Bryan Caplan  

More on the Modality of Monogamy

PRINT
The Keynesian shell game... Extracting Information from Re...
Tyler:
I say both men and women are understating their number of sexual partners.  Contrary to what is portrayed in this chart, I postulate an American male average of about four.  I do not agree with the common claim that American men will overstate their number of partners.
In the GSS, males report an average of 14.19, women an average of 4.76.*  If you mean the median, then males report a median of 3, woman a median of 2.  For men, 4 partners is the 62nd percentile.  So perhaps the results are compatible with your guesses.

P.S. The modality of monogamy is not an artifact of age.  Monogamy is modal for 25-44, 30+, and 40+, for both men and women.

Update: Fans have calculated the total number of partners for the main characters on Friends.  It's high.

* Excluding the "989 or higher" bin.


Comments and Sharing






COMMENTS (22 to date)
Emily writes:

The CDC reports data on this as well, and you can see breakdowns by age/education/etc., as well as lifetime vs. past-year. Table 16 also reports on other nationally-representative surveys: http://www.cdc.gov/nchs/data/nhsr/nhsr036.pdf

Anon writes:

There are a fair number of bogus pipeline studies that show men slightly overstate number of previous sexual partners while women significantly understate the same thing.

http://www.tandfonline.com/doi/abs/10.1080/00224490309552164#.VMEtBy7w-7Y

Mean technically should be the exact same, but in bogus pipeline studies, it's a tad *higher* for women. This is partly because of the biological tendency for older men to pair with younger women. Hence, women in the data set have previously paired with older men (who wouldn't be captured in the data set). There will also be much more variability (e.g. much longer tails) in men's responses (if truthful), which could affect the results one way or another (the idea that men's median is higher than that of women is absolutely laughable - I hope no one actually believes this is the case).

The idea that men would understate makes absolutely no sense, and anyone making that claim has a sore misunderstanding of both psychology and biology.

In fact, bogus pipeline studies have shown that men even overstate infidelity, which I find absolutely fascinating.

http://link.springer.com/article/10.1007/s11199-013-0266-3

Richard writes:

Does anyone know what counts as a sexual partner here? Do you actually have to penetrate or does it include lesser activities too?

Sam Hardwick writes:

Why exclude the "989 or higher" bin?

Miguel Madeira writes:

"the idea that men's median is higher than that of women is absolutely laughable - I hope no one actually believes this is the case"

Seems very logical to assume the men's median being higher than women's median - imagine the effect of prostitution over these statistics; or simply the classical (and perhaps old-fashioned) scenario of married men having affairs with the village/office "slut".

In these cases (that I suppose realistic) you will have the men's median being higher than women's median (men will have something similar to a bell distribution, and women perhaps something like a Pareto distribution)

Anon writes:

@ Miguel

You display a misunderstanding of both statistics and biology.

Pros are such an insignificantly small percentage of the overall population that the so called "sex worker effect" has virtually no impact on the data.

Additionally, the scenario you present is more media manufactured than grounded in truth. Although, I would argue that even the media doesn't really play up that stereotype too much. Much more common is the stud vs. nerd male comparison.

Anyway, I'm wasting my time writing any of this as the empirical research has already been done:

http://www.ncbi.nlm.nih.gov/m/pubmed/19627993/

"By examining published, empirical data we show that men and women consistently differ in the shape of the distribution of the number of sexual partners. The female distribution is always relatively narrow-variance is low-with a big majority of women having a number of partners close to the average. The male distribution is much wider-variance is high-with many men having few sex partners and many others having more partners than most females."

I don't have anything else to add.

Anon writes:

Actually - I'll add one other thing. In my opinion the reason most people are hesitant to believe that the female median is > than the male median is the same reason people consistently lie on the self-reported studies. Men are reluctant to believe that their significant others have had more experience than they have(even though it is almost certainly the case for >50% of men), and women would prefer this not to be known as well. It is a form of "ignorance is bliss" that is implicitly agreed upon.

Miguel Madeira writes:

"Pros are such an insignificantly small percentage of the overall population that the so called "sex worker effect" has virtually no impact on the data."

Imagine that a prostitute has 4 clients a day, working 20 days a month, 10 months a year, 10 years in a live. If all clients were new clients, she will wave 8000 clients. Of course, she will probably have the same client several times, but if a prostitute will have, lets say, 1000 diferent clients in her life, it is enough for a small percentage of prostitutes to have impact in the statistics. Or, more probably, to distort the statistics - because a statistic by random sample will much more easily to have the answers of clients of prostitutes than of prostitutes (sorry by my messy English)


"By examining published, empirical data we show that men and women consistently differ in the shape of the distribution of the number of sexual partners. The female distribution is always relatively narrow-variance is low-with a big majority of women having a number of partners close to the average. The male distribution is much wider-variance is high-with many men having few sex partners and many others having more partners than most females."

I think this only shows that variance is higher for men than for women, not that the distribution is more asymmetrical (in the sense of having a big difference between the average and the median) for men than for women.

Robert Simmons writes:

Have you considered hetero- versus homosexuality may explain the differences between men and women? Every indication I've seen is that gay men have more sex partners than anyone, and gay women less. Does the survey differentiate at all on that?

Duncan Earley writes:

I suspect Males only overstate if the have had very few, say 0 or 1 or maybe 2, but past that they probably don't overstate as why bother? Saying 6 instead of 5 doesn't seem that much different.

Also it would depend on who is asking. If it was a female or a male doctor I suspect men would actually understate.

Miguel Madeira writes:

www.pnas.org/content/97/22/12385.full.pdf

"One of the most reliable and perplexing findings from surveys of sexual behavior is that men report substantially more sexual partners than women do. We use data from national sex surveys and studies of prostitutes and their clients in the United States to examine sampling bias as an explanation for this disparity. We find that prostitute women are underrepresented in the national surveys. Once their undersampling and very high numbers of sexual partners are factored in, the discrepancy disappears."

Miguel Madeira writes:

In this case, even if the sample has a realistic share of prostitutes, the exclusion of "989 or higher" is enough to distort the results (by the way, what is the gender distribution of the "989 or higher"?)

Brian writes:

Anon,

I have to admit I'm not following your reasoning. My reaction is the same as Miguel's, who happily has supplied supportive links. If there are a small number of women who have a huge number of male partners, that would guarantee that the female median is less than the male median. If those females are not even included in the survey, that would guarantee that the reported female mean is also lower, even without reporting bias. Yet you seem to be saying something very different. Can you explain?

Anon writes:

@Miguel - like I said, the prostitute/sex worker effect has been thoroughly debunked

"Dr. Gale is still troubled. He said invoking women who are outside the survey population cannot begin to explain a difference of 75 percent in the number of partners, as occurred in the study saying men had seven partners and women four. Something like a prostitute effect, he said, “would be negligible.” The most likely explanation, by far, is that the numbers cannot be trusted."

http://www.nytimes.com/2007/08/12/weekinreview/12kolata.html

Also, Miguel, if you actually read the study I posted, instead of merely glossing over the abstract, you can see they show that male sexual history is clearly right skewed, not just greater variance

@Brian - I'm not sure exactly what you're having trouble with? It's quite simple to jump into excel and make a small model that shows that even taking into account the sex worker effect, any sort of significant right skew in the data for men (which has been clearly shown) will presuppose a greater median in partners for women than men

Anon writes:

@Miguel - also wanted to add that study is completely meaningless as it's asking a question that's already been answered. We don't have to suppose a prositute effect when looking at AVERAGE number of partners because we have already shown (see the study I linked in my first post) that the discrepancy btwn men and women's AVERAGE reported sex partners disappears when using a bogus pipeline study. The cause of the discrepancy has already been shown to be the tendency for men to slightly overstate and women to significantly understate sexual partners in self reported surveys. So, the AVERAGE question is answered. When we look at the MEDIAN question (which sadly the bogus pipeline study didn't release the numbers for, to my knowledge), all we have to do is understand that men's sexual history is more right skewed than that of women (already proven through epidemiological studies) and we can be almost 100% certain that women will have greater MEDIAN as well.

Again, I find it fascinating that so many people are reluctant to believe all of this, even with the simple facts laid out.

Miguel Madeira writes:

An example who you can have higher variance in men and, at the same time, the male median being higher than the female median:

10 males with 0 sex partners
20 males with 1
30 males with 2
20 males with 3
10 males with 4

average: 2
median: 2
standard deviation: 1.155


49 females with 1
33 females with 3
8 females with 4

average: 2
median: 1
standard deviation: 1.125

You can argue that I invented these specifical numbers to come to the conclusion; of course - but my point is exactly to show that, if the male and the female distribution are of a diferent type (a simmetrical distribution for men - a few in the top and in the botton and much in the middle - and an assymetrical distribution for women - many in the botton and a few in the top) you can have, at the same time, a higher variance in men but the male median higher than the female median.

You can have also this in a different model - imagine that women have a symetrical distribution and men have an assymetrical distribution with many on the top and few at the botton (also a realistic assumption, I think - there are more "cool kids" than "nerds"):

20 men with 0 sex partners
80 men with 5

average: 4
median: 5
standar deviation: 2

30 women with 3
40 women with 4
50 women with 5

average: 4
median: 4
standard deviation: 0.775

[I calculated the standard deviation with the Excel function STDEV.P)

Miguel Madeira writes:

[My last comment was posted before reading the coments of Anon]

Anon writes:

"(a simmetrical distribution for men - a few in the top and in the botton and much in the middle - and an assymetrical distribution for women - many in the botton and a few in the top)"

Except this is literally the exact OPPOSITE of what epidemiological studies have shown.

http://www.ncbi.nlm.nih.gov/m/pubmed/19627993/

Again, read the above link closely, and look at the full study if you have access.

"many on the top and few at the botton (also a realistic assumption, I think - there are more "cool kids" than "nerds"):"

Again, this is literally the EXACT OPPOSITE of what the data shows us. Male variability is RIGHT SKEWED. You can find countless on countless examples of this.

I put together a chart that makes it easy to understand. Based on empirical data THIS is what it should look like. (Note, i even added the pro effect, even though in real life it would be MUCH MUCH smaller than this)

http://i.imgur.com/LPdovWO.png

Miguel Madeira writes:

«Dr. Gale is still troubled. He said invoking women who are outside the survey population cannot begin to explain a difference of 75 percent in the number of partners, as occurred in the study saying men had seven partners and women four. Something like a prostitute effect, he said, “would be negligible.”.»

Imagine that 15% of men are or were clients of prostitutes (some statistics point to a number like this). Lets assume the average number of "normal" sex partners during life is 4; then, to explain the 4 to 7 difference, the clients of prostitutes had to have been with 18 different prostitutes (during their life); it is indeed an high number, but perhaps not totally unplausible.

Note that the very small number of prostitutes is not relevant for this point - the relevant number is the number of clients.

Anon writes:

[Comment removed for yelling and rudeness.--Econlib Ed.]

Anon writes:

"to explain the 4 to 7 difference"


Miguel, while I understand your point, there is actually no difference to explain. The bogus pipeline study shows that avg for women is slightly *higher* than men (due to woman normally pairing with slightly older men that wouldn't be in the sample).


I took the liberty of screencapping the relevant graph from the study

http://imgur.com/8UJBL7d

In summation, I believe the evidence shows that:

* The reason that there is a discrepancy when behavior is self-reported is because people lie

* There is no discrepancy to explain when both genders are truthful (other than why female # average is slightly *higher*

Hence, in my opinion there is no reason to fixate on the pro effect. It is statistically insignificant (as shown above).

Cheers

John T. Kennedy writes:

Typo: You dropped the t from the end of the last word.

Comments for this entry have been closed
Return to top