Undercover statistics

Sex and statistics may sound like unlikely bedfellows, but stats are one of the only ways we can actually get an idea of what's going on behind closed doors and under the covers, and we can peek at how human behaviour is changing over time. So what do the numbers say, and can we really trust them? Professor David Spieglehalter is a statistican at Cambridge University, and he explained some of the surprising things we can learn from crunching the numbers to Khalil Thirlaway...

David - Statistics are the way we can actually find out what really is going on or, at least, as best we can.  So we don't just have to listen to people's stories, their anecdotes, their boasts maybe, we can actually try to do the numbers.  Now sometimes we can work out the numbers from official statistics, really good statistics - you know, we know how many babies are born.  It used to be an interesting statistic to correlate birth records with marriage records and find out, for example, what proportion of women were pregnant when they got married - it was usually about 30%, so sometimes we can get some very good numbers like that. Other times we really just have to ask people about their behaviour, what they get up to, but then there are good and bad ways of doing that.

Khalil - How reliable are these various methods of asking people about their experiences?

David - Well, there are some terrible methods. If you just put up a survey, an online survey or a little insert in a magazine, as there used to be and asked people to fill this in and send it in if they felt like it, you're going to learn nothing.  Even if you get a big response rate - you know thousands - ten thousand people might respond, it still only says something about those ten thousand people, which is okay, but you don't really learn anything about what's going on in the general population.  If you really want to have an idea about what's going on in the whole country, you have to try to get a representative sample to respond.  You can do random sampling techniques, you know using standard survey methods, the problem then is getting people to actually answer the questions and getting them to give reliable answers.  And there's all sorts of methods that people have gone about to try to do that.  The big UK survey is known as Natsal - it costs about £7 million each time.  They do it every ten years and that relies on face to face interviewers going into people's houses, gaining their trust - they get about £30 a moderate fee for taking part - but they build a little relationship, not a very personal relationship but, by assuring the person that this information is valuable, it's good for planning health services, it's a contribution to society, they get around a 66% response rate, which doesn't sound great but it's much better than most surveys.  And the other crucial thing is that when the questions are asked now, the interviewer doesn't hear the responses - it's all done on a computer.  So the person being interviewed enters their answers to the questions in and then the computer's locked down and the interviewer never know what's actually said.  So when you're asking what could be slightly embarrassing  questions or difficult questions about age of first sex, how many partners you've had, have you had concurrent relationships, as it's known, when you've got two or more sexual partners going on at the same time - people will answer those sorts of questions.

Khalil - Surely the responses to these surveys do depend on honesty on the parts of the participants?

David - Yes, and there's various checks built in within the interview to check the person isn't contradicting themselves from previous answers.  There's also checks amongst the whole survey to correlate with previous surveys.  Interestingly, if you asked 30 year olds in the year 2000 when they first had sex and you ask a random group of 40 year olds in the year 2010 when they first had sex, you should get roughly the same answer.  But there's one check the data absolutely always fails.  It's a mathematical fact that if you've got an equal number of men and women of the same age distribution and you ask them how many sexual partners have you had in your lifetime, then the average, that's the mean average (the total number of partners divided by the number of people), should be the same for both men and women because each partnership requires a man and a woman.  We're talking about opposite sex partnerships here.  But it always in surveys men claim to have had a larger number of sexual partners on average than women do.  Now, as we said, this is a mathematical impossibility but the gap is lower than it used to be.  It always used to be that men said they had twice as many as women did which is so wonderfully impossible, and the gap's gone down, but it's still there for the whole lifetime.  And this is a bit of problem, of course, for sexual searches and a lot of effort has gone into trying to identify why this might be the case.  It could be just be because men boast and women are very reticent and there's a suggestion that men tend to round up and women tend to round down.

Now there is some evidence that psychologists have got of what's called "social desirability bias," just the fact that women might be slight less willing to acknowledge multiple sexual partners, and there's this wonderful experiment that some American psychologists did where the randomly allocated students to three different ways of responding to a sex survey.  And one of them was very confidential, one of them had the threat of exposure that they knew that one of their friends, one of their other students would pick up the form and take it away and, in the third arm, they were actually wired up to a lie detector actually it wasn't a lie detector, it was a fake lie detector - the little pen going wiggly, wiggly and all this stuff, but they thought they were.  And they showed that for women, not so much for men, that if the people wired up to the lie detector acknowledged having a greater number of sexual partners and actually then matching the number that the men had said.  So not a big effect, but enough to suggest that that is also part of the reason.

Khalil - What do these data tell us about how our sex lives have changed over the years?

David - Yes, I think this is one of the most interesting things.  Now, for example, the big British survey's been done three times, so 1990, roughly 2000, 2010 you can start looking at the trends and even if there's some uncertainty about the numbers the trends, you think, are fairly reliable.  And there are some, you know I suppose, slightly surprising trends for example, the frequency of sex has gone down over the last twenty years - that's the one people often pick on as being interesting.   And that's even among couples who are together, couples 16-44 year old, opposite sex couples, on average in 1990 had sex five times every four weeks, then it went down to four and now it's three and so all sorts of reasons.  Of course, I'm a statistician, it's not my job to say why this happens but the people, the actual researchers who did the survey have, under considerable media pressure, come up with their sort of ideas of the theory of why this might happen in that they think that quite a lot of the responsibility is just, people's lives have changed so much.  Our interconnectivity, our smartphones or our ipads or whatever are there all the time right through the evening - even into bed maybe and so, this constant connection with the outside world, you know, they think that might have reduced that time of just personal intimacy.

Khalil - Not now darling - I'm playing angry birds...

David - Or I'm watching the latest box set or something or other.