Times, I know it would seem there any biography of her which is in the book eminently Turin's by listeners straight here. And it was written during the First World War and it contained long essays about four eminent Victorians. Three of them were comprehensively demolished as pretentious and altogether unpleasant. Florence Nightingale, of course, got much more favoured treatment. How objective it really was.
I'm not quite sure. But it did emphasise very strongly careful planning and also the very wide ranging nature of her interests. Both juries, both in the outskirts of what's known Istanbul in the outskirts of Istanbul. Both that, but also very wide ranging interests and activity after issue returned from the crime. It raises the question of the nature of the statistical work that she did. And I think once. Point is that. Computational. Facilities are very limited in those days.
Log tables, slide rules, I don't know how many people in the audience remember a slide rule is a piece of wood. And if you listen very carefully, you could show that two times two with something between three point nine nine and four point one. And if you had a very sophisticated slide rule, as I expect to go, did, you might get another digit coming out. But a. Otherwise, computationally, she would have been very limited. This accounts of the fact, I think in part that that her.
Writing the little obits, I've read lips very strong emphasis on graphical methods of presentation done in a very skilful and careful way. Models of Tyra say that a presentation of quite complicated pieces of data. In addition to contact from her confinement house.
She was in touch with Cantlay in Belgium, who is the leading perhaps the leading biostatistician of the 19th century, Kateri go see inventor of cauterise index body mass over height squared, which is this is still a primary to choose in contact with Castaway and about various aspects of the more technical side, the statistics. I think that. Dr. LaDawn. Well, training find their computational point, a computational cereal's would have been limited. As I said, he the shooter had French curves.
I don't know how many people know what a French curve is in the sense that I'm using eight other senses or might think of it, but. French curve is a shaped piece of wood, which can be used in effective to fit spine functions to 55 cubic seim shrines to empirical data. And it was still in when I first started statistical work just more years ago than I want to think. But I did the use of a spy was expected to be one of the things one would have expected to know.
I was working in in an engineering department. And spine's way spy Unfitting was used with these French curves to fit smooth curves to parallel functions. So that was one side. But the main development. In the development of the more technical side of statistics, Hany is very, very strongly on developments in computation. And it wasn't until about 1870 or 1880, which is towards the end of the Florence Nightingale period and my inactivity.
It wasn't till then that the Hattan Brunswick calculator was invented and perfected in, I think, by collaboration between a Russian and a Swede. And the sound of the broughton's, the calculator, the grinding sound it made, echoed through the lecture halls of London and the universe is no doubt in others and working in many fields, the sound of the Bundesliga calculator was still to be heard as late as 1970 or so.
I'm going to say some more. About concretisation, what it would do would be to add up a large number. It had no memory, but it would it would add up a large number of numbers and that squares and their protest. So you could form some square some of products, some scale of two variables. Some are products of them in one cumulative operation on the machine. But nevertheless, there is no storage. And so when one, for example, inverted a small matrix, I remember inverting a seven by seven matrix.
But the best debateable numerical methods available should have been about nineteen seventies. Nineteen fifties. And it took eight hours to avert a seven by seven matrix because every step had to be checked. Before before I tried to do that calculation, I had quite dark brown hair. You can see now it suddenly turned grey from the stress of doing that calculation. So computation is always. Driven. What part of statistics is commonly and widely used?
It's very exceptionally that one that one can use when in the past use methods that were not very favourably computationally. So the history of the development of statistical theory and statistical methods is very, very much tied with the nature of computation. And in particular, the nature of computation, which is relatively painlessly available. Now, we've heard great lecture. I hope people have written up so we can study in detail. Thank you again, Deborah, for it. Thanks, David.
I believe that David Spiegelhalter has also got a few comments that he'd like to make, so I'll just pass on to David Spiegelhalter. Okay. I hope you can hear me all right. Yeah. Are you going to excuse me being a bit nervous? I always I'm nervous speaking with David Cox around ever since he was my p h d examiner back in the 14th century, sometime when I had dark hair as well. And this is a fascinating lecture.
There are also some details by Florence that I didn't know about. One could see the window and I really enjoyed it. I'd like to focus on one particular aspect, and that's really tied up with the pandemic that we're in at the moment where, as Debra mentioned, the scientists and the statisticians have been working there. What's itself. And, you know, it's just working so hard.
I mean, every other slight aside, I do hope that this is appreciated by people, just how hard, you know, everyone has been working during this period. And for a lot, you know, either for the same salary or for no money at all. It is. And the stress involved in it. So I. But I think in a way, I'm sure Florence Nightingale would have relished this opportunity to bring together a data policy, and she should have been it right in the middle of it.
I'm sure, because she as I mentioned to my when I did this lecture last year, she was like, I you know, one woman. Pressure groups know collecting data influencing all over the place. You knew everybody all from her bed, writing letters, sending presence and, you know, extraordinary energy and effort that she put in. But she was concerned with policies. She she wanted to change things. She wanted to make things happen. And that's what I. That's on your last year I said, which you use Twitter.
But even now, I really wonder, would she have been content to be a statistician who tries to gather evidence? Of course. And to that that can be used by policymakers. But would she have been actually advocating for specific policies? I would love some people's opinion. Look, David's opinion crystals in anyone's opinion on this, because I kind of think it's an it is an important issue.
It's something I'd been wrestling with all throughout this whole pandemic. I've been dying like other statisticians say. They died like Jan and others, beating a lot of media work. And I've been trying all the time to steer this middle way. I just don't feel I don't feel happy to say this is what should be done.
All I want to be able to do is to try to explain what the evidence is and, if necessary, appeal for better communication of the evidence and better evidence and to go out there and actually collect the data. But at no point where I'm always trying to avoid it, in spite of always being asked, well, what do you think should be done? What do you think about locked? Do you agree with it or not? I just I'm not going to say and I've been trying to keep that line the whole way through.
And I wonder if Florence would have been happy to stick to that line if she was sitting there in her bedroom as as Deborah said, you know, being an absolute role model for lockdown. She'd been there quite possibly doing online interviews with News Night, which she had been happy just to explain about the evidence or which you've been telling everybody what she thought should be done.
And I suppose I'd just like to leave that as a slightly open question, my ideas and maybe just my particular feeling. One thing I really like about the statistical profession is that they do not tend to be advocates for what should be done. They're advocates for better data and evidence and trying to understand things are, you know, to see the world as it is. But they do not tend to be people who say, well, this is then what should be done.
And I think that is a good role sometime to say putting saying my opinion immediately. And I might find myself actually. I would hate to disagree with Florence. I don't got to have a dad do that. But some I do think that this is an important issue that she brings that. Yes, that her model isn't the right one, is the one we should follow or not. So. Okay, so just to finish off, I'd like to say thank you so much, Deborah. A fantastic lecture. I really learnt a lot. Great slides.
And I love the tie in with what's going on at the moment, all the work at Imperial and elsewhere and the role of statistician's in all of that. Wonderful. And so I had a great arc to it now. Lovely, lovely structure. And so with that, I'm going to stop rabbiting on and let Jen take control of whatever we're going to get asked next. Thank you very much indeed. Thank you very much, David. Yeah, you make a very good point.
I'm constantly being asked the same thing. They want to give you an opinion on what should or shouldn't be happening and very much just trying to run. Let's just talk about the evidence. We had one question come in asking about whether or not you feel some of the statistical results have been hidden from the community. And I'm guessing it's maybe in response to kind of it that what we think the transparency has been like. Do you think we've been seeing all of the data that has been available?
Shall I start on that? I think there's been a real balance between doing what we normally do and putting things out slowly through peer review papers. So our first react data, we didn't get out there because we were doing it through normal channels. And actually, the speed of this, you can't. And we see it with the vaccine studies that some data goes out there and then people have it.
Has that been peer reviewed? But. Companies have to tell their shareholders with the studies that I'm sharing data monitoring committee for where we're still having debates about what the right channels of communication would be. And there's a real tension between immediacy, scrutiny, and you certainly would keep things absolutely hidden that if the others have got better solutions to that. I'd love to hear them. Yes. Do either of the Davids have any comment on that? Oh, yes, I do.
Yes, I did. I'll begin by David, divided by bodging, first open my big mouth first hand. This is incredibly important and am I so strongly believe as you as we all do in openness and transparency about evidence and data. But of course, and many people within the we got to distinguish, I think, between government and the organisations that are working within government is absolutely crucial. And I distinguish that very strongly. And I spend my time criticising as loudly as possible.
What I feel are lapses on behalf of what we've got as the political adviser, the number ten comms group, et cetera, which essentially are representing what I feel are the politicians, the government. And at the same time, I do tend, I think, quite reasonably, to defend very strongly the role of the organisations that are collecting the data and analysing data, presenting it to the policymakers and to to a really quite extraordinary extent, getting it out to the public. I know.
I'd like to point to you know, I'm sorry, I'm conflict of interest on now non-executive director on the board of the U.K. Statistics Authority. So I would say that the AU and are doing a fantastic job at this, but the AU and ESA are doing a fantastic job. I think you're getting the data out. Just Bozman is new analysis, use a requested datasets and so on public health thing. And that dashboard was pretty crummy to start with, but they've had a fantastic team on that and now it is extraordinary.
And the masses of hits I was on, I'm on it, you know, cut us for so every day I'm I'm clicking on it and checking every day. You know, it's a really fantastic resource. And and to provide an API so everyone could drag the data off that. So to go into other Web sites, the BBC one is very good. And all these other Web sites that people are constructing to to analyse that data.
Excellent transparency. And and it's not just transparency. And in terms what you call fishbowl transparency, just going black and putting stuff out is massive pdaf that you can't get there you can't do much with. And this is transparency in a way that enables people that's I think empowers people. Now joined by security centre and which is, you know, keeps a really low profile.
And I mean, I should say, because the people may not realise that the director of that is Claire Gardner, who was might be a student who's a Bayesian statistics and works at Imperial. So which I think sets a very good standard. And actually, the data that are used that the basis for the tearing decisions is out there. You know, it's being reported very, very quickly now. It is. It is. There is a 79 page PDAF, but the data files are underneath.
It's not easy to find particularly, you know. What's going on in every single area and all the stuff, the masses of it. And and and I've been saying all day, as I say, all these billions being spent on tests and trace vast amounts on, you know, management consultants. Couldn't they have spent some more money on a decent website, given all the sacrifices we are all making continually? And we should be able to know exactly why these decisions are being made about us.
So I would appeal it. I think there has been great efforts made. But more could be done. But it is extraordinary what we have got access to. I think I think compared. I'm not talking yet. I'm not talking about the use of graphs in briefings so that we could get onto that later. But I have to control my language, of course. But but in terms of the transparency, the data, there has been a very good example shown. And I think basically within the agency is a very huge amount of goodwill to do this.
David coughs Do you have any comments to add to that? No, I feel, sadly totally on the fringe. Well, outside the fringe of this, I just look at the telly. I'm not a bad thing with the sanity. But like what I say, you know, I think it's. Often trying to put too much on one side or top or slide, take it away again. And it's an old trick. The older used to be very common in medical issues.
If I went to any 30 or 40 years ago, people would put up very complicated slides and take them away again before anyone has a chance to read them. I see far too much in that. I don. So I really criticised the work that's underneath. Yeah, that that's a that's a separate issue. That's the actual presentation to the public. Seems to me often astonishingly, very poor. David, I'm very. I think you're absolutely right.
I think it's a great shame because they're very good analysts and people within government, within agencies. But then when it comes this public presentation, it's some of the worst examples, I think the sort of stuff that if it was done by first year, each student at a test lecture, we'd be there. Apsley bellowing out loud and saying this is unacceptable as part of a presentation. And then, you know, lots of coloured lines, which I would within it with a legend somewhere.
So, you know, some awful stuff. And you know, this one recently on the. Which I've been ranting on about on the main page, which is explaining the winter plans terribly important. And it's some ghastly thing knocked up in Excel when you can't even tell whether whether the scale is linear or logarithmic or that somebody is used to make it look science. And to put some colour rate and, you know, rather than actually to inform anybody. And it's not under any standards whatsoever.
Which given there are so many good people, you know, you know, in you know, in government agencies who know how to do good visualisations. I really I find it very upsetting. This is stuff that comes from there and is much better that. OK. So I went there are a couple of people with hands up, but I'm going to try and do this and I've been trying to keep an eye on the order in which things have come in.
I think it's quite interesting. As with a Florence Nightingale lecture, we've had a couple of questions come in on this idea. I tend to balance one of which is saying that daughter attended a mass open day and there were no women speakers. And when they asked why there were no women, because the comment was how many famous women mathematicians can you name? I'm going to ask you to go through them right now. But there's a question of do we always need to make sure that there's a woman on a panel?
And we've also had a question from Denise Leavesley to David Cox, a festival. She says it's a fantastic book. Thank you very much, Deborah. But, David, your celebrated in the community for your research collaborations with large number of women. And Denise is interested as to whether or not this has been a deliberate decision on your part and how you feel it has contributed to your thinking or not a deliberate decision. Just a great, great, great fortune to work with.
Very able people. Both genders, genders. Incidentally, college. The appointment of sequestration's certainly had nothing to do with gender at all, and we ended up with not far short of them, mixture of men, women on marriage. Not because of the new policy. So I tried. Right. Anybody who's got some. I'm sure I can just work with them and that we can together get productive results better then make good on her. But conscience, gender bias. No, I don't have one.
I think it's a really interesting point at this sort of positive discrimination and unconscious gender bias, and should we always make sure there's a woman on the panel? Deborah, do you have any comments on that? I do. I think the point at which people apply for a job, you want to appoint the best people.
The the thing beneath that is making sure that the level playing field so that women get a fair crack at doing the things that might get them to being appointed to a lecturer or a chair and so on. Where I would take a rather different line is when you're looking at particularly committees or public facing things, then absolutely you want balance diversity and whether that's RSS council for which we need nomine.
And so we we've just elected one batch. We need nominations by the end of this month for the next batch. The about the last thing I do as outgoing president is to chair the present nominating committee for Cylvia successor. We need diverse nominations for that, because if you haven't got those nominations, you're not drawing on the best.
And so wherever you haven't got diverse nominations, you need to look and ask why on a panel we are sitting here and in this day and age, it would be wrong if this was for men. And it's, I think, particularly helpful in that, except for things like celebration, women's mistakes. If it's all women, there's at the moment, this is a rebalancing role for that. But, you know, I'm very conscious that that is by no means the other diversity.
We're sitting here for white people. So clearly, there are some things need doing to change that. So, yeah, so that's where I'd have to say that them certainly if you want to attract your best students to university and you put an all male panel and then you wonder why they go somewhere else. Yeah, I thought partitions bear to think it through a little bit better than that. But Baker. But I mean, it's one of the reasons I do feel very proud to be part of the statistical profession in that.
In my experience, there isn't a need to make a big effort to get to a gender balance on research teams or organisation. It just it just happens because there's just really as of now and there's a big day there, especially with the younger generation coming up who are majority females in the majority. I think the younger fellows for the old you know, the old males like her. Like me. It's Dave flips around. But that's going to be changing.
So I really like the fact that with the way not a great effort is needed to get that that sort of that balance in statistics at the moment, which is I find I think it's something hugely to celebrate. Would be different in maths, I think. But it just shows, I think, what a wonderful subject we're in that we don't have to make this effort. But as Deborah is so rightly said, this is only in terms of gender balance, as many other characteristics as well.
Concerned about. Yeah, I actually did a little bit of an analysis looking up membership and the gender split. And, you know, you very much see that in the older generations it's very male dominated. But as you filter through, it's becoming more 50/50. And I do think that naturally, as time goes on, there will be more women taking president's roles and things like that. So, yeah, I'm quite encouraged by it. So we've got a couple of hands up.
So I'm going to go to Len, first of all, who had a hand up for a little while now. So I believe Leon will meet you. See, you can talk. Len, thank you. I'd like to respond to David Spiegelhalter question about would she and then present a problem and ask you have any of you know, any means of doing anything about her or any answer to it? And it would she have given specific direct advice to policymakers like cabinet ministers?
Yes, and she did. And her brief to the pull of to the what's called Cubic Space Committee, which was part of the poor law bill of 1867. She certainly gave very specific advice on that. And that can be seen as the first step towards getting to a national health service, that is of establishing quality care for the poorest, not those who can pay for it. So that was that was very important.
But the problem I want to re raise and we're getting away from the issues you've been talking about now is going back to the Crimea and war. We don't have to my knowledge, correct me if you know something, a really decent count of the war dead. The Nightingale's analysis and the royal commission analysis were both based on data from the Army Medical Department, a nightingale often said how they were conflicting reports of up dead.
And of course, Alexander Tullock collected data. It wasn't just the official people. And she did know in places that the number of people buried was greater than the number of people who died. And and was it suggesting anything went wrong. But clearly there were false in the record keeping. But you see, her data for those wonderful polar area charts is based on what got to the Army Medical Department.
And that's only people who got admitted to Hospital B at a regimental hospital or a general hospital. But men and officers who died on the battlefield, they're not counted there. Now, when I was in the UK last February, March and was planning to stay longer, I intended to go to the National Archives because they have the reports of the adjutant general, which the British Library doesn't have, and that might help to fill in those gaps.
But I have never seen a, you know, a table which gives you deaths in regimental hospitals, in General Hospital, blah, blah, blah. And deaths on the battle ground and our deaths of doctors and nurses and other people who were there doing other kinds of things. So we've never had a really full count on on on that subject. Anyone got any ideas or is willing to undertake to do it? So I can I bodging Len? Lovely to hear from you. I'm sorry I haven't got any ideas.
I don't know about these data sources, but I would like to draw a modern parallel because the crucial thing about this is that you can't interpret data unless you know the process by which it was collected. And the modern parallels that, you know, the daily deaths reported on Kofod off. The people who have had a positive test then died within 28 days of any cause. And so people who caught it in a sense and dropped dead before they got tested won't be included in the daily count.
They won't be in the dating statistics. They'll be in the death registration's if if they decide that they, you know, put it on the test. That's difficult without being tested. So there's deaths that are not occurring there. Again, people who got Koban, they got run over by a bus three weeks later. And what also will be included as a Kovar death. And so it is rather important ways to know what is being counted.
And I think this this whole pandemic, although obviously there have been examples of wonderfully sophisticated modelling and computation, hugely elaborate analysis, so much of it has come down to the absolute basics with which night and go would have been so familiar. Which is kind of tabulating data and understanding whether you can believe it or not.
And at what question you're trying to answer, because if you're trying to say Deskovic covered directly caused death, that's different to monitoring deaths. So if you're clear what the question is, then you can be clear with the data you've got is fit for purpose. Don't. David. Yet. David Cox, if you've got any comments. I don't know. OK. So I think that there's a really interesting sort because we're only about five minutes left aside now.
So there's quite an interesting comment here on sort of practical statistics. Well, we've also had a comment that saying that Florence Nightingale would definitely be telling us to wash our hands. But then there's a question talking about there's a tendency for statistical training to become increasingly mathematical to the exclusion of practical problem solving.
Have you got. What practical experience do the panel members consider essential for the next generation of statistical consultants and how best we can communicate that to the public? And there's a sort of similar question saying how there's a lot of mathematical statistics where they may never have performed an experiment or survey in their life. And then we've got investigators on the other side who require statistical training.
And what's happened to Nightingale's recommendation that practitioners do the education? Just take your comments on that. Well, could I make a comment? Yes. That's the general principle, is you teach people things by developing out of what they already know or interest. So if you are interested, if you have students who've come to study at university, study of that and are deeply interested in pure mathematics and nothing much else, then you have to stop it.
Probably Luti and gradually steer them towards the notion that he's not beyond all of this. Mathematics are very, very interesting and important scientific problems, which you can which they may be having to do something useful in that. The mathematics and develop into into application. For most people, that would be an absolute disaster. You have to start with applications. Those are to some extent reached. Appreciate. Appreciated. But the student in question.
You have out of that into some general ideas, out of the out of the narrowing of interest into the unknown. And so it depends very much way you present your audience is snarky and it's not a mix. You're in trouble because you might have to try and appeal to the different categories of people before you do the more details with mystical ideas. And of course, it depends enormously also whether you've a connexions to explain the Hollow's states or they have that very,
very small and that that makes a big difference. Can I button and they won't. I've been working this air for nearly 50 years and I've finally actually learnt the wisdom that David has just expressed. It's taken me a long time. I knew I should have listened to him a lot earlier. But I know absolutely that I grew up in the mathematical route.
You know, Jar's joined the pure maths until he got too difficult and went through the probability and then mathematical statistics and and then that's the route I took. And that's what I taught for years. And that's how I worked with papers I wrote. And it's only recently, I think that I really flipped around and produced the book. Anyway, the point about this book, The Office Statistics, is just my like myself attempts at redemption, at completely flipping the whole thing around.
And as David said, it depends where people are coming from. I was going to call it starts with data scientists because I wanted people to not come up through the mathematical route, and that's why probability doesn't come until two thirds of the way through. So the island's statistical inference putting them together is, you know, near the end of the book, because there's so much you can do through problem solving. Actually, problem driven all the time. Question driven. Problem driven.
You can do such a vast amount. I'm so amazed me how much you could do with all this mathematical stuff. And so I'm I'm you know, I've had my Damascene moment, you know, a few years ago. And now I have completely flipped around how I think statistics should be taught except for the people, you know, who have come up through that mathematical route and they have to be brought in in a different way. So I think implicit in what David Cox was saying was that you kind of need both.
Know even saying wherever they start and you need to sort of edge them in the other direction. And this may be the time to say that 40 years ago, I was doing a call since Disclaim France, the. It was a joint project as a region calls. We had Coxon. Hinckley has our course text. And what John France said to us, who is my lecturer, was this is a book you won't get everything out of the first time.
You need to reread it next year and again in five years or so, because the more you have practical experience, the more you'll realise what he's trying to say. And equally, I would say those doing applied work from time to time need to get to see the wood for the trees. So we wouldn't be upset with covered if we hadn't had people dealing with other infectious diseases. I keep the details different, the modelling design.
So you've got to keep both in tension, which certainly are both today if it's a done throughout their career. But I've still got a copy of Hickox and Hinckley, and when I get back into my office, I shall be rescuing and bringing it home. Got mine to. OK. So we're pretty much at time and I'm aware that I need to give Crystal a few moments to say thanks to everybody. And the final goodbyes. But are there any final comments from any of you that you'd like to make?
That's around at the panel discussion. I just want to know what a pleasure it's been. Yes, and thank you again for the lecture. Fantastic. And thank you to Oxford for putting this together, because it's been the most wonderful event. And though I'm very sorry that we're not going for tea and cake and water and Oxford and a dinner for some of us, I think we've probably had a very different and in many ways richer discussion
than we might have done if we've done it when it was originally planned. So thank you for reinventing it. I'm one of the last things I will say is one of the advantages of doing it online is that everybody writes their questions in the box and we have got questions that we haven't been able to cover. I don't know if there are going to be any arrangements, but it may be possible. So I apologise for any questions that couldn't come up.
But it may be possible to put those to our panellists after the event and perhaps write it up somewhere. And I will have to leave that to Oxford to see if it's possible. Sorry if I've given you a task that it's simple, but there are some really interesting questions that we didn't get. And I apologise for that. But thank you very much. A really interesting panel discussion. Three of you have been incredible. And I will handle far final goodbye.
OK. Thanks to all of you for taking part and for spending this time being so extra special, thanks to Deborah. She did first the lecture, then the panel. But it's been really great to get your views on all of this. Both looking back at what Florence did and what she might have done if she were alive now, as well as bringing all these contemporary issues, because I think I really genuinely think that mathematics and statistics specifically is just the ultimate transferable skill.
You can move or work on one thing and to another and and thereby meet such interesting people, as well as helping to answer interesting questions. It's been a tremendous pleasure for us to host this sit in a different way than we originally planned, but nonetheless bring it to everybody. And hopefully we were able to open this up to a wider audience that wouldn't have been able to to come had it been specifically in Oxford.
So, you know, there are some silver linings in this. Deborah has said that very clearly. She is willing to make her slides available. They are currently in a form that is too big to easily distribute. So we will have to do a little bit of I.T. compression of that before people end up hanging, trying to download them. But we we have your contact details. And so we will make those available. You saw the link already in the chat. So do visit the virtual exhibition.
If you haven't had a chance to already, there's lots of interesting stuff there. It's like you get to stand in a virtual room and and choose what you focus on off the wall. It was a tremendous honour for Beverly Layne who helped organise this and to go along and actually see and choose amongst the pieces of her original correspondence and huge thanks to Paleo for sharing those with us to Beverly for having organised stage one and then stage two.
And to everybody who's been involved to bring this to. So thank you all very much. It seems like it's the middle of the night, even though it's only five p.m., but really best wishes to everybody. Thank you for taking part. And we look forward to next year. When I would. Thanks very much.