← More or Less
More or Less
Romanian Crime
00:28:21
Are the statistics put forward by UKIP accurate, and are Romanians responsible for on...
Loading summary
Transcript98 lines
- [00:00]
A
Thank you for downloading More or less from the BBC. This is the version first broadcast on Radio 4. Here's Tim Harford.
- [00:08]
B
Hello and welcome to More Or Less, your weekly look at numbers in the news and in life. This week, we look for the most spurious statistical correlations we can find. Psychologist Gerd Gerenze tells us how to be savvy about risk. And speaking of risks, is lightning really a mass killer? Something like 24,000 people a year worldwide get killed by lightning. So it's 24,000 a year. It's a major hazard, yes. But first, are Romanians responsible for a crime wave in the run up to the European elections? Nigel Farage, leader of the UK Independence Party, Ukip, got in some bother over comments he made, repeated here during an interview on LBC Radio.
- [00:52]
A
I was asked a question.
- [00:53]
B
If a group of Romanian men moved in next to you, would you be concerned? And if you lived in London, I think you would be. Mr. Farage has backpedaled somewhat from that. But he and other senior Ukip politicians have tried to make the case that there is a problem with Romanian crime, and they've used statistics to do that. The question is, has Ukip got the facts right? Ukip put out a full page advertisement in the Daily Telegraph and that advert makes three statistical claims. Now, one of them's nearly true, one's a bit weird, and the third is just wrong.
- [01:24]
C
28,000 Romanians were arrested in the last five years in the Metropolitan Police area alone.
- [01:33]
B
This statistic is nearly true. It's a couple of years out of date. And when more or less looked at it last year, we realised that this was referring to 28,000 arrests, not to 28,000 people being arrested. And the difference is, we could be talking about a much smaller number of people getting in trouble with the police again and again. But basically the claim is correct. Although it's slightly context free, the advertisement doesn't give us a sense of whether 28,000 arrests is an unusually large number. But we'll come back to that. Now for the second claim.
- [02:05]
C
92% of all ATM crime in London is committed by Romanians.
- [02:12]
B
Right. Now, this is the weird one and Charlotte MacDonald is here to help me puzzle it out. Hello, Charlotte.
- [02:18]
C
Hi, Tim Sharla.
- [02:19]
B
What is an ATM crime? Is it stealing cash machines with a bulldozer? Or is it illegal begging beside an atm? Or is it mugging people at a cash machine? Or is it stealing pins? What is it?
- [02:31]
C
I think you're right to ask, because it's easy to Misunderstand statistics when you make assumptions about what exactly they mean.
- [02:37]
B
And what do they mean?
- [02:38]
C
Well, not much really. There's no criminal offence of ATM crime and so there are no official statistics on the subject.
- [02:44]
B
So where did the number come from?
- [02:45]
C
UKIP were quite transparent about that. It's something a police officer said a couple of years ago in an ITV documentary. Here's a transcript.
- [02:54]
B
The fact is, 92% of all ATM fraud we see in this country is committed by Romanian nationals. Very, very tight communities, very tight gangs.
- [03:05]
C
We went to the police officer's team, the dedicated Czech and Plastic Crime Unit in London, who told us the figure was based on police intelligence at that time.
- [03:15]
B
So what does that mean?
- [03:16]
C
We asked and were told they couldn't explain how they worked it out. The information was too sensitive to release.
- [03:22]
B
But presumably they do have something in mind when they say ATM crime.
- [03:27]
C
They say it could mean any criminal activity that takes place at a cash machine, including stealing car details or even distraction theft. This is not just in London, by the way.
- [03:37]
B
So what we know with certainty is there's a specialist police unit and they are very worried about Romanians monopolizing some kind of cash machine related crime. But they won't tell us more than that. Right, well, actually, that is weird.
- [03:50]
C
On to the final statistic. 7% of all crime across the 28 EU member states was caused by 240 Romanian gangs.
- [04:02]
B
Wow. True.
- [04:04]
C
No, not true at all.
- [04:06]
B
So where was this one plucked from?
- [04:07]
C
Okay, UKIP cites Europol, the EU's law enforcement agency. But UKIP mangled the statistic. It's not that Romanian gangs commit 7% of all crime, it's that Europol counted up all the criminal networks they could find and they said that just under 7% of them were Romanian or mostly Romanian. The Times newspaper, in contrast, got it right. Here's what they said.
- [04:32]
B
About 240 organised crime groups from Romania have been identified by Europol, accounting for 6.7% of the total number of criminal networks active in Europe.
- [04:45]
C
And remember again, this is just a head count. We don't actually know whether the Romanian networks are big and influential or just quite numerous.
- [04:52]
B
Ok, so we've examined the UKIP claims. One of them is roughly true, one of them is totally wrong, and one of them, the one about ATM crime, points to a real issue, but it's oddly narrow, it's hard to define, and it's based on numbers the police won't share with us. And I suppose the bigger question is taken as a whole rather than zooming in on whatever poorly defined crime we choose. Is there a problem with Romanian citizens and criminality?
- [05:17]
C
I submitted a Freedom of Information request to the association of Chief Police Officers, which hold national police records. But this won't surprise you. They couldn't give me the figures that I wanted.
- [05:28]
B
Why not?
- [05:28]
C
The data on European foreign nationals is pretty patchy. But there are two things I can tell you. First of all, out of EU foreign nationals, Romanians have been convicted of more offences than any other nationality except Poles. And there are seven times as many Poles in the UK than Romanians. So it's not unreasonable to say there's an issue here. I've also been able to work out the types of crimes for which Romanians are convicted.
- [05:52]
B
And what are they?
- [05:53]
C
Theft. And by a large margin, about three quarters of the offences for which Romanians were convicted related to theft and a smaller proportion for burglary. And that's in the most recent year of data that I have available.
- [06:06]
B
So what are the trends? Is the situation getting worse or better?
- [06:09]
C
Well, Romania's Ambassador to the UK, Dr. John Ginga, says crime committed by Romanians is going down. He says policing statistics show that in the first three months of 2014, the number of Romanians convicted in the UK was about 1,500, compared to about 1,800 in the same period in 2013, which is a reduction of about 15%.
- [06:30]
B
And are the crimes in question serious?
- [06:33]
C
Well, you can get an idea of that by looking at the numbers of Romanians in prison. We know that in London alone, Romanians were charged with an offence around 10,000 times over five years. But there were only five hundred and eighty eight Romanians in prison in the entire UK on the 31st of March this year. That suggests not many people get convicted of serious crimes which carry a long prison sentence.
- [06:57]
B
Thank you very much, Charlotte. I hope you're sitting comfortably and paying close attention. We are about to describe a classic brain teaser that never fails to stir controversy. Listen carefully. Hello and welcome to let's Make A Deal, a game show that aired on NBC in the US in the 1960s. The host, Monty hall. Yes, that's his name, is showing the contestant three doors. Behind one of the doors is the top prize, a brand new Cadillac. Behind the other two doors are goats. Not what you want. The contestant has to choose between door one, door two or door three. She thinks for a while and then chooses door one. Now, to make it even more tantalizing, Monty Hall. The host walks up to door three and throws it open. It's a goat which means the Cadillac must be behind door one or door two. Monty now turns to the contestant and offers her the chance to switch doors. But what would you do? Remember, door one and door two are still unopened. Door three has a goat behind it. The contestant chose door one to start with, but now the host, Monty hall, is offering her the chance of swapping, should she do it. This puzzle features in a new book called Risk Savvy by Gerd Gigerenza, director of the Max Planck Institute for Human Development in Berlin. And I caught up with him to find out whether the contestant should swap or stick with her choice of door one and what that choice tells us about risk and uncertainty. Now this is a fantastic and well known among probability nerds brain teaser. So give us the standard answer, which I think people will be surprised by.
- [09:01]
D
Most people intuitively say no, I stick with my door partially because they think it's a 5050 chance.
- [09:08]
B
There are two doors, one or two,
- [09:10]
D
and if I switch, I regret it, so I rather do nothing, stick with it. The surprising answer is that in this textbook problem, the best choice is to switch.
- [09:21]
B
So why would that be? I know it's counterintuitive, but the chance that the Cadillac is behind door one was originally one in three. And when Monty hall opens door three and shows the Cadillac isn't behind door three, that chance doesn't change. There's still a one in three chance that the Cadillac is behind door one. And so the contestant can give herself a two in three chance to win by switching to door two. Or let me show the logic in another way. Imagine there were a hundred doors and still only one Cadillac. The contestant picks a door. Monty hall opens 98 other doors and none of them have the Cadillac behind. And so the contestant can stick with her original door. Very unlikely to be right. Or she can switch to the only other door that's still closed. And if the original door didn't have the Cadillac behind it, then the other door definitely will. Well, so much for the most controversial and counterintuitive probability puzzle around. But Gerd Geigerenza, not so fast. Are we sure it really is a probability puzzle?
- [10:29]
D
That's the typical story. But now the real twist comes. The Monty hall problem is about risk. Everything is known and Monty has to follow rules.
- [10:42]
B
For example, the rule that he always has to offer the contestant the chance to swap. And maybe in real life he doesn't. Maybe he's actively trying to fool the contestant. If the rules are known, then we can calculate probabilities. But if they aren't known. We can't simply solve the problem using probability theory. And it turns out that in the real Game show, Monty hall didn't always offer people the chance to swap. Sometimes he played tricks, offered them money to swap. Mixed things up in all kinds of ways. This changes let's make a deal from a world of risk in which probabilities can be calculated, to a world of uncertainty in which they can't in an uncertain world.
- [11:24]
D
It's an illusion to think that probability theory will give you the optimal answer. This illusion is also called the turkey illusion. So the story is assume you are a turkey. It's the first day of your life, A man is coming, and you fear he will kill me, but he feeds you. Next day you fear again he will kill me, but he feeds you. Third day, the same thing. By all standard mathematical models, Bayesian and otherwise, the probability that he will feed you increases every day and a day hundred. It's higher than any time before. But it's the day before Thanksgiving and you're dead meat. Remember the last financial crisis? Before the crisis, the certainty that it will go on like this, suggested by the rating agencies and by the bank's calculations, went up and up and up until everything went down. So what we need is both probability theory for situations where we can calculate the risk, but also smart heuristics or rules of thumb, and good intuition in a world where we cannot calculate that.
- [12:40]
B
One of the stories in Gerd's book is about Harry Markovitz, who won a Nobel Memorial Prize in economics for working out a complicated theory for how to invest different amounts of money across a diverse portfolio by weighing up the different risks involved for each option. Many banks rely on this method, but what's interesting is that when it came to investing his own money, Markovitz ignored his complicated theory and instead used a very basic rule of thumb called 1N. So, for example, dividing your money 5050 between stocks and bonds, or dividing your money 1/3 each way between property stocks and bonds. Researchers have examined which works better using Markovitz's Nobel Prize winning method, or the simple rule of thumb that Markovitz himself used to 1N.
- [13:29]
D
In many of these experiments, 1N made more money according to common indicators than Markowitz optimization.
- [13:37]
B
So the simple rule of thumb is better than the optimal rule.
- [13:41]
D
If a problem is highly unpredictable, as in the stock market, if you have many options, so many parameters to estimate and little data, then make it simple.
- [13:57]
B
That was Gerd Gigerenza, author of Risk Savvy. You're listening to more or less. Our email address is more or lessbc.co.uk and I'm Tim Harford. The music of Beethoven there evoking thunder and lightning in his 6th Symphony. If you were listening to the Today programme last week, you might have heard an item about how solar activity might be triggering lightning strikes on Earth. Anything that can help us predict the severity of lightning has got to be useful. It's something like 24,000 people a year worldwide get killed by lightning. So it's 24,000 a year. It's a major hazard.
- [14:42]
E
Yes.
- [14:43]
B
But John Humphries wasn't the only one surprised by that statistic. Our inbox lit up with questions from listeners, essentially all asking questions, can this be true? 65 people a day killed by lightning. Well, James Fletcher's here. Hello, James.
- [14:57]
A
Hello, Tim and James.
- [14:59]
B
It does sound implausible, given that being struck by lightning is the standard metaphor for something that's nasty, but also extremely unlikely.
- [15:06]
A
It does. But if you're in doubt, you can do a quick sanity check by searching an online news aggregator for lightning deaths. And if you do, you'll see that people are indeed being killed all the time around the world. But getting a more accurate number is not so easy. There are only about 20 countries who keep an accurate track of lightning fatalities. In the UK, for instance, we know that up to 60 people are hit by lightning each year and on average around three are killed. So we have those statistics. But lightning is a lot more prevalent in Asia, South America and Africa, where billions of people live. And we have very sketchy data on lightning deaths in those places.
- [15:44]
B
So where does this figure of 24,000 come from? Because it does get quite widely cited.
- [15:49]
A
It is a figure you hear a bit. And the number comes from a conference paper published in 2003. The authors are two lightning experts who tried to come up with a reasonable way to estimate global deaths. The challenge is that while the good data comes from industrialised countries, lightning's more dangerous in less developed places because cars
- [16:08]
B
or houses with metal plumbing provide some protection against lightning. And what's really dangerous is working all day out in a field somewhere. So we'd expect rural or agricultural poorer societies to lose more people to lightning.
- [16:22]
A
And you can add to that the fact that the standard of medical care won't be as good.
- [16:26]
B
So what we're looking for to get our statistics straight is a place with rich country statistics but poor country lifestyles.
- [16:33]
A
And these experts found a very good candidate.
- [16:35]
B
Where's that?
- [16:36]
A
The United States of America a hundred years ago. It was a rural society with very few cars and naturally no access to modern health care, but which nevertheless kept statistics on lightning strikes. Now, the death rate from lightning back then Then was about 6 deaths per million people every year. The authors then applied that to the 4 billion people that they estimated were living in areas of the world that are susceptible to lightning strikes. Lightning's a very tropical thing, after all, and that's how they got to 24,000 deaths. But the authors are very open about the fact that that's a rough calculation.
- [17:11]
B
So are there any more recent or better estimates?
- [17:14]
A
Well, one of the authors of the earlier estimate, Ron Holler, has been involved with a more recent attempt to come up with a better number. In extrapolating from some of the new improved statistics from certain developing countries most affected by lightning, this new method produces a new estimate. Global lightning fatalities are on this method, around 6,000.
- [17:33]
B
Well, that's a quarter of the size. So which number's better?
- [17:36]
A
Well, they're both educated guesses and I asked Ron Holler which he'd go for.
- [17:40]
E
I don't really know whether it's closer to one or the other. Every day I look at reports of lightning fatalities and injuries around the world and some days large numbers of cases come in and I said, well, maybe 24,000 is a reasonable number and other days or weeks it's a little bit quieter, and said, well, maybe it's really down around 6,000, I just don't know. I would probably go for the higher number.
- [18:03]
A
Other people I contacted favoured the lower 6,000 figure. So I guess that's just a sign of how little we really know about this.
- [18:10]
B
Well, thank you, James. You're listening to More or Less.
- [18:16]
A
They say it's your birthday.
- [18:22]
B
Last week, you may recall that our loyal listener Sheila was recovering from celebrating her 60th birthday on a weekend she'd had to wait a long time. Her 20th, 30th, 40th and 50th birthdays were all becalmed on weekdays and she wanted to know whether other people might have to wait even longer to celebrate a nice big round number birthday as an adult on a weekend. The back of my trusty BBC envelope said, no, most people will have a round number birthday weekend before they turn 60, and nobody has to wait longer than that. But we did invite further comments from listeners and I'm delighted to say that we got them.
- [19:04]
F
Hi, I'm Ollie Lee and I'm a recent economics graduate.
- [19:07]
B
Ollie sent a spreadsheet and this elaboration.
- [19:10]
F
I used my computer to make A list of every date of birth between 1 January 1900 and 16 May 2014, when the last episode of More or Less was broadcast. That's 41,775 days, including leap days.
- [19:24]
B
I suggested that Sheila was unlucky in having had to wait until she was 60, and Oli agrees. His trusty spreadsheet says that 93% of people born since 1 January 1900 will have a weekend round number birthday as an adult at or before the age of 50. The 7% who have to wait until they're 60 were all of them, like Sheila, born on Mondays. But what about people born on leap days on February 29th if we let them celebrate on March 1st? Well, they're much like anyone else.
- [19:58]
F
Now, we know that leapers do age by one year every 12 months, but let's say that they can only celebrate if the year has a 29th of February. And since 1900 there have been 29/29 of February. 21 of them will have a round number birthday by 67 of them have to wait a bit longer until their 70th, 80th or 90th birthdays. But anyone born on 29 February 2004 has to wait until their 140th birthday before they can celebrate on a weekend. And by that point they'll have probably retired and so won't need to worry about being at work on their birthday.
- [20:30]
B
There you go. Save the date. Saturday 29 February 2144. Ollie's analysed every birthday from the beginning of the 1900s, so clear out the inbox. There's no topping that. Or is there?
- [20:45]
G
Hi, I'm Adam Jacobs. I'm a medical statistician and run a stats consultancy company for my day job. Centennial years are not normally leap years, but every 400 years they are. So 2000 was a leap year.
- [20:58]
B
Aha. I see. Now, Ollie was our definitive source for 20th century birthdays. But people born in the 19th century had to cope with the fact that there was no le year in 1900. We skipped straight from 1896 to 1904. Well, does that change things? Adam?
- [21:15]
G
The irregular pattern of leap years over most centennial years can mess up the otherwise regular progression of days of the week. So it can take longer to reach a significant birthday. The longest it's possible to have to wait for a significant birthday at a weekend is 100 years. If you were born on a Monday between March 1842 and February 1848, then you had to wait 100 years. The pattern repeats three centuries out of four. So those who will be born in the 2000-40s may also have to wait until they're 100 before celebrating a significant birthday at a weekend.
- [21:50]
B
Adam Jacobs and Ollie Lee, Uber listeners, many thanks to you both. And if you have a good question for us, or indeed a good answer, then please email more or lessbbc.co.uk. Here at More or less, we're starting a campaign to ban margarine. That's partly because butter is delicious, but it's mostly because of the side effects of margarine. We've seen convincing evidence that margarine consumption causes divorce. It's all in a cool chart on a website. Let me just bring it out. Yep, totally, totally convincing. All I need is some way to convey to you just how convincing it really is. Aha. Ben, could you come in for a minute?
- [22:42]
E
Ok.
- [22:43]
F
Right.
- [22:43]
B
Ben Crichton is more or less his violist in residence, and Ben is going to play some swooping, elegant glissando notes. You can do that, right, Ben?
- [22:53]
G
Yes.
- [22:54]
B
Hey, Ben, what's the difference between a viola and a chainsaw? I don't know, Tim. You tell me. If you have to, you can get a tune out of a chainsaw. Ok, let's just get to the glissando, shall we? They'll mimic what these graphs look like. For example, if it was a graph of something increasing steadily over time, a graph that starts low on the left and ends high on the right. It's going to sound like this. Ok, so here's what the graph for per capita consumption of Margarine in the US from 2000 to 2009 sounds like. Falling over time. And here's what the graph for the divorce rate in the US state of Maine over the same period sounds like. And put them together and you'll be able to hear. They correlate perfectly. Thank you, Ben. Now, you've probably seen graphs like this too, in your favourite newspaper or website, and you'll agree that when two things line up perfectly like this, it's time to take action. Ban margarine to save marriages. Who's with me?
- [24:11]
E
My name's Tyler Vegan. I'm a student here at Harvard Law
- [24:14]
B
School, and Tyler runs the website which published this graph. So I'll let him explain its importance.
- [24:20]
E
What's kind of fun about it is it allows people to be their own scientists for a few minutes because they get to come up with a hypothesis, like maybe there is something. For example, the margarine and the divorce rate statistic. Maybe when there's more margarine in the house, it's more likely to cause divorce, or there's a link between some of the molecules and margarine or something. And we get to think about that for a second and then realize that we have no basis for that in reality, that there's just nothing that can confirm that for us. And we kind of can reject our own hypothesis outright. And it's fun to be able to do that.
- [24:50]
B
For a moment, it all seemed too good to be true, didn't it? Tyler's site is called Spurious Correlations, and he set up a computer program to mine datasets and find things which seem to be correlated. He graphs them and he publishes the graphs on his website. So you can see that the age of the winners of the Miss America contest. Correlates with murders by steam, hot vapors and hot objects. You can also see that worldwide non commercial space launches correlate with, of course, sociology doctorates awarded in the United States. As Tyler says, if you think about these for a minute, or even if you think about them for a second, the idea that there's any causal relationship seems ridiculous.
- [25:51]
E
I've seen a lot of headlines, especially sensationalist ones, where it might say scientists find connection between X and Y. And in a lot of those situations there might be a correlation or even a mathematically insignificant correlation between two things. But it's really important for us to be critical of whether that correlation exists, whether we can show that there's actually statistical significance, and whether there's a causal mechanism. A lot of my charts illustrate situations where there isn't a statistically significant correlation, but it looks like there is because of how I plotted them on a graph. And when you only have maybe 10 points or 10 variable sets or numbers to go by, it's not that hard to find overlapping lines that kind of curve or vary together. And just because two things vary together on a linear scale doesn't mean that they're actually correlated or that they're actually causing each other.
- [26:40]
B
Tyler has given us a humorous way to remind us of the golden rule of statistical thinking. This is certainly something we take very seriously here at the BBC. In fact, if I just open the studio door, you might be able to hear.
- [26:54]
C
Staff are reminded that correlation does not equal causation and that sometimes correlation isn't even really correlation. And remember, Alan Yentaub's homemade scones are on special in the canteen today.
- [27:08]
B
I'm sure as you're listening, your mind is already churning over various fantastical explanations as to why margarine consumption causes divorce, or for that matter, why divorce causes margarine consumption. So please send your theories to more or lessbc.co.uk. we'll read the best ones out next week. And for lots of other spurious correlations, check out Tyler's site@tylervegan.com or just search for spurious correlations. That's all we have time for this week and as always, Our address is BBC.co.uk more or less. Our email address is more or lessbc.co.uk. we'll be back next week. Until then, goodbye.
- [27:55]
A
More or Less was presented by Tim Harford, the Financial Times undercover economist, and the programs made in association with the Open University. The producer was me, James Fletcher with Charlotte Macdonald and Laura Gray. And special thanks to Christian Halstead, who stepped in as a viola stunt double at the last minute. You can download many more programs for free from Radio 4. You can find them at BBC.co.ukradio4.com.