lederhosen (![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
lederhosen) wrote2007-08-16 12:34 pm
lederhosen) wrote2007-08-16 12:34 pm
Does not compute
Unless you've been living under a rock, you've probably encountered the "men have more sexual partners than women" paradox at some stage. If you haven't, the gist of it is this: surveys that ask men and women about how many opposite-sex partners they've had in their lifetime generally find that the typical man has almost twice as many as the typical woman. (See this NY Times article for some examples.)
What is going on here? If our population is roughly 50/50 male/female, a naive look at the situation tells us that the numbers should be equal; any time a new couple forms up, both those figures should go up by the same amount, right?
There are several factors that contribute here. Some of them are well-known and get trotted out every time this sort of thing comes up, but there's a sneaky one that hardly ever gets a mention and really should.
The standard ones:
Sampling error - are you getting a representative cross-section of the population? If you base your conclusions on Warren Beatty and a nunnery, you're not getting a good picture of the overall male and female populations.
Reporting error - people misremember stuff or actively lie, and if the survey's not carefully worded they might have different ideas about what counts. Given different attitudes to male and female sexuality, it wouldn't be surprising if men had a higher reporting rate than women.
Mean vs median - a lot of these statistics are based on the median, which isn't the average; if you have three women reporting 0, 0, and 3 partners, and three men reporting 1 partner each, the female median is 0 but the male median is 1. (More discussion of this here, although it makes an error that I'll get to in a moment). However, even stats based on the mean do show a disparity.
The sneaky one that doesn't get mentioned much:
People are born, age, and die.
Let's imagine a village with a population of 100 men and 100 women, aged 0 to 99. Each year on January 1, one male baby and one female baby are born, and the two oldest inhabitants drop dead on their 100th birthday. To keep things even simpler, let's assume they're all monogamous for life - each person has one and only one mate, and always from within the village. That's pretty darn symmetrical - far more so than the real world - but I'm going to introduce one little point of asymmetry.
Suppose that when they pair up, the average man is 20 and the average woman is 25. In fact, to keep things simple, let's suppose that those are the exact ages that apply for everybody in the village, and as far back as our records go. What does this mean to our sociologist?
If you poll everybody in the village, you'll find that 20 of the males (those aged 0-19) have had no sexual partners, and the other 80 have each had one, giving an average partner count of 0.8. If you do the same with women, you'll find that 25 of them haven't yet found a partner, so their average will only be 0.75.
That's not much of a difference, but we can make it bigger. Most of these surveys don't cover all ages; this CDC study, for instance, surveyed people between 20 and 59. What happens if our sociologist follows suit?
All the men in the 20-59 bracket have exactly one partner (average 1.0), but only 35 of 40 women in that group have (average 0.88). If you narrow the survey to cover only people between 20 and 39, which is quite believable for work of this sort, the gap gets even bigger: 1.0 vs 0.75, making men look 33% more 'promiscuous' than women.
In real life I don't think the average disparity in ages is quite as large as five years, so the size of this effect would be a little smaller - but it's certainly enough to be important.
(If you're looking at how many partners people have in their lifetimes, this objection doesn't apply, but a lot of the reported data doesn't work on that basis.)
What is going on here? If our population is roughly 50/50 male/female, a naive look at the situation tells us that the numbers should be equal; any time a new couple forms up, both those figures should go up by the same amount, right?
There are several factors that contribute here. Some of them are well-known and get trotted out every time this sort of thing comes up, but there's a sneaky one that hardly ever gets a mention and really should.
The standard ones:
Sampling error - are you getting a representative cross-section of the population? If you base your conclusions on Warren Beatty and a nunnery, you're not getting a good picture of the overall male and female populations.
Reporting error - people misremember stuff or actively lie, and if the survey's not carefully worded they might have different ideas about what counts. Given different attitudes to male and female sexuality, it wouldn't be surprising if men had a higher reporting rate than women.
Mean vs median - a lot of these statistics are based on the median, which isn't the average; if you have three women reporting 0, 0, and 3 partners, and three men reporting 1 partner each, the female median is 0 but the male median is 1. (More discussion of this here, although it makes an error that I'll get to in a moment). However, even stats based on the mean do show a disparity.
The sneaky one that doesn't get mentioned much:
People are born, age, and die.
Let's imagine a village with a population of 100 men and 100 women, aged 0 to 99. Each year on January 1, one male baby and one female baby are born, and the two oldest inhabitants drop dead on their 100th birthday. To keep things even simpler, let's assume they're all monogamous for life - each person has one and only one mate, and always from within the village. That's pretty darn symmetrical - far more so than the real world - but I'm going to introduce one little point of asymmetry.
Suppose that when they pair up, the average man is 20 and the average woman is 25. In fact, to keep things simple, let's suppose that those are the exact ages that apply for everybody in the village, and as far back as our records go. What does this mean to our sociologist?
If you poll everybody in the village, you'll find that 20 of the males (those aged 0-19) have had no sexual partners, and the other 80 have each had one, giving an average partner count of 0.8. If you do the same with women, you'll find that 25 of them haven't yet found a partner, so their average will only be 0.75.
That's not much of a difference, but we can make it bigger. Most of these surveys don't cover all ages; this CDC study, for instance, surveyed people between 20 and 59. What happens if our sociologist follows suit?
All the men in the 20-59 bracket have exactly one partner (average 1.0), but only 35 of 40 women in that group have (average 0.88). If you narrow the survey to cover only people between 20 and 39, which is quite believable for work of this sort, the gap gets even bigger: 1.0 vs 0.75, making men look 33% more 'promiscuous' than women.
In real life I don't think the average disparity in ages is quite as large as five years, so the size of this effect would be a little smaller - but it's certainly enough to be important.
(If you're looking at how many partners people have in their lifetimes, this objection doesn't apply, but a lot of the reported data doesn't work on that basis.)