← More or Less
More or Less
03 Sep 2010
00:27:50
How reliable are life expectancy figures? Can cycling ever be safer than driving? And,...
Loading summary
Transcript76 lines
- [00:00]
A
Thank you for downloading this week's More or Less podcast. Here's Tim Harford.
- [00:05]
B
Hello and welcome to More or Less. If numbers were food, we'd be ready. Steady cook. This week, we'll be whipping up a feast from the most unlikely war. Bad science. Bicycling. Yes again. And death.
- [00:19]
C
It's 316 times more difficult to kill 10 times more people.
- [00:24]
B
But first, amid this summer's political argument about public sector pensions described as unaffordable by Nick Clegg, journalists began, quote, life expectancy figures. Which is fair enough, because the longer we live, the less affordable our pensions become. I understand there is an upside to living longer too. The trouble is, there are several ways you can work out life expectancy and over the years, few have proved very reliable. Listener Paul Sweeting wrote to us to query figures used in a BBC News Online report.
- [00:53]
D
It was an article I saw which was talking about the increase in pension age and talked about an increase in pension age from age 65 to 66 and then said, for comparison, the current life expectancy for men is 77 and
- [01:04]
B
it's 81 for women, according to BBC News Online. Then a man retiring at 66 can expect a mere 11 years rest after over 40 years of toil. But that's wrong, says Paul Sweeting, who, when not listening to more or less fills the lonely void by working at the University of Kent, where it turns out he's professor of Actuarial Science.
- [01:28]
D
The number they quoted is what's called a period life expectancy from birth. Say it was calculated using 2008 mortality data. So what you would do, you would say look at each age and see how many people there were of that age throughout the year. So say how many 30 year olds there were and then say how many 30 year olds died through the year. And that would give you a mortality rate for 30 year olds in that year. And you do that for every individual age. So you'd have the mortality rate for every individual age in 2008. What you then do is you use all this data, string it together and say if someone was born now and they experienced the same life expectancy as everybody did at every age in 2008, then we'd expect them to live 77 years. So it's looking at life expectancy based on no improvement in mortality rates at all. And that's very conservative. If you look at what's called cohort life expectancy, which considers the mortality that you would probably experience or you'd be expected to experience as you move through your life. That suggests that you'd have a life expectancy from birth of 88 or 89 years for men rather than 77, and about 92 years for women rather than 81.
- [02:43]
B
So we have to distinguish between what's called the period life expectancy, which doesn't allow for future improvements in mortality rates, and the cohort life expectancy, which does allow for these improvements. To understand these statistics a little better, we thought it might be useful to personify them. So we set the newest member of the more or less team, Wesley Stevenson, a challenge to work out his own life expectancy. Wes is here now. Wes, how long have you got?
- [03:10]
E
Well, the best case scenario, it seems, is that I could live to be a thousand.
- [03:13]
B
Wes, that's the first sentence you've uttered on the programme and it's ridiculous. You're in denial about your own mortality. Man.
- [03:20]
E
Well, perhaps I am. The whole question has made me a little bit nervous. But first, the usual health warning on this. There are obvious problems applying averages to an individual, but putting that to one side for now, Paul Sweeting's actuarial tables suggest that as a 34 year old, I should expect to live to 86. That's another 52 years. And it's interesting to see how much my life expectancy has improved since I was born. In fact, Professor Sweeting thinks I've done pretty well to get this far.
- [03:47]
D
The fact you've actually made it as far as age 34 and congratulations for doing so. Thank you, means that you have already got past some of the hardest parts of the mortality curve.
- [03:58]
E
When you say the hardest things of the mortality curve, am I lucky to be here?
- [04:04]
D
There's a few interesting peaks in mortality that happen early on in your life. You've got the fact that mortality in the first year of birth is still on higher than mortality for age 2, 3 and 4, because clearly when anybody moves from the security of the womb into the real world, it becomes much harder. So there's that issue. And then for men in particular about age 15, mortality rates start to increase as testosterone kicks in and people start taking more risks and driving too quickly, getting into fights and so on, and you get what's called the mortality hump around late teens, early 20s. So mortality rates for men peak at around 20, 21, then drop down again as men reach their mid 20s, and then start rising again up until age 121, or whatever ultimate age you choose.
- [04:51]
B
Still, it's all a bit vague and uncertain, Wes. If Moneybox's Paul Lewis and I Set up that massive pension scheme we keep talking about. Small differences in life expectancy could bankrupt us.
- [05:00]
E
Yes, one extra year of life expectancy for the people in your pension scheme and adds, on average, 3% to your liabilities. Pension schemes are trying to get better at calculating life expectancy. In fact, they have to. The pensions regulator now requires companies to make realistic assumptions about it, so they'll be getting into the detail where you live, how much you exercise, your income, your health and, most importantly, whether you smoke or not. Now, the best I could find from my BBC pension scheme was that its members are expected to live on average to 90, which is not too bad. But I needed an answer that was tailored to me. So I called Professor James Waupl, director of the Max Planck Institute for Demographic Research in Germany. He's got his own model for working this out.
- [05:44]
F
Life expectancy has been increasing in the past in countries doing well, like England, by about two and a half years per decade. And two and a half years per decade, that's about three months per year. So it's really quite remarkable, six hours per day. So we use that as the basis for projecting someone born 34 years ago might live on average. That's the 89.6. And then I took into account the fact that you were in good health and had good health habits, so I adjusted your life expectancy to increase it by four or five years based on your good health.
- [06:12]
E
And this is the average. This is the mean, isn't it?
- [06:15]
F
That's the mean average. So you might die tomorrow, you might die at age 100, and so there's a small probability of dying tomorrow, there's some probability of dying at age 100, and then you just average every possible time of death weighted by the probability of that time of death. It turns out that the 50, 50 point, called the median, that's the age that you have a 50% chance of reaching, which is about three years higher than the mean. So if the mean is 94, for you, the median, the 5050 age would be 97. And the reason the median, this 5050 age, is higher than the mean is there's some chance you might die soon, within the next few years. But if you make it to 94, you're not going to make it to 196 or something like that. You might make it to 100, 110. So there's not too much space to live after 90, but there's a lot of space to die before 94. That's why the mean is less than the Median.
- [07:02]
E
And what about the mode? When do the most people die?
- [07:05]
F
Yeah, the mode. The most likely age of death, that's about three years higher than the median. So for you, the single most likely lifespan is probably something like 100.
- [07:16]
E
So the short answer to how long I'll live is 77 if you take my period life expectancy at birth, 86 if you take my cohort life expectancy now 90 if my pension scheme is to be believed, or 100 if you take Professor Valpel's word for it.
- [07:32]
B
At the beginning, you mentioned 1,000.
- [07:35]
E
Ah, well, yes. That estimate came from Dr. Aubrey de Grey, who is chief scientific officer at a charity in California called Sens. They're out to promote research into regenerative medicine, which they say will slow aging. His theory is that technology will advance at such a pace that. That by the time I reach old age, there will be enough technology around to keep me alive for another 30 years, which in turn will be enough for further technological advances which will keep me going for another 30 years and so on. He told me, over a slightly fuzzy line from the Swiss Alps that it's not as crazy as it seems.
- [08:11]
G
Now, you may intuitively think that it's very unlikely that we would ever improve the therapies against aging rapidly enough, but really that only means improving therapies maybe three or four times faster than we already are improving them. So that's not terribly scary, really. So then the question is, okay, at what point in the future are we going to reach that longevity escape velocity, that rate of progress? And I think that we've got maybe a 50% chance of getting there within about 25 years from now. That means you're going to be 59 and I'm going to be, what, 72. So we've got a fair chance of actually benefiting from those therapies and thereby being able to postpone the ill health of old age and the diseases and mortality of old age as long as we like.
- [08:59]
B
I understand the logic, but why a thousand?
- [09:02]
E
Well, Dr. De Grey says that something will kill you eventually, even if you keep the effects of aging at bay.
- [09:08]
G
A thousand years is a number that sounds very much as though it was pulled out of the. But it wasn't really. I have indeed predicted in the past that people will have an average lifespan of around 1,000. And that's the number you get if you just calculate how long people would live on average, if they maintained indefinitely, however old they were. The same risk of death each year that young adults in the industrialized world have Today you know, risks of death from car accidents and so on.
- [09:38]
B
Dr. Aubrey de Grey from Sens in California. Well, Wes, sounds like you could be with us for many years to come. As long as you're careful.
- [09:46]
E
Yes. Don't worry, Tim.
- [09:47]
B
I'm always careful.
- [09:48]
E
See you next week.
- [09:57]
B
Wow. Well, Wes did say it always was a problem to apply averages to the individual you're listening to, more or less. In association with the Open University, that was Wesley Stevenson and I'm Tim Harford. Last week we studied the surprisingly patchy statistics on cycle helmet safety. We got lots of emails on the subject, not all of them polite. And it's clear that this is a topic which gets a lot of people very excited. Even before our program was broadcast, one of my loyal Twitter followers, Tweety Mike, sent me a message saying, please, please prefix both ends of your story with the truth that helmetless cycling is safer than both driving and walking. Well, we didn't, partly because we simply don't have the data on helmetless cycling anyway. But Mike's claim did catch our attention. Is cycling helmeted or not, really safer than driving or walking? Not according to the Department for Transport. Their latest data do suggest that cycling is safer than walking. Between 1998 and 2007, on average, there were 33 cycle fatalities per year per billion kilometers travelled by bike. Over the same time period, there were 42 pedestrian fatalities per billion kilometers walked. But look at serious injuries and minor injuries and the picture is reversed. Cycling is more dangerous per kilometre traveled and cars are very much safer still, at least for the people inside them. Just 2.6 deaths per car occupant per billion miles traveled by people riding in cars. So that seems pretty open and shut. Cycling is over 12 times more dangerous than riding in a car, although walking is even deadlier. But is that fair? I spoke to Dr. Jennifer Mindel, a health researcher at University College London. She told me not to take the Department for Transport figures at face value. One question is the recording of deaths and injuries. If you're in a car, that's clear cut. But if you're walking to work, fall over and break your arm, that may well not be recorded as a travel related injury. And by the same token, an accident on a BMX bike on an off road track will often be recorded by the hospital as a cycling injury and will then end up in the travel statistics. A second question is whether it's right to measure deaths per kilometre travelled. Another possibility is to measure them per hour. If you spend an hour cycling, is that more dangerous? Than spending an hour driving. This makes cycling look a lot safer, although whether it's sensible depends on what you want to measure. Defenders of the per hour measure say that people tend to spend no more than an hour commuting, no matter what mode of transport they use. On the other hand, if I want to know whether I'm safer driving or cycling across London from my home to White City to record more or less, then per kilometre is the measure I want to use. Even then, says Jennifer Mindel, I'm not necessarily looking at data that will answer my question.
- [12:50]
H
You've got to compare like with like. So, for example, you want to compare walking or cycling with driving along similar roads. Driving on a motorway or a trunk road is about 10 times safer than driving along the roads you would need to get to to get from your home to White City. But we can't make that comparison. We don't have those figures. Secondly, we know that young men are the group that are by far the most likely to be involved in a traffic collision, whether they're cyclists or drivers. A large proportion of cyclists are young and are male, whereas for drivers, the high increased risk for young drivers is hidden by the fact that most drivers aren't young men. What we really need to know to answer your listener's question is, is what are the risks of driving, walking and cycling along similar routes for people of the same age and sex group? And that's what we don't know.
- [13:50]
B
One of the things that really surprised us when we did our item on cycle helmets. I'm a cyclist, my producer is a cyclist. We have warm feelings towards the idea of cycling, but we were so surprised at how many people had a really set position both on the issue of cycle helmets and more broadly, on the issue of cycle safety. There seem to be some real fundamentalists out there on either side. When you've been studying this, have you encountered this? Are you coming from a particular angle
- [14:17]
H
as a public health doctor? I'm coming from the position that, yes, there are some injuries due to cycling, but the benefits of cycling outweigh the risks about 20 to 1. There was a study done 30 years ago that showed that there are 20 years of life gained by the regular activity of regular cycling for every year lost by injury or death. The other important thing about encouraging people to cycle is that there's now growing evidence, both for pedestrians and cyclists, of this concept we call safety in numbers. The more people there are who are walking or cycling, the safer it is for everybody. So, for example, in London, cycling rates of More or less doubled, but the number of injuries and deaths hasn't changed, so the rate has actually fallen substantially.
- [15:05]
B
Are there any countervailing factors? You do seem to be coming a little from one side here.
- [15:11]
H
I think we need to remember that driving in the UK is one of the safest places in the world to drive. Our figures are amongst the best, but it also depends on who you are, how experienced you are and what sort of road you're driving on. And the problem has been that people have not been making this like, for, like comparison. So I suppose the position I'm coming from is that the risks of cycling have been exaggerated. If you look across all age groups, across the whole country, the risk per year for UK drivers of death is 1 in 30,000. Now, that's slightly less than it is for uk cyclists at 1 in 23,000. But if you look at French drivers, their risk is more than double. The UK cyclists, their risk is 1 in 10,000 in the UK. Yes, cycling probably is more dangerous for a cyclist than driving is for a driver overall, but the difference between them is much less than people think. It's the same difference as driving in France instead of the uk, or driving on suburban roads instead of trunk roads and motorways. And nobody thinks twice about making that shift. And perhaps we should stop exaggerating the risks of cycling.
- [16:21]
B
Jennifer Mindel. Another way to think about this is to look at those cyclist casualty figures. 33 per billion kilometers travelled, that's not that high. About one death per 30 million kilometres. So if you've cycled 30 million kilometres and haven't suffered a fatal accident, we'd like to hear from you. Our email address more or lessbc.co.uk.com and I'll be cycling home tonight with care, but also a reasonable amount of confidence. If you want to find out more about that or any other item on the program, please Visit our website, BBC.co.uk more or less now, at the beginning of the summer, an Oxford neuroscientist, Professor Dorothy Bishop, announced a new prize for science reporting. Where do we sign up? Was our first response. And then we discovered it was a prize for the most inaccurate report of a piece of academic work. And we decided this was a prize we'd rather not win. Well, statistical reporting and science reporting aren't quite the same thing, but there is a big overlap and similar issues often arise, so we thought we'd poke our noses in. Is science writing in the UK so bad that it deserves a wooden spoon prize? We spoke to Natasha Loder, Science Correspondent at the Economist and chair of the association of British Science Writers, and to Dorothy Bishop herself. I began by asking Professor Bishop why she'd established her prize.
- [17:41]
I
I put forward an Orwellian prize for scientific misreporting because I really got furious at reading accounts of published work that were purely inaccurate and thought long and hard about just perhaps writing a critique of a paper that was inaccurate and then thought, well, no, this might be a better way of really sort of striking back against the people that really don't seem to put much care into what they write.
- [18:05]
B
Natasha, science reporters stand accused. I mean, are things really that bad or is that an exception?
- [18:12]
A
I do think that in judging the prize, you are going to have to speak to the journalists concerned, because lots of pieces of science journalism are not actually based simply on the articles themselves. You may actually end up talking to a whole bunch of scientists. In fact, you may talk to someone who would summarise on the basis of this paper what 10 years of work would mean. So you may find in some cases that you can't necessarily judge the piece of reporting just against the science reporting itself.
- [18:39]
B
Dorothy, isn't that fair? The job of science journalists isn't to summarise a particular paper faithfully. It may be to do. To do more broad work.
- [18:50]
I
Indeed. And I would start by saying I'm not tarring all science reporting with a negative brush at all. This is by no means an attack on the profession, it's an attack on the individuals that misreport things. And I accept in my criteria for the award that there are mitigating circumstances in particular press releases, or indeed researchers who themselves want to sort of hype something up just in order to get it into the papers. And it can be difficult for journalists sometimes to know when that's the case, as opposed to when it's really a genuinely new, exciting breakthrough.
- [19:23]
B
Natasha, do you think that the scientists have just misunderstood what journalists are trying to do?
- [19:28]
A
I think there's plenty of examples of where scientists have misunderstood what we're trying to do or even how we operate. I think, having read Dorothy's blog, I think I'd really like her to invite her to come and actually work with me for a week and actually write an article for the Economist, because I do.
- [19:44]
B
It's not that you can't refuse.
- [19:46]
A
I do think there's a sort of misunderstanding about how things work.
- [19:49]
I
But what about when a scientist does a piece of work and it's completely misrepresented? What are we supposed to do?
- [19:55]
A
Well, I would suggest that you Complain to the Press Complaints Commission.
- [19:59]
I
Well, I think it's rather more fun having an Orwellian prize myself, Natasha, More
- [20:02]
B
or less listeners are interested in the subject of statistical reporting in particular, which really is a subset of science journalism. I mean, do you think that statistical reporting is particularly badly done?
- [20:12]
A
Yeah, there is often a misunderstanding about relative and absolute risk, and that trips up journalists all the time. I don't think the more experienced health writers would have that kind of problem, but I could be wrong.
- [20:23]
I
But there's a real relevant difficulty here, though, that. I mean, in my area where we're working, say, with developmental disorders or neurological conditions, a lot of scientists do papers where they show that there's some difference between two groups. So they show that there's difference, say, between children with autism, children without autism or whatever. When you talk to a journalist, they immediately want to know, well, what's the relevance of this? And quite often, to be honest, there isn't immediate applied relevance. You're finding out more about the nature of the condition.
- [20:48]
A
So why are you doing the research if it's not relevant?
- [20:51]
I
It is relevant, but it's not necessarily immediately applicable. You're trying to find out more.
- [20:55]
A
Couldn't you just say, well, it's not immediately applicable, but what we would like to do one day is.
- [20:59]
I
And then it gets reported as this is what you're going to do.
- [21:02]
A
Well, I mean, what's the problem with that? I mean, you've been given a research grant from the government, which is funded by the public. The last time I looked, to actually do a piece of research which is supposed to have some relevance to the public. I don't have a problem with journalists asking you to defend why that's relevant. Scientists really don't like being put on the spot and being asked about what the relevance of their research is. But I'm afraid it is a question I feel I have a duty to ask.
- [21:24]
B
Now, Dorothy, you announced your prize in June. It's been summer, it's all been quiet. When are you going to announce the results?
- [21:31]
I
In January, I was thinking, because that seemed to give enough time for things to happen.
- [21:35]
B
Well, I'm sure you'll get in touch with us then. We'll be interested to see what the nominations are. Dorothy Bishop, Natasha Loder, thanks very much. Now, Barack Obama announced the symbolic withdrawal of U.S. troops from Iraq this week. Although 50,000 troops will remain in the country for now, the insurgency in both Iraq and Afghanistan has ebbed and flowed. And as in Vietnam and Malaya, as well as lower profile wars such as Insurgencies in Colombia and Peru. It's proved a tremendous challenge for conventional armies to deal with scattered guerrilla forces. So what do we know about such guerrilla insurgencies? How they start, why they have sticking power, and how they end. Well, for an answer, why not turn to a physicist? Sean Gorley is a member of the Complex Systems Group at the University of Miami and is using his research to start a new consulting firm called quid. Working with other researchers, he's been looking for patterns in the data emerging from Iraq. And he's found that the same patterns that emerged from the Iraq war also apply to other insurgencies across the world and across history. He's even been briefing the Pentagon, and I asked him to brief us too.
- [22:40]
C
So the first thing that we looked at was the numbers of people that were killed in each attack. And I guess when we first started this, the thought was that there'd be sort of an average size. Like, you know, there'd be seven or eight people killed, and, you know, there'd be a few attacks that were quite large, but they would sort of tail off very, very quickly. And it would kind of look, I guess, like a bell curve or a Gaussian distribution.
- [23:00]
B
So a Gaussian distribution. I think a lot of people would know that as the normal distribution.
- [23:03]
C
Yeah, right, right, a normal distribution. So your bell curve. So the heights of people would be well described by a bell curve where you have an average of 5 foot 10 and a standard deviation of a few inches. And so basically, with a bell curve distribution, you're not going to see people that are 10 or 12ft tall. So what we found when we looked at the distribution of attack sizes in Iraq, what we found was that it was not a Gaussian curve, not a bell curve, it was a power law distribution.
- [23:28]
B
Effectively, what we're talking about is if the attack's twice as big, it's half as likely. If it's four times as big, it's four times less likely. That sort of law. Were you really surprised to find a power law?
- [23:38]
C
The fact that any kind of mathematical distribution existed was perhaps the biggest surprise.
- [23:42]
B
And actually it was a bit more specific than simply being a power law, wasn't it?
- [23:46]
C
Absolutely. So the distribution of attack sizes in Iraq was a power law distribution with an exponent of -2.5. The way to think about that is it's 316 times more difficult to kill 10 times more people. So if you go from, you know, 10 to 100, it's 316 times less likely, which could be a proxy for more difficult. That 316 times more difficult is something that we see across conflicts.
- [24:11]
B
So what's the explanation for that pattern?
- [24:13]
C
That's sort of the next step was to sit down and say, well, look, we know we've found this distribution that characterizes conflicts around the world. What's creating it? You want to take a step back and say, well, what are we looking at here? An attack size is a proxy for the strength of the group carrying out the attack. And so what you're really looking at is a distribution of group strength. And group strength is a product of the dynamics that create groups. So, you know, what we're looking at is really the autonomy and decision making processes of the different groups that are involved in the system.
- [24:40]
B
David Petraeus, the general who is credited with leading a successful U.S. strategy in Iraq. After several years where the strategy was far from successful, he's now in charge in Afghanistan. He seems to be a smart guy. He's got a PhD, he's surrounded by advisors, other generals with PhDs. If he phoned you up and said, sean, I appreciate these insights from academia, what should we be doing differently as a result of your study, what would you tell him?
- [25:10]
C
I think the fundamental thing is looking at what the dynamics of an insurgent ecosystem really is. And I think there are two key features of that are the idea of groups coming together to coalesce and groups breaking apart to fragment. And so, you know, one of the real questions is, why can an insurgent ecosystem take on and, you know, largely compete against the strongest army the world has seen? And that's where you got to go step back and say, well, what are the group dynamics that are in play? And what you have is effectively many, many strategies. You have many, many groups with many, many strategies. And each one sort of has in some ways his own kind of genomic profile. And certain ones are more successful than other ones. But there's sort of the idea that there are many, many bets. And one of these bets becomes successful, more and more resources come into play, and more and more resources are kind of piled up on top of it. And this group becomes successful. And then what happens, of course, is it becomes successful to the point where it becomes a target. And the target comes from the US military side and it breaks it. But when it breaks, it doesn't break in half. It fractures and shatters. And then the pieces are then redistributed back through the system. And why that's so effective is that you've got a continually evolving system where learning is maintained within a group, and when that is fragmented, the learning and the innovation and the strategies are then redistributed back through the system. And so you get a continually evolving, fluid dynamical system that is very effective at taking on and defeating a more structured and stable opposition.
- [26:37]
B
Sean Gorley of the University of Miami. That's all we've got time for, but we do have a request. Next week, we'll be looking at the way maths is now taught in primary school and why most parents simply wouldn't recognise the way their children are taught compared with what they learned themselves. Now, we occasionally get emails from people who say they hated maths at school and couldn't understand a thing, but who nevertheless love More or less. But that means there may be people listening right now who don't know how to do long division or long multiplication, and more to the point, who never did. So, if you didn't understand a decimal place of what was being said to you in primary school maths, please get in touch. We'd like to see whether modern mathematical methods mean more to you. Our address, as always, more or less@BBC.co.uk or you can contact us via our website. Also the place to subscribe to our podcast, listen to our archive or read more. It's BBC.co.uk more or less. We'll be back next week, so please keep your questions and your comments coming. Until then, goodbye.
- [27:40]
A
More or Less was presented by Tim
- [27:42]
H
Harford, the FT's undercover economist. The producer was Richard Knight and the editor was Richard Varden. It was made in association with the Open University.