘an organisation who signs up is effectively saying they’ve nothing to hide.’

Interestingly, one user did not see the value of the National Statistics ‘kite mark’, stating that the statistics themselves might be perfect . . . but the interpretation and spin put on them is the problem.

The focus group participants reviewed the pillars (Trustworthiness, Quality and Value) and principles for each within the refreshed Code. Overall they found the pillars a bit confusing and that Value, in particular, required further explanation.

Some words and concepts throughout needed unpacking like ‘robust.’ Fair enough. Robust is a word woefully overused in the civil service and what exactly does it mean: solid; good; meaningful?

They also felt the idea of ‘public interest’ required clarification. For example, sometimes ‘the public is interested’ is conflated with ‘the public interest’.  One user suggested ‘national interest’ or ‘public good’ were more meaningful terms.

All in all, a great experience for us.

To hear that in general the nine panelists do trust statistics and question what they read and expect government to follow the Code was insightful.  Their comments were invaluable and will result in a clearer document.

Thank you to Sense About Science for facilitating and to those in the focus group.

Code of Practice Stocktake Report

Today we publish the Code of Practice Stocktake Report.

Why did we carry out the Code Stocktake? After all, the Code seems to have served the public and Government well since it was published in 2009 – at least, this was a comment we heard from many of the people we spoke to as part of the Stocktake. But the world has moved on since 2009, especially in terms of the use and dissemination of data. We also had a sense that sometimes the Code is seen in Government as an inhibitor and a barrier. Therefore we wanted to take stock of how flexible the Code was to these changes and hear user views on how the Code works in practice.

Engagement was very much at the heart of the Code Stocktake. Substantial effort was devoted to successfully communicating through workshops, presentations and meetings across the UK. I want to thank those that contributed their invaluable thoughts and comments and the team that carried out the Code Stocktake – Richard Laux, Johanna Hutchinson and Tegwen Green – for their hard work and commitment. This rewarding and collaborative exercise has enabled the development of the proposals set out in the Code Stocktake report published today; proposals that have enhancing statistics very much at their heart.

We started the Code Stocktake by exploring what the Code of Practice is for. We concluded that the Code ought to be valued less as a set of specific rules and more as behavioural norms for providers of statistics. We heard how the Code ought to provide statisticians with a framework and a set of principles to help navigate the balance between serving Government and the public good. We have summarised these drivers as three simple, high level outcomes: trustworthiness, quality and value.

With these outcomes in mind, we explored the extent to which the Code could do an even better job as a guide to behaviour to those providing statistics to society.  We concluded that while the Code has been OK, it could be improved. It should be being clearer about what it aims to achieve.

For example, we ran a few sessions asking people who use the Code regularly (for example, statisticians within the GSS) to name the qualities they admire in statisticians. Things like objectivity; willingness to produce statistics openly and honestly even if the stories they paint a picture the Government may not want to hear; capacity to communicate complexity in a clear, convincing way.

These are human qualities that underlie the existing Code; yet somehow this human dimension is a bit hidden, and the current text comes across as a set of strictures. Reframing the Code around the principles it seeks to achieve, and why, would help move it towards being a more empowering and transformational document. This Code could then be supported by guidance (not rules) particularly in areas where users of the Code feel guidance would be most valuable.

This leads us on to a further question of scope. Here we distinguish between scope and positive influence. We want to maintain a formal regulatory scope as official statistics, which is what we’ll review and where we’ll expect full compliance.  But we’ll advocate the Code as a set of principles that can be applied more broadly, to a wider set of numerical outputs than just ‘official statistics’, but at the discretion of the providers. This reflects conversations we’ve had with analysts who said it’d be great if we highlighted the capacity of the Code to provide a guiding light: to enable trustworthiness in a wider range of published outputs in difficult or ad hoc situations.

To put it simply: the outcomes of trustworthiness, quality and value are universal aspirations for all people publishing analysis and data. We can’t see any reason why they shouldn’t be able to draw on the Code’s principles as a guide, regardless of whether they are producing statistics or another kind of analytical output. Boundaries aren’t really helpful; it’s the outcome of public value that matters.

So our aim is for a more agile, flexible refreshed Code, which provides a framework, a set of principles, to ensure that statistics and other types of numerical information meet the highest standards of trustworthiness, quality and public value. The detail behind this vision is provided in the Code Stocktake report, which we’ve published as an exposure draft to invite comments and observations on our proposals. So do get in touch and let us know what you think.