Public Administration and Constitutional Affairs Committee
Oral evidence: The Work of the UK Statistics Authority: Scrutiny Session, HC 781
Ordered by the House of Commons to be published on 21 October 2021.
Members present: Mr William Wragg (Chair); Ronnie Cowan; Jackie Doyle-Price; Rachel Hopkins; Mr David Jones; John McDonnell; David Mundell; Tom Randall; Lloyd Russell-Moyle; Karin Smyth; John Stevenson.
Questions 1 - 40
Witnesses
I: Professor Sir Ian Diamond, National Statistician, UK Statistics Authority; Ed Humpherson, Director General for Regulation, Office for Statistics Regulation; Sir David Norgrove, Chair, UK Statistics Authority.
Written evidence from witnesses:
– [Add names of witnesses and hyperlink to submissions]
Witnesses: Professor Sir Ian Diamond, Ed Humpherson and Sir David Norgrove.
Q1 Chair: Good morning and welcome to the Public Administration and Constitutional Affairs Committee. This morning the Committee is holding a session into the work of the UK Statistics Authority and the Office for National Statistics. The Committee is very much looking forward to hearing from our witnesses today, given the public’s increased exposure and reliance upon data and statistics on a day-to-day basis throughout the Covid pandemic.
I am going to ask our three witnesses to introduce themselves for the record, please, beginning with Sir David Norgrove.
Sir David Norgrove: I chair the Authority.
Professor Sir Ian Diamond: I am the National Statistician.
Ed Humpherson: I am the head of the Office for Statistics Regulation.
Q2 Chair: Thank you. My first questions are directed to Sir David Norgrove. Sir David, how would you define your tenure as Chair of the UKSA? What challenges has the organisation faced and how have you responded to them?
Sir David Norgrove: I have been thinking about this because my time as Chair comes to an end in March next year, so I have been looking back. I was thinking about how it felt when I was at my pre-appointment hearing with this Committee just under five years ago. I suppose at that stage there were two main concrete things in prospect. One was that I could see that John Pullinger was likely to retire as National Statistician and the other was the census.
On the first, I tend to think that the most important duty of a Chair is to have the right chief executive, and so I saw it as extremely important to get the right replacement for John and I am delighted that Ian is in post because I think he has answered that question resoundingly successfully.
The other was the census and, as with so much else here, I do not think that I or the board can claim credit for the success of the census. This is very much a collaborative thing led by the board but, also, delivered by the executive. I think the census is shaping up to be an outstanding success, so I think those two concrete things are very important.
Q3 Chair: At that pre-appointment hearing, you said that during your time as Chair you would want to work with the National Statistician to raise the status—I think you said—of the role played by statisticians and economists in Government. How successful do you feel you have been in achieving that?
Sir David Norgrove: I think the status has never been higher, and that has been helped in a way by the pandemic. The way that statisticians have responded has been just extraordinary and the delivery of existing statistics and then the speed and flexibility of the response to the new demands has just been fantastic I think, so I don’t think they have ever been more highly regarded.
Q4 Chair: In terms of statistics, also at that pre-appointment hearing, you hoped that through your tenure as chair statistics would somehow look different, be more open, be better and indeed be timelier, and how successful would you judge those things to be?
Sir David Norgrove: I think that statistics have changed hugely in the last five years. I do not claim credit for that. I think it is mostly down to the executive, but existing statistics have been transformed, I think, both through new methods and through the use of administrative data and the response in the pandemic illustrates that. Also, there are lots of new statistics, and again shown by Covid, and timelier too.
If you think of monthly GDP statistics, that has been quite a remarkable achievement. Also, statisticians are now much more outward looking and much more responsive to user needs. Thinking about the past, I was struck that at my first meeting with the then Governor of the Bank of England, Mark Carney, he criticised ONS, or challenged us, for commenting on statistics when they are published. We now have people appearing on the “Today” programme, so people are much more externally focused.
The other thing I would mention, which we have not really touched on, is the way that OSR is now much more firmly established as a credible regulator with a strong, independent voice. Those are the things I would point to.
Q5 Chair: What challenges do you foresee for your successor?
Sir David Norgrove: I think there is going to be a continuing challenge around maintaining the scale of change. We are just at the beginning of a revolution, I think, in statistics using administrative data. The creation of a platform within ONS I think is going to make ONS the hub for Government data. That is going to lead to challenges that are technical, political and ethical. That is one stream of challenges and you might want to go into that in more detail if you wish.
Another challenge is around maintaining the independence of statistics and the quality, value and trustworthiness through OSR.
Q6 Chair: Thank you. We will certainly be exploring those topics as we go through the session, Sir David. Could I just ask Sir Ian—and then, Sir David, you may wish to comment as well—about some the leadership within the ONS, noting that in the last six months two of the ONS’s deputy National Statisticians have departed—a significant change to leadership. Are you able to comment on that and how resilient do you feel the organisation is?
Professor Sir Ian Diamond: I will start by saying that I think it is incredibly resilient and then I will comment. When you have a very successful organisation and people working at a very high level, they will automatically be attracted and be attractive to other organisations. Both Iain Bell and Jonathan Athow have been headhunted to other organisations, giving them new roles that they will find exciting. Would I have liked them both to go within a short period of time? Obviously not but, at the same time, when people have those great opportunities you cannot always put a great timing on them.
Let me just turn to the resilience. First, of course we are recruiting as quickly and properly as we can. Interviews for both posts will be held in the next few weeks and I can report strong fields.
The second point I would make is that we have also, at the same time, been investing over the last years in the quality of our director cadre. We have a very strong group of directors who have been able to step up and take on roles, so that no one outside would see any lacunae as a result of those moves. I have to say, Chair, if I may, that I suspect some of those directors at a very high level are going to move on in the next few years to promotions. They really are very good.
I would also add, finally, that of course at this time last year we appointed Sam Beckett as our first Second Permanent Secretary, so on the economics side - of course, Sam is the joint head of the Government Economic Service - we had extra leadership on that side as well. Sam and I enjoy working together and have built a very strong working relationship which, again, supports both the resilience of the organisation and, I would suggest, the ability for us to manage this short interregnum without those two director generals, but we will soon be back to full complement and full steam. Thank you.
Chair: Sir David, do you have any comment?
Sir David Norgrove: A couple of things. I think the quality of people over the last five or six years has steadily improved. There has been a fair amount of change in personnel at all levels and, as ONS has gained greater national prominence—as well as prominence within Government—that has led to attraction of higher-quality people.
The second thing is that the way that the organisation works together has improved as well. My sense is that John Pullinger inherited quite a siloed organisation. Under his leadership and that of Ian, the organisation pulls together in a much more collaborative way than it has ever done in my experience.
Q7 Chair: Sir David, if I could move on to the letter that you sent to the Committee. You suggested that there was a need for a “firmer central controlling mind.” How could the ONS and the National Statistician take a firmer and more central role within a decentralised system of UK statistics?
Sir David Norgrove: I was writing that in the context of lessons learned from the pandemic. That is the context from which I am making that point.
I do think the statistical system responded extraordinarily to the pandemic. It did hit nails on the head but there was a certain amount of fumbling around for the hammer at the beginning. It took a lot of effort, particularly by Ian—using his position and his personality—to bring things together. The point I would make is that in preparing for a future crisis of whatever kind, people need to think much more clearly in advance about how data is going to be organised and who is going to be in charge and responsible for bringing it together and thinking about what is really needed.
Chair: Thank you. Sir Ian, any comment?
Professor Sir Ian Diamond: No, I think that is a fair point David made. In another—and I very much hope we don’t have another—similar situation I do think there is an advantage in there being one person who is working across government to take those views, in exactly the same way, if I may, that Sir Patrick Vallance has with regard to science right across Government.
I do think, in the situation we have been in over the last 18 months, that David Norgrove’s comments are entirely relevant and good.
Q8 Chair: Sir David, just reflecting on that—because it is my understanding that there isn’t a particularly formal agreement underpinning this notion of a former and more central controlling mind—what have been some of the recent advantages, and perhaps drawbacks, of the ONS taking a more central role?
Sir David Norgrove: If I think about the pandemic, one of the issues there is that our provision of health services is very complicated, with a lot of different bodies. Particularly earlier on, there were bureaucratic issues, as well as issues of genuine disagreements about statistics and about data, that would have been resolved much more quickly had the National Statistician been clearly in charge of how things were to be done.
Just to reinforce the point that we have both made would be my central conclusion. I really do not see any disadvantages from that in a pandemic. In the normal course of events, having a single person in charge of the whole thing in detail would not work but in a crisis I think that is essential.
Q9 Chair: Developing this theme still further, perhaps, but in terms of heads of profession within Departments, Sir David, do you think those heads of profession are adequately supported by the National Statistician?
Sir David Norgrove: My sense and observation is that both Ian and Ed, who plays a role in this too, as head of the OSR, are very actively supported by them.
Professor Sir Ian Diamond: Chair, if I—
Chair: Sir, Ian, yes, I was going to bring you in. I want to bring you in to give your perspective on that as well.
Professor Sir Ian Diamond: I apologise for jumping the gun on that one. We have put a lot of work into that. It is something that I am incredibly keen on, not only in the context of the Government Statistical Service but also in the context of the Government Analysis Function, which of course brings together statisticians, economists, operational researchers, actuaries, geographers and social researchers.
In so doing, we meet regularly, both the heads of professions of statistics and the departmental directors of analysis—not together but in separate meetings—and we discuss items of interest. We discuss standards. We discuss the issues that are happening to individuals and we build a network. That means regularly sitting down—and I am afraid these days having a virtual cup of coffee—with heads of profession. I am hoping soon to make it back to real cups of coffee with people. I know that people feel comfortable getting in touch with me if there is something they want to raise, and we move very quickly to address it.
There have been occasions when that has moved to ensure coherence across the four Administrations of the UK, for example, when we got together a little earlier this time last year to make sure that everywhere had a similar definition of what was, sadly, a death from Covid. I do think the family built the network and people feel empowered to make a phone call if they feel there is something we can do to support.
Q10 Chair: I will bring Ed Humpherson in at this juncture. In July of this year, OSR shared a statement on data transparency and the roles of heads of profession for statistics. That statement said that “It is vital that senior leaders support heads of profession to do their jobs effectively and that decisions to release statistics and data are free from political interference.” What practical steps can senior leaders take to ensure that their decisions to release statistics and data are free of any political interference?
Ed Humpherson: Senior leaders can do three types of things: things at the level of principle; things at the level of professionalism; and things at the level of process. The most important of those is principle. For us there is a fundamental principle that is part of the code of practice that there should be equality of access to data and statistics that are used in public debate.
To support transparency, we would want senior leaders, permanent secretaries, directors general across Government to endorse and promote that principle, so that, when their Department or their Ministers are quoting data and statistics, those data and statistics should be freely available and published in an accessible and clear way.
Secondly, I think those senior leaders can continue to enhance the role of heads of profession—these professionals within their Department—and recognise that, if the head of profession is saying, “There is some data here that we ought to publish because it has been released publicly,” they immediately confer the privilege on the head of profession to make that call.
The third thing is at the level of process where I think there are some things that some Departments do well, which others could learn from. Some Departments have an established process of publishing what are called ad hoc statistical releases. These are things that aren’t planned but which would come out when there is an issue that emerges. Other Departments don’t do that. Whenever we encounter a Department that doesn’t do that, we say, “Well, look at the Department for Work and Pensions or the Department for Education. They have a really good system of ad hoc releases. You should adopt that process as the norm.”
I think there are things at the level of process; things about resolving how the publication of data should bypass what are called the grid slots, the process of planning government communications, because it is something that has been used publicly. It is already in the public domain. It doesn’t need a grid slot.
Therefore, yes, I think we would say principle, professionals and process would be the thing senior leaders could do.
Q11 Chair: Thank you. Sir Ian, who has ultimate managerial responsibility for heads of profession? Is it yours or the Department permanent secretary?
Professor Sir Ian Diamond: The Department permanent secretary and I provide mentoring and overall professional and intellectual leadership.
Q12 Chair: Have there been any instances of tension between the two?
Professor Sir Ian Diamond: No, to be absolutely honest, there hasn’t. I would feel entirely empowered to ring up or contact a permanent secretary if there was such a tension, but there has never been one.
Q13 Chair: That is helpful. Just as a final reflection from this opening set of questions, perhaps to Sir David first: how can senior leaders in Government promote a culture that values good use of data?
Sir David Norgrove: I think the primary responsibility lies with the statisticians themselves to produce data and statistics in a way that compels attention, is clear and understandable and meets the needs of policymakers, whether those are officials or Ministers. How can Ministers themselves and senior officials promote good use of statistics? The obvious things are about the way they behave; how to encourage them to behave better. As Ian and Ed have both said, it involves a whole range of different approaches and a constant effort.
By and large, I think people in the British Government do use statistics pretty well and we focus on the issues that arise. Sometimes it feels there is a lot but if you look at that in the context of the sheer volume of statistics that are quoted by Ministers and others, I think people get a bum rap if they are accused of misusing them. Mostly my experience is that when statistics are misstated it is usually cock-up rather than conspiracy. It is not usually egregious. It is usually because they are under pressure or they have misunderstood something. Then people get themselves into a hole and we have to help them dig themselves out but by and large I think Ministers use statistics pretty well.
Chair: Thank you. Ed Humpherson, your observations?
Ed Humpherson: I very much agree with David on this. Whenever I have spoken to a permanent secretary about an issue or, indeed, to a Minister, there is a core understanding of the importance of getting it right. We don’t get push back at that level. We don’t see lots of examples of nefarious attempts to misuse statistics.
I think there are sometimes issues around some things I have already mentioned, things getting seized up in processes and probably a lack of prominence being given to the principles of the code of practice. It is not deliberate nefariousness. It is more an issue perhaps of neglect and, even then, that is very isolated.
Sir David Norgrove: If I could just add one thing?
Chair: Yes, you can, Sir David.
Sir David Norgrove: I do not think that in my time I have come across a case where a Minister or an official has tried to change a statistic for political purposes.
Q14 Chair: Okay. Not necessarily changing a statistic but selecting statistics for greatest effect. Sir Ian, your observation on that question?
Professor Sir Ian Diamond: My observation is very similar to David’s. The first point is that statisticians do have a responsibility to be clear that they are providing insight, not data deluge. To provide that insight requires a culture that says the statisticians want to be—and are—involved in conversations with their policy colleagues from day zero. That is certainly something that Tamara Finkelstein, who heads the policy function across Government, and I are absolutely passionate about.
In that way, we are able to have good conversations about what are the questions for which there are answers to be required, and then we can assess, first, whether we have the data to be able to do that and, secondly, if we do not have the data to answer those questions what we need to do to get it if that is possible. That is super important.
Where we have had real success in the last 18 months is when a member of the ONS has been embedded in a different place—for example, in the Covid taskforce—and really being able to say, “What are the questions we need an answer to? Then we can provide independent and rigorous answers to those questions.” That has enabled us to get a position where Ministers and others are very keen to get the information and to use it properly.
The final thing I would say is that there is a responsibility on everybody when they are presenting statistics to present them well. I had a conversation last week with the Chancellor of the Duchy of Lancaster about good practice in presentation of statistics. He invited me to put together a short piece on good practice and we sent that to him yesterday.
The important things are good presentation, good insight coming from good questions and then all leaders—such as yourself, Chair—being prepared to say how important they see the benefits of accurate, rigorous data in informing policy.
Chair: Thank you. I think we can all agree with that, Sir Ian, and on that note I am going to go to Ronnie Cowan, please.
Q15 Ronnie Cowan: Could I ask a brief supplementary on the topic we are covering here, because I am really surprised to hear you say you don’t think politicians misuse statistics for their own ends? I am quoting an article here from 3 October this year in The Guardian. It says, “Politicians and senior civil servants have been publicly called to account for the misuse of Government figures on 47 occasions by the official statistics watchdog… Sir David Norgrove, the UKSA’s chairman, wrote publicly to five MPs, four Scottish MSPs and a member of the Senedd… while other issues were raised with civil servants by Ed Humpherson.” Therefore, are they misusing those figures or just misinterpreting them?
Sir David Norgrove: I did not say that they never misuse them. All I was saying was that, given the vast quantity of statistics that are published by Departments and by Ministers, the number of times we intervene could be a lot larger than it is and that, in practice, I am surprised at how little we have had to intervene than how much.
We have intervened more during the pandemic and made more comments than we did in the years before that, partly because of the huge increase and the interest in statistics but, also, because of the pressure that the system has been under, both in terms of defending the policies but also just in terms of getting the data out and understandable. I am not saying people don’t misuse statistics—they do—but it could be a great deal worse.
Ed Humpherson: While those cases certainly do not give grounds for complacency—that is absolutely clear—they also indicate that among the biggest problems, the one which is most recurring in that caseload where we have written publicly about ministerial reuse, it is about a Minister in any part of the UK—and we have cases not just in England but in Scotland, Wales and Northern Ireland—quoting a number that is not otherwise publicly available. That is the most common thing. We step in and we say, “The information should be available. The public has a right to see on what basis you are making that statement.” That is by far and away the most common example. In fact, even since Sir David wrote to the Committee in February on this issue of transparency, we have subsequently highlighted publicly eight significant and prominent cases of unpublished data being used and a lack of transparency. To us, transparency is by far the predominant area of concern and it is an important one.
Q16 Ronnie Cowan: There is little or no point in organisations like your own producing accurate, timely information and statistics if they are not going to be understood by the people it is presented to. In that vein, the ONS Date Science Campus in partnership with 10 Downing Street offers a data masterclass. How far does this data masterclass go in addressing statistical illiteracy across the UK Government?
Professor Sir Ian Diamond: Can I respond to that? I am so glad that you are aware of the data masterclass. Modesty—as I do a fair bit of it myself—prevents me from commenting on the quality of it, other than to say many people across Government, across the civil service, ambassadors, senior civil servants, permanent secretaries have commented that they have enjoyed it, that they have found it very useful and informative. We are now in the process of rolling it out to a wider range of the civil service, particularly through the campus for civil service training.
It has been a very good initiative. We are very proud of it, but I am not going to pretend that you could do one data masterclass, flick a switch and suddenly everything is perfect. I am hoping in the next few weeks to do some short presentations for Ministers, which will come from the masterclass. If you had the time, I would be delighted to welcome you to one of those.
Chair: I would do that, yes.
Ronnie Cowan: Yes, we are on.
Professor Sir Ian Diamond: It is important that we recognise that this is not, as I said earlier, flicking a switch. We have to be motivated and dedicated to many ways of improving data literacy, not only within the civil service but in our public as a whole.
One of the things that I am very keen on, which we are doing, is: for many people learning about data, in school or in further education or wherever, it is about having relevant examples. Across Government, we have some of the most relevant and beautiful examples to show people the beauty and benefits that data can bring. We need to be doing that as well. I am very proud of the data masterclass, I am very proud of the take-up, but I am not pretending it is a switch that has saved everything that needs to be done.
Ronnie Cowan: Do you have anything to add to that, Mr Humpherson?
Ed Humpherson: Masterclasses are great, and we are very happy to see them being rolled out. Yes, really good.
Q17 Ronnie Cowan: It is interesting that you were talking about the civil service, talking about senior members of Government. We are all bombarded with information these days through social media. A lot of is completely uncontrolled. Trying to sort out the good information from the bad information is an issue we have. In Finland, schools are teaching kids how to interpret statistics so as they can do it themselves. It is a lot easier than just trying to stop people putting bad information out there. Teach people to understand statistics from a very early age. Is there anything in the Government’s mind that you know, or within your own mind, to take what you are teaching and take it into universities, take it into schools as well?
Professor Sir Ian Diamond: I do think we need to be doing that. That was my point about getting good examples. Certainly, I spent some decades teaching in higher education, and it was that use of relevant examples that was able to enthuse people who initially did not think they could do mathematics or quantitative work at all. I said, “Actually, you can do this, and look how it will help you in your future life.” That is something that we do need as a nation to grasp. I would not say that the Office for National Statistics can be the sole group that enables that to happen, but we will be doing everything and continuing to do everything that we do.
For example, we have a relationship with the Royal Society Advisory Committee on Mathematics Education. Again, I have spoken to Sir Martin Taylor to say, “Look here, we do need to get practical, easily understandable examples throughout the school system.” I do think that Sir Adrian Smith’s report—again, on mathematics education—is an important one because it talks about everybody doing mathematics, but mathematics being relevant to what they were studying at that time. I think that is so important.
Ed Humpherson: I like the question you asked, Mr Cowan, for two reasons. The first is that you talked about helping people sift out the bad information and focus on the good information. I think that is a good way of thinking about this bombardment of information we have. The core of our work at the Office for Statistics Regulation is to help people do that by making sure the good information sings out, that it shines out, that it is not misused, that it is available, that it is transparent, as I was saying earlier, that it meets the highest standards of trustworthiness, quality and value.
I always think in this debate about misinformation, people focus too much on the misinformation. The best way to combat it is by getting the good information, the sorts of things that the ONS produces and Government Departments can produce, make sure that gets out and it is available and is transparent. In the pandemic, at its peak, the coronavirus dashboard had 76 million hits in 24 hours. That is 76 million people getting to the good information. That is the best antidote, so that is the first thing.
The second thing I liked was that you talked about critical thinking, not about statistical literacy. I am always a bit nervous about people talking about statistical literacy because I think it proceeds from the assumption that we—the person speaking—know things that other people do not know and ought to know. It implies there is a deficit in public knowledge. I do not think that is right. There may be some abstruse elements of statistics that not everybody understands, but I think the public’s capacity to appreciate and understand statistics is much higher than that statistical literacy narrative gives credit to, if the statistics are communicated clearly, as Ian has said, in an engaging way. The pandemic has completely demonstrated that that is the right way to look at it, that people do have an appetite and a capacity and an interest in statistics.
Professor Sir Ian Diamond: Ed mentioned the dashboard, but at the same time, perhaps the place that statistics has often been most seen in the last 18 months has been in the slides presented at the press conferences. I would have to say that over time those improved greatly. Indeed, the Office for National Statistics embedded a small team into the centre of Government to help with the presentation of the data and to make sure that those presentation slides were being presented in the most accessible way. That is important, and we need to continue that as much as we can.
Q18 Ronnie Cowan: That leads me very nicely into my next question to Sir David. UKSA’s “statistics for the public good” strategy has now been in practice for the last year. How effectively do you think it has been implemented to date, and what plans are in place for further implementation?
Sir David Norgrove: It is a case of so far, so good. The first stage is the embedding, in a sense, of the principles: radical, ambitious, inclusive and sustainable. I am glad to say that as I talk to people in ONS and more generally in the GSS—and I have had a series with Ian, a series of meetings with statisticians in each major Department—people talk about those words and challenge themselves about how they are delivering against those principles.
We have just come to the end of the first year. What I can see is a continued drive to implement those principles. We will no doubt come on to some examples, but I would pick out two. One is the implementation of the new data platform, which is being created by ONS. The Government have already put substantial money behind that and, subject to the spending review, I hope we will carry on and develop that to become a wonderful data hub for Government, bringing joined-up data to allow joined-up Government. Without joined-up data, I do not believe you can have joined-up Government. That would be radical and ambitious, and it is radical and ambitious.
The second thing I would mention is the Inclusive Data Taskforce, whose recommendations were published a few weeks ago. They have been received well, and people are taking them seriously. I hope we will discuss those in a minute, too. Those are just two examples, but I could cite many others.
Q19 Ronnie Cowan: To all three witnesses here, is there enough focus on the statistical literacy of the general public within this five-year strategy of statistics for the public good?
Sir David Norgrove: As Ian and Ed have just described, we cannot take primary responsibility for educating people in the use of statistics. Ed, though, went on to say that educating people about statistics themselves is not necessarily the best way into this and may underestimate people’s understanding of statistics.
I see our primary job as making sure that the statistics and the data that are being produced meet the public need, first of all, and secondly, that they are presented in an absolutely clear, transparent, powerful way, so that people do not have to work hard to understand them, that they are clear and comprehensible. We have made some progress with that, but there is a lot further to go.
Ed Humpherson: If I could just supplement that, it depends what one means by statistical literacy or focusing on statistical literacy. If it means there is a deficit in the public, that the public misunderstand things and need to be encouraged out of their ignorance, I think that is a wrongheaded way of thinking of it. If it means the producers of statistics—ONS and other parts of Government—need to have a very, very strong understanding of how best to communicate in a way that engages, a focus on statistical literacy is absolutely essential.
If I could just highlight a superb piece of work done by the Economic Statistics Centre of Excellence, ESCoE—it looked at public understanding of key economic statistics. It did a very extensive survey and then focus groups. What it found was that understanding was quite good of things that relate directly to people’s lives. People did have a good understanding of inflation. They did have a good understanding of interest rates. They found it harder to engage with a more abstract concept like GDP. They found GDP was not presented in a way that engaged them very well. Interestingly, they also said they were not entirely convinced by the concept of unemployment, because they felt there was more unemployment in communities that they lived in than the figures appeared to reflect, and that was to do with the way unemployment is defined in the statistics.
Those sorts of insights about what people find easy to grasp and what they find counterintuitive are absolutely essential for producers of statistics to embrace and engage with and think, “How do we work with that to make sure that the real message of the statistics lands properly?” That is the way to think about statistical literacy.
Sir David Norgrove: This is slightly off your point, but one very pleasing thing is how high the trust is in the statistics that ONS produces and in official statistics generally. We do a survey every couple of years—last year’s was postponed. In the last survey, the trust in the statistics that we produce is running in the high 80%. It is hard to see how it could be much higher.
Professor Sir Ian Diamond: I agree with everything that has just been said. David mentioned the Inclusive Data Taskforce, and I do believe that we have a real responsibility to make sure that our data are presented and communicated in a way that is accessible to all members of our society, and also to ensure that all members of our society have a voice in our data. That is something that we are committed to ensuring over the next little while, and that links in with the literacy piece.
Q20 Ronnie Cowan: I have to just briefly say to the point that Mr Humpherson made that I am not keen on GDP being used as a measure of success, and I wonder if doing something about wellbeing, hand in glove with it, would be the comparison to make. That is maybe for another day, because we are up against the clock here.
Professor Sir Ian Diamond: Recognising the clock, I will just say that we have been having conversations about expanding wellbeing; indeed, some indicators around climate change, alongside that of the standard GDP.
Q21 Ronnie Cowan: That is very good to hear. Thanks very much.
Moving to Sir David and Sir Ian, how are you working to strengthen the link between the policy and the analytical professions in the civil service?
Professor Sir Ian Diamond: Could I kick off on that? It is such a good question. It is an incredibly important question. For me, good policy I always feel does come from extremely good and relevant data, but for those data to be relevant and properly put together, they need to be cocreated by analysts and policy people working closely together. That is something that I am passionate about, as is, I would have to say, Tamara Finkelstein, who leads the policy group right across Government. We are working together to build that culture, and indeed we are recording a conversation together just to make that point even more.
The other thing we are doing as a result of Lord Maude’s recent report is a stocktake of what the quantitative understanding is across the policy profession so that we are able to understand any interventions that need to be made, and also to work on enduring that those conversations, which have to start at day zero of the development of any new policy, can be informed by people who understand the kinds of questions they need to ask as well as the interaction between the two. It is an important question and something that is a very active and rapid piece of work that Tamara and I are doing together.
Ronnie Cowan: Mr Norgrove, do you want to jump in there?
Sir David Norgrove: The only other thing I would add is something I have referred to a couple of times before: the integrated data service, which sounds a technical and boring thing, potentially, but I think it has the potential to revolutionise our understanding of the way that different Government policies interact and affect people. For example, at the moment, it is very hard to see how the policies of the Ministry of Justice in terms of looked-after children interact with educational outcomes or longer-term outcomes of employment. Joining up the data will allow that understanding to be created and I think will have a fundamental effect on the way that data drive Government decisions.
Q22 Chair: I was going to ask about the integrated data programme later on, but it is more appropriate to do so now, given it has been mentioned. Has there been any feedback from data users on the integrated data programme yet?
Professor Sir Ian Diamond: Feedback is perhaps almost a little early, given that the private beta only went live on 1 October. Having said that, what we are doing, which is bringing data from many areas—not just administrative data, sometimes it is data that have been generated from surveys in the research community or in other areas—together and enabling them to be linked together and to address very, very big questions, often some of which cut across different Government Departments. Indeed, most policies, in my experience, cut across Departments.
While, as I say, the private beta version of this only came into process just a few weeks ago, it builds out of something that ONS has been doing which is the secure research service, which has enabled the linking together of many data. I just draw to your attention to the public health data asset that we have put together during the pandemic, which has large amounts of health data and socioeconomic data and vaccine data, for example, which has been widely used by many users across Government and indeed in the research community, all with proper ethics, all with proper approvals to demonstrate that the work is in the public good. It has been able to help our understanding particularly of some of the inequalities that have been demonstrated, sadly, in this pandemic.
Just one example. By linking death certification with census data and then also being able to link in housing data, we have been able to understand the inequalities that have existed in mortality due to Covid-19 and to demonstrate the extent to which some of that is driven by disadvantage, some of it also by other factors. Being able to link those data together enabled us to bring insights that have enabled strong policy in many areas of Government. The feedback that we have had from people for being able to do that has been extremely positive.
Chair: Thank you. Sorry to interrupt you, Ronnie. Thank you.
Q23 Ronnie Cowan: That is fine.
My last question to you—I think you can see where we are going to go with this, but I will give you the opportunity anyway to say it—is what do you see as the benefits and limitations to including statisticians or analysts in the development of policy? The reason I ask that question is—I am not going to get overly political in this Committee—I see things that I think are blindingly obvious that policies should be put in place, yet they are not put in place. I am wondering, is it because the statistics are not there, or because they are not being listened to?
Professor Sir Ian Diamond: I do think that even though things may appear very obvious, it is always quite good to see what the evidence is that would support them, not only because it may or not support them but because it may direct exactly how you are going to intervene. You might think we want to do policy X, but how you do that seems to me to be perhaps a slightly more complex situation. Then I would argue you do need strong monitoring and evaluation of that intervention of that policy, which does require the statisticians to be involved from right at the beginning to plan the scientific evaluation so that you are able to see how that policy works and whether it needed fine tuning to make it even better than you would have thought it when it seemed obvious to you.
Ronnie Cowan: Do the other two witnesses wish to make your pitch?
Sir David Norgrove: The trouble is you are asking for human perfection, perhaps. People are driven often by emotion, aren’t they, as well as other pressures. It is up to the statisticians and the analysts sometimes to force themselves into the debate as well as to be invited in.
Ronnie Cowan: Thank you very much.
Q24 Karin Smyth: I want to move now to the Inclusive Data Taskforce; Sir Ian, to you mainly. A phenomenal amount of work has been done over a very difficult period, so well done to everybody on that.
The question is, how do you intend to use your role now as the National Statistician and head of Government analysis to encourage other departments to take on some of the recommendations of that work?
Professor Sir Ian Diamond: An important point, if I may. In a number of ways, is the answer. First, we worked very, very closely with the Government Equalities Office in the response that we have made to it. That is going to be part of how we drive things forward. We have involved the Government Statistical Service and the Government Analysis Function in all aspects of the development of both the response to the taskforce, which was a completely independent group that Dame Moira Gibb led brilliantly, but also in the development of our operational plans, which we will release in January.
It will be for me to say, “Look here, this is super important for all my fellow statisticians”, and driven by the heads of the profession within Departments, I will also be taking the opportunity to raise issues, where appropriate, with, for example, permanent secretaries. I do think we need to recognise that this is a real opportunity for us to ensure that every member of society has a voice in our data but also that we build trust. One of the things that came across from the Inclusive Data Taskforce was that for many people in marginalised communities there is an issue of trust. It is not so much trust in the Office for National Statistics but trust in Government more generally. I believe that this is an opportunity for us to demonstrate that we care about you and we care that your data are used, and this is what we can deliver.
One of the things that our own work has done while the taskforce was going on was our engagement with different communities to maximise the response to the census in March of this year. One of the things we said to people was, “You have to fill in your census form because this is the only way we will find out about your community.” Now it is incredibly incumbent on us to go back to those communities, now that we are starting to get results, and say, “Effectively, you gave us your data. This is what your data says about your community.” I will be making sure that where those data are relevant to other Departments across Government, that we are sharing them, particularly to some of the Departments that have particular interests in working with those groups who are most marginalised in our society.
Q25 Karin Smyth: How have you found other Departments’ receptiveness to that approach?
Professor Sir Ian Diamond: 100% supportive.
Q26 Karin Smyth: Good. I think you have slightly touched on it, but we are also interested in spreading this practice and how important you think it is for all data producers to utilise this approach.
Professor Sir Ian Diamond: I think it is important, because we need to recognise that there are different parts of our society that work in different ways, and we absolutely need to ensure that if we are being inclusive in our policies, we need to be inclusive in our data, and we need, therefore, to be able to say what is going on in particular small geographies or particular communities. We need to be able to do that in a way that properly informs policy, or policy may well be developed entirely brilliantly but not appropriately for different parts of our society.
Ed Humpherson: Can I just add something from a regulatory perspective here? I think that we, as a regulator, can play an important role in supporting and delivering the roll-out of the Inclusive Data Taskforce recommendations. We talked a bit today about the work that we do, which is very much in the public eye, about the use of statistics, but underneath that, on a day-to-day basis, my team is out there reviewing the production of statistics across all parts of Government in all parts of the UK. We are making sure the standards are being adhered to. We will build into that the extent to which statistics on the prison population or on farm incomes or on whatever it is that we are looking at are implementing exactly these messages.
We have a tagline, which is, “Statistics need to move beyond the tyranny of the average.” If you just focus on the average, you miss the experiences of people in places and communities that are different from the average. We will certainly be wanting to see those recommendations, and we want to be a mechanism for rolling them out.
Professor Sir Ian Diamond: It is incredibly important. If I just may, one final point. For very, very good reasons we have spent a lot of time working with survey data, but when you get to particularly small groups, you can end up with only having maybe two or three respondents from a group, and then, frankly, you are losing representativeness of that group.
I remember one survey in the 2000s that I was involved with. It was a UK-wide survey, but it had 150 households in Scotland. It is entirely unclear to me how you design a survey with 150 households and then say you are making estimates for anything in Scotland. It is super important not only to focus on the inclusion but also to look at the methodology of any data collection to ensure that we are being inclusive in what we are doing.
Q27 Karin Smyth: Thank you. That neatly follows to my next question, which is about the methodology around gender, sex and transgender questions. The Inclusive Data Taskforce provided a recommendation on the capture of data on sex, gender and gender identity, which does contrast with the recommendations produced by the Scottish Government on data collection. The question is, what impact do you think that could have on the comparability of data between administrations?
Professor Sir Ian Diamond: It is an important question. Clearly, we recognise the jurisdiction of Scotland to take those decisions. One of the things we now have to do is to build a way of making comparisons between the two ways of collecting the data. I think it is going to be possible to do that by using—there are two questions being asked, because I think it should be possible to make some progress, but I recognise the challenges that exist. Since Scotland took its decision, we have started to work on that challenge, and I would be happy to keep you updated as we move our methodology forward.
Karin Smyth: That would be helpful. Thank you.
Q28 Mr David Jones: The pandemic has raised the profile of the work of the ONS and UKSA through the implementation of the Covid-19 infection survey and various other sources of data provided to reflect the effects of the pandemic. Do the ONS and UKSA have the capacity to deal with the extra attention that has been focused upon them and their work both by the public and Parliament?
Sir David Norgrove: Ian will be able to answer from the executive. I think we do. In a sense, though, the demand for data and statistics is almost infinite. There are occasions when we have to cut our coat according to our cloth and set priorities, but that is true of any organisation. I would say that within the priorities that we have set and the business plans that we have set, we have delivered what we expected to deliver, and in many ways, more. I pay tribute, as I did in my letter to the Chair some months ago, to the statisticians in ONS and in the wider GSS in the way that they have responded to the pandemic in very difficult circumstances.
Looking ahead, a lot will depend on the settlement we get in this public expenditure round. I think that question more broadly might be asked again after that announcement.
Professor Sir Ian Diamond: This does give me an opportunity to acknowledge the incredible work that all my colleagues have done over the last 18 months to be able to stand up at pace new surveys, to work with stakeholders to make sure that we were collecting the data that were needed, and to do some incredibly exciting things with new data sources, ranging from traffic cameras through mobility data through telephony data. A massive number of things were done.
Is that in the long term sustainable? As David said, we will have to see what the public expenditure round says, and we will continue to make prioritisations. I would say that we have managed extremely well to continue to deliver not only these new data sets that were required but also to continue to deliver, for example, monthly GDP and labour force statistics, and to do so in a way that has reflected the challenges of the pandemic. It has been an incredibly tough 18 months—I would not pretend it has been any more tough than it has been for everyone else—but colleagues have responded extremely well, and we have worked very, very hard to build a culture of can-do but one that is also characterised by support for people and ensuring people are able to look after themselves as well as feeling supported by ONS.
Q29 Mr David Jones: How do you intend to maintain and strengthen not only your actual independence but the perception of that independence as the demand for statistics continues to grow?
Sir David Norgrove: I would agree that that is an absolutely vital question because it goes to the point about trust. The independence of our statistical system is vital to public trust in the use of their data, and particularly over the coming years as data are used in more creative ways to answer the questions in the way that we have already discussed. Maintaining public trust and being trustworthy absolutely goes to the heart of what we are about.
My first point on that would be that maintaining independence is not just a matter about the chair and the board or indeed the National Statistician and the Director General for Regulation. My sense, and my strong belief, is that independence of the statistical system runs through our statistical service like a stick of rock. I think that is the best protection we have. Indeed, it has been tested perhaps as never before during the pandemic, and my feeling is it came through extremely well. I am very reassured about that.
The other dimension to it is the people who might threaten the independence. I do think that our politicians—and the protection of this Committee is very important in this, and indeed this hearing—recognise it, that people do value at all levels the independence of our statistics. I think that is widely recognised. Of course, we have to take specific actions too, and I would instance the ending of pre-release access for Ministers that happened a few years ago. I think that was an important step. Since then, and I am sure it will continue, through OSR and with my interventions too, we have taken action without fear or favour wherever we spot misuse of statistics or lack of transparency. I hope and believe that will continue.
The final point I would make is that regulation in the future is going to become more complicated. I am not sure it will necessarily become more difficult in terms of the threats, but it is going to become more complicated as we get joined-up data and use of administrative data, and as model-building and algorithms and artificial intelligence become more prevalent. OSR will need to increase its capacity, probably in terms of numbers but certainly also in terms of capabilities, in order to cope with it. I think we have made a good start, for example, in the excellent review of the exam algorithms fiasco last year, but that is just the start.
Professor Sir Ian Diamond: I think it was an important question. You said, “both the actual and the perceived independence.” On the actual, I have never felt threatened at all. I just wanted to say that very clearly. On the perceived, clearly, for us, one of the most important things that we must do is to be transparent in everything that we do, particularly with analysis, so that we are making it very, very clear how we have done things and why we have done them. Effectively, through the integrated data service, much of what we do will be reproducible, and I think that is super important, that anyone can see what we have done, how we have done it and where those numbers come from, and demonstrate that the question is sensible, and this is the answer and we have come to it in a professional, rigorous and independent way.
Q30 Mr David Jones: What are you doing to identify any gaps in data as we progress through and hopefully move out of the pandemic?
Professor Sir Ian Diamond: The first thing to say is there are many gaps in data. One of the things that we were talking about serendipitously just yesterday is the need to do two things. First, to be asking, “What are all the questions that we are going to need to be asked over the next little while?” I will give you an example. If we look at levelling up, we are talking to Andy Haldane, who is just coming in, to say, “What are the real questions to you, particularly at a sub-national level?” We ask questions, but what you should then not do, I would submit, is then just make sure you have the data to answer those questions. What you need to do is say, “What are all the data flows that we need so that we can then ask further questions, so that we can then ask the questions that nobody has thought about yet? Can we make sure that those data flows are super important?” That is effectively what we are doing: constant horizon-scanning of particular areas and the data we have. What are the data we need? At the same time we are not stopping at, “These are the data that will answer the questions we have just at the moment.” What data flows can we make sure are there so that we can ask the questions we have not yet thought of, and doing so in consultation with many stakeholders?
Q31 Mr David Jones: The ONS and other statistics providers have responded to the pandemic by providing data at a considerably faster pace than usual. Can that pace be maintained? Is there a balance to be struck between quality on the one hand and pace on the other?
Professor Sir Ian Diamond: The first thing I would say is that we are not in the business and we will never be in the business of a quick and dirty answer. It is incredibly important to say that all our data match a high quality. Having said that, we have done things much faster in arenas than we have in the past. Some of that I think is because we have taken on board things that might have taken us a bit longer to have implemented but which we have moved forward much better. Some of that has been automation. For example, our weekly opinions survey, which we stood up at the beginning of the pandemic and is now running fortnightly, goes out on a Wednesday and we close it the following Monday morning and get the results out by Monday afternoon. A lot of that is done completely through automation. There is no problem with the quality, but at the same time we are doing it at pace.
Some of the faster economic indicators that we use—for example, those around gas prices or around vacancy data or using payroll data—yes, they are fast, yes they are incredibly accurate. What we need to recognise is that they may not be, for example, GDP. They are indicators and they are entirely accurate and high quality in what they are, but they are not, if you like, part of the bigger industry of producing many data for, in this case, GDP. I would point to many of the analyses that we have done. For example you were very kind about the coronavirus infection survey. We can produce the basic statistics, and we do produce the basic statistics, on a twice-weekly basis, but some of the much more complex statistical analysis, which is incredibly high level and which we do in partnership with The University of Manchester and the University of Oxford, requires, sometimes, two or three weeks of statistical analysis and use of high-performance computing. One thing I can assure you of is that those data are only released when the quality has been properly assured and is high quality.
The one thing I can assure you is all data you see from the ONS is of a very high quality when we can produce things at pace. I think it is entirely right and proper that we do so, and I think the last 18 months have demonstrated how much we can do. We will not ever go down a line of, if you like, trading quality against speed.
Mr David Jones: Thank you very much.
Chair: Thank you very much. I am going to go to John McDonnell, please.
Q32 John McDonnell: Can we come to a few more detailed questions about data ethics? It has been a running thread through most of your commentary so far. What are some of the current ethical challenges identified by the UK Statistics Authority Centre for Applied Data Ethics and how is the centre identifying upcoming potential challenges that face the research community, and how does it propose to deal with them on an ongoing basis? Can I ask maybe Sir David to kick off on that one?
Sir David Norgrove: I cannot talk about the Centre for Data Ethics. I think it is better for Ian to talk about what the centre is doing. My reflection is that ethical issues are becoming ever more pressing and important as people’s data is brought together, and so maintaining confidentiality and security is absolutely at the heart of our plans. For example, with the census now, the aim is that people should be able to create their own sets of analyses so rather than publishing analysis in tables, which are then very inflexible, allow people to create their own tabulations of data. The risk there is that they are able to create a tabulation that allows somebody to be identified, and a huge amount of work is now going into ensuring that automatic systems are in place to prevent that.
That is one kind of ethical question. The other kinds of ethical questions are around the use of people’s data in a way that meets the public good, and that is one of the reasons why I am so grateful for the objective that we were given in our founding legislation, “Statistics for the public good”, and that is why our strategy is called “Statistics for the public good.”
Professor Sir Ian Diamond: Ethical use of data seems to me to be at the heart of everything that we do. We use public data in virtually all of our work and we do so—effectively through you and through this Committee and through the Parliament—through the permissions of the public that their data will be used ethically and in a way which people can see is for the public good. I think it is incredibly important and it is something that we are absolutely passionate about which is why we set up the UK Statistics Authority Centre for Applied Data Ethics, why we also set up the National Statistician's Data Ethics Advisory Committee which looks at all projects that we do and judges whether they are ethical or not.
David has already addressed the question of anonymity. He has addressed it with the census. I would also point out that we are working closely with The Alan Turing Institute at the moment to develop ways of producing data which absolutely preserve anonymity completely, but which allow the analysis to take place properly. That is, I think, an important point moving forward.
A second point is the question around whether we can demonstrate in any particular piece of work that it is in the public interests and in the public good. Now, I do not have an algorithm for doing that; nor do I believe there is one. That is why we have both the ethics committee and a completely independent approvals panel which looks at all proposals to use data, and which basically asks the question, “Would any normal person think this is in the public interest that this work is done and is there a very clear description—coming back to my point about transparency—of what the work is and what it will do that people will understand?
There is no algorithm, but the thing I would say is it is incredibly important that both those committees are informed by a really good process of public engagement. We need to listen—and we are—to the public all the time about what they would feel is good use of data. If I may say so, too often people have just said, “We just want your data”, rather than why we want your data and how your data will be used and will it be used in a way which may not impact on you directly but may impact on our family, or may impact on your society, or may impact on other society, and it will be preserved anonymously and very, very safely. When we move into that conversation about what will be done with data, people tend to have a much better idea of what we should do.
The issue of ethics is very much one of anonymity, public interest, and how data will be used. Those are the things we look at all the time. I do think there are increasing challenges around health data and I was asked, alongside the Chair of the Association of Medical Research Charities to give some advice to the Department of Health around the use of GP data following some issues in the summer around opt-out. My advice there has been very much around the same sorts of questions around public engagement and around really understanding and working with the public at an early stage before you decide what you are going to do. To that extent, in early November we are having a summit on public engagement which will include a number of the key players from the health community as well as the Office for National Statistics, so the research councils, but critically also members of the public.
John McDonnell: Just on future challenge—
Ed Humpherson: Sorry, can I just come in there?
John McDonnell: Yes, please come in, yes.
Ed Humpherson: I would like to refer to two pieces of work as a regulator, which bear on this. The first is, one of the features of the world of data abundance that Mr Cowan is referring to is that we have this tension between, on the one hand this kind of utopian idea of the power of data to unlock new services and new insights and so on, AI, and all those sorts of things, and on the other, this anxiety that those tools may not be working in the interests of individuals. Those tensions came together when we looked at the process by which exam grades were set last summer, the so-called algorithms that set A-levels and GSCE results. When we looked at that, we found that the principles that underpin good official statistics—trustworthiness, quality and value—are also relevant to the application of a model in that sense, so there was a context.
This was not a story where a technical algorithm went wrong because it was badly coded or the model in some way went rogue; it was not that. In a way, the model succeeded technically. It failed in terms of public legitimacy. It failed in terms of the explanation of what it was doing, how it handled outliers, how it gave people opportunity to appeal, all of those things which really are the bedrock of people being able to engage and feel that they can trust the system being set up; it failed in terms of legitimacy. I think it is very important to remember in all of this that all the things which underpin good statistics will also underpin good use of algorithms.
The second thing I wanted to mention was that we recently published a review of the applications to the committee that Ian described to use data for the public good, because we wanted to see how researchers are describing the public good. When they say, “public good”, what do they mean? It is a really very, very interesting set of outputs. By far and away the most significant area that researchers say meets the public good, is when they will use data to improve policy. Much less prominently are things like saying we will do the research which will help inform the public. I think that is really something we can take away and work with the research community on because that is just as important as the important policy. So, two things—our work on the exams I think highlights these issues and, secondly, our research into the public good I think does highlight some of the ways that people think about the public good, and how it could be more rounded. Thank you.
Q33 John McDonnell: That is very helpful. I do not want to delay matters but one of the ethical challenges possibly for the future that has been raised previously is around the commercial use of data. What engagement do you have on advising around the ethics of that?
Professor Sir Ian Diamond: Commercial data has been incredibly helpful during the pandemic and we have developed good uses of commercial data. What you are asking, though, is about whether your data is going to be used to make money for a major corporation. That seems to me to be a seriously ethical issue to which there is no right answer, to be honest, because many people—and, again, this is where pubic engagement comes in–would say it might be very good to use health data which have, if you like, the ability to identify people who could be candidates for clinical trials to develop a new drug which would have an opportunity to impact on rare disease for example, but a rare disease which is extremely life-debilitating for those people who unfortunately have that disease.
Now, that is a different conversation perhaps than one about someone who just wants to make an extra few pounds from selling people’s data on to someone else. I do not think I have got an algorithm that tells me where those ethics lie, but I do think we need to have a public conversation before we start doing anything to make sure that the public understand exactly how their data are likely to be used and we feel that there are the permissions from the public to take part in that.
Sir David Norgrove: Can I add one thing to that? There is not, I don’t think, a generally understood set of tests that would be applied in any circumstance to the use of data. That is partly because we are still at the beginning of this issue, but it is also because so many people have got their fingers in this pie. Before the pandemic, we recognised that the world and his wife—perhaps I shouldn’t put it that way. We recognised that there were many multiple organisations trying to produce codes of ethics for the use of data. We recognised this and had a series of meetings, and I cannot remember, Ed, but were there about 15 organisations involved in producing ethical standards?
Ed Humpherson: It was of that order—a dozen or 15.
Sir David Norgrove: Everybody was saying much the same thing but in different ways. There is a question here of how, as a society, we move to a situation where there is a common public understanding of the tests that you apply when data are going to be used.
Q34 John McDonnell: Where it is needed. We will most probably come back to this at a later date. Ed, could I come to you on the issue around the casework load at the moment. According to yourselves in the OSR casework for 2021, the authorities considered nearly three times the number of casework pieces than the previous year. Is the OSR equipped to deal with that level of complaints at the moment?
Ed Humpherson: I believe we are. The first thing to say is that we dealt with 323 cases, which was this threefold increase compared to any previous year, but we did so more quickly. Our median time to respond was 10 days. In the previous year it had been 13 days and before that it had been 20 days. So we sped up despite the increase in the volume. That is testament to the huge efforts that my team made but it also illustrates that we are able to respond. We think of these things that come to us from Members of Parliament, from members of the public or from our own internal analysis which lead to concern as the most important thing that we do. It is the public face, standing up for statistics, which is our job so we would always prioritise it.
I think we have shown that we can handle it. We did restructure a little bit to be able to do it. We could see the surge coming because we could see this huge public appetite and interest in statistics and data, so we restructured a little bit. We did slightly refocus our work, but we are certainly able to cope with it and do it well. My evidence of doing it well is not simply the metric of time taken, it is also the impact that we have. We have had a wide range of impacts. We have secured the publication of unpublished data. This point is about transparency, which I am afraid I am banging on about today, but it is so important. When we have stepped in we have always been able to secure that the Department have published data that was previously unpublished.
We secured quite significant changes in the production of statistics. Something we have not talked about is our review of ONS’s population statistics and how they measure sub-national populations. We have driven huge improvements in the presentation and communication of test and trace statistics. We filled gaps. For example, for the first time the Home Office published statistics on people who have no recourse to public funds. They have that residency status. That was following an approach from the Chair of another parliamentary Committee. We looked into it and worked with the Home Office and we secured clarification.
Not only did we deal with it, but we coped with the load. Not only was it done more quickly, but we secured impact. Having said that, if we see a further surge—if we see that level being increased—I think there would come a point where I would say this is not a short-term surge from the specific circumstances of the pandemic, it is a more sustained shift in the level of public engagement with statistics and data. Then I would be going back to the board to say, “If you want us to respond to this with the care and attention that it deserves, you will probably need to allocate more resources into it.” I do not think we are quite seeing that at the moment. We publish quarterly information. For the two quarters of the year that we are now in, the volumes are well above any previous year but not at the levels of during the main pandemic.
Q35 John McDonnell: Why has the amount of internally generated casework reduced in comparison with previous years?
Ed Humpherson: It is true that it is reduced. It was 39% of our cases in the previous year and only 16% of our cases were internally generated this year.
I should explain to the Committee what that means. We initiate our cases when a member of the public or Member of Parliament comes to us and raises a concern. That is what initiates the case, but we can also do that ourselves through our horizon-scanning, web scrape, and in various ways if we spot something we will open a case. That is just as important as the things that come to us from the public. So that is what we mean by internally generated cases when we ourselves spot things. Why has it fallen? One thing to point out is that the year when it was 39% of our cases there were 43 internally generated cases; this year when it was only 16% of our cases it was still 51 cases. We increased the number of internally generated cases from 43 to 51 and the reason the percentage dropped is that we had this surge of external input.
What I am trying to convey is that we did not slacken off from doing our horizon-scanning. We continued to do so and in fact did so more and initiated more cases, but we had far more interest from Members of Parliament and from the public. Some of the things that were obvious to us that we need to look into, started as internal cases and then members of the public came to us for example with things to do with aspects of the daily briefing slides. They started internally but within a day we had the postbag.
Q36 John McDonnell: Let me get this absolutely clear. Nearly half of the cases in 2021 related to quality, reliability and trustworthiness of the statistics. If I am right, this is the first time that this has been the most common category of casework. What do you think has contributed to this and what can be done to improve the perceived quality, reliability and trustworthiness of the statistics?
Ed Humpherson: This a really interesting pattern. In the past, probably the most significant area is people being concerned about misleading uses of statistics. If you look at the report that we published recently you will see that in the previous year it was the Department for Education that was the leading Department involved and the concerns there were all around whether the Department was presenting statistics on school funding, school performance in a way that is appropriate. That has been the steady diet ever since we have been doing this work.
This year the surge has been of people worried about the quality, the trustworthiness of the statistics and so on, not how they are used. I think what that is telling us is that we had a huge growth in public interest and public appetite, not for the sort of policy debate but because they wanted to understand the numbers. They were seeing the numbers on the 10 o’clock news, they were seeing them on the front page of their websites, they were seeing them in their Twitter feeds—the numbers were much more prominent and much more at the heart of people’s engagement with the pandemic and this extraordinary rate of access of the coronavirus dashboard. That is mostly people looking in their local area: “What are the case rates in my local area?”
What you are seeing there is not people interested in the toing and froing of political policy debate, they want to know about the numbers. I think that is why you see our caseload being very much dominated, nearly the majority of the cases, for the first time with this kind of thing. They want to know, “Are the numbers reliable? Are they giving me a fair picture of what is going on nationally, regionally and locally?”
What to do about it? The good news is it is fairly clear what to do about it. The issues that people are raising, they are asking question like, “How do I make sure I understand what is meant by a Covid death?” or “I have seen these figures about hospital occupancy, I do not really understand what they mean.” This goes back to our conversation earlier about communication of statistics, it is do with the way producers present the statistics, explain where they are from, explain what they are, understand that the public as user, what those user needs are and respond to it. Explain, for example, the difference between the different sources on Covid mortality and what they mean and the reason for those differences.
The first part of your question is why, and I think the reason is that members of the public have just been engaging with those data in a way they haven’t before. What to do about it? I think it is the code of practice for statistics. Come back to that, implement that and you will start to address those issues.
Q37 John McDonnell: You might have to go big on Sir Ian’s data masterclass, well beyond the sphere you were originally planning.
Ed, in response to the release of the 2021 casework report, questions have been raised in the media about the need for investigation into the oversight of Government communications. Does the casework review suggest that is necessary?
Ed Humpherson: The casework review suggests that there are ways to improve the presentation and communication of statistics for sure. I think you are referring perhaps to some statements made by Full Fact, the fantastic fact checking organisation who are close partners of ours and we work very closely with them. I wouldn’t have a view on wider Government communications, which are a lot more than presentation and communication of statistics. I do want to, though, reiterate a point that I have made a couple of times already. When a number is quoted publicly, publish the underlying data. That should just be embedded into departmental practice and it is not at present. That is what we are going to drive for. That is the element of communication that I would want to hammer home.
Sir David Norgrove: It needs to be quick. It needs to be very soon after the use of the data and not weeks afterwards and without requiring Ed or me to write to somebody to say it should happen.
Ed Humpherson: The thing about it is that we are not trying to catch anyone out with this. If somebody is asked a question and they give a quantified answer, it is not like we are saying, “Ha, that wasn’t quite right.” We are just saying, “Publish it, make it easy for people to see what you are saying.” It is quite frustrating, I think, that that is not just the norm.
Q38 Karin Smyth: We want to move to the dual roles around production and regulation of population estimates. To Ed and then Sir David, in 2019 you will be aware that our predecessor Committee recommended that UKSA took steps to improve and demonstrate its independence to minimise vulnerabilities around regulation and production within one body. Could you outline to the Committee how you have followed up on this recommendation, the steps you have taken to provide clarity on the separation of ONS and UKSA?
Sir David Norgrove: First, I don’t think we needed, in reality, to improve the independence. The independence was there and is there. Both I and Ed would, and do, defend the independence of OSR fiercely. Not that it has ever come under threat. I do not think anybody has ever tried to challenge it. I do not think it is about the reality of the independence, the Committee’s concern was about the perception of independence given the dual role that the authority exercises. Ed and I have worked together to develop a series of changes there to illustrate that and Ed can talk you through them.
Ed Humpherson: We have been very keen that the separation and the independence that we know is how we practice our work is seen more clearly externally. What have we done there? The first thing is we have clarified the independent communications voice. We have our own Twitter account, our website is now much stronger, and it is clearer to see that it is a separate website. We use our voice, in fact my voice, in the public media a lot more. We do all of that to say, “Here is a separate entity which is making its judgments by reference to the code of practice for statistics and not by reference to what the ONS directs.” That is what we are looking to convey.
The other thing that is very important to bring out here is that those are things which are, to some extent, presentational and underlying that is the decisions being made, the work that we do. We do regulate ONS, and we regulate ONS in the way that we regulate any producer of statistics, whether it is the Scottish Government, Welsh Government, HMRC, the Department for Work and Pensions, Public Health England—we look at ONS in the same way. What we always try to do is to celebrate when good work is done and challenge when things need to improve. ONS has had its fair share of celebration and of challenge, and I can outline that for you, if you like.
Finally, we have also done some sort of administrative things. For example the minutes of our decision-making committee are now published regularly. They are available and publicly transparent for people to see. The key is to judge us by what we do, and I would say we have been as harsh in our challenge to ONS as we have to any producer in the UK over the last year.
Sir David Norgrove: To some extent it is a matter of time. Funnily enough OSR does not exist in the legislation—actually even ONS does not exist in the legislation, but OSR was only created a few years ago in 2018 or maybe 2017. It is a new organisation, and it has taken a little bit of time to establish the brand. Brands, in my experience, are established by the substance of what they represent. The last few years have shown the substance of OSR’s work, the interventions are now widely reported and OSR as an organisation is now being reported in the press. That will, over time, reinforce the perception of independence.
Q39 Karin Smyth: Just to clarify, Ed, you regulate in exactly the same way so there are not examples of difference in how you regulate any other reporting body or the ONS?
Ed Humpherson: We expect full compliance with the code of practice. We conduct assessments and reviews that test that, we report them publicly, we write letters publicly highlighting the good things that have been done and also the areas to improve. We have been very positive, for example, about the infection survey, very positive about faster economic indicators. Meanwhile we have challenged quite trenchantly on population estimates. The risk that the ONS was not listening sufficiently to some of the evidence that suggested that in a local area there were problems with some of its estimates. We highlighted problems with business statistics being not timely and we have asked ONS to put in place an improvement programme. We have highlighted that some of the processing systems are not fit for purpose. I would say that we have been completely comfortable with issuing the challenge.
I should say, because Ian is of course here with us, that this is not an adversarial contest between the two of us. We highlight challenges, ONS takes them on board, looks to use them as a driver of improvement, sets out their improvement commitments and does so very openly and in a very good spirit. I am fairly clear in my mind that this is a consistent and balanced approach to regulation.
Q40 Chair: Thank you. A concluding question from me for all three of our witnesses. As we reflect upon the most recent census, what do you see is the future for the census in this country, its continuation or indeed the form it might take? If I can go first of all to Ian Diamond, please?
Professor Sir Ian Diamond: Thank you very much. Just as a reminder, we have agreed with the Government that we will come back in 2023 with a set of recommendations on the future of the census. My intention is that we will do so and the project that is working hard to make that assessment is going very well.
If I may, let me just say a word or two about that project. What it does is ask the question—and it comes back to a question that was asked earlier by David Jones about quality against speed—“Can we provide local population estimates”, which is one of the main things a census does “and can we do that accurately?” Let us not give up any quality but can we do that accurately and in a very timely way, say monthly, using administrative data from a variety of sources. That is the major focus at the moment of the work. I would have to say that work is going well.
I think it is possible that we will get some good results there and we intend to produce some experimental estimates in 2022. Those will be, of course, calibrated against the current census and also against any other estimates that we have. The bottom line is that there is no one administrative data source that provides the answer. There is no magic bullet here. Therefore a lot of the work is about how we amalgamate data from different data sources to enable us to decide what is actually a population estimate. We use data from many sources, such as council tax or utilities data, which come from local authorities or from the private sector and others such as data from HMRC on payroll that enable us to understand populations.
We also need to estimate migration, and we have made a lot of progress internationally there. The big challenge is internal migration. There is no law for you to move your GP when you move, therefore identifying internal migration requires different sources. We are making good progress there.
There are then questions, Chair, which need to be addressed that are about some of the other questions that a census provides us with information on, which are not covered universally on any administrative source at all. I might, for example, point to occupation or religion, questions that we decide to collect or indeed the granularity with which we collect data on ethnicity. There has then a question about how and whether we will be able to provide that kind of information in an ongoing way without some form of census or series of very large surveys. That is what we are aiming for 2023.
Chair, if you had the time and wished to get a bit of a workshop on where that project is going, we would be delighted to give it to you, but we are aiming to make experimental estimates of monthly local population, estimates in 2022 with a final report and recommendations in 2023 at the same time producing the results from what I have to say was a brilliant census in March 2021. I thank every Member of this Committee for their support and every member of the public who has kindly filled their questionnaire so assiduously.
Chair: Thank you. There have been a couple of invitations for this Committee, and we look forward to having them in writing and we will look favourably upon them. Thank you for that. Ed, do you want to come in, please?
Ed Humpherson: Yes, to say this is a decision about what is produced, I would not be involved with that, but I can say that whatever ONS does produce to estimate demographic characteristics—the population—we will regulate it. We will assess it and we will highlight positives and we will highlight the areas for improvement and any gaps. That is our role in this.
Chair: Thank you very much, and a final reflection on that from David Norgrove, please.
Sir David Norgrove: We have had a census every 10 years since 1801 so it will be a big step if we decide not to have another one. I think we should all recognise that a census is only accurate at the time that it is taken, and it gets progressively out of date. There is a lot to be said for finding a substitute source that allows us to have more up to date information more quickly. However, I do believe that we will only do away with the census if ONS can produce data of equivalent quality. I trust ONS and the board when they come to make that recommendation to do that with the impartiality and independence that they always show.
Chair: Thank you, that is a nice way to end our session this morning. I thank our three witnesses for their time, insight, expertise and, above all, for their public service. Thank you very much.