On BBC Sounds, there are podcasts to help you look after your body and your mind from increasing your immunity to feeling more confident or tips on how to focus. Sorry, what were you saying? If it matters to you, it matters to us. Feel good inside and out with What's Up Docs and Complex with Kimblee Wilson. Listen on BBC Sounds. Hello and welcome to more or less your indefatigable guide to the numbers all around us in the news and in life. I'm Tim Harford. This week not one but two statistical national treasures, Sir John Curtis and Sir David Spiegelhalter and some big questions. Is it true that when Covid hit the UK, a one week delay in imposing lockdown led to 23,000 deaths? Has the leveeing of VAT led to the closure of more than 100 private schools? Is there really a quiet revival of Christian worship? And do more than 10 million families rely on X as their main source of news? First loyal listener Quentin got in touch to ask about a number that's been doing the rounds in newspaper headlines recently. Here's an example from the mail. More than 100 private schools forced to close after Labour's controversial tax rate on fees. The story claimed that... A total of 105 independent schools have closed due to Labour's controversial VAT rate on fees. In case you're wondering what a VAT rate is, I'm afraid it's not an exciting heist. It's just this. Private school fees used to be exempt from VAT but in opposition the Labour Party promised to remove that exemption if elected. They were and they did. The standard 20% rate has applied from January 2025. So is it true that 105 private schools have closed as a result? Our sauna correspondent Tom Coles has been looking into it. Hello Tom. Hello. Are you sure people are going to remember that one, Tim? None of us will ever forget it. So anyway, this stat is sourced to the ISC, the Independent Schools Council, based on their own list of schools that have closed. But you can cross check most of that list against a Department for Education database called Get Information About Schools. Hmm. What do you use that for? Not to get information about schools in Scotland, annoyingly, but the Scottish Department for Education told me three private schools have closed, though three have also opened. So let's call that quits for the sake of simplicity. OK, so what is the database with the highly literal name Safe England and Wales? That 104 independent schools have closed in the 18 months since Labour came to power, promising to make those schools pay VAT. At least one more has closed that's not on my spreadsheet. Something that was fine. Not really. No. Especially if you're using it to judge the effect of the introduction of VAT. Because if you count school closures in this way, in the last 18 months of the Conservative Government, the months immediately before the ones we're talking about, 108 schools closed. So imposing VAT has meant the rate of school closures has actually fallen. Yeah. What's more, under the Labour Government, 165 private schools in England and Wales have opened. The school count is up. Thanks be to VAT. OK, so this number is, well it's a number, but it's silly to connect these closures to VAT. Wait a minute, Tom. Why are you still here? Thanks Tim. When you lift the lid on this silly number, you find some interesting stuff about the state of the independent sector. For a start, 18 of the closed schools are independent special schools, educating disabled kids and kids with special educational needs. The vast majority of the kids in these schools, more than 90%, have an EHC plan, which means that local authorities are paying the school fees and they're exempt from VAT when they do that. These shouldn't be part of a count about VAT at all and they're also in a booming sector. The reason the overall number of independent schools is increasing is because loads of special schools are opening, 146 over this same time period. If we turn to the mainstream independent schools, another 12 of those by my account were also specialist schools where the state picks up the bill. Some of those are also tiny. For example, five of the schools have a capacity of five or fewer pupils. Three of the schools closed because there was only one pupil in each and they moved on to bigger schools. Interesting. So, if you take out those schools which are clearly nothing to do with VAT, what do the remaining schools look like? A really mixed bag. Seven of the schools had a capacity of more than 500 and eight of them were selective. But a lot of these schools were quite small, about half have a capacity of 150 or fewer. The majority of the schools are primary or prep schools and that's to be expected because the falling birth rate in England and Wales means there are fewer primary age pupils to go around. They're also half full the last time they counted the pupils. So VAT or no VAT, it is hard to run a school when there are fewer school aged children? Yeah and VAT won't have made it easier. But there are still cases where the school closures were clearly nothing to do with VAT. For example, one prominent school to close was a selective all girls school in Croydon, the old Palace of John Whitgift School. But its closure was announced in September 2023, so quite a way before the Labour election victory. Unsurprisingly, lots of the schools that closed did cite VAT as a contributing factor to their closure alongside minimum wage increases, the removal of charitable status on business rates and inflation hitting the cost of running and maintaining, in many cases very expensive to run and maintain, old buildings. That and the demographic changes. Many of these schools were financially precarious already. One talked about financial problems dating back to 2019. I spoke to the head of a small Islamic Montessori primary school that had found things harder and harder since Covid and VAT was the last thing on a long list of things. Just putting my economist hat on for a moment, in any healthy market you're going to see some businesses closing and other businesses opening. And it seems like this is what's happening in the private school sector and of course VAT will play a role in some cases. Obviously it's no fun to be a teacher or a pupil or a parent at a school which shuts down suddenly. But Tom, do we have a sense of whether taken as a whole the private school sector is faring better or worse than in the past? I think worse pretty clearly. While the number of school closures doesn't tell you much, when you take account of the size of the schools, more capacity is being lost through closures now than in the recent past. What about the actual change in the number of pupils in private schools? What do we know about that? That is the key figure and something we actually encourage people to bet on. Yes, the bet. We have had emails asking us who won. Yes, in September last year, politics sub-stacker Sam Friedman and Maxwell Marlowe of the Adams Smith Institute struck a bet on how much the private school pupil numbers would decline by. Pupils across all schools accounted every January, so the starting point is January 2024 before Labour came to power. They're counting the pupils about now to get the answer and we'll get those numbers in the summer. At halfway point, January 2025, just as the tax came in, pupil numbers were down by 11,000 or about 2% of the total pupil count. Thank you, Tom. The Covid-19 pandemic might be over, but the mulling over what happened isn't. While we were off air, the Covid-19 inquiry published its second report, focusing on the first wave of the pandemic. Many of you saw the resulting headlines. The trail of bad decisions and delays that led to 23,000 avoidable deaths. Covid chaos in Boris Johnson's government led to 23,000 deaths, damning inquiry finds. And many of you emailed in to ask where this number came from and is it right? I think it would be understandable if you took those headlines to mean that the inquiry had done some in-depth research about the first wave of the pandemic and after much scrutiny of the information available, their top assessment was that lockdown would have saved 23,000 lives if introduced a week earlier. So would it surprise you to learn that it's not based on new research and analysis? It's simply quoting a piece of research by academics at Imperial College, published back in 2021. That research was itself debated at the time. Here with me to explain this in more detail is Professor Sir David Spiegelhalter, a man who seemed to spend most of the pandemic talking to more or less. He began by explaining that the figure of 23,000, an estimate of the number of lives that would have been saved if lockdown had begun a week earlier, is just the midpoint of a broad range of estimated outcomes. That 23,000 has got an uncertainty interval from 10,000 to 32,000. But even more important, any statistical model, as everyone knows who does this stuff, is based on numerous assumptions and they drive the conclusions and yet they can be challenged. In George Box's famous quote, all models are wrong and some are useful. And this particular model, the Imperial models for the effect of these what's called non-pharmaceutical interventions like lockdown have been strongly contested. One reason is that lockdown was announced on March 23, 2020. The Prime Minister Boris Johnson also announced a range of measures a week earlier on the 16th of March. He asked that everyone stop non-essential contact with others and to stop all unnecessary travel. He asked people to start working from home where they possibly could and that they should avoid pubs, clubs, theatres and other such social venues. So it's actually clear that daily infections peaked before the lockdown. Things were getting better before the lockdown. So if the estimated number of daily infections were already going down before the lockdown, what was going on? For Spiegelholter, it's about the change in behaviour that was already taking place. In such a model, as I said, has to make many assumptions about what was happening and in particular what the effects of the voluntary measures that were introduced on March 16 encouraging people to stay at home, actually closing schools as well, which were done. And these we now know not at the time had actually a very strong effect. And perhaps the Imperial College model didn't fully take account of this. The point is you have to make an assumption about what happened during that week. And the Imperial model makes a quite a strange assumption of assuming that there was a sort of gradual effect on the number, the average number of people that each COVID infection goes on to infect. And they assumed that it declined slowly, steadily over that week, rather than having a sudden change. And this assumption means that if you do posit, well, what if we were locked down a week earlier? The effect of that is a lot bigger than if you had assumed that the effect of the voluntary measures was pretty instantaneous too. While Sir David thinks the R number, the number of people who each person will infect, would have been falling more quickly in that crucial week after voluntary social distancing measures were introduced, but before the Prime Minister instructed most people to stay indoors. The Imperial research assumes that voluntary measures were less effective, and the R number fell more slowly before mandatory restrictions. The data was patchy then, working out how many people were infected was hard. Researchers were using surveillance data, such as hospital admissions, and only later in the pandemic would we be regularly testing people. The point is this just shows that even among experienced statisticians and researchers, modelling is hard. But even if you accepted the model's findings about saving 23,000 lives, that only describes the first lockdown and the first wave of the pandemic. David points out that the Imperial paper itself acknowledges that if the first lockdown had been sooner, the second wave would probably have been bigger. The team at Imperial College made a perfectly reasonable attempt to model the impact of a new virus at a time when we didn't have good data, but it is only one of several attempts that have been made by academics, and this specific number of 23,000 is not a new synthesis or consensus based on what we now know in 2025. David says that this 23,000 stat is a modelling estimate, not an established fact, and he worries that the inquiry report has turned it into one. In the actual report, it said that modelling has established that the number of deaths would have been 23,000 fewer had they locked down a week earlier. In the press release, they changed language a bit and said that modelling shows that there would have been 23,000 fewer deaths. But any self-respecting statistician would not give it that weight, he says. It means that there's no one in the inquiry who actually understands statistical modelling at all, and I think for something that's costing hundreds of millions of pounds and taking vast amounts of effort and time, it's pretty bad that they don't have internal statistical advice which could stop them making such a massive mistake. David has contributed to three other public inquiries, including into the serial killer Dr Harold Shipman and the infected blood scandal. And they all had dedicated statistical teams that I hope would detect this sort of stuff, and I'm just amazed that this gets through. Our thanks to Professor Sir David Spiegelhalter. We asked the inquiry whether they had a dedicated statistical team, but we couldn't get an answer. However, they did give us a long statement that contained the following section about the 23,000 figure. The figure of 23,000 cited in the Inquiry's Chair, Baroness Hallis' report, is from a paper included in evidence, authored by leading data scientists, statisticians and experts in the field of infectious disease epidemiology. The paper concludes that, of the control measures implemented, only national lockdown brought the reproduction number below one consistently. If introduced one week earlier, it could have reduced deaths in the first wave from an estimated 48,600 to 25,600, thereby providing the 23,000 figure. On Monday 5 January, a debate was held in the House of Lords on the government's continued use of the social media website X, formerly known as Twitter. This was in the wake of reports that its inbuilt AI chatbot, Grock, was being used to create sexualized images of women and even children. Since then, the media regulator, Ofcom, has launched a formal investigation into X over the matter. But during the debate, Baroness Ruth Anderson, responding on behalf of the government, argued that X was an important source of news for many people in the UK. Not only are 19.2 million people registered, British citizens registered with X, but 10.8 million families use X as their main news source. That's more than any other social platform, which I found to be genuinely extraordinary. Extraordinary, maybe, but is it genuine? Do 10.8 million families use X as their main news source? Are children up and down the country gathering on the sofa while mum and dad read out their live feeds? Our reporter Nathan Gower saw this number and started digging. Hello Nathan. Hi Tim. So what have you found? So this figure apparently comes from government commissioned market research data, which isn't publicly available. So there's no way for you to look at the underlying data? Well apparently not. That's a shame because it's always my first step on this kind of story. You find out how many people were asked, what kind of questions, whether there's any kind of inconsistency in the data, that sort of thing. Luckily, we do have other publicly available data. This actually comes from Ofcom. Every year they conduct a major piece of research, a big survey of the population about their news consumption habits. Well that sounds useful, what does it say? So I had to dig into the big underlying spreadsheets to find the data. This is literally your job Nathan. I do it happily. It turns out they ask respondents what their single most important news source is. Now the government claim was about X being the main news source for 10.8 million families. Now to me that sounds like a close synonym for most important news source. Agreed. So what do the Ofcom figures say? According to their survey, 3% of adults said that X was their most important news source and it was the same proportion, 3% for 12 to 15 year olds as well. So that all equates to about 1.8 million individuals in the UK. This is much lower than the government's figure of 10.8 million families. And families, famously I would argue, consist of more than one person. So this implies it's more like 20 to 30 million individuals. So nearly half the country. So the national regulator says that just under 2 million people regard X as their most important news source. Well the government seems to be implying that the figure is more like 20 million or 30 million people and that's based on secret data that we can't see. That seems to be about it. And also Baroness Anderson made another claim that more families said X was their main news source than any other social media platform. But the Ofcom figures again tell a different story. They say that Facebook is the most popular social media platform for news with 7% of adults naming it as their most important news source. TikTok, YouTube and X, they're all tied on 3%. And it's the same story for 12 to 15 year olds. Only 3% of them say that X is their most important news source. Okay so that claim seems to be wrong as well. So who's top for news when you look at all sources? The BBC is top. They've got 32% of adults naming one of its various outlets as their most important news source. It's also top for 12 to 15 year olds and that's got 15% of them naming it as their most important source. I imagine that's mainly loyal more or less listeners. Unfortunately Tim, there is no data to back that up. What does your heart tell you Nathan? That they are indeed loyal listeners. Unfortunately, Ofcom's data says they're watching the telly. TV was the top source with 36% of people naming a TV station as their most important source of news. Social media in general was 21%. Other internet sources such as news websites are on 13%. Radio is 9% and daily newspapers that's about 4%. So what did the government have to say about all this? So they told me that a correction would be made in Hansard, the official report of all parliamentary debates. They're going to correct the official parliamentary record because of you. Nathan, you have left your fingerprints on history. I can die happy. They told me that the record would be changed to include 10.8 million families use X as a news and information source. So they're now just saying 10.8 million families get news and information from Twitter or X. They're no longer saying it's their main source of news. That's right and notice that they've added information in there. That's a change. Also, they're still saying 10.8 million families. But because they won't show us the data, we don't know how that equates to individuals. For what it's worth, Ofcom's figures say 8.5 million individuals use X to access news. Nathan, thank you very much. You're listening to more or less. And if you were listening last week, you'll have heard a whole lot of more or less. Five special programs looking at the stats of the nation. As a few of you spotted, one of those stats was actually wrong. In the first episode on Monday, talking about the importance of the NHS in terms of state spending, IFS director Helen Miller said in an unscripted discussion that the NHS employed 6 million people. That's actually how many people are employed in the public sector in total. The NHS employs more like 1.5 million. Sorry. And thanks to everyone who wrote in. Our email address, as always, is more or less at bbc.co.uk. Loyal listeners will remember that back in July last year, we talked about a report from the Bible Society called The Quiet Revival, which claimed that church going in England and Wales went up 50% between 2018 and 2024. This claim was based on two sizable polls from Ugov. What's more, this Christian revival was being led by young people, particularly young men. As we said at the time, there was good reason to be skeptical about these results. The Church of England and Catholic churches weren't reporting an increase in attendance, for example. But Rianan McAleer, the Bible Society's director of research and impact, stuck to her guns. While our data certainly has curiosities in it, and more questions can be asked, I don't think there is enough evidence to suggest that it's totally rubbish. Well, more contrary evidence has since arrived. You see, back in July, we didn't have the 2024 data from the British Social Attitude Survey. This survey covers the whole of the UK and is generally considered the gold standard in surveys. You actually have to say those words when you describe it. It's tracked church going since 1983. And if the Christian revival is real, you would expect to also see it in this data. What's more, at the end of last year, gold standard survey analyst, Professor Sir John Curtis took a look at the numbers. He is Senior Research Fellow at NatSEN, the National Centre for Social Research, which runs the BSA. The Bible Society found that monthly attendance at Christian services rose from 8% in 2018 to 12% in 2024. It's quite a big increase. Obviously, there would have been a dip for the pandemic. What did the British Social Attitude Survey show? The truth is that the British Social Attitude Survey effectively shows the reverse pattern. So if we go back, for example, to our 2018 survey done before the pandemic, we had 12% of people saying that they attended a Christian service, so somewhat higher than the YouGov survey. Dipped during the pandemic. It's come back up a bit, but it's now running at around 9% of people saying that they attend a Christian service once a month. So we've got the same numbers. We've just got the numbers the other way around. Basically, although there was been something of recovery from the position during the pandemic, the long-term decline in attendance at a Christian service of worship still seems to be going on. And there isn't any evidence in our data of any kind marked revival in Christianity in Great Britain. John says the same goes for the supposed revival in church attendance among young people. There's no sign of that either. Although he's keen to stress that the picture is very different if you're talking about all religions and not just Christians, young Hindus and Muslims are much more likely to attend religious services than their Christian counterparts. One interesting thing about the quiet revival thesis is that it's not actually based on a revival in Christian identity. If you actually look at the proportion of people who identify with a Christian religion, it's running at 40% in our data before the pandemic. It's also running at around 40%. Now, actually the decline in identity does seem largely to have stopped. It's just that people are even now less likely to attend, even if they do identify with a Christian religion. Though that 40% figure is almost exactly replicated by the Bible Society data in the two surveys. In other words, what the Bible Society seemed to have found is an increase in attendance, despite the fact that identity is the same and despite the fact that their figure for identity is very similar to ours. The Bible Society points out that the BSA survey changed methodology in 2020 from in-person interviews to online or phone, which could disrupt the long-term comparisons. John says the survey is still a random probability survey and it's been carefully waited so the samples match. But being so far as perhaps there is a difference between the samples, it should, if anything, push our data in the direction of also perhaps finding higher levels of participation. Because for example, when you look at our data on participation in elections, for example, or certain other phenomena, you do find that the sample that's been generated online is perhaps a little more participative, a little more engaged in general in social and political life. So that suggested, if anything, that the change in methodology might have led us to anticipate our numbers going up rather than the figure still showing a decline. Let's go back to the central claim. UGov polling for the Bible Society suggests monthly church going went from 8% up to 12% and the British social attitude survey suggests it went from 12% down to 9%. I have to say, I've looked at the tables that UGov provided to the Bible Society. It's perfectly clear that the Bible Society were reliably reporting the data that were acquired for them by UGov. So I don't think there's any reason to believe that the Bible Society in some way or another were putting a particularly favorable light from their point of view on the data. Those were the data they got and the interesting question is why did they get the data they got, given various other sources of evidence, suggests that perhaps this doesn't seem to be the story. Our thanks to Professor Sir John Curtis. That's all we have time for this week, but please keep your questions and your comments coming in to more or less at bbc.co.uk. If you fancy a challenge, you can take the Open University statistics quiz. Just search out the more or less website and follow the links to the Open University. Until next time, goodbye. For decades, now I'm shining light on the secret techniques criminals use to steal your money. With insight from guest experts and the real people involved in these scams, so you can see the fraudsters coming before it's too late. That's the new series of scam secrets. Listen now on BBC Sounds.
