Sunday, April 15, 2007
A recent John Tierney blog entry cited a much discussed study of online dating behavior (PDF), which suggests that shorter than average women have it easier in the online dating market, with the ideal height being about 5'2 - 5'3, and taller heights incurring increasingly greater costs.  However, women clearly vary in height well outside this range, so something is responsible for so much variation in height being maintained. Here I examine a fairly obvious, partial reason: being somewhat taller than average pays off if a woman is specializing in a niche that depends on sex appeal, in particular sex appeal that incorporates a fair amount of moxie / aggressiveness.
To test this hypothesis, I used celebheights.com to determine the heights of the women listed in the Maxim 2006 Hot 100 ranking. It's somewhat of a "sample of convenience," but it's hard to argue that the women there aren't sex symbols or that Maxim has an irrational bias toward women of a certain height. I could only find data for 77 of them , which you can see below the fold. 
The above frequency distribution by two-inch blocks (up to and including the right endpoint) shows that the heights of sex symbols are approximately normal (skewness = 0.07, or essentially symmetrical), with mean = 65.6 in. and SD = 2.8 in. A representative sample (PDF p.10) of 1,371 US women aged 20-39 showed that their mean height is 64.1 in., making sex symbols on average 1.5 in. taller than the average American woman. Using any reasonable estimate of variance in the general female population, a two-tailed t test shows that this difference in means is significant (t = 4.5, p less than .0001 -- perhaps lower; my calculator cut it off there). Clearly, being a bit taller than average helps in becoming a sex symbol. But does it matter even within the sex symbol set? Not really. Here is rank in the list as a function of height:
The Spearman rank correlation coefficient between height and rank in the list is 0.14 but is not significant (p greater than 0.2). So, being a bit taller than average probably helps you get noticed, but it is not of central importance. Note, though, that the non-significant relationship is not due simply to a restricted range for height in this sample -- the SD of 2.8 in. is close enough to the commonly used estimate of 3 in. for the general population.
Having established the greater average height of such women, what accounts for this pattern? This tends to be the most bullshitty part of a report, so I'll just throw out some guesses, and readers can add their own two cents in the comments. (Apologies for not researching / citing this section, but my hunch is that it'd be easy to dig up a few references for and against each of the guesses, as the "why" is always harder to pin down.)
Conjectures that assume there's something sexy about tallness per se:
- Longer legs. My eyes feast upon other parts, so perhaps in the comments the leg men can explain the appeal of long legs.
Conjectures that assume height isn't sexy per se, but that it covaries with some sexy other traits:
- Greater height could reflect a higher degree of male-typical hormones, which would aid the female in projecting the less passive / "take what she wants" attitude that characterizes sex symbols.
A conjecture that assumes height isn't related to sexiness at all:
- Greater height is necessary to stand out from the crowd and intimidate other women in the entertainment industry. Maybe the average guy couldn't care less about a woman's height, but to break through and survive in a cut-throat industry, sheer height helps.
Whatever the reasons turn out to be, it's clear that there are reproductively prosperous niches that taller than average women are suited to, so that some genetic variation in height will be preserved despite males' apparent greater interest in females who are 5'2 - 5'3. Presumably the same is true in the other direction: former rockstar Shakira says here that being petite (she is ~5'1) causes men to act protective around her. Petite women might have a more sprightly, giggly appeal; so, alleles for shorter height could be preserved either due to the sexiness of girly girls or due to their receiving more protection and investment from others because of their pedomorphic stature.
 Oddly, the authors phrase this as a trade-off of height vs. income -- i.e., how much more income would a 5'10 woman have to earn compared to a 5'4 woman in order to garner the same amount of male attention? This makes sense when looking at female preferences since most female online daters are concerned principally with the guy's height and income, and thus face a trade-off should a particular guy not excel at both traits. It makes zero sense to look at male preferences this way, though, since most men are interested in many other physical traits before height even enters their mind, and income and power are not sexy to most men.
 I coded the women by rank only, so if a reader wants to know which data are missing, I'll upload the Excel sheet, and they can see for themselves. Almost all of the missing data are from the bottom of the ranking anyway, so the more important points are accounted for.
 I took the cited height for granted if it was an integer; otherwise I browsed through the text of the entry and discussion to decide which number was most accurate, sometimes rounding up, sometimes down, and sometimes keeping it as is. (19 of 77 data-points are non-integers, so I didn't round indiscriminately.)