During her 30-year civil service career, national statistician Jil Matheson has seen a lot of change. The UK Statistics Authority chief discusses the ever-evolving roles of data-crunchers with Suzannah Brecknell
Jil Matheson has been a government statistician for all of her working life, joining as a graduate trainee and working her way up to become national statistician and chief executive of the UK Statistics Authority (UKSA) – the person charged with oversight of the UK’s official statistics system. Yet she identifies just three ways in which the role of a statistician has changed in that time: “What we do; how we do it; and who we do it for!”
That might suggest the statistician’s job is unrecognisable from the days when Matheson joined the civil service, but she says the fundamental aim remains the same: “Using well developed methods, tools and techniques, to give a real insight into what’s happening in the world”. Within that, of course, there has been a lot of change – not least because people have become interested in different aspects of the world: “So from the ‘90s onwards [there was] increased interest in migration statistics, for example; or more recently, there was more interest in wellbeing – understanding how people live their lives and their experience of living their lives.”
The skills needed by a statistician have changed, too: it’s increasingly important that statisticians can not only understand numbers, but also communicate their meaning. “It’s no longer enough just to be an expert modeller or methodologist,” says Matheson, explaining that this change has come about in part because of the change in who government statisticians work for. They remain employed by civil service departments, but in 2008 the Statistics and Registration Service Act set out that the production and publication of official statistics must serve the public good. “That’s quite a profound change from 20 years earlier, when statistics were essentially about the needs and interests of the government of the day,” says Matheson. “Thinking about public good means there’s a whole new audience.”
One of the key ways in which statisticians communicate with these new audiences is through the official commentaries that accompany statistical releases. These are supposed to give context and explain the significance of the data, but a 2012 report from the UKSA – which assesses all government statistics against a code of practice, to judge whether they can be judged National Statistics – said commentaries are the aspect of statistical releases which most often cause concern for assessors. Commentaries “attracted more requirements [for improvement] in assessment reports than any other aspect of the code,” said the report: “Some official statistical outputs lack commentary altogether, but more often it is the limited helpfulness of what is written that needs to be addressed.”
Matheson notes that this may because “the tradition in some places has been that statistics are produced for a particular expert and known user, often somebody in your own department, and suddenly – or it feels like suddenly – you need to think about a much wider audience.” This brings in non-experts, the media, and decision-makers in the wider public and other sectors, and Matheson notes that “conveying sometimes very technical stories to that non-expert audience isn’t something that comes naturally to some people: people who’ve been recruited because of their great facility with numbers.”
To help out, Matheson’s team at the UKSA aims to support statisticians with both general training, and tailored advice on particular statistical releases. They also look for support outside government – “We have some great friends in the media and elsewhere who act as critical friends, giving feedback on our releases” – and Matheson encourages other civil servants, such as those in private office and policy roles, to play their part in driving demand for better communication on stats. “If you are briefing a minister, then you want something simple and clear too,” she says, “and you have a right to expect that from your statisticians.”
There are, of course, many examples of good communication within the Government Statistical Service (GSS) – the profession encompassing all statisticians and headed by Matheson. “The one that stands out in my mind is the work that the Ministry of Justice did following the  riots,” she says. The department was “incredibly responsive”, putting together statistics quickly to meet public demand, and then “they presented it clearly” at briefings where statisticians spoke directly to the media – another important aspect of communicating statistics. While more statisticians now have media training, she says, “some departments are still a little twitchy about letting their statisticians talk directly to the press.” This reflects a broader reluctance to let officials talk to the media, she says, “but those who have tried it really see the value and the importance of it. For statisticians, being able to answer questions that the press have, being able to explain how the numbers have been produced and what they show, is a really important part of building trust and helping good reporting.”
Spin and statistics
Politicians also play a big part in communicating government statistics, and their input hasn’t always been conducive to a fair and honest debate. Matheson seems resigned to the fact that politicians often skew the statistics, slightly side-stepping CSW’s question about whether some periods of her career have been worse than others in terms of political spin. “Politicians are going to say what they want to say, and they will use the evidence in the way that they want to use the evidence,” she says, though she adds that “what has improved enormously is the independent publication of statistics, so those claims that politicians make can be checked and verified.”
She also thinks we could improve trust in statistics and politicians themselves by removing any form of pre-release access – the practice of giving ministers statistics prior to their public release. Ministers currently see statistics 24 hours before publication, but Matheson thinks that if the practice must remain, it should be reduced to just “one hour, with the statistician responsible for the numbers briefing directly the minister or the press office or the chief policy maker who may be expected to comment.” (See news).
How they do it
Statisticians’ audiences and the kind of topics they investigate have changed, then – and another big change is in progress. While traditional methods of surveying and research remain, statisticians are increasingly using other sources of information to build insight into what’s happening in the world. One such source is the administrative data which departments produce as they go about their daily operations. The ONS has a team looking at how it can use this kind of data more effectively, with a particular focus on how administrative data can be used in the 2021 census.
When the coalition first came to power, Cabinet Office minister Francis Maude spoke of entirely scrapping the census and relying on administrative data to gather population data. Instead, following consultations and on Mathesons’s recommendation, a hybrid approach is being used: retaining a ten-yearly census, and supplementing this with statistics gathered from administrative data in the intervening years.
In fact, Matheson strongly believes that government could and should make more use of administrative data. “One of the things that has frustrated me when I look at the civil service is that there is so much fantastic data and it’s an under-used resource, both for individual departments and collectively,” she says. There are, of course, legal limitations to the ways which data can be shared and re-used, but there are also cultural barriers and concerns that data quality remains poor – one of the main reasons why she advised against using administrative data alone for the census. “That quality,” she says, “will improve when the data get used.” When senior leaders “are looking at the data and using the data to make operational decisions” – when the data is seen as crucial to the business – it will “get the attention that it really deserves”, she believes, driving improvements in quality and making it more valuable for statisticians and researchers as well as departmental decision-makers.
Alongside the drive to use administrative data is the emergence of ‘big data’ – the availability of very large databases, often gathered via digital means – as a resource for statisticians and policy makers. Again, a team in the ONS is exploring the potential of big data to inform national statistics – for example, using mobile phone data to track travel-to-work patterns could be “really, really powerful if you’re a transport planner or a local authority”.
The ONS team is also exploring the limits of the data, says Matheson, noting that another challenge for the statistics profession will be “how to distinguish statistics [that can be used] for decision-making because they’re well-founded, they’ve got a strong methodology and they’re reliable, from all the other data that’s out there.”
There’s another challenge: public attitudes to this kind of data re-use. The NHS’s care.data scheme, designed to use anonymised health data to improve healthcare, has been put on hold following criticism of the information campaign accompanying it and strong opposition in the media. Will it ever be possible to win the public over when it comes to using their data for reseach purposes? Matheson believes so, but says it will be a question of not only communicating the benefits, but also having clear governance and strong security arrangements “in place and available for inspection”.
The ONS, she notes, “has a fantastic track record in protecting the privacy and confidentiality of the data that it collects because we only ever collect data for statistics and research purposes; we’re not interested in individuals.” Nonetheless, she acknowledges that people may be more comfortable sharing “vast amounts of personal data” with supermarkets than with arms of government. This, she suggests, is partly because people can “see a direct individual benefit” when they share their data with private companies – tailored special offers, for example – but “for stats and research, you’re appealing to benefit for the country or benefit for medical research or benefit for future generations, rather than a direct transactional benefit.”
The value of statistics
If the role of a statistician has changed so much, have attitudes to statistics also changed? Matheson believes the public are increasingly interested in statistics (she reports that one or two taxi drivers have recently expressed an interest in her job, rather than groaning or trotting out the quote about “lies, damned lies and statistics” – a good job, as she says she’d like to “throttle” people who use that quote). In the civil service, too, there is a “recognition that data and statistics now play a much more central role in policy making, in evaluation and in operational decisions.” No longer “just a back office function, [statistics] can actually drive decisions and help with policy.”
She believes this change has accelerated with the open policy making agenda: “For open policy you need open data, and the two things I think have developed together.” But there’s some way to go, she says: there needs to be more understanding of the “real value of data and statistics”. Improving this understanding will be a priority for the future, she suggests, saying: “I’m sure my successor will want to pick up that theme.” For as we speak Matheson is in her last week as the national statistician; from next month, her post will be held by John Pullinger, currently the librarian and head of information services at the House of Commons.
The appointment of an external candidate (albeit one who began his career in the civil service) to replace Matheson could be seen as justifying concerns raised in 2011 by the outgoing UKSA chair Sir Michael Scholar about the state of the government statistics profession. Sir Michael wrote to the cabinet secretary and the head of the civil service, noting that the statistical profession in government was “larger but flatter” than in previous decades – a situation which, he suggested, would “perhaps particularly concern you when, as part of your succession planning, you look around for a successor to Jil Matheson in some years’ time. The field will thin because there are so few senior posts for statisticians."
CSW doesn’t ask Matheson to comment on the field of applicants for her post – but she recognises Scholar’s picture of a large, flat profession, and shares his concerns about that picture. “Of course I would like to see more [statisticians in senior posts],” she says. “I think the civil service would benefit from having senior people with that kind of facility.” She is a little elliptical when asked how this situation came about, suggesting that it’s “something about what departments value” and pointing to the civil service reform agenda – which aims to “increase the skills and diversity of senior decision making in departments”. The implication is that it’s another case of the civil service valuing generalist skills above specialist or operational ones.
So how can the problem be addressed? “There’s a lot of potential in the Civil Service Capabilities [Plan],” she says: this highlights “the need to strengthen the role of professions” and “recognises that that requires training; requires stronger career management and talent management; requires recognising the variety of skills that people can bring [to senior management].” Not only would supporting more statisticians to reach senior levels in the civil service drive better use of data, she says, but it could also help address another challenge facing the service: increased competition to recruit people with statistical and numerical skills. As all employers try to use data more effectively, the service will need to make sure that “we have a really good offer for people who’ve got those skills – and that will include pay, but it will also include career prospects.”
Matheson is leaving behind a very different Government Statistical Service to the one she joined in 1975: nowadays, communication matters as much as modelling; the primacy of public good is enshrined in legislation; and mobile phone signals are becoming as important to the statisticians’ toolbox as surveys and clipboards. She has an evident sense of pride in this service, in the “men and women across government who are really responding to that agenda [of public good] with enthusiasm and integrity”. Her professional satisfaction has come from helping others to accurately understand the world, and she seems confident that in the years to come government statisticians will still have the skills to fulfil that role – not just for fellow civil servants, but for many other parts of society.
The challenge for the civil service will be to sufficiently value these statisticians. Asked for her final thoughts and advice for other civil servants, she says firmly that it must be understood that statistics “really matter”. She’s referring to those men and women in the GSS: “The skills that statisticians have are essential to both operation and policy, and to public accountability,” she explains, “and departments need to really make the most of those skills.”
See also: news