You are here
Is the average torso among women the most attractive?
Is this the most attractive torso in women?
This is a rough sketch of the average torso of a sample of adult women that was found most attractive in a study by Donohoe et al.(1, pdf), more attractive than the average torsos of “super attractive” women such as Playboy centerfolds, high-fashion models from the 1920s and 1930s and high class Australian prostitutes/escorts.
This study employed a different methodology from most studies in the genre to date. The authors produced sketches of torsos with random combinations of shoulder, waist and hip widths – all widths found within samples of ordinary women – and had sets of images rated for attractiveness by male students.
The authors offered three reasons for why the “super attractive” samples fell below the attractiveness of the average among ordinary women:
- The “super attractive” women may actually be less attractive. But the authors think that this is unlikely.
- The “super attractive” samples had lower absolute waist and hip circumferences. The authors speculate that they did not have enough images where narrower waists and narrower hips were paired with the right shoulder length and the right dimensions for others parts of the torso that were not manipulated.
- Other features not considered by the authors – such as face, breasts – could make the “super attractive” samples overall better looking.
The “super attractive” samples are easy to address.
- Donohoe et al. need to look at the frequent prevalence of masculinized women among Playboy Playmates, thanks to the bisexual Hugh Hefner.
- Donohoe et al. also need to look at high-fashion models. These women are super-attractive indeed … but to the homosexual male fashion designers who find the looks of adolescent boys appealing.
- There is also literature showing a positive correlation between masculinization in women and a tendency toward promiscuity,(2–5) and thus we expect lots of women with below average femininity among prostitutes or escorts.
A reason why Donohoe et al. think that it is unlikely that their “super attractive” samples actually have below average attractiveness than the average among ordinary women is presumably that if these women were not very attractive then they would not end up where they did. This reasoning is easy to address for the fashion models samples. Regarding Playboy magazine, Hefner had the advantage of coming up with a pioneering product that caught on in the absence of competition. Airbrushing, crafty posing and breast implants among a lot of Playboy playmates since the 1990s onward would also help Playboy magazine remain a name to reckon with in spite of declining sales since competing publications in the internet age are unlikely to have the funds to recruit high-priced feminine beauties that Hefner can recruit but doesn’t because he is into masculinized bleached brunettes with fake breasts.
Donohoe et al. found strong support for stabilizing selection for the average, but what does one expect from their methodology? Nature will select against or equip organisms to sexually select against phenotypes comprising of random deviations from the average, and their stimulus set does indeed contrast the average torso with torsos featuring random combinations of shoulder, waist and hip widths.
We know from other research that above average femininity is a powerful correlate of beauty in women. Variation in shoulder, waist and hip widths all together does not capture much of the femininity variable; feminization does not change these three widths in a non-random manner; and feminization changes other aspects of the torso along with these width – e.g., the abdominal region gets elongated with feminization; see shape changes resulting from increasing attractiveness, clearly strongly overlapping with above average femininity.
In short, Donohoe et al. did not have the design to detect a component of attractiveness related to deviation from the average. Selection for averageness and selection for a specific type of deviation from the average will tend to cancel or oppose each other, and one needs proper methodology to detect both components. In practice, directional selection will have an upper bound resulting from selection for the average.
All indications are that if we start with the average face and body in women, which are attractive, and feminize them somewhat, then we improve upon their attractiveness, but there is an upper limit to how much the attractiveness can be increased by greater feminine exaggeration. There are some other components of physical attractiveness, most extensively described for the face, that correspond to increased attractiveness with some deviation from the average.
In their samples of ordinary women, the waist-to-hip ratio (WHR) varied from 0.52 to 1.15, with an average of 0.78 and standard deviation of 0.12. They found that peak attractiveness corresponded to a WHR of 0.75, which is in the middle of the range, just as shoulder-waist and shoulder-hip combinations in the middle of the range were most attractive, but as the response surface below shows, there is nothing about having a 0.75 WHR or for that matter any WHR that maximizes attractiveness. In other words, peak attractiveness was related to a combination of a specific waist circumference and a specific hip circumference that produced a WHR of 0.75 rather than any other combination of waist and hip measurements that had a WHR of 0.75.
To understand these graphs (see study pdf for details), imagine you are looking at a mountain ahead of you when you see the image on the left, with its outline tracing attractiveness. The image on the right shows the view when you see the mountain from top. The solid diagonal line on the right image corresponds to a WHR of 0.75 and the dashed line below it corresponds to a WHR of 0.7. Thus the peak attractiveness is for a WHR of 0.75 found at the peak of the mountain. Descending below the peak of the mountain diminishes attractiveness even if the WHR remains 0.75.
Here is the waist-hip response surface showing the placement of different groups. The ordinary women samples comprised of women ranging from age 19 and over. The average for women between the ages of 19–45 was a height of 5-foot-4, a waist circumference of 77cm (30 inches) and a hip circumference of 101cm (40 inches).
A = Playboy centerfolds (from measurements provided by Playboy magazine), B = 1920s high-fashion models, C = 1990s high-fashion models, D = Australian escorts; E–H = ordinary Australian women, E = 19–24 years, F = 25–44 years, G = 45–64 years, H = 65+ years. The data on high-fashion models were taken from a study by Byrdbredbenner et al. discussed here.
The ordinary women sample ranging in age from 25–44, F, was the closest to peak attractiveness, but the sketch of the average (and also most attractive) torso shown above, which describes the average for group F does not look like it describes the average for group F, i.e., there is a problem with properly translating these proportions to a sketch.
- Donohoe ML, von Hippel W, Brooks RC. Beyond waist–hip ratio: experimental multivariate evidence that average women’s torsos are most attractive. Behavioral Ecology 2009; Epub, ahead of print; doi:10.1093/beheco/arp051
- Cashdan E. Hormones, sex, and status in women. Horm Behav. 1995;29(3):354-366.
- van Anders SM, Hamilton LD, Watson NV. Multiple partners are associated with higher testosterone in North American men and women. Horm Behav. 2007;51(3):454-459.
- Mikach SM, Bailey JM. What distinguishes women with unusually high numbers of sex partners? Evolution and Human Behavior. 1999;20:141-150.
- Ostovich JM, Sabini J. How are sociosexuality, sex drive, and lifetime number of sexual partners related? Pers Soc Psychol Bull. Oct 2004;30(10):1255-1266.