← More or Less
More or Less
Scottish referendum polls
00:28:12
Tim Harford talks to pollsters about how they are trying to gauge the political mood...
Loading summary
Transcript95 lines
- [00:00]
A
Thank you for downloading More or less from BBC Radio 4. Here's Tim Harford.
- [00:05]
B
Hello and welcome to More or Less, the programme which reveals the secrets behind the numbers in the news and in life. This week, the triumph of Countdown and the curse of Strictly Come Dancing. But first, we bring this program to you. Amid a flurry of Scottish referendum poll results, yes leads in Scott's poll shock, the Sunday Times headlined last weekend as it published an Internet opinion poll conducted by YouGov which showed the yes campaign was ahead for the first time by 51% to 49%. When the don't Knows were excluded, there was pandemonium in the media, a sharp reaction on financial markets and the leaders of the three big Westminster parties, each of which supports the continuation of the United Kingdom, cancelled Prime Minister's questions and dashed to Scotland in an attempt to shore up up the Better Together campaign.
- [01:01]
A
But then three days later.
- [01:03]
B
Hello, Ruth.
- [01:03]
A
Hi, Tim. A new poll was published showing the Better Together side was six points ahead, excluding the Don't Knows, which is a pretty solid margin, well above the reported margin of error.
- [01:14]
B
Two days passed and then low.
- [01:16]
A
Another YouGov Internet poll showing this time the no vote in the lead. And an ICM telephone poll for the Guardian also showing the Better Together campaign ahead, but only by two points, which is within the margin of error. And so the polling continues.
- [01:33]
B
So what's going on? What are the polls really showing us? How are they conducted and are they reliable enough that they should be causing these political and financial aftershocks?
- [01:43]
A
It's important to understand something about these polls, many of them, including, as you've mentioned there, Tim, The Sunday Times YouGov poll that caused all the fuss. Are Internet polls now, Usually when we
- [01:52]
B
think of opinion polling, we imagine phoning people up or knocking on their doors and trying to get a random sample of the population at large. But Internet polls work quite differently. I asked Anthony Wells of YouGov to explain.
- [02:07]
C
Traditional polling is the theory behind it is all about trying to get a random sample. So you go to the whole of the country in some way from telephones or addresses or whatever, and get a random group of people and they will, by laws of probability, reflect the national population. Internet panel obviously isn't like that. It's very, very different theory behind it. What you're doing is almost building a model of the British population so it's got the correct number of people of each gender, the correct spread of age, the correct spread of social class, the correct spread of political viewpoints who are willing to take part in surveys for a small payment. So if you've got 300,000 people and you know you're a bit short of little old ladies, then you'll spend your time running recruitment campaigns to try and get more little old ladies on your panel so you can build up representative samples more easily.
- [02:53]
B
Just to clarify, that More or Less does not endorse the use of the little old lady cliche, Anthony, but just sticking with it, hypothetically, the little old ladies who answer Internet polls, how do you know that they are representative of little old ladies as a whole?
- [03:08]
C
Well, in some ways they won't be. Little old ladies who are on the Internet are more likely to be affluent, for example, but those are things we can wait for and control for. What we can't control for is, for example, they're also likely to be more au fait and more comfortable with technology, and those things are just inescapable.
- [03:24]
A
In a nutshell, then, surveys of an Internet panel need quite a bit of tweaking to make them reliable.
- [03:29]
B
And that tweaking is based on historical experience. And because the referendum is pretty much unprecedented, we don't have much historical experience
- [03:38]
A
and the polls are so close. Anthony Wells says he can't be sure what the result will be, but he says he is certain that in the weeks leading up to last Sunday's YouGov poll, he saw a move towards yes.
- [03:51]
C
We're not confident that yes or ahead. It could be yes ahead, it could be no ahead. There could be neck and neck, but we are very confident there's been a real and sustained and large change. There are actually individual people in that sample who are now saying yes, when six months ago they were saying no, or they were saying don't know. Real individual people who have actually changed their mind compared to what they thought six months ago.
- [04:12]
B
Anthony Wells from YouGov, another of the pollsters is Martin Boone, the director of icm. He's not sure we have enough experience to be confident in the current opinion poll numbers.
- [04:24]
D
The easy part is making a poll sample, an online poll sample, look like it's a true representation of an eligible population, in this case, obviously, the Scottish electorate.
- [04:34]
B
So, what, you just adjust for age, you adjust for income, you adjust for things you can measure. But there might be other biases. Oh, yes. That you can't control for.
- [04:45]
D
Yeah. Well, can we control for them? The art of polling is all about seeking to control for other perceived biases. So ICM over the last 20 years has been, I guess, most famous for controlling for a theory which we've called the spiral of silence. So in the 1990s, it applied to the Conservative Party, where it was perceived to be unpopular, the nasty party. And our method sought to appease people in many ways who might have felt that they were embarrassed to admit that they were a Conservative voter. And we employed specific techniques. One of them was called waiting by recall of past voting. Another one adjusting our numbers by giving people back to parties that they previously voted for if they said they don't know or they would refuse to tell us on this occasion. So we controlled for that. Now, in Scotland right now, I believe that it could be true to a limited degree that there's a kind of patriotism spiral of silence going on, that perhaps some people think it's unpatriotic to say that they're a no voter, maybe they're inclined to silence.
- [05:48]
B
So just to be clear, you are currently adjusting for that?
- [05:51]
D
I'm not controlling for a specific anti. No spiral of silence, because this is an unprecedented referendum. If I could peg my data back to a previous Scottish independence referendum, life would be good, believe you me. But I can't, because there hasn't been one. And that's kind of one of the black holes which has really made the life of pollsters such as myself much more difficult on this occasion.
- [06:16]
B
And the polls have become part of the political argument. Both sides have pointed to the polls and used it to bolster their own case. Are you concerned that activists might actually get on the panels and try to skew the results the way that would suit their particular campaign?
- [06:37]
D
I think it's been a long held concern of mine that online panels could indeed be infiltrated by supporters of one political party, or in this case, yes or no supporters. Indeed. Dare I admit, I was talking to the leaders of both the yes and the no campaign last month, and I basically thought that and said to them that if they weren't trying to infiltrate online panels, then what on earth are they doing? But actually, having said that, I looked at ICM's own online panel yesterday to see how many new recruits we'd had in Scotland of late, believing that there would be quite a few of them. But in actual fact, we only had 240 new recruits on our panel in Scotland in total since January of this year. And that's not an unusual number. So we would have had equal numbers of people in other areas across England.
- [07:23]
B
And the panel is what, thousands? Can you say how many are on the panel?
- [07:26]
D
That's a very difficult question to answer. Polling companies who are doing online polling in advance of the referendum are probably dependent on a relatively small pot of Scots who are being interviewed by all the different polling companies at some point. And that is a big concern for me. This will be the first time when a big electoral event in the UK has been predominantly and almost exclusively poll using one methodology. So, yeah, we're dependent on a pot of people which is defined, but we don't know quite how big it is. But in my view, it won't be big enough. And in that lies a real danger for the accuracy of the polls in advance of this referendum.
- [08:04]
B
This time next week, if the result of the referendum is a big surprise, it's very different from the polls. Would you personally find that a shock?
- [08:14]
D
No, I think this referendum has the potential to be a polling Waterloo, the biggest since 1992, when the polls got it wrong. So I know I can't say I'd be surprised. I very much hope that weapon companies do the market research industry justice by getting it, at least in the right area. I think, and hope that the best that we can as an industry hope for is that we dodge a bullet. But it's entirely possible that the bullets do start spraying our way.
- [08:42]
B
ICM's Martin Boone, one of many people who will be nervously following the Scottish referendum on Thursday. Now, before we leave the topic of Scotland, we felt it was important to lend a hand to Nigel Farage, leader of the UK Independence Party. Mr. Farage may be in favour of independence for the UK, but he doesn't think independence from the UK is a good idea for Scotland. Loyal listener John Tees pointed out to us that Speaking on the US TV channel MSNBC last week, Mr. Farage was worried about an independent Scotland's future down the road. I see it as a country that would face very serious economic problems. The welfare cost of Scotland are huge. Well over 50% of the people in Scotland are living on benefits of some kind. Can it be that more than 50% of people in Scotland are living on Benefits? We called UKIP's press office and they told us they were having trouble finding out where Mr. Farage got that number from. But not to worry, because he is absolutely right. According to the Department for Work and Pensions, 60% of Scottish people receive some sort of support from the state. That sounds high, doesn't it? But we're not sure that number means what he thinks it means. The phrase living on benefits conjures up those cliches of workless households and welfare scroungers, but it actually includes pensions, child benefit and tax credits. For people in work, 60% of people in Scotland get one or more of these benefits. And the figure is precisely the same in England. It's slightly higher in Wales at 64% and in Northern Ireland at 66%. So if this situation is a problem, it's a problem for the UK as a whole. A more useful figure might be the total benefit spending per person, which in 2011-2012 was over £3200 per person. In Scotland, that is indeed higher than the UK as a whole. But the gap's small and it's shrinking about 2% in 2011-2012, down from 7% five years earlier. You're listening to More or Less. Our email address is more or lessbc.co.uk, the place to write to us. If, like John Tees, you have something you'd like to call to our attention. The Channel 4 programme Countdown was recently recognised by Guinness World Records as the longest running television game show. It has 32 years, 70 series and 6,000 episodes under its belt. What better time then to celebrate the greatest moment in the history of Countdown? Some might say the greatest moment in the history of TV broadcasting itself. Most people think of Countdown as a game of words and letters, but for us, it's all about the numbers. In the Countdown numbers round, contestants are given six numbers and challenged to add, subtract, multiply or divide them together to produce the target a random number above 100 and below 1000. Let's go back to 1997. James Martin and Gerald Foster were tussling in the heats and their numbers were 3, 6, 25, 50, 75 and 100. The target was 952. Gerald reckoned he could make 953, but James thought he could hit the target of 952, and went on to explain to maths whiz Carol Vorderman how to do it. It's easy. 100 plus 6 is 1006 times the 3 is 318. Then simply multiply 318 by 75.
- [12:34]
A
Multiply it by 75.
- [12:35]
E
Yes, please.
- [12:36]
A
Multiply this by 75.
- [12:38]
E
Yes, please.
- [12:39]
B
Which gives you. Oh, hang on. 23,850. Which, after subtracting 50, gives. Oh, here's Carol Vorderman doing the sums.
- [12:55]
A
23,000. Sorry, I've gone now. 23,800.
- [13:00]
B
And then just divide by 25. Simple.
- [13:03]
A
And divide this by. By 25. Do you know, I think you're right. That's incredible.
- [13:20]
B
Quite, quite brilliant. So, of course we had to call up James Martin, you have to multiply 318 by 75.
- [13:30]
E
That's right.
- [13:30]
B
This is the moment that reduces the great Carol Vorderman to asking for a calculator. You didn't know the answer to that?
- [13:36]
E
No, neither of us knew it. And of course, when the cameras are running, it's almost impossible to actually think through something like that. So, no, I didn't know what that big number was going to be. I'd only seen it just towards the end of the 30 seconds, so I wasn't 100% sure it was going to work at all.
- [13:48]
B
I think that is fascinating, though, because people will naturally assume that you've done all the sums, but actually you didn't need to solve the sum to know that the final answer would give you what you needed.
- [13:57]
E
Yes, indeed. Exactly.
- [13:59]
B
You might, at this stage be wondering how James could get the right answer if he didn't actually know what 318x75 was. He actually didn't need to. He knew that what he needed to do was multiply by three and subtract two. And his flash of inspiration was, well, he had 75, 25 and 50. 75 over 25 is three and 50 over 25 is your minus two. So multiply by 75, take away 50, divide by 25, that multiplied by three, take away two, and that's true, even though you don't know what the big numbers in the middle of that calculation actually are. Of course, this isn't the kind of maths most people can follow. It's certainly not the kind of maths most people can do in a television studio. But then, James is no ordinary person. Now, you are a statistician at the University of Oxford.
- [14:57]
E
That's right. Well, I work in the statistics department, really. I would call myself more a mathematician. I work in probability theory.
- [15:03]
B
But isn't that. Isn't that kind of cheating to go on Countdown and to sort of be a professional mathematician?
- [15:08]
E
I think so. But then you'll find you come up against lots of people who are international Scrabble players and the like, who have tremendous skills at the word. So you take your advantage where you can.
- [15:17]
B
I think now, as far as I know, you are the only man in history to reduce Carol Vorderman to this sort of giggling, hysterical wreck, asking someone to pass through a calculator. How did that feel at that moment?
- [15:33]
E
It was quite fun at the time. It's a very relaxed studio. At least it was when I was on. It was in the days of the great Richard Whiteley, and it was all very good fun and very relaxed, so I was enjoying myself a lot.
- [15:42]
B
James Martin of St Hugh's College, Oxford Countdown legend. You can probably find that clip on YouTube. I may tweet it later. You're listening to More or Less, and I'm Tim Harford and. And one of the stories that registered a giant blip on the more or less suspicious data radar a couple of weeks ago came from the BBC's own website an article with the headline food
- [16:07]
A
poverty experts issue Malnutrition Health warning. More people are suffering from malnutrition as a result of worsening food poverty.
- [16:17]
B
Experts have warned and went on to
- [16:20]
A
say that conditions like rickets were becoming more apparent because people couldn't afford quality food in their diet.
- [16:28]
B
So these are pretty serious claims directly linking food poverty to specific diseases such as rickets. Ben Carter's here. You've been taking a look at this. What have you found, Ben?
- [16:38]
F
Well, Tim, we can start by blowing up this claim that rickets is becoming more apparent. A study of Victorian diseases by the Health and Social Care Information Centre shows that the number of people being diagnosed in hospital with rickets has gone down by a third in the last couple of years.
- [16:54]
B
So why was it reported as having gone up?
- [16:56]
F
Well, John Middleton, vice President of the Faculty of Public Health, was behind the Ricketts quote. He told me that he was talking anecdotally and conceded that the data painted a different picture.
- [17:08]
B
Right. Well, the situation's different, though, when we count the number of people being admitted to hospital with malnutrition over the past year.
- [17:15]
F
Yes. The number of recorded hospital admissions with malnutrition as a primary or secondary diagnosis has gone up from just under 4,000 in 2009-10 to nearly 7,000 in the year ending April. This.
- [17:28]
B
Now only a few hundred of these are primary diagnoses where someone is admitted
- [17:32]
F
to hospital and malnutrition is the main reason for that.
- [17:36]
B
So most of them are secondary diagnoses and that's when someone's admitted to hospital with, say, a broken leg and whilst there, the doctor might make an assessment and say, you're also suffering from malnutrition.
- [17:47]
F
Which means that a lot depends on how alert doctors are to the possibility of malnutrition.
- [17:53]
B
I spoke to Professor Marinos Elias, he's chair of the Malnutrition Action Group of Bapen, which is the British association for Parenteral and Enteral Nutrition.
- [18:03]
G
I have plotted those figures on an extended period ranging down from 2000, and there's been a five fold increase in the numbers of cases coded for malnutrition. And that clearly can't be due to a five fold increase in malnutrition in our society. Although there may have been an increase in in segments of society, such as the older people that are getting more prevalent in our community. There are one or two explanations. I think it does have to do with awareness. There's certainly been campaigns from governmental and non governmental organizations to do with malnutrition and hydration. And there has been over the years, a result of all these campaigns and initiatives, an increase in the proportion of screening that has been taking place in hospitals, and that will increase awareness.
- [18:56]
B
So this increase may well be, at least in part because doctors and health professionals are looking more closely. People who are diagnosed with a disease are coded by the hospital. It helps the hospital and the health service as a whole keep track of how many people are suffering with a particular illness.
- [19:12]
F
But coding also serves another purpose. Hospitals work on payment by results. And if malnutrition was listed as a secondary diagnosis, then that could sometimes result result in a hospital receiving more money.
- [19:24]
B
Okay, so we know that the number of recorded cases of malnutrition is on the rise, and we've heard some potential reasons for that. Is food poverty having an impact on the rising numbers?
- [19:36]
F
Most people would assume that the more deprived you are, the more likely you are to suffer from malnutrition. And the data seem to back up that assumption. But while the most deprived people in society do suffer from malnutrition more than the least deprived, there isn't a great deal of difference. 40% of malnutrition cases occur in the three most deprived deciles, but 30% of cases occur in the four least deprived deciles.
- [20:01]
B
But Ben, we need to be careful here because the deciles actually represent areas of the country rather than people. So some of them might be poor people who are living in middle income areas.
- [20:12]
F
But when you look at other diseases in the study, things like TB and measles, there is a much more marked breakdown. Nearly 60% of cases in those diseases occur in the three most deprived deciles. Some poor people might live in more affected affluent areas, but the vast majority of inhabitants certainly aren't food poverty victims. So why are they suffering with malnutrition? Here's Professor Elia.
- [20:35]
G
Again, one factor that needs to be considered is that malnutrition increases with age, progressively increases with age. And it's not clear if the least deprived, perhaps more wealthy upper socioeconomic classes that tend to live longer than the lower socioeconomic classes develop malnutrition at a higher age. So if one was to adjust for age between the upper and the lower degrees of deprivation, perhaps a bigger gradient might become apparent.
- [21:10]
B
As Professor Elia points out, malnutrition is much more common in the elderly. More than half of the 612 primary diagnoses recorded last year were in people over the age of 60. But in the past year only 35 children or teenagers were given a primary diagnosis of malnutrition. This is predominantly an affliction of people who are already sick for other reasons, and particularly the elderly. Well, thank you very much Ben. You're listening to More or Less. Our website is BBC.co.uk more or less where among other things, you can check a previous item we did on the show about food poverty and food banks. Our email address is more or less bc.co.uk please drop us an email and tell us what's on your mind or set us some really hard sums Strictly's Back Some loyal listeners to More or Less may also be loyal viewers of Strictly Come Dancing, a long running television show in which a group of slightly famous people pair up with professional dancers, learn to dance and then compete in a no holds barred dance fight. There can be only one winner,
- [22:31]
G
but
- [22:31]
B
should this happy sound actually strike fear into the hearts of the husbands and wives of the show's contestants? Because reportedly there's a dark underbelly to the popular BBC one show the Celebrities
- [22:46]
A
Cursed by Strictly Come Dancing spotlights Strictly Come Dancing Romance Curse strikes for the 15th time. This time it's Rachel Riley.
- [22:56]
B
Curse of Strictly strikes again as Susanna Reid and partner separate to spell it out. The Curse of Strictly is the tendency of happily married celebrities to join the show, fall for their dance partners and then dump their husbands or wives. Who would have thought that fit? Attractive and wealthy people forced to share an extended and intimate physical challenge together with very few clothes on would find some sexual chemistry. It's extraordinary. So is there really a curse of Strictly? After all, relationships do sometimes break down, even celebrity relationships. Just because things happen to Strictly contestants doesn't mean that Strictly is to blame. And another thing, when you start to look at events that the newspapers use as examples of the Curse of Strictly, they're pretty disparate. They include two dancers falling for each other and later splitting up, contestants who split with their existing partners without beginning a relationship with anyone else, contestants who have affairs but not with anyone on the show itself, and so on. In one case, two people met on the set and dated for several years. That doesn't sound like much of a curse to me, but I'd like to be sure. So I commissioned John Moriarty, a mathematics lecturer at the University of Manchester, to take a look at the problem.
- [24:13]
H
We can look at some numbers to answer that question.
- [24:16]
B
That's what I like to hear. This is more or less excellent. Go for it.
- [24:19]
H
We could look at married people and we could look at unmarried people. Married people are easier to look at statistically because there's more data available from censuses. So among Strictly contestants, there have been, sadly, two married people who have separated in the same year as their appearance on Strictly.
- [24:38]
B
And just reaching back into my celebrity gossip mag, they were cricketer Mark Ramprakash and maths goddess Rachel Riley. So two marital splits among 150 celebrities. Let's do the maths, Jon.
- [24:49]
H
Of course, in Strictly, people are paired up, and so there have been 150 celebrities. So multiply that by two, because we're considering both the contestants and the dancers. So that gives you 300 person years. The UK national divorce rate in the last census was about 0.3% of the population per year.
- [25:08]
B
So 0.3% times 300 person years. Incidentally, that's also the length of the great Bruce Forsythe's showbiz career. That gives us an expected rate of just under one divorce among the Strictly contestants and dancers. But Strictly's actually seen two marriages end. Should we be surprised at that? Maybe not. John Moriarty has been back to the history books.
- [25:33]
H
If you look at the US Census in 1900, you can rank the professions there by the divorce rate, and top of the list is actors and showmen. If you look at the US Census in 2000, then top of the list of divorce rates is dancers and choreographers.
- [25:51]
B
Well, well, well. In light of this finding, John Moriarty redid his calculations.
- [25:57]
H
If we take into account the correction for the fact that essentially people in this context are dancers and we all know about dancers. Okay, then we'd apply a factor of 2.6, so taking us to an expectation of about 2.3 divorces in that time.
- [26:14]
B
So really, the number of celebrity divorces on Strictly Come Dancing, given the number of celebrities and dancers, is almost exactly what you would expect given what we know about divorce amongst the dancing population.
- [26:30]
H
May I say that's exactly right.
- [26:33]
B
So there's the mathematical proof. There is no cursive Strictly. Just a bunch of attractive, flirty dancers. I'm sure that will be of some comfort to the husbands and wives of this series contestants. But who else should be worried? Let's go back to that U.S. census with its list of divorce rates by
- [26:50]
H
professional to give just a few other examples. Near the top we have jobs like bartenders, telephone operators, telemarketers, waiters and waitresses, entertainers and performers, sports and related workers. At the bottom we have a whole panoply of engineering jobs. And it really is very striking. It seems to be near the bottom. It's really every other job that is some sort of science or engineering job.
- [27:13]
B
So if you want a stable marriage, marry a scientist or an engineer.
- [27:19]
H
On average, that would be a good choice.
- [27:21]
B
Excellent. Excellent. John Moriarty from Manchester University's maths department. And that's all we have time for this week. Please keep your comments coming in to more or lessbc.co.uk. you can download the programme for free at BBC.co.uk more or less listening to it on a continual loop until the same time next week. Until then, goodbye.
- [27:45]
A
More or Less was presented by Tim Harford of the Financial Times. The producer was Ruth Alexander. You can Download many more Radio 4 programmes at the Radio 4 website. BBC.co.uk radio4.