Undercover Economist

The problem with sexed-up statistics

‘Statistics tell us nothing until we understand what is being counted in the first place’

Men think about sex every seven seconds. Eighty-four per cent of women are emotionally unsatisfied with their relationships. Single people in the United States have more sex than married people do. Sixty-nine per cent of people over the age of 35 have had extramarital affairs. People have 40 per cent less sex now than they did 20 years ago. Truth, or myth?

A new book by statistician David Spiegelhalter, Sex by Numbers, runs a statistical comb through our collective sex lives. His book is largely designed to teach us about sexual behaviour — who is doing what with whom and how often — but along the way he manages to impart some important statistical lessons too.

Lesson one is that statistics tell us nothing until we understand what is being counted in the first place. To ask how old people are when they start having sex, or when they stop having sex, or how many sexual partners people typically have, we need a generally agreed definition of “having sex”.

We should not take for granted that we all mean the same thing when we talk about sex. Just ask Bill Clinton, who notoriously declared “I did not have sexual relations with that woman, Miss Lewinsky.” When it later became clear that he had received oral sex from her, he apologised for giving a misleading impression but maintained that “my answers were legally accurate.”

Clinton’s carefully chosen words were in tune with the way most people used language. A survey of US college students conducted in 1991 found that only 40 per cent of them reckoned that oral sex counted as “sex”. (The US Senate implicitly reached a similar conclusion in clearing President Clinton of perjury.)

While Clinton exploited ambiguity, modern scientific surveys of sexual behaviour try to eliminate it. According to definitions used by the UK’s well-regarded National Survey of Sexual Attitudes and Lifestyles, Natsal, Clinton did have sexual relations with Miss Lewinsky.

A second lesson is that we should pay attention to whether statistical work has been done carefully or casually. Consider Time Out magazine’s finding that people have sex 10 times a month if they are in a relationship, though only five times a month if they are married. This is twice as much as more credible surveys have found. As Spiegelhalter observes, Time Out’s method can only tell us about the sexual claims of people who go out of their way to fill in sex surveys. Spiegelhalter is similarly dismissive of the “Trojan US Sex Census”, which announced that Los Angeles was the most sexually active city in America with 135 sex acts per person per year. While a great source of publicity for a manufacturer of condoms, the people who fill in Trojan’s survey don’t represent the rest of us.

Some of the most famous sex researchers are also limited by a lack of representative sampling. Alfred Kinsey found that 37 per cent of men had a homosexual experience resulting in orgasm; Shere Hite reported that 95 per cent of women experienced “emotional and psychological harassment” from their male partners. The underlying research here was politically groundbreaking but we cannot have too much confidence that these numbers are correct.

Hite, for example, distributed questionnaires through university women’s centres, abortion rights groups and other women’s groups; the response rates were less than 5 per cent, making it unclear whether respondents were typical of women as a whole. Kinsey was on the lookout for interesting sexual case histories and so sent his researchers to prisons and to bars famous for being gay meeting places. He may well have captured a broader range of sexual behaviour as a result but at the cost of a representative sample. As the great mathematician John Tukey once told Kinsey, “I would trade all your 18,000 case histories for 400 in a probability sample.” If the aim is to judge what is going on in the population as a whole, Tukey was right.

To revisit the factoids in the first paragraph: most are unproven, the results of unrepresentative surveys. The “seven seconds” claim is an urban myth and provably nonsense. But the final discovery — that we are having 40 per cent less sex — is true. According to the rigorously collected Natsal survey, heterosexually active people aged 16-44 typically had sex five times in the past month back in 1990. By 2010, the number had fallen steadily to three times. Perhaps the next Natsal survey will be able to figure out why.

Written for and first published at ft.com.