Friday, 3 September 2010

A brief rant about waist-to-hip ratio

This is a bit off topic, but it's a good illustration of William James' notion of the psychologist's fallacy and it addresses a pet peeve of mine.

Evolutionary psychology is becoming more and more popular and the media is one of its biggest fans.One thing that annoys me is how quickly and uncritically people latch on to these stories and use them to justify the status quo. One of the most popular stories is that men prefer women with small waists and big hips. This is measured using the Waist to Hip Ratio (WHR). The WHR is the circumference of your waist divided by the circumference of your hips. The links below will tell you that men are irresistibly drawn to women with WHRs of .70. This number is apparently imbued with evolutionary significance because prepubescent girls have WHRs close to 1 (their waists are the same size as their hips), while post-pubescent girls have WHR less than 1 (waists smaller than hips); and also because low WHRs are associated with a good hormonal balance. One thing that makes this idea attractive is that it conforms to our modern, western experience - many women who are considered to be extremely attractive have low WHRs and it's difficult to generate examples of women who are famous for their beauty, but who have high WHRs. This evolutionary angle legitimizes our society's standard of attractiveness. We assume that everyone else basically shares our own preferences (the psychologist's fallacy), so, rather than this result simply telling us something about modern, western mens' judgments of attractiveness, there is the irrisitable pull to generalise this preference to ALL men.

To give a sense of the way the popular media handle this topic, here are a couple recent stories about the WHR: 1 (this one includes exercise tips to help women appear to have a more ideal WHR ratio) 2, and 3 (this one also claims that "men's perfect lovers come with a waist-to-hip ratio of .70", implying, I suppose that WHR ratio influences how good you are in bed??). Science reporting is rarely subtle and these articles are no exception. They talk about "males", "females", "mate preference", and "evolutionary" indicators of fertility. This language suggests to the average reader that these results are universal. That they reflect the preferences of people in general. But, does the research behind the headlines support this universality?

Evolutionary psychology, like most other branches of psychology, tends to lack cultural population validity. This problem is well-summarised in Henrich et al. (2010; see also here for interesting commentary on the original paper). The gist is that we can never take culture for granted in psychological research; there is no realm in which it is safe to assume that an effect is universal. The burden of any serious evolutionary psychology research program must be to establish the generality of their results across cultures. It doesn't matter how cool the evolutionary angle is - oh, look, this co-varies with fertility!!. It doesn't matter how obvious the effect seems to us. If male preference for women with low WHRs doesn't obtain across cultures then it's not universal. This isn't to say that there couldn't still be an evolutionary component to our preferences. It would be remarkable if there were not. But, genetic contributions to behaviour are complicated. So, failure to establish the generality of a preference for low WHR doesn't necessarily imply that men aren't sensitive to information that conveys fertility in potential partners. But, it does mean that there is not a universal reliance on this one particular type of information. It is quite likely that a whole lot of cues interact in a complex system of perceived attractiveness, to the extent that it doesn't make much sense to isolate one variable. So, anyway...

What IS the evidence for a low WHR ratio preference across cultures? Well, it's actually quite muddled. Westman and Marlowe (1999) provide a pretty good intro to the evidence for the WHR preference, so I'd recommend their paper for a quick overview. They point out that the  majority of studies on WHR rely on American undergraduates, although there is also evidence for a similar preference in Hispanic, British (although see below), and American-Indonesians. Some researchers (e.g., Singh, 1993) suggest that this preference is universal across cultures (p. 305). But, rather than jump straight into a statement of universality, Singh says something a bit more measured. He claims "the fact that WHR conveys such significant information about the mate value of a woman suggests that men in all societies should favor women with a lower WHR over women with a higher WHR for mate selection or at least find such women sexually attractive." That last bit is interesting. It merely suggests that men shouldn't find women with low WHR unattractive. This is a very different argument than the oft repeated universal preference for low WHR.

Unfortunately, Singh's perfectly reasonable prediction has morphed into a presumption of universal preference for low WHR. This means that we hear little about evidence that contradicts this assumption. But, as it happens, there is quite a bit of evidence against this claim. Westman & Marlowe (1999) tested the effect of weight and WHR on perceived attractiveness, health, and suitability as a wife in the Hadza of Tansania. The men in that society showed no preference for women with low (.7) or high (.9) WHR, but they did show a distinct preference for heavier (cf. thin) women. Yu and Shepard (1998) also failed to find an effect of WHR on attractiveness among the Matsigenka. Swami et al (2007) looked at WHR preferences among males in Spain, Portugal, and the UK.In all three countries BMI, not WHR, accounted for the most variance in perceived attractiveness. WHR influenced attractiveness judgments for Spanish and Portugese, but not British men. However, even in the Spanish and Portugese samples WHR accounted for only about 18-19% of the variance, while BMI accounted for over 70% of the variance in perceived attractiveness. This paper also has a great summary of methodological issues with prior WHR studies (e.g., the use of two dimensional line drawing, failing to control for BMI). Cornelissen et al (2009) looked at patterns of British male gaze fixation during attractiveness judgments of pictures of women. Men tended to look at the upper abdomen and face, not the hip or pelvic area. The pattern of gaze fixations matched they way men evaluated the same pictures when estimating body fat, and did not match the way men evaluated WHR. Reading these papers suggests a lively debate in the literature about the universality of low WHR preference. I am not an expert in this area, and these examples don't even scratch the surface, but they do indicate lack of consensus on the generality of the low WHR preference.

So, what does WHR even mean, evolutionarily speaking? Most people seem to argue that low WHR indicates a good balance of estrogen to other hormones, which is important for fertility. Fertility, undoubtedly, is essential to evolutionary fitness but 1) WHR isn't going to be the only cue to fertility and 2) there are other important characteristics that may account for more variance in reproductive success in some situations (e.g., if the vast majority of women in a certain age range are fertile). Cashdan (2008) looked at actual average WHRs in a variety of cultures, mostly non-Western. She found that the average WHR was > .80 (remember, .70 is supposedly the magic number). Cashdan pointed out that androgens and cortisol both increase abdominal fat in women (increasing WHR). But, higher levels of these hormones are also associated with increased strength and stamina, which come in handy in less than optimal circumstances. She says: "Waist-to-hip ratio may indeed be a useful signal to men, then, but whether men prefer a WHR associated with lower or higher androgen/estrogen ratios (or value them equally) should depend on the degree to which they want their mates to be strong, tough, economically successful, and politically competitive" (p. 1104). This suggests that it's possible to construct a perfectly reasonable evolutionary account for why men might prefer a high, rather than low, WHR (i.e., given a stressful environment where strength and stamina matter). The variables that dominate in a particular situation will likely depend on a number of specific environmental and cultural conditions. In other words, it's complicated.

This story, unlike the one about low WHR preference, doesn't seem to reflect our (modern, western) experience, so it's less likely to catch the popular imagination. We don't tend to think of male attraction based on female heartiness, but we also live in a particularly rich culture where we don't spend a lot of time physically searching for / killing food or building shelters. So, here's the psychologist's fallacy again. Evolution is complicated and the features that confer fitness are necessarily dependent on context. This means that it's not too difficult to think of a number of plausible evolutionary explanations for a particular phenomena. The preferred explanations are most likely going to be the ones that fit with our current experience, but this doesn't make them better explanations.

18 comments:

  1. Dixson (2009 - "Sexual Selection and the Origins of Human Mating Systems") wrote what I consider the authoritative chapter on human sexual dimorphism (ch 7: Human Sexual Dimorphism: Opposites Attract), in which he summarizes the research, including most the things you cite and rather a lot else, with particular attention to evidence from non-western, pre-literate cultures. I'm not an expert on this either, but he really and truly is. He says:

    "...men prefer female WHRs ranging from .67 to .8. Values within in [sic] this range are rated as more attractive and marriageable than are higher WHRs (.9-1.0). A WHR of .7 is not a universal preference, therefore, and we should not expect it to be so. Recall that, in Finland, health women of reproductive age have WHRs ranging from .67 to .8. It is reasonable to expect that sexual selection might have favored male preference within this range, and that the same might apply to different populations worldwide."

    I think the methodology has left a lot to be desired and I think mainstream media and a particular subset of evolutionary psychologists have done a piss poor job of talking about this phenomenon, using it to justify the "young, nubile, nulliparous women" line. But I also think it's a pretty robust finding. The popular understanding of it is clearly shallow and informed by cultural attitudes about gender rather than by the science, but that doesn't affect the overall validity of the finding per se.

    You'd like "Mother Nature" (Sarah Blaffer Hrdy).

    ReplyDelete
  2. Thanks, for the recommendation - I'd come across Dixson as well, but I couldn't get my hands on any of his articles before writing this. I'll check out that chapter.
    This is a complicated issue. On the one hand, there is a lot of evidence that, in judgment tasks, men from many societies rate pictures of women or (more commonly) simple line drawings of a female form as more attractive when they fall into this range. But, then you also have evidence using similar items (and sometimes using better items, i.e., pictures of real women) that show no effect of WHR when BMI is properly controlled for. In other words, the evidence is mixed. Also, has anyone actually looked at how this preference in dumb judgement tasks related to actual behavour? Do women with low WHR marry more frequently / have more children than women with higher WHRs? It is essential to establish this or the evolutionary angle is dead in the water.
    We know that many, many men show a preference for low WHR, but many men do not. It may be that most men show this preference. And that's fine. But it seems like shaky ground for building a simple explanation based on evolution. My problem is with the notion that "it is reasonable to expect that sexual selection might have favoured a preference in this range." Yes, it is reasonable to expect something like this. But, it is tremendously easy to construct these evolutionary just so stories. And, we should have a really, really, high bar for invoking evolution to explain behavioural effects. The fact that there is not, in fact, a universal preference for low WHR suggests that, whatever the actual genetic basis for the preference when it is expressed, it's not a simple story. Think about all the press about "the bonding hormone" that's been around lately. What was once a simple story about how having more oxytocin made you more committed to relationships (simple linear story) is now a complicated story involving interactions between genetics and cultural context (see an interaction between hormone expression and Western and Eastern social norms). So, to me, telling the simple linear evolutionary story about WHR preference mis-represents the evidence b/c it implies universality. I'd be interested to hear a more nuanced evolutionary explanation that places this preference in the context of a network of competing / complementary preferences.

    ReplyDelete
  3. The biggest problem with this kind of garbage study is that it asks men what their perfect lovers are, not the ones they actually had.

    This has two difficulties: one that men will inflate their fantasies, and two that most of these studies are carried out on college kids who have had maybe one bed partner in their lives no matter how much they brag otherwise. Asking is not the way to do research, is it?

    What needs to be done is to get men in their 70s and ask them not what they fantasize having, but what they actually had. Let's face it, there's an awful lot of stumpy-legged, hairy-lipped, thick -aisted women running around. Fess up guys, someone's spreading those genes.

    Could you imagine doing any other research like this? Dietary research? Ask people what they eat, surely that will be accurate enough ... !

    ReplyDelete
  4. BMI was developed as a population heuristic. At its inception, Quetelet (the 19th century mathematician who invented it) was explicit in stating that it's useless at the individual level. Thus, we're compelled to strongly question the methodology of research that relies on BMI at the individual level more than a century and a half later.

    Also see Dixson et al (2010) on preferences in Papua New Guinea, (2010) on New Zealand vs. U.S, (2007) on Cameroon, and (2007) on China.

    Rilling et al (2009) [PDF] also present a refined hypothesis based on multi-viewer-perspective 3-D models vs. the usual line drawings.

    Another interesting finding is that of Karremans et al (2010) in a paper called 'Blind men prefer a low waist-to-hip ratio [PDF].'

    @Janis "The biggest problem with this kind of garbage study is that it asks men what their perfect lovers are, not the ones they actually had."

    The biggest problem with this criticism is that it assumes all men are capable of "having" any woman they desire. Of course that isn't the case. Do you suggest a study in which men are allowed to "have" a woman from a range of WHRs to control for this?

    ReplyDelete
  5. @Janis & @Andrew - Yes, the failures in validity abound. The entire notion of preference is problematic here since any evolutionary explanation must account for actual behaviour.

    The blind men study is a weird one. It would make sense for an evolutionary bias towards low WHR's to manifest at the level of the visual system because the advantage to men is knowing who is a fit mate before hopping in the sack. So, it might be a stretch to expect such a bias to carry over into another modality. To the extent that it does show up in blind men, one might interpret the preference as cultural, rather than anything else (it's not as if blind men are not enculturated to have particular ideals of beauty, even though those ideals are not visual).

    Also, there is evidence that blind people actually have poor senses of touch (because a lot of our sense of touch works in concert with constant visual information). The common belief that blind people have extra sensitive senses of touch seems to stem from the ability to read braille (which is certainly impressive). But, this skill can be explained through practice and training, rather than enhanced sensitivity. This might suggest that blind men would actually have a preference for an exaggerated WHR (if there was a real preference for low WHR in the population) as a greater threshold would be required to detect a difference.

    Now, I don't think these are necessarily the correct interpretations of these effects, but it just shows that there are some problems of interpretation going on here. Basically, I'm just throwing around ideas.

    ReplyDelete
  6. Male Female Health Risk Based Solely on WHR
    0.95 or below 0.80 or below Low Risk
    0.96 to 1.0 0.81 to 0.85 Moderate Risk
    1.0+ 0.85+ High Risk

    ReplyDelete
  7. I agree that jumping to conclusions is (always) an issue. But is your summary merely that although there is (quite a lot of) evidence that men prefer a (low) WHR this has not (yet?) been proven to be universal?

    There is also evidence that women prefer a lower waist-to-chest WCR ratio in men, it may not be universal (I have no idea) but it doesn't stop it being a useful marker for fitness or beauty. I may not like it _personally_ as perhaps my WCR doesn't line up with the ideal, but I'm willing to accept the evolutionary fitness explanation until a better explanation comes along.

    Your comment on the blind study doesn't make sense. It doesn't matter if the blind people are overcompensating, under-compensating, idealizing or dreaming. It's still a finding. That paper seems pretty solid in saying that men prefer lower WHRs. Scientifically it doesn't matter for _that_ study _why_ this is the case, but it is definitely interesting that blind men tend in the same WHR direction as more culturally (presumably) sensitized non-visually impaired people. Saying that it doesn't cross modalities is silly, really. Most sexual and physical fitness markers are cross-modal (smell, skin texture, voice timbre, ease of movement, etc).

    Imagine for a moment that the "universal" WHR is spread over a bell-curve centered around .70. Understanding why deviations around that mean occur would be interesting (for example it could be that some populations have greater SDs around the mean than others and that would in iteself be important to study) but ignoring the fact that there _is_ a mean value is side-stepping the science, not negating it.

    I'm sorry you don't seem to like the finding and perhaps neither do I, but that doesn't make the science _wrong so far_. Better explanations are always possible.

    ReplyDelete
  8. Read: "Who's afraid of Charles Darwin?" On feminism and evolutionary theory, author: Griet Vandermassen.

    ReplyDelete
  9. In premodern Japan the ideal feminine body was a cylinder - there are bumpers and pads in the traditional kimono to make the silhouette completely flat all around. So anyway...

    ReplyDelete
  10. RE: Japan

    OTOH desexualisation of women through mode of dress in strongly patriarchal societies isn't exactly unprecedented. See also: Burqa.

    ReplyDelete
  11. Andrew Badenoch of Evolvify posted an interesting article on this topic:

    http://evolvify.com/female-attractiveness-waist-hip-ratio/

    FTA:
    "As a white, heterosexual, male, I am part of a group that represents a scant 1.47% of the human population that is supposed to prefer “big butts” marginally less. I submit that any supposed difference in this regard is a cultural influence that may correlate to race (because cultural influences often correlate to race), but is ultimately an exaggeration of a shared human nature. Further, I proffer that my preference for “abs” over “butt” (and/or “breasts”) is closer to the instinctual default."

    ReplyDelete
  12. "there is (quite a lot of) evidence that men prefer a (low) WHR this has not (yet?) been proven to be universal?"

    Well, first of all, several studies dis-confirm a universal WHR preference. It's not a matter of time until universality is established - we already have evidence that some groups do not show the preference or show a different preference.

    Second, one of the missing links (heh) in the evo psych argument here is that preference translates into successful procreation. Does the fact that you are attracted to women with a particular body type mean that you are more likely to end up having kids with someone of that body type?

    The first variable to consider is how likely we are to be able to pair up with someone who has an "ideal" body type. Today, we have quite a bit of choice in potential partners. There are a lot of people in the world and moving to different cities and countries is relatively easy compared to thousands of years ago. So, maybe these days we can be picky and "shop around" for someone we like and who meets some version of our preferred body type.

    But think about the situation thousands of years ago when humans weren't that common and lived in much smaller groups. To what extent would men have been able to pair up with someone according to their preferences for a low WHR? I would guess that some men would be able to act on this preference, but the others would end up with whoever was left (to my knowledge it is rare that women in this type of small scale society end up unpaired).

    The second variable is how much more likely is a man to have kids if he pairs up with a low WHR woman than a high WHR woman? In other words, assuming that preference leads to sex, how much have you increased your chances of successful reproduction by pairing with someone with a .7 WHR compared to someone with, say, a .8 WHR?

    I reported some evidence about the relationship between WHR and hormone balance, but this relationship does not in and of itself equate to increased fertility (which is a ridiculously complicated issue). If having sex with someone with an "ideal" WHR ratio doesn't lead to better reproductive outcomes, then an evolutionary argument doesn't make any sense. Of course, this will be a very difficult relationship to untangle since we can't exactly perform experiments to test the effect of WHR on reproductive success.

    It may be the case that there is a reproductive advantage, but the evidence isn't there yet. Until it is, it is irresponsible to make strong claims about WHR and evolutionary advantage.

    ReplyDelete
  13. Oy Vey. I think we may all be talking at cross-purposes, or at least confusing each other.

    By universal principle I (and I hope any person who calls themselves a scientist) doesn't take that to mean every possible instance.

    By universal, I mean a universal statement like "gravity makes water run downhill". The fact that it can be shown that in specific circumstances (e.g. superfluid helium) this is not true, does not make it less of a generalized 'lesson' about how most liquids behave.

    So far it seems that men _IN GENERAL_ and _AFTER_ smoothing out outliers and (relatively) short-lived fashions prefer (not require, insist on or get) a WHR of 0.7.

    Is the argument against universality because there is a Mr. Botham in Borneo who's into BBW and likes plump women with a WHR of 2.1? (ie that there are outliers?)

    OR is the argument against universality because you believe there is no bell curve with a mean of 0.7 at all? (ie there the mean is irrelevant?)

    There's no such thing as the average person, but people do have two arms on average.

    ReplyDelete
  14. Of course all laws have scope. Since I'm not a physicist I don't know the specific scope of the law of gravity, but, sure, if you observe something outside the scope of the law then you shouldn't expect it to conform to the law's predictions.

    This is entirely different than within population variability ("there's no such thing as the average person"). If you're dealing with a law then there is no variability in conforming to the law within its scope (i.e., an object will move through space according to the law of gravity 100% of the time if the object is observed within the scope of the law) . If there is such variability, then you don't have a law.

    Now, let's face it, in psychology we're not dealing with laws, but we might have more or less stable or predictable behavioural biases.

    Within-group variability in WHR preference doesn't hurt the argument that for the most part men prefer women with low WHR. That's just normal, within-group variation.

    But, this is not the type of evidence I cite that casts doubt on the universality claim. If the evidence shows systematic differences in preference related to other measurable factors (e.g., BMI, food scarcity) then you are no longer allowed to talk about average preferences, because preference is mediated or moderated by other factors. Continuing to act as if there is a single population with normally distributed variability is incorrect both mathematically and conceptually.

    ReplyDelete
  15. Sabrina:
    -----------------------------------
    But, this is not the type of evidence I cite that casts doubt on the universality claim. If the evidence shows systematic differences in preference related to other measurable factors (e.g., BMI, food scarcity) then you are no longer allowed to talk about average preferences, because preference is mediated or moderated by other factors. Continuing to act as if there is a single population with normally distributed variability is incorrect both mathematically and conceptually.
    -----------------------------------

    That's well said and I now at least understand why you disagree.

    I think your argument can be used the other way round: the fact that the measure is subject to differences caused by other factors is _more_ reason to talk about an average. It might be meaningless to you in one context, but very meaningful to someone else in another.

    Think about family size. In the US the average family size is circa 2.6 people per home, in India it is approximately 4.8 people per home. There's a big variability due to lots of factors: food scarcity, culture, wealth, education, access to contraception.

    Those average numbers matter even though they are averages and codependent with other factors. The averages inform policy, provide evidence of underlying problems (daughter/son ratio for instance), help determine spend and social planning. Also they provide a feedback mechanism of cultural norms (rightly or wrongly). I am unlikely to have 2.6 people in my family, but the number matters nonetheless.

    Continuing the analogy of family size and WHR If you look at somewhere like Europe, the birth rates and family sizes tend to converge around an average (the deviations reduce) over time as countries converge their norms and share living standards, culture and so on. I don't have a link to historical data but if you are interested have a look at http://www.nationmaster.com/graph/peo_ave_siz_of_hou-people-average-size-of-households

    An average WHR _within_ a society (or societal groups like "nations in Europe", or "english-speaking-countries") matters because it has 'descriptive power' assuming no huge variations in other factors.

    I understand your underlying point that the average is often mis-used and over-stated as a 'norm' or producing a 'finding' when in fact it does nothing of the sort, especially when talking about society and culture (to paraphrase "Culture is what you can get away with"). But that's not the same as saying that it's not useful to know what the is. That there are variations of samples gives us _more_ reasons to think in averages and distributions, not fewer reasons.

    I think our discussion can be boiled-down to:

    * Men report _their_ WHR ratios in women
    * When averaged the WHR ratios tend to 0.7
    * Sloppy media reports this as an "ideal" WHR or worse calls this a "universal ideal"
    * In fact there are clusters around different WHR means for different societies and groups

    Thanks for the exchange.

    ReplyDelete
  16. I like your family size example and I agree that it would be equally informative to report within-culture averages for WHR.

    And, yes, if the media could possibly show some nuance in reporting this, that would be even better!

    ReplyDelete
  17. What annoys me is how WHR is touted as an indicator of health or fertility. What a joke! I have a straight up and down body, meaning that there isn't much of a difference between my waste & hips. I am a healthy weight, and the only way to make my waist smaller would be to literally have my ribs removed! LOL! And guess what? When my bf & I decided to get pregnant, about 2 weeks after going off the pill and *trying* about 4 times, I was officially knocked up.

    So much for evo-psych bullocks.

    ReplyDelete
  18. I am a woman and I have a WHR of .70.

    Whenever I wear clothing that accentuates my WHR, straight men pay attention to me - favorably. I get flirted with, complimented, touched, receive favors, asked out, etc. much more often than when I wear bagging or loose-fitting clothes that hides my WHR. I find this behavior to men to be true no matter what country I am in, or from what culture/subculture those men come from.

    This is not to say that women with a lower or higher WHR cannot or are not considered attractive by men. This is not to say that the same men who find women with WHR of .7 to be attractive will not also find women with a WHR of .8 or a WHR of .6 to be attractive. This is not even to say that men who find a woman with a WHR of .7 and a so-so face attractive will not find a woman with a WHR of .85 with a GORGEOUS face more attractive. WHR is by no means the ONLY measure of attractiveness in a woman. Compounded with the (many) other factors that indicate attractiveness and health, WHR may be only a small percentage of why or what a man finds attractive about a particular woman. However, it IS ONE factor that tends be attractive for men, and I feel it is intellectually disingenuous to write it off simply because we may not like it. Also, consider that FIRST we had biology; THEN we had culture. Culture is derived from our biological preferences. We are materialistic driven culture, because acquiring resources is hard-wired in humans and in ALL organisms. Acquiring resources improves chances of survival AND of reproductive success, and the survival of our offspring. Of course, we are not biologically hardwired to like iPhones, per se, but iphones are a resource, and we are hardwired to seek having more resources as opposed to less.

    As others stated, this seems to be general rule of thumb. Of course there are outliers. Traits that lend themselves to survival or reproductive success affect gene pools and populations NOT individuals.

    ReplyDelete