You are here
Using adult actresses and nude models to infer what heterosexual men prefer in women’s looks: Part 2
Previously I posted my response, published in the August 2009 edition of the Archives of Sexual Behavior, to an article by Voracek and Fisher (2006) on using adult actresses and nude models to infer what heterosexual men optimally prefer in women’s looks. When I posted this response, I forgot to check whether Voracek and Fisher replied back. They did. So now I am posting their reply and my comments on it (highlighted).
Data are the Natural Enemy of Hypotheses: Reply to Holland (2009)
Martin Voracek1 and Maryanne L. Fisher2
- Department of Basic Psychological Research, School of Psychology, University of Vienna, Vienna, Austria
- Department of Psychology, Saint Mary’s University, 923 Robie St., Halifax, NS, B3H 3C3, Canada
Maryanne L. Fisher
Email: mlfisher @ smu.ca
In our study (Voracek & Fisher, 2006), we analyzed available online data (as of mid-2001) for a sample of 125 elite (i.e., highly popular and successful) 1990s porn actresses from a leading European adult media company (Private Media Group Inc., http://www.prvt.com/).
Our central finding from this unobtrusive-measures design was the following set of associations. In this sample of porn actresses, lower body mass index (BMI) (an androgenousness cue) was correlated with more frequent movie starring, whereas several curvaceousness cues (waist-to-hip ratio, waist-to-bust ratio, and bust size) were not. Conversely, in the same sample, curvaceousness cues (lower waist-to-hip ratio, lower waist-to-bust ratio, and larger bust size) were correlated with more frequent magazine starring, whereas the androgenousness cue (BMI) was not. Of note, the association of lower BMI with more frequent movie starring remained even after we statistically controlled for models’ age, magazine starring frequencies, and curvaceousness cues. Similarly, the associations for each of the curvaceousness cues with more frequent magazine starring remained after we statistically controlled for age, movie starring frequencies, and the respective other curvaceousness cues.
While it is a truism that correlation does not equal causation, correlations do have causes. A pattern of correlations, found in the same sample and as distinct, differential, and stable as the above set of associations, is striking, and, in all likelihood, it reflects some underlying causes. Specifically, in our case, it led us to speculate whether visual cues to female physical attractiveness might indicate domain specificity. Thus, in our original discussion, we proposed that cues of androgenousness (versus curvaceousness) may be particularly salient for evaluating the attractiveness of moving (versus posing) female bodies.
In Holland’s (2009) comment on our work, he takes issue with the general approach we used, calling it “problematic.” Specifically, Holland asserted that porn actresses or nude models were an allegedly poor source of information about the features underlying female physical attractiveness, as evaluated and preferred by heterosexual men. Although we are in agreement with Holland in that the majority of previous physical attractiveness research has limited ecological validity, we do not agree with Holland’s assertion that sexually explicit media (especially those available on the open market) lack validity and are inappropriate for the study of female physical attractiveness. We do not stand alone in maintaining this perspective; his stance is clearly in the minority, as indicated by the number of contemporary, authoritative appraisals and research examples pertaining to this question that we cited when discussing the background of our own research (Voracek & Fisher, 2006, p. 298).
A minority stance is not necessarily in the wrong. In Voracek and Fisher’s article, the authoritative appraisals and research examples involve analyzing fashion models and Playboy centerfolds or assertions that male-oriented pornography can provide clues about the ultimate in men’s preferences regarding women’s physical attractiveness. A simple look at the faces of top female fashion models and many Playboy centerfolds will reveal above average masculinization among them, in sharp contrast to the well-documented preferences of most men and most women for above average femininity in women’s faces. How come the authorities never took this into account? I will get to the porn stars shortly.
Most notably, we referred to Salmon and Symons’s (2001) well-known statement, reiterated here, that adult media’s “characteristic features have been shaped in free markets by the cumulative choices made by…men…who have ‘voted’ with their money” (p. 56). More research examples along these specific lines have been reviewed in a recent monograph about consumption, as viewed from the standpoint of evolutionary psychology, which was published after our article (Saad, 2007, pp. 228–235). It is precisely this inevitable logic concerning the constraints and dynamics of free markets that are shaped by consumers who vote with their money that is relevant here. Market forces constitute the ultimate validity of commercial, sexually explicit media as an informative source for female attractiveness research. Pornography as an industry is about successfully selling visual enactments of male sexual fantasies. If these products were to miss this target, male consumers would not spend their money and adult media companies would become bankrupt. As for the source of the data we analyzed, we made it clear that we did not choose a company at random, but rather an undoubtedly successful and, in fact, a leading company. Private Media Group Inc. was founded in 1965, is active in over 35 countries, was the first adult entertainment company traded on the NASDAQ stock market (in 1999), was chosen by Forbes Global as one of the Top 20 best small companies in the world (in 2002), and reportedly owns the largest digital online archive of adult entertainment content in the world (www.prvt.com/company.php; http://en.wikipedia.org/wiki/Private_Media_Group).
Holland sets aside all of these important facts and does not even mention our central finding (namely the associations summarized above). No alternative analysis or interpretation of our data were offered nor were novel data (observational or experimental) presented that would yield opposing evidence to our specific findings. In essence, in his comment Holland assembled various points which were not new in this research context nor central to our findings; his more or less related points did not threaten the validity of our findings. However, due to the potential interest of readers, we address the major points raised by Holland in turn below.
I have set aside none of the points in Voracek and Fisher’s original article.
Market forces and top female porn stars
I noted that
Voracek and Fisher assumed that the physical appearance of top-ranked porn stars/nude models reflects what men generally find most attractive in women. However, most women will not pose nude, let alone participate in pornography. Therefore, as a first approximation, highly ranked porn stars/nude models reflect the best looking among the minority of women who are willing to pose nude or indulge in pornographic acts, not the best looking women per se.
Thus I have clearly responded to the issue about market forces. The market forces are selecting, as a first approximation, the better looking among the minority of women willing to pose nude or participate in pornography. Voracek and Fisher’s argument about market forces would only be meaningful if a random/representative sample of young adult women attempted to pose nude or participate in pornography and then male consumers of erotica selected the best looking among them, but why should we believe that a representative sample of young adult women are involved to start with?
Central finding of Voracek and Fisher (2006) on top-ranked porn stars and nude models
Voracek and Fisher accused me of not mentioning their central finding, but I wrote:
Therefore, Voracek and Fisher’s conclusion that androgenousness cues are salient for attractiveness evaluation of moving bodies whereas curvaceousness cues are salient for posing bodies could very well be substituted by the hypothesis of feminine women being less willing to participate in hardcore pornography/disinhibited sex acts, not that less feminine moving forms are preferred by men.
Thus, I have clearly referenced their central finding by referring to their assertion that “androgenousness cues are salient for attractiveness evaluation of moving bodies whereas curvaceousness cues are salient for posing bodies.” This assertion by Voracek and Fisher was based on a higher level of masculinization among the porn stars compared to nude models, inferred from their reported bust-hip-waist measurements and reported BMI.
Voracek and Fisher argued that I have offered no alternative analysis or interpretation of their data. They are wrong.
Alternative interpretation of the data of Voracek and Fisher
I cited literature that masculinized women are more inclined toward promiscuity, on average, and hence greater inferred masculinization of porn stars compared to nude models cannot be ascribed to men’s preferences for greater masculinization in moving bodies when there is an alternative explanation that feminine women are less likely to be porn stars and less likely to indulge in the more extreme/disinhibited sex practices that could lead them to porn stardom. This is clearly an example of an alternative interpretation, which Voracek and Fisher completely ignored in their reply, and this alternative interpretation is based on data that oppose the authors’ central assertion.
I also cited literature showing that the majority of the women analyzed by Voracek and Fisher had a reported BMI below the threshold of a healthy level of body fat in women and that most of these women were well below the peak attractive BMI for young adult women as reported in many studies. Voracek and Fisher completey ignored the BMI data I cited, which provide an alternative interpretation of their data:
If the pornstars’ reported data are correct, then most of these women do not have a BMI that is optimally attractive to most men and hence the central assumption of Voracek and Fisher that top porn stars offer a clue about the ultimate in men’s preferences regarding female physical attractiveness is invalidated. Indeed, why should a large number of men be attracted to women with sub-optimal body fat levels, sub-optimal from the standpoint of health, fertility and fecundity? If top pornstars represent the ultimate in female physical attractiveness, then why do so many of them reportedly weigh much less than the optimum preferred by men in many studies?
If the pornstars generally possessed a body mass optimally attractive to most men but their reported data are incorrect, then the statistical analyses of the authors, solely based on reported measurements, are not meaningful.
Alternative analysis of the data of Voracek and Fisher
I clearly presented an alternative analysis of the data of the Voracek and Fisher, which I shall address more fully later:
The weights of pornstars and nude models (particularly Playboy playmates/centerfolds) are generally underreported.
Too many pornstars and nude models have breast implants, rendering their reported bust measurements not useful.
Novel data presented by Holland
Voracek and Fisher said that I included no novel data. It is not necessary for me to include novel data to refute their article/thesis. But I did include novel data. I included a link to a zipped folder of images of a) porn stars that had appeared in the Private Media video series focusing on their top porn stars and b) top American porn stars between 1995 and 2005. These topless images show a large number of these women having breast implants. Voracek and Fisher cannot argue that this zipped file represents no novel data. A picture of a topless woman contains more information about the relative size and shape of her breasts than a 2-digit number representing her bust circumference (which offers no clue about whether the woman has breast implants).
First, Holland noted a “substantial overrepresentation” of models from Eastern Europe in our sample and further noted that we did not address this observation. From our sample description, it is clear that the sample was ethnically and geographically fairly homogeneous, as more than 90% of the actresses were Caucasian and 85% from Europe. Although, as mentioned, the proportion of East European actresses was sizable (38%), it was not conspicuously or surprisingly large when one considers the company’s European background. Holland overlooked the fact that we analyzed data from an European company, which, as one might expect, almost exclusively features European actresses. If one was to examine an American company, the resulting data would very likely contain fewer East Europeans. Therefore, the fact that European models were used by a European company warranted no attention in our original article.
A “European company” is misleading since Europe is a large geographic mass comprising of many nations. The Private Media Company was started in Sweden, by Berth Milton, in 1965. In the early 1990s, the company’s headquarters were moved to Spain, and the company is incorporated in Nevada, United States. In short, the Private Group had plenty of time to establish a recruiting base centered around Scandinavia, including Britain, Northern Germany and the Netherlands. This broad area would provide a lot of [natural] blondes, which seem to be popular with the male consumers of pornography, judging by the large number of porn stars dying their hair blond.
But after the break-up of the Soviet Union, pornographers started to recruit a lot of their women from the destitute Eastern European nations. Why? Voracek and Fisher can’t just dismiss this as a European company seeking European models. Physical appearance is an issue and Europeans are not ethnically homogenous. A random sample of Russians or people from the Ukraine will readily be physically distinguished from a random sample of the Dutch or Irish.
Voracek and Fisher ignored the obvious in their original article and refused to address it in their response to my criticism: for a given amount of money, better looking women can be recruited from poorer nations that have women of similar international appeal as richer nations. For Voracek and Fisher’s assumption – that the top porn stars represent the ultimate in female physical attractiveness to heterosexual men – to hold true, they need to assume that women’s willingness to pose nude or participate in pornography does not vary with level of attractiveness, which is not true. More attractive women will demand more money for the job and very good looking women will have more opportunities for mainstream success and are expected to preferably avoid nude modeling and porn work. Therefore, the substantial recruitment of Eastern European women after the break-up of the Soviet Union is an important issue that Voracek and Fisher did not consider and, properly interpreted, it invalidates their claim that their choice of models represents the ultimate in female physical attractiveness for men.
Note that the direction of hair dyeing (more often darker to lighter) and aspects of physical modification (e.g., straightening/making nose finer) among Eastern European women attempting to make themselves more attractive, thereby better approximating the physical appearance of Northern Europeans, is more consistent with a greater frequency of attractiveness among northern rather than eastern women. Hence, it is unlikely that the large number of Eastern European women recruited by the company reflect a higher frequency of attractiveness among eastern women.
Since the authors mention American porn companies using fewer Eastern European women, I asked someone who ran an adult video store if he could shed some light on the matter. He contrasted two American XXX adult companies: Vivid Entertainment and Devils’s Film Inc.:
Steven Hirsch, William Asher (Owners)
S. Hirsch, Custodian of Records
Vivid Entertainment, LLC
3599 Cahuenga Blvd., West
Los Angeles, CA 90068
Custodian of Records
Devil’s Film Inc.
2425 Ventura Blvd. STE 110
Woodland Hills, CA 91365
Vivid Entertainment, the largest adult video producer in the world, typically produces standard XXX pornography, whereas Devils’s Film disproportionately produces extreme hardcore pornography, described as “gonzo,” where the focus is on genitals and extreme sex acts. Devils’s Film uses a lot of Eastern European women, whereas Vivid uses far fewer. Why the difference? Obviously, a higher prevalence of poverty in Eastern Europe is a big reason. It is not difficult to find comments such as the following, presumably by individuals with the requisite experience:
Eastern European girls
Devils Films and a lot of other companies hire East European girls for their productions. Although they look stunning most of the time and they are willing to do the raunchiest stuff in the bizz... scenes with those girls mostly don't do it for me.
So often it is sooo clear that those girls are only doing this stuff to pay the bills. In scenes they hardly take the initiative to do something spontanious. The look at the cameraman or director with the questionmark "what must I do next?" The best example is the ffm scenes where the girls wait their turn to be fucked in stead of really join the action. When they are ordered to lick another girls ass, the hardly touch the spot with their tounge.
I have seen scenes from companies like sineplex.. where those east european girls look like they have pain when they are taken from behind.
No... they are beautiflull and of course their are also some good scenes every now and then buit general... I Fast forward the Devils Films and Sineplex Stuff.
The point should be clear: physical attractiveness is not the sole criterion of porn stardom, but willingness to engage in a broad repertoire of on-camera sexual activities and the enthusiasm with which one indulges in these sexual activities are also obvious important factors, willingness to indulge in more disinhibited/extreme sex acts being affected by a woman’s poverty level. Therefore, Private Media’s heavy use of Eastern European women is not inconsequential, especially in light of the extreme/statistically atypical sex acts that are often portrayed by this company (discussed shortly).
Holland’s claim that actresses’ weight was underreported is not new to this area of research, as indicated by the accompanying references he cited. However, is there compelling evidence to support the veracity of this assumption? In general, we think not. Numerous studies in the sports sciences have reported a similarly low (i.e., <20) average BMI within samples of young female athletes (for data on female fencers, see Voracek, Reimer, Ertl, & Dressler, 2006), so there is no reason to expect that lean, young women with a BMI of 18 or 19 would inevitably have emaciated looks. Interestingly, Holland also cited a study of female fashion models (Tovée, Mason, Emery, McCluskey, & Cohen-Tovée, 1997) which found a lower average BMI (17.6) than we did for the porn actresses sample (18.4). Presumably, some women who aspire to become models do not and they may later become porn actresses. If this is the case, then these converging measurements appear quite fitting and credible. Holland also criticized previous work of ours (Voracek & Fisher, 2002) as being naïve in that we allegedly relied upon the underreporting of body weight in Playboy centerfolds. However, his affirmative reference (Szabo, 1996) was an analysis of merely 11 centerfolds during 1994–1995 of the South African [sic] Playboy edition, whereas we analyzed 500+ centerfolds from 1953 to 2001 of the magazine’s U.S. edition. Therefore, his only supportive evidence is irrelevant. These facts aside, even if there were indeed a consistent underreporting bias in model’s weight, the results of our studies (Voracek & Fisher, 2002, 2006) would be entirely unaffected, as both studies were correlational and correlations are invariant under linear transformations of the variables from which they are calculated (e.g., underreporting, or negative bias, of weight).
Voracek and Fisher claim that there is no compelling evidence that the weights of the porn stars and nude models, including Playboy Playmates, are underreported. Let us look at how they rationalize this assertion.
They cited data on young female fencers to argue that BMIs between 18 and 19 don’t look emaciated, but I never made the claim that women in this BMI range look emaciated. The median BMI in their sample of porn stars was 18.3 and the lower limit of the range was 15.2. Thus, I asked, “Does this mean that a substantial minority of top-ranked porn stars/nude models look as thin as the typical high-fashion model and that some look like they are starving?” Voracek and Fisher didn’t answer this question.
Voracek and Fisher then brought up a sample of female fashion models, in a mid-1990s study where they had an average BMI of 17.6, to show some kind of convergence of the reported BMI of female fencers, fashion models and porn stars, ignoring/being unaware that fashion designers currently prefer models with a BMI in the neighborhood of 16, that female fencers are athletes that are not selected to be aesthetically pleasing to heterosexual men, and female fashion models cater to the aesthetic preferences of male homosexual fashion designers.
I cited Szabo (1996) as an example of an author who argued that the weight of Playboy centerfolds is underreported. Voracek and Fisher dismissed this as irrelevant because Szabo based his claim on 11 centerfolds from the 1994-1995 South African editions of Playboy magazine, whereas Voracek and Fisher analyzed 500+ centerfolds from the magazine’s U.S. editions, covering 1953 to 2001. This is a very misleading claim. Szabo took the trouble of comparing the reported body weights of the Playboy centerfolds with what their pictures suggested, whereas Voracek and Fisher simply took the measurements Playboy magazine listed and never bothered to check whether the pictures of Playmates suggest weights as low as Playboy reported them to be.
Later, I will gather some pictures of Playboy centerfolds and their body weights, as reported by the magazine, and let the readers decide for themselves. For now, I believe it suffices to say that if Voracek and Fisher believe that this Playboy model, described by the company as 5-foot-7.5, 36A-26-35 and 108 pounds, indeed weighed 108 pounds at the time the photograph was taken, then why do they even bother to study the attractiveness of women?
Voracek and Fisher argued that if the underreporting of models’ weights has been consistent, then the results of their studies would be unaffected because correlations are invariant under linear transformations. Whereas correlations are indeed invariant under linear transformations, only the following results would be useful: lower BMI among porn stars compared to nude models and a thinning trend among Playboy centerfolds over a time period.
Holland also claimed that, owing to a supposedly high prevalence (up to 50%) of breast implants, the utility of actresses’ bust measurements was limited. Holland ascertained these prevalence figures from nude pictures, available online and related to the ongoing Private video series “The Private Life of [porn actress name],” but concurrently conceded that there was “an element of subjectivity in discerning breast implants from pictures.” Since Holland provided no methodological details, no independent observers were involved, and the assessments were unverified, there is reason to doubt the validity of his findings. Holland’s statement that “breast implants are not consistent with making moving bodies look firmer” is surprising, to say the least. In fact, we expect that cosmetic surgeons will emphatically disagree with his claim. If this were really true, women in general would not utilize breast implant surgery (Fisher & Voracek, 2006); average bust size in our sample was not conspicuously large (slightly less than 86 cm), suggesting that women must be considering more than size when they opt for implants. Holland also overlooked that our data were collected up to mid-2001 and thus comprised of actresses from the 1990s, when breast implants much were less prevalent than in the late 2000s (Swami et al., 2008). In contrast, the “Private Life of…” series he consulted mainly has release dates around 2005 and later and thus does not overlap with our database. As set out above, even if there was a consistent positive bias in model’s bust size (through breast implants), our results would be unaffected, due to the invariance of correlation coefficients under linear variable transformations.
Voracek and Fisher said I provided no methodological details and no independent observers about my claim of a high prevalence of breast implants among porn stars (conservative estimate of 43% in one sample and 50% in another sample) and hence there is reason to doubt the validity of my claim. Well, I provided them with pictures in a zipped file. All they had to do was look at these pictures and search for more pictures of these women and evaluate my claim, but they didn’t do this.
I did claim that there is an element of subjectivity in a visual assessment of breast implants (the methodology), i.e., a conservative estimate of 50% prevalence of breast implants could be 40% or 60% in reality, but substantial prevalence of breast implants among the women in question is self-evident. If Voracek and Fisher can’t see a high prevalence of breast implants among these women, then, again, why are they bothering researching women’s beauty? The reader will object that this is not a scientific response, but what can be done short of going back in a time machine and taking an MRI/equivalent of each model/porn star? The reader will at least note that I have presented better data: pictures of topless porn stars (which usually reveal breast implants easily) vs. Voracek and Fisher’s measurements, which they took off of Private Media’s website (do not allow assessment of the prevalence of breast implants).
Voracek and Fisher expressed surprise at my statement that “breast implants are not consistent with making moving bodies look firmer.” They said that cosmetic surgeons will emphatically disagree with his claim and that if this were really true, women in general would not utilize breast implant surgery. Voracek and Fisher quoted part of my sentence. The full sentence reads, “Whereas Voracek and Fisher argued that androgenousness cues are salient for attractiveness evaluation, breast implants are not consistent with making moving bodies look firmer.” This statement was a response to their assertion that men prefer more masculine and hence firmer looks in moving bodies (porn stars in action). Since these pornstars with breast implants are not replacing naturally large breasts with implants but replacing small breasts with implants, they are creating a situation where their implant-filled breasts are bouncing around more notably than their naturally small breasts would while being filmed in action, which is obviously not consistent with a firmer look; the pornstars are overwhelmingly young adult women, not needing to take care of sagging breasts. Voracek and Fisher also bring in breast implants among ordinary women, but the purpose of these implants is primarily to look good in clothes, which involve many settings where there is little movement of the torso and breast movement is somewhat restricted by clothes.
Voracek and Fisher said that the average bust size in their sample was 86 cm, i.e., not conspicuously large, and hence these women must be considering more than size when they opt for implants. What would the non-size factor(s) be? The implants are not consistent with providing a firmer look while the torso is moving during sex acts. A non-size factor would be shape, but note that when women achieve a modest bust size after breast implants, then they had to be near-flat-chested naturally, and an obvious reason for the implants in this case would be to look more feminine. Therefore, in light of evidence showing a positive correlation between promiscuity and above average masculinization in women, one could certainly propose that more feminine women are less likely to become porn stars, which is an alternative explanation of the data that the authors interpret as masculinization being more salient to the attractiveness of moving bodies in women. Again, since femininity is a very powerful correlate of attractiveness in women, why would one expect top-ranked pornstars to be among the most attractive women—once more, take a good look at the pictures of the top-ranked pornstars I have referred to in my analysis.
Voracek and Fisher said that their data were collected up to mid-2001, whereas I used many Private models from around 2005, and that breast implants became more common in the 2000s compared to the 1990s, suggesting that I sampled more women with breast implants. Let me revisit the samples I used.
Among the top AVN award-winning pornstars from 1995 to 1999, I counted 10 of 18 (56%) with obvious breast implants; between 2000 to 2007, this figure was 17 of 35 (49%).
The Private Media top models that I used were featured in the series titled “The Private Life of [porn actress name].” At least half of the Private Media top porn stars that I used had been active in the porn world and working for the Private group before 2001 even though the videos commemorating their career, featuring various scenes of theirs over the years, were produced later. I had a sample of 42 of these top porn stars. Among the earlier 21, I counted 12 (57%) with obvious breast implants; among the later 21women, I counted 6 (29%) with breast implants.
Therefore, in both samples the prevalence of breast implants is lower, though substantial, in more recent years, and these data were made available to Voracek and Fisher but they ignored it.
I also counted breast implants among Playboy centerfolds in the 1990s and, ignoring ambiguous cases, about half of them turned out to have breast implants. Voracek and Fisher will dismiss this as subjective but let them have a look at topless pictures of Playboy Playmates and then tell me that the prevalence of breast implants among them isn’t substantial.
Voracek and Fisher again reiterated that if there was a consistent positive bias in models’ bust size (through breast implants), their results would be unaffected, due to the invariance of correlation coefficients under linear variable transformations. But this would only apply to the bust measurements of porn stars vs. nude models. The more important issues are ignored by them, namely that a) the high frequency of breast implants suggests that pornographers are not able to recruit enough women with the femininity that most men desire in women, thus strongly suggesting that many of these women are far from the ultimate in attractiveness to heterosexual men, and b) that a high frequency of breast implants among women whose naked torsos are prominently moving in the movies is not consistent with the greater salience to heterosexual men of firmer, less feminine moving bodies.
Finally, Holland asserted that anal intercourse is uncommon in the general heterosexual population and yet it is frequently depicted in productions of Private Media Group Inc. He perceived this incongruity (which we argue to be false) as another clue as to the inappropriateness of the data we utilized. There are two important issues that warrant discussion. First, the rise in the depiction of female receptive anal intercourse in pornography since the late 1990s is well-known (http://sexuality.about.com/od/sexinformation/a/anal_sex_stats.htm) and not specific to this company. Second, Holland fails to quantify what he means by “uncommon.” For example, uncommonness in statistical significance testing may be defined as occurring <5%, which is the conventional significance criterion, or in genetics, as <1% which is the cutoff point between common polymorphisms and rare allelic variants. Closer scrutiny of Holland’s only affirmative reference for his claim (i.e., Laumann, Gagnon, Michael, & Michaels, 1994, p. 130) reveals that 13.2% women (ages 18–59) in cohabitations and 7.6% women in marriages reported any occurrence of anal intercourse in the last year (based on a national probability sample, collected in 1991 in the U.S.). So, then, is anal intercourse truly “uncommon”? Relatedly, there is ample evidence for cohort, period, and cultural effects in this sexual practice which, in general, is more common than oftentimes assumed (Voeller, 1991). It has been estimated that, in terms of absolute numbers, about seven times more heterosexual women than homosexual men engage in receptive anal intercourse (Halperin, 1999). As for period effects, more recent surveys (2000s versus 1990s) yield higher prevalence estimates. For example, according to the U.S. National Survey of Family Growth one-third of women (ages 15–44) reported ever having had anal intercourse (Leichliter, Chandra, Liddon, Fenton, & Aral, 2007), and one U.S. study of female heterosexual STD clinic attendees (Satterwhite et al., 2007) reported prevalence estimates of 18% (1999–2000) versus 7% (1993–1995). As for cohort effects, younger survey respondents (college-age U.S. women) likewise report higher prevalence estimates than respondents from previous birth cohorts (Baldwin & Baldwin, 2000: 23%; Flannery, Ellingson, Votaw, & Schaefer, 2003: 32%). As for cultural effects, heterosexual anal intercourse appears to be more prevalent in some cultures outside of the U.S. (Jaeger et al., 2000). For example, household surveys from Brazil indicate that for 40% of rural and 50% or urban heterosexual couples it is a normal part of their sexual repertoire (Morris, 2004, p. 232). Therefore, contrary to Holland’s assertion, anal intercourse is not uncommon among heterosexual couples.
The highest prevalence of anal sex in the heterosexual population that Voracek and Fisher have come up with is 50% of urban heterosexual couples in Brazil. Assuming that the data are from a representative sample—they cited a book written for popular consumption by the general public—this still falls far short of all Private Media Group actresses indulging in anal sex and nearly 100%, if not 100%, of their pornographic videos and magazines featuring anal sex. In addition, the higher prevalence of anal sex in some non-European societies is of little relevance because the consumers of the products of Private Media Group are predominantly in Western societies.
Voracek and Fisher noted that I did not quantify what I meant by uncommon when I referred to anal sex among heterosexuals. I referred them to a study by Laumann et al.; they simply had to look it up They noted that in this study, based on a representative sample of Americans in the early 1990s, 13.2% women (ages 18–59) in cohabitations and 7.6% women in marriages reported any occurrence of anal intercourse in the last year. They then questioned whether this means that anal sex is uncommon among heterosexuals. Of course it is. Most heterosexuals never engaged in anal sex in the last year or ever, in sharp contrast to just about all Private Media productions featuring anal sex.
Note that most studies cited by Voracek and Fisher present the anal sex statistics in term of ever having indulged in anal sex. This is misleading when it comes to assessing an interest in anal sex among heterosexuals because many heterosexuals that have indulged in anal sex have done it only once or twice, out of curiosity, and then did not carry on with it because they did not find it enjoyable; some others indulge in it rarely. One can also safely assume that a number of heterosexual individuals have indulged in anal sex, not out of interest in it but because a partner insisted upon the act. Conversely, some heterosexual individuals may be desirous of anal sex but have not been able to find a willing partner, but as we shall see from the available data, the majority of heterosexual individuals in Western societies are not interested in indulging in anal sex.
Laumann et al. provided detailed statistics about the prevalence of anal sex. They reported that during the last sexual episode, indulgence in anal sex was reported by 0.9% of women aged 18-24, 2.4% of women aged 25-29 and 1.2% of women aged 30-32. These are the statistics that are more useful for assessing stable interest in anal sex in the heterosexual population. These figures are well below the 5% commonality threshold that the authors mentioned, which should be clarified. In general, if the probability that the observed difference between two samples due to chance factors involving sampling is less than 5%, then one can assume that the two samples are significantly different due to non-chance factors. Voracek and Fisher are welcome to compare any of the following: Private Media pornstars having ever indulged in anal sex in movies produced by this company vs. heterosexual women in general having ever indulged in anal sex, the probability of a Private media pornographic video picked at random depicting anal sex vs. the probability that a heterosexual individual indulged in anal sex during the last sexual episode; the prevalence of anal sex in the heterosexual population will be substantially lower. The highly relevant comparison is the statistic assessing how staple the depiction of anal sex is in Private Media videos vs. how staple a practice it is among heterosexuals—the difference is an order of magnitude.
Voracek and Fisher mentioned a rise in anal sex among heterosexuals in the 2000s compared to the 1990s. Again, their data sources present the statistic in term of ever having indulged in anal sex, which, as noted above, is misleading because many of these individuals have just tried it once or a few times. Let us consider some studies that have educated people about the risks of consuming illegal drugs and found increased drug use among those imparted knowledge of these drugs compared to those who were not, in spite of their being designed to discourage illegal drug use—some people will just try drugs to feel their effects; all that is needed is for someone to tell them about these drugs. In contrast, over the time period that the authors have cited an increase in the prevalence of anal sex, the mainstream media and secular education in schools and colleges have increasingly exposed the general public to a variety of atypical sexual practices, including anal sex, in a non-condemnatory, neutral or positive manner—hence, it should not be surprising if the proportion of people having indulged in some atypical sexual practices has increased; part of the increase would result from greater likelihood of reporting because of reduced social stigma rather than an increase in indulgence, but the ones more likely to previously have lied about indulging in the practice would disproportionately not be regular practitioners. The crucial statistic is the proportion of the population that regularly indulges in these atypical sexual practices or has a stable interest in engaging in these atypical sexual practices. With respect to anal sex, this proportion is considerably lower than the near-100% depiction of anal sex in the products of the Private Media group since its inception. So the question remains: why should we assume that heterosexual men who regularly consume Private Media Group products or those of a similar adult media company are representative of heterosexual men?
Voracek and Fisher cited Daniel Halperin on the claim that “about seven times more heterosexual women than homosexual men engage in receptive anal intercourse.” This is highly misleading. Women comprise roughly half the population, men who have sex with men comprise roughly 2.5% of the population and homosexual men comprise roughly 1-1.25% of the population. So there are about 20 times as many women as men who have sex with men, and 40 times as many women as homosexual men. So what is the bright idea behind comparing absolute numbers? Moreover, as seen from the statistics provided by Laumann et al., many women have indulged in anal sex only once or twice or indulge in it rarely. Consider the indulgence of women in anal sex during the last sexual episode, cited above. How do Voracek and Fisher think this figure compares to that among homosexual men?
Since Voracek and Fisher bring up homosexuals, there is one other issue we need to consider. Some people experience same-sex attraction but do not indulge in homosexual behavior, whereas some have voluntarily indulged in homosexual behavior without experiencing same-sex attraction—the proportion of individuals that have either voluntarily indulged in homosexual behavior or experienced same-sex attraction at some point of their lives is 15-20%, whereas only around 5% identify as homosexual or bisexual. Then, when we statistically analyze what, if any, categories underlie the distribution of same- and opposite-sex sexual interests, we infer two categories, best described as lifetime-exclusive heterosexuals vs. those that are not. In other words, some self-identified heterosexuals (who are predominantly heterosexual) do not belong to the heterosexual taxon. And then we note that that these self-identified heterosexuals, who do not belong to the heterosexual taxon, are also disproportionately indulging in anal sex; also this.
In short, there is no reason to believe that anal sex is a common practice among heterosexual couples, especially “truly heterosexual” couples.
All in all, the issues we discuss above in response to Holland’s comments seem to prove the truth in the maxim that data are the natural enemy of hypotheses (attributed by McGrath, 2005, to Goethe, but searched in vain by us in his Maxims and Reflections; von Goethe, 1998).
Data can confirm or refute hypotheses and hence it should not be stated that data are the natural enemy of hypotheses, but let us look at who has the data to back one’s claims.
The original article by Voracek and Fisher features:
- A short review of some of the existing literature on women’s physical attractiveness.
- It mentions problems with the ecological validity (real world applicability) of the existing literature.
- The assumption, based on market forces, that top-ranked porn stars represent the ultimate in female physical attractiveness for heterosexual men.
- An analysis of the reported height, weight and bust-waist-hip measurements of porn stars and nude models.
- The claim that for attractiveness evaluation of women’s bodies, androgenousness clues are more salient for the moving form and curvaceousness clues more salient for the posing/static form.
I have not disputed problems with the ecological validity of the existing literature. But I have argued that their assumptions behind the study are seriously flawed, their methodology is poor, their statistical data are poor, and there is a far better explanation of their find regarding a lower frequency of feminine women among porn stars compared to nude models.
Assumptions leading to the study
Voracek and Fisher considered market forces, but refuse to consider, even after my initial response, non-looks factors—a) enthusiasm with which they indulge in sexual activities; b) poverty, which affects willingness to indulge in more extreme/disinhibited sex practices—involved in the selection of porn stars and the achievement of porn stardom; market forces are selecting among the minority of women who are willing to pose nude and act in pornographic movies. They also assume that regular consumers of Private Media products are representative of heterosexual men.
Poor methodology and poor statistical data
Voracek and Fisher did not bother to actually look at the women they analyzed to assess whether some of their reported statistics are plausible, which is important in that the women were not measured by them or their associates. Two relevant measurements here are bust size and body weight. Even after my pointing out, Voracek and Fisher are unwilling to acknowledge that a visually observable high frequency of breast implants among the women in their [and similar] samples makes the bust measurements of limited utility and that their body weights are often underreported. Hence, some of their statistical data are poor.
An alternative explanation of porn stardom
Voracek and Fisher were initially presumably unaware, and after my pointing out still refuse to address, that physical masculinization, a factor powerfully associated with diminished physical attractiveness of women, tends to increase disinhibition, promiscuity and interest in atypical sex practices, thereby alternatively explaining both the lower frequency of femininity/curvaceousness among porn stars compared to nude models and invalidating the assumption that top-ranked porn stars must necessarily represent the ultimate in female physical attractiveness to heterosexual men.
- Baldwin, J. I., & Baldwin, J. D. (2000). Heterosexual anal intercourse: An understudied, high-risk sexual behavior. Archives of Sexual Behavior, 29, 357–373.
- Fisher, M. L., & Voracek, M. (2006). The shape of beauty: Determinants of female physical attractiveness. Journal of Cosmetic Dermatology, 5, 199–204.
- Flannery, D., Ellingson, L., Votaw, K. S., & Schaefer, E. A. (2003). Anal intercourse and sexual risk factors among college women, 1993–2000. American Journal of Health Behavior, 27, 228–234.
- Halperin, D. T. (1999). Heterosexual anal intercourse: Prevalence, cultural factors, and HIV infection and other health risks, Part I. AIDS Patient Care and STDs, 13, 717–730.
- Holland, E. (2009). Pornographic actresses are a poor choice for assessing what men optimally prefer in women’s looks: Comments on Voracek and Fisher (2006). Archives of Sexual Behavior. doi:10.1007/s10508-009-9470-1.
- Jaeger, A. B., Gramkow, A., Sørensen, P., Melbye, M., Adami, H. O., Glimelius, B., et al. (2000). Correlates of heterosexual behavior among 23–87 year olds in Denmark and Sweden, 1992–1998. Archives of Sexual Behavior, 29, 91–106.
- Laumann, E. O., Gagnon, J. H., Michael, R. T., & Michaels, S. (1994). The social organization of sexuality: Sexual practices in the United States. Chicago: University of Chicago Press.
- Leichliter, J. S., Chandra, A., Liddon, N., Fenton, K. A., & Aral, S. O. (2007). Prevalence and correlates of heterosexual anal and oral sex in adolescents. Journal of Infectious Diseases, 196, 1852–1859.
- McGrath, J. J. (2005). Myths and plain truths about schizophrenia epidemiology: The NAPE lecture 2004. Acta Psychiatrica Scandinavica, 111, 4–11.
- Morris, D. (2004). The naked woman: A study of the female body. London: Jonathan Cape.
- Saad, G. (2007). The evolutionary bases of consumption. Mahwah, NJ: Erlbaum.
- Salmon, C., & Symons, D. (2001). Warrior lovers: Erotic fiction, evolution and female sexuality. London: Weidenfeld & Nicolson.
- Satterwhite, C. L., Kamb, M. L., Metcalf, C., Douglas, J. M., Malotte, C. K., Paul, S., et al. (2007). Changes in sexual behavior and STD prevalence among heterosexual STD clinic attendees: 1993–1995 versus 1999–2000. Sexually Transmitted Diseases, 34, 815–819.
- Swami, V., Arteche, A., Chamorro-Premuzic, T., Furnham, A., Stieger, S., Haubner, T., et al. (2008). Looking good: Factors affecting the likelihood of having cosmetic surgery. European Journal of Plastic Surgery, 30, 211–218.
- Szabo, C. P. (1996). Playboy centrefolds and eating disorders: From male pleasure to female pathology? South African Medical Journal, 86, 838–839.
- Tovée, M. J., Mason, S. M., Emery, J. L., McCluskey, S. E., & Cohen-Tovée, E. M. (1997). Supermodels: Stick insects or hourglasses? Lancet, 350, 1474–1475.
- Voeller, B. (1991). AIDS and heterosexual anal intercourse. Archives of Sexual Behavior, 20, 233–276.
- von Goethe, J. W. (1998). Maxims and reflections (E. Stopp, Trans.). London: Penguin.
- Voracek, M., & Fisher, M. L. (2002). Shapely centrefolds? Temporal change in body measures: Trend analysis. British Medical Journal, 325, 1447–1448.
- Voracek, M., & Fisher, M. L. (2006). Success is all in the measures: Androgenousness, curvaceousness, and starring frequencies in adult media actresses. Archives of Sexual Behavior, 35, 297–304.
- Voracek, M., Reimer, B., Ertl, C., & Dressler, S. G. (2006). Digit ratio (2D:4D), lateral preferences, and performance in fencing. Perceptual and Motor Skills, 103, 427–446.