Fun with statistics
In the course of this article describing a research study on marriage, fidelity, wages, and divorce, Tracy Corrigan makes the following observation:
Munsch found that during a six-year period, an average of 3.8 per cent of male partners and 1.4 per cent of female partners admitted to cheating in any given year. So either men cheat mainly with single women, or women are economical with the truth.
The reality, given the level of divorce rates, is probably that both are prone to fibbing: in fact, I can’t help feeling that it goes with the territory. In every survey, in every country, men claim to have more ”“ often many more ”“ sexual encounters than women.
In Britain, that’s an average of 12.7 heterosexual partners over a lifetime, compared with just 6.5 for women. Except that it is logistically impossible for the average man to have more partners than the average woman.
Perhaps Corrigan is correct and people are lying. But it is hardly logistically impossible for the average man to have more partners than the average woman, with just a few people doing the lying.
I’m not even sure it’s improbable, although it might be. Corrigan is assuming a one-on-one encounter pattern. But it is certainly possible—and perhaps even likely—that a lot of the unfaithful men, or sexually active men in general, are having sex with the same women, who are each having sex with a lot of men.
In other words, there might be a pool of what we used to call loose women (not to mention the professionals known as prostitutes), who are each sleeping around with a lot of men. This may be logistically difficult, but hardly impossible. In fact, it’s the sort of thing that used to be standard in my high school long ago, and perhaps in yours—many of the sexually active boys were actually sexually active with a fairly small number of girls.
Is it really so hard to imagine that something somewhat similar might still be going on re infidelity, and could account for a certain amount of the general disparity? After all, we have adventuresses such as Christina Saunders, who aspired—and succeeded, if we are to take her word for it—in achieving a belt-notching history of sexual conquests that would make Casanova stand up and take notice (of course, who was actually the conqueror and who the conqueree is a disputable question).
If more women than men are inclined to be sexually sedate in terms of numbers, and yet there is a certain subset of women who are exceedingly active like Christina (after all, compared to men, women are physiologically capable—at least in theory—of bouncing back rather more quickly), then it all seems quite mathematically possible—that is, if that subset of extremely active women are inclined to underestimate their numbers of sexual encounters (unlike Christina, who seems to have kept count). That’s all that would be necessary for the averages to exhibit the observed skew.
Then again, maybe everybody’s lying.
[NOTE: I fully expect that readers more well-versed in math and statistics than I am may find a flaw in my argument. Go right ahead.]
I don’t think it’s mathematically impossible: remember, an average (mean) is not the same as a median. Rather than a mathematical proof, let’s take an example:
Alan, Brad, and Charlie each have sex with Donna, Eileen, and Florence, respectively. But all three of them also have sex with Gail.
The average number of male sexual partners is obviously 2. For women, the average number of partners is 1+1+1+3, all divided by three=2.
What may be happening is that women with large numbers of partners are misunderestimating, or maybe the sample is skewed in ways that tends to leave them out.
If we are talking about infidelity among married partners, there is a simple explanation for the discrepancy: married men are having sex with unmarried women. These unmarried women would not be included in the subset of married women who are unfaithful, because they are hot married.
There is another possibility: women have multiple sex partners more than men. I find this possibility unlikely, as from my experience and from what I read, men are more likely to have multiple sex partners than women. But I could be wrong.
david foster: a skewed sample is another definite possibility. Are prostitutes being sampled, for example?
Was that a Freudian slip?
“These unmarried women would not be included in the subset of married women who are unfaithful, because they are hot married.”
Correct to:
“These unmarried women would not be included in the subset of married women who are unfaithful, because they are not married.”
Just to throw another curve ball into the discussion: how many married men are having sex with other men as well as with women? (Same question in reverse for women.)
Another curve ball: how well did the survey explain what constitutes sex?
I note that many high-school students these days, for instance, don’t believe that oral sex counts as sex. I seem to remember that you even had a former president who failed to consider an act of fellatio as either a sexual exploit or infidelity.
Did the surveyors explain whether a grope, or a kiss, or a massage with a “happy ending” would count?
“Except that it is logistically impossible for the average man to have more partners than the average woman.”
Nope. Simple counter-example: Four men, four women. Two women have sex with just one man (different ones), other two women have sex with all four men. Number of partners for men: 2, 2, 2, 2. Mean-mode=2. Number of partners for women: 1, 1, 4, 4. Mean=mode=2.5
In my own experience (and I’m not bragging here; I’m actually embarrassed and contrite to admit this, but it is germaine to the conversation) women are no less willing to have affairs than men these days. The advent of the internet has made it quite easy. Women generally seem to fall into affairs, not with “tall, dark and handsome”, but rather with someone who will talk with them. It seems to me there are an awful lot of women in that situation and I would argue that women are not being truthful about having affairs.
My $0.02 (adjusted for inflation),
I’m not well versed in statistics, but I’ll claim not to be completely ignorant.
Let’s consider a representative small town with 1000 men and 1000 women. Over their lifetimes the men will have 12700 liaisons and the women will have 6500. Something doesn’t add up.
But Neo’s point about prostitutes is an astute one. Suppose that 1% of the women are prostitutes: 10 in all. Suppose each of them has 600 different clients, i.e. one a night for a career of almost two years. That pretty much takes care of the discrepancy. (Alternatively, a single prostitute might have four clients nightly over a four-year career: 4*365*4=5840.)
I can readily believe that retired prostitutes who are no longer doing sex work might not reveal the details of their past. IMO the prostitution interpretation is plausible but needs additional evidence.
I got it from Agnes
She got it from Jim
We all agree it must have been
Louise who gave it to him
Now she got it from Harry
Who got it from Marie
And ev’rybody knows that Marie
Got it from me
http://www.youtube.com/watch?v=EDeRYmB4t6Q
(Alternatively, a single prostitute might have four clients nightly over a four-year career: 4*365*4=5840.)
Oops! Obviously a single prostitute can’t have 5840 different clients in a town of 1000 males.
Details, details…
But if we grant that Neo’s “loose women” are also reticent about their past after their time of greatest sexual activity, the overall argument is not indefensible.
Have to disagree with Bob M’s counterexample. He’s right for the women, but for the four men their number of partners is 2-2-3-3 for a mean of 2.5, same as the women.
Remember every man has sex with the two women who give it up to all, but two of those men are also having sex with the two women who have sole partners.
Seems that back in HS there were fewer sexually active women, but that they had many different guys wanting to sample their wares.
I also vaguely remember a study years ago that claimed (sorry to bring it up, but…) that among those in their mid to late teens, a greater percentage of young men had had sex with female cousins than young women with their male cousins. The explanation was that the girls had encounters with more than one of their male cousins.
The average number of partners must be nearly the same.
Consider any number of men and a roughly equal number of women constituting the entire population. Arrange the names of the women in a column on the left and the names of the men in a column on the right.
Draw a line between the name of the woman and the name of the man in every unique sexual partnering. That is, if Betty and Frank had sex either once or many times one line is drawn between their names. Betty counts as one partner for Frank, and Frank counts as one partner for Betty.
The average (mean) number of partners for the men is the number of lines divided by the number of men. Similarly, the average number of partners for the women is the number of lines divided by the number of women.
The number of lines is the same in both calculations, and the number of men and women is nearly the same. Hence, the two averages must be nearly the same.
Dave: but not if most women have fewer partners then most men and there’s a small number of women with enormous numbers of partners, and those women are the ones who are lying by underestimating the number of their partners (or those women do not appear in the sample for some reason, depending on the sampling methods—for example, if they are prostitutes and somehow the sampling methods are not reaching them.)
That is my basic point.
As a metric of how people actually behave (as opposed to what they say about their behavior), the data is grossly flawed.
Neo’s observation about prostitutes/”loose women” is ingenious and plausible, but there are other plausible interpretations.
The bulk of the discrepancy need not necessarily be due to a single factor.
Well . . . I have worked in large office buildings, and large industrial settings, each with about a 50/50 split of men/women. I can tell you that when you get large groups of men and women working together, there’s
oops, somehow posted before I was ready.
Well . . . I have worked in large office buildings, and large industrial settings, each with about a 50/50 split of men/women. I can tell you that when you get large groups of men and women working together, there will be a bunch of the men having sex with a bunch of the women. You can count on it. The numbers will be more-or-less equal and the majority of them will be married. The women are fibbing to keep from appearing to look “loose.”
Haven’t done that since high school.
@ gs As a metric of how people actually behave (as opposed to what they say about their behavior), the data is grossly flawed. How do you know that?
ELC Says:
‘@ gs As a metric of how people actually behave (as opposed to what they say about their behavior), the data is grossly flawed.’ How do you know that?
Via my previous posts in the thread, and Dave’s post.
Urrhh…anyway you slice it between one and four percent of married people cheat on their spouses every year.
This seems to be a reasonable amount. I’m not sure it’s necessary to go any further into the numbers.
Instead this seems a distraction from the real issue, the personal and societal impact of infidelity: divorce, STD’s, child neglect, depression, suicide.
There are only a few ways for the actual number of average partners to be different. If they are split 50/50 and only having sexual intercourse between the two groups then the average number of partners across the group *must* be equal. Two groups of ten, One guy with ten partners means all ten women can only have one partner – each side equals out to an average of one partner. The Median value is most likely quite different, but the average isn’t going to be.
If there are more of one gender than the other then it can get out of balance, at the sample size we are talking here we know the sexes are roughly 50/50. Add in another female to the above example and the men are still at one and the women slightly less.
If they are having intercourse amongst thier own groups that would skew it too. If you had one of those males have intercourse with another then the woman’s average would still be one but the men at 1.2
There is lying too, if the guy said he had 15 partners and only five of the women admitted to having intercourse then the numbers are .5 and 1.5 – which is so far from the truth that the only conclusion you can make is that you can’t really do a study on this.
And lastly they could just have borked the whole thing up. They may have gone to a male homosexual sex party on one side and an abstinence vigil for the women. If so then they just screwed up.
One would expect *some* minor differences (answers recorded wrong, people not remembering accurately, etc) and that is why you have things like confidence intervals, margin of error, etc. But that much shows something is done incorrectly or there are a whole lot of homosexual males compared to females. My guess is lying. If done well the “prostitute” factor shouldn’t occur and be part of the “noise” that gets filtered out (unless prostitutes make up a large enough part of the female population that they are not dismissed as part of the study – which would be an “interesting” result).
Well, going off of cheating I actually know happened, a whole bunch of guys slept with a small number of girls– nothing like having the concept of “run a train” explained to you by someone for whom it is a hobby and something they’re proud of. (*gag*– and yes, multiple women.)
Miss one or two of those women, assuming everyone kept count and told the truth, and your numbers will be hugely skewed.
In younger days, a bunch of friends who grew up together frequented the same night club for a few years. A group of females did the same. Many of the friends of both sexes had sex many times.This, of course, was the seventies.
As a single guy now for 14 years, I have noticed that yes, the cheating (hooking up) is happening , at a seemingly increasing rate. I know husbands and wives who go out with other wives / husbands, and toward different sections of town. Don’t forget swingers’ groups. They are larger in number than most would believe. And a lot goes on there. Not that I personally know anything first hand, but i do have this friend who has told me about it.
I know couples whose kids have grown, and live alone together in the same house, in different areas. Only two friends I have known that married young are still married to their original spouse. Some are divorced from their third.
When one looks at the questions being asked, one knows immediately that they are not reporting the arithmetic mean. As others have pointed out, the arithmetic mean is always going to be the same. So it’s not an interesting question to study. No one’s going to run an experiment to see if 2+2=4 for even for hippos, ice cream cones, and baseballs. We know the answer.
(Unless they are specifically testing for fibbing, in which case they would be reporting the data with that discussion.)
As the results are in decimals, they can’t be modes or medians in the simplest sense. You can’t have .8 of a sexual partner (Right. Go to town. Have fun with that, folks. It’s too easy a setup.) But if you have a bunch of answers clustering around 8 and 9, with many more 9s than 8s, it’s reasonable to call that 8.7 or something.
So whether the median or mode is being measured, the small group of women having many more partners than most other women gives you unequal numbers for the two sexes.
“Miss one or two of those women, assuming everyone kept count and told the truth, and your numbers will be hugely skewed.”
No, they will not (unless the study is horribly fraudulent – which I will allow). It’s not like you are the first person on the planet to ever see that this could occur.
Large scale studies that use a random sample of some sort throw out a certain percentage of the outliers – that is those women are *not* counted unless they end up being a significant portion of the population.
While it is MUCH more complex than this you note that on an even distribution (which this most likely is not, lots of 0-5 but not many 20+’s, but it makes the idea easy) that there are few 0’s and there are few 100’s so since they “contaminate” the averages you simply do not count them. These things get expressed in things like confidence intervals, p values, and a number of other ways. On curves that tend to be unbalanced (in this case true – lots of 0-5’s and a VERY few 100+) it isn’t the first time that has been seen either – the methods are more complex but not really that hard to execute.
It’s not like mathematicians have never seen what is described and are shocked when it occurs. There are quite good methods to take care of it and, if done even remotely properly, if those outliers make a difference then they aren’t outliers. That brings up a whole host of other issues if those *are* true and the study passes much scrutiny at all. It means a large enough portion of women are truly whores, sluts, free, or whatever term you choose are quite common.
strcpy – that would allow for them to NOT miss the women I mention, and for everyone to be telling the accurate truth, but still have the skewed numbers.
If they are throwing out the outliers, then they don’t even have to miss the women with, ah, freaky tastes– and if only one in, oh, a thousand women does this, and one in a hundred men is interested in it, the men who do that exclusively would have “normal” numbers, men who did it once would have normal numbers, and the women would be outliers.
Just remembered– there was a study out of GB that showed that women had a higher rate of getting drunk before hooking up; if both are drunk during a hook up, enough to not remember what happened, the man might assume he got lucky and the woman might assume nothing happened.
A pair of women I knew for a few months managed to “rack up” about two hundred partners each, with them arguing out what counted and what didn’t, and levels of proof. (They were having a contest–a ‘ho down.’) Up goes the number for the guys by two, and theirs goes through the roof.
Ooh, started looking around for the source of that quote, since it was lifetime instead of married cheating like the study — this study of 30k+ folks found that more than one in five women aren’t having sex, while only 15% of men aren’t. (that’s the biggest number difference in that study– the yearly partner average is like .8/1.3 female/male)
Don’t forget to take death into account.
I contend that high count women have a higher death rate per year than normal women or than high count men.
It doesn’t take too many dead prostitutes to drop the women’s count.
The same would likely hold true for cheating spouses, with female cheaters having a higher death rate than male cheaters.
Ooh, good point.
There are a LOT of different options before we start assuming “those asked are lying”– well, unless we start with the assumption that folks are lying!
Okay, first I’ll admit to having cheated. Now let’s look at how all this can work out. The following occurred over a 30 year period:
Woman 1: Single. Ended up being a lesbian, too. Not that this is statistically significant. Just sayin’.
Woman 2: Single, with many, many partners. She alone can accomodate the statistical gap that you see. All by herself.
Woman 3: Married. Her first time. Statistically, that doesn’t put a dent in the prior sampling, but will actually reinforce the probability (or something like that, I’m trying to sound smart).
Woman 4: Married. Her first time. Hmmm…
Woman 5: Married. Her second time. But her first time was with a woman. Gads, now it’s getting complicated.
Woman 6: Divorced. She had remained faithful to her husband while married.
So what does all this mean? Just that there are whole lots of combinations out there, many of which can either prove or disprove the statistical results. Of the married women in the group, all have since divorced, two are remarried and one even went back for another round with me after the remarriage. All of the women had different motivations and different patterns
We can be loose with the truth, but it’s also probable that the probabilities are probably not so problematic. I have no problem accepting the statistics as stated.