Arnold Kling  

My Favorite Example of a Biased Statistic

PRINT
Who's On Your Shoulder?... Election Prediction: Voting w...

In a column pointing to research findings on a variety of topics, William Saletan includes


The richer the country, the lower the ratio of male-to-female promiscuity.

In no country is the ratio of male-to-female promiscuity measured in an unbiased manner. The true ratio of male-to-female promiscuity is determined entirely by the ratio of females to males, which means that it is probably close to one. If your survey says otherwise, then your survey responses are biased. In Learning Economics, this essay is reprinted as "The Omniscient Voyeur."

Nearly-omniscient Tyler Cowen provided the unbiased pointer.


Comments and Sharing


CATEGORIES: Economic Methods



COMMENTS (6 to date)
Richard O. Hammer writes:

Once I read an official hardcover family-sociology textbook, written by a real professor at a big government-university in the Midwest. It compared data of family behaviors in many societies. The more "civilized" a society becomes the more that society succeeds in restricting sexual intercourse to marriage, it said. There is more unrestrained sexual intercourse in less civilized societies. This observation seemed plausible to me since my own experience confirms it (I am a single man living in the most civilized empire Earth has yet endured).

But back to the quote from William Saletan. Isn't he saying something similar?

bob writes:

I haven't read the study, but by "promiscuity," couldn't they mean median rather than mean number of partners for a given population? There's no reason why those would have to be equal for both males and females.

I doubt that this is so much of a biased statistic as it is a basic downfall of public opinion polls with sensitive topics: sampling and respondent psychology.

Sampling meaning, you could be talking to men that have had more partners (Ron Jeremy) and women that have had less partners (Mother Teresa) or vice versa.

Pyschology meaning the nature of the question is such that it could cause someone (male or female)to answer what would be "acceptable to society". I wonder if it would be safe to say: men would say greater than or equal to their real number of partners and women would say less than or equal to their real number of partners? (I said I wonder!)

But let's be realistic, the way you guys are thinking about it, you're treating the events as though they are mutually exclusive which is a phallusy...

5 women could have slept with 2 men each but that doesn't necessarily mean they didn't sleep with the same two men and only those two men. Meaning the women had 2 partners each and the men had 5 partners each.

Am I missing something?

ed writes:

I wondered about this, too. I assumed it meant something like "the number of promiscuous men divided by the number of promiscuous women," for some definition of promiscuous. Then the factoid could be true if, for example, poorer countries feature a lot more sex with female prostitutes.

spencer writes:

From my experience living in Latin America as a young man where males of all social classes frequented prostitutions much more frequently then they do in Anglo-Saxon societies I really have to question the reason for this assurtion that the male to female ratio is about the same in all societies. From what I know about muslim societies where women are protected by the family even more then in Latin societies I suspect the case is even less so in the Middle East then in Latin America.

What is your source for your statement that the data is bad.

Have you tried to compare your data with the prevelence of prostitution?

Don writes:

Spencer,

If you thought about it for a moment, you'd reconsider your question. In the Latin America anecdote you relay, the behavior you claim to have observed would lead to many females having a low number of partners and a few (prostitutes) having a staggeringly large number -- leading to a mean number that's exactly the same as for heterosexual men. This is identically true as a matter of logic, if we confine ourselves to the population of heterosexuals.

Comments for this entry have been closed
Return to top