Sunday, December 17, 2006

Dept of dog bites man: hot chick geeks   posted by agnostic @ 12/17/2006 08:46:00 PM
Share/Bookmark

Oh, of course it's surprising to see a hot girl blathering about science fiction in a wine bar!

Lately Razib's been taking a lot of heat for posting a totally harmless observation that he saw a girl, whose physical attractiveness was in the top 2%, prattling on about science fiction books in a locale where people are supposed to put on their best "I'm cool" face. My initial response is here, but there were several more responses to the original post that completely missed the mark (responses so far from: retrospectacle, bushwell, aetiology, pharyngula, ethics in science, and zuska). So, I figured I'd post a more elaborate rejoinder, chock full of logic and data to counteract the largely fact-free emoting thus far. In brief, Razib was right to be surprised -- surprise!

Rationality and facts below the fold...

Razib's original shock was the result of applying Bayes' Theorem to a real-world situation. Let Hf = "a person X is a hot female," where "hot" indicates top 2% status, and S = "a person X is a hardcore sci-fi geek." Razib heard some babbling about sci-fi first, and then made an unconscious guess / assumption about how probable it was that the speaker was a hot girl. In the above symbols, he computed P(Hf | S) = P(Hf ^ S) / P(S). Now, P(Hf) is easy to calculate: in the 15-64 age group, the male : female ratio in the US is 1 : 1 (cite), and Razib said the girl was in the top 2% for looks, so P(Hf) = 0.01. Not surprisingly, there are no easily found data on the percentage of sci-fi geeks who are also hot girls. I did, however, find lots of data on the sci-fi geek / female intersection. In any event, the only case where such Bayesian reasoning would have been not empirically worthwhile is where hot girl status and sci-fi geek status were independent. We'll look first to see if girl status (hot or not) and sci-fi geek status are independent, and then look at how introducting hotness would affect the results.

I know it will come as a Bayesian shock those who just landed here from Mars, but sex and sci-fi geek status are not independent: males are clearly overrepresented. This is true not only for the readers and consumers of sci-fi but for the producer-creators and editors as well. It is even more true for closely related fields like role playing games (RPGs) like Dungeons & Dragons. This is so obvious that I enumerate the data at the end, where "dog bites man" material belongs. Let's first ask why the observation offended so many? Perhaps on hearing that someone correctly applied Bayesian reasoning to accurately predict whether female X was hot or not given her hardcore sci-fi geek status, she felt she would be assumed to be non-hot by others if her nerd status were known. However, this is no cause for concern, since the most accurate guess we can hazard about a girl's looks is based on just looking at her. If female X is top 2% hot, she will be in no jeopardy of being thought average or ugly simply because she is a huge sci-fi geek.

Another complaint was that Razib was really trying to say he was surprised that an intelligent girl was hot. Of course "intelligent" and "sci-fi nerd" aren't interchangeable, despite overlap. Compare: "I overheard X talking about their business trip to Tuscany, and how well the PR campaign with Vespa was going -- you'll never guess, but X was a hot girl!" In fact, I wouldn't be surprised at all. The median IQ level of PR execs and sci-fi fans is likely not appreciably different, but I'm open to reading data to the contrary. This situation produces no surprise since females are well represented in PR, including quite attractive ones. As Dennis Mangan pointed out, attractive intelligent girls are more likely to join PR-type fields since their good looks are more of an asset there compared to geeky occupations. No one who's been to college will be surprised that there exist smart females, nor that the male-female ratio is close to even as long as one doesn't stray too far from the mean (say, within 1 SD). What's shocking is that this girl had atypical interests even for intelligent females, rather than say, express a fascination with the haute cuisine of the diverse French provinces.

Yet another complaint was that Razib's correct use of Bayesian reasoning would perpetuate a particular stereotype, i.e. that sci-fi geeks and hot people don't overlap much. This stereotype might keep away prospective sci-fi geeks if they were also hot: if the sci-fi community has a deserved reputation for being disproportionately unattractive loners, the hot people might want to steer clear of this hobby. Thus will the stereotype perpetuate its existence. But what, really, is the alternative -- to lie to prospective geeks, or perhaps just keep mum? If so, the prospective geeks will figure out sooner or later that joining a Star Trek club will kill their reputation and leave posthaste, so discussing the stereotype openly will not result in fewer geek members than hiding or denying the stereotype. How did the stereotype begin, then? Did someone roll the dice and the result came up that "sci-fi nerds shall consist almost entirely of non-hot people?" Hogwash. There's something about the hardcore sci-fi existence that appeals to a certain group of people, damn few of whom are hot -- perhaps because sub-hot loners turn to certain hobbies for want of a coterie of same-sex buttkissers and opposite-sex conquests, in virtue of not being very physically attractive while growing up. Similarly, death metal fans are not randomly drawn from the population, and they're biased such that we'd be surprised if we met a Miss Congeniality / Southern Belle type at a death metal concert.

Considering the various complaints, I can't find any reasonable objection to Razib's observation. Again, it must be that one of the faulty objections discussed above, or similar ones, is the real reason that so many got so miffed so easily. It can't be that they actually thought a hardcore sci-fi geek was equally likely to be male as female, let alone introducing the hotness variable. But just to be thorough, here is some data I discovered with ease by Googling. The results won't shock, so if you take them for granted, you'll only be bored by reading further. They are included only to show that dishonesty in the service of calumny will not be tolerated by those of us who still care about keeping "oughts" out of the process of empirical observation. We will call "bullshit."

Starting with the creators and editors of science fiction, here is a long list of stats. Let's consider a few examples. Men and women make up 58% and 38%, respectively, of the Science Fiction Writers of America, and 79% and 21% of the Horror Writers of America. Men and women account for 74% and 26% of professional-market editors for sci-fi-related literature. Men are overrepresented among the winners of nearly all major sci-fi literature awards, whose winners are typically 75% male. (The sole exception is the radical feminist, ideological Tiptree Award.)

As for reader preferences, this study of middle schoolers found notes:

The females reported a stronger interest in Romance, Friendship, Animal Stories, Adventure, and Historical Fiction, while the males reported stronger preferences for the categories of Sports and Science. In addition, the results indicated that the male respondents had a stronger preference for non-fiction than did the female respondents.

This report on the preferences of high schoolers has two tables (Tables 8 and 9) that summarize what genres boys and girls indicate as their favorite. Three surveys were done: one in 1982, another in 1990, and a final one in 1997. The authors don't report all data, just those genres that garnered at least 10% of the responses. Across all three surveys, "science fiction" and "fantasy" never met this threshold among female responses, while "romance" and "mystery" did all three times. At the same threshold, boys consistently liked "adventure" and "sports" -- "science fiction" made the cut twice, and in the year it didn't, "fantasy" did. If we collapse "science fiction" and "fantasy," then boys consistently like such books. The authors also note yet another special case of greater male variance: "Boys show more diversity in the genres they like," and indeed a comparison of Tables 8 and 9 shows that male preferences are more evenly distributed across various genres than female preferences, which are more bunched around the most popular genres. That jibes with most people's personal experience: when you're on the metro or subway, take a look and you'll notice that females tend to conform to a single norm (recently, The Da Vinci Code); whereas males are more fragmented, preferring the manly genres of war, sports, and cars, as well as wimpy geek fare such as sci-fi and popular science.

This review from the American Library Association notes that:

In terms of gender differences, boys showed a markedly higher interest for science, science fiction, nature, and magazines than girls, while girls' responses were considerably higher than boys' for mystery, historical fiction, and friendship and families. The small sample size [N=23; agnostic] makes these differences slightly suspect, but the similarity between this and other studies is noteworthy.

Continuing:

Greenlaw (as cited in Todd 1988) worked with slightly older children (grades 4 to 12). She gave a questionnaire to 1,240 students, asking what they liked to read and how they chose their books. The data were classified according to genre, with the results showing an overall preference for mystery, fantasy, jokes/humor, adventure, and sports. There were marked gender differences as well. Boys chose science fiction, sports, how-to-do-it, and jokes/humor more than girls, while girls more often selected mystery, romance, biography, fantasy, and—by far the greatest gender distinction—poetry.

Most of the other studies reviewed corroborate the main story, but their publication date was pre-Women's Liberation. The fact that they largely agree with the picture afterward shows that changing sex roles in society will not make girls freely choose to read sports, science fiction, and war stories with the same frequency as boys. Badgering and compulsion only are suitable solutions to engineering an equality of preferences.

Nor is the pattern unique to the US. Here is a review of studies from New Zealand, whose apt sub-title is "Identifying the reasons why female high-school students do not like reading science fiction books." If we look at the oldest students (Form 6 = 16-17 y.o.s) in order to best gauge adult preferences, Appendix Item 5 shows that science fiction books rank 2nd of 13 in popularity among boys but 11th of 13 among girls.

Branching out from books, this review on movie preferences shows that males and females are also split when it comes to sci-fi on celluloid. Subjects (264 males, 296 females) were asked to list up to 15 of their all-time favorite movies; the movies were then grouped into Top 25 and All Movies categories. Table 7a shows the Top 25 movies as judged by males and females. The largest sex difference is in "romance" (31% of Top 25 movies rated by females but 4% of those rated by males), and the next largest is in "science fiction: these made up 20% of Top 25 movies rated by males but 4% of those rated by females. As for all movies listed by males and females, "science fiction" shows the largest sex difference: these account for 10.7% of male responses but only 4.6% of female responses.

I'll confess that I'm not very into sci-fi or fantasy, and as I've done this research, I got the hunch that "science fiction" may be a bit broader than non-afficionados assume. It seems like the sex differences should be even greater than the admittedly large gap, but perhaps there are less geeky sub-genres of sci-fi / fantasy that appeal to girls. So I looked at the ultra-geeky sci-fi / fantasy hobby of role-playing games (RPGs) like Dungeons & Dragons. Surprisingly, this was the only area that had easily retrievable market research reports. This summary of market reports confirms that hunch: the study with the most controlled sampling found that 81% of RPG players are male (two less controlled studies found 77% male and 91% male). In one of the less controlled studies, it seems that females were much less likely than males to perform the uber-geek role of DM (I looked it up too; it means "dungeon master," the person who directs the course of a particular game):

Although 9% of the respondents were female overall, just 4% of the respondents classified as DMs were female. Furthermore, 30% of female respondents indicated that they never DM, while just 7% of male respondents indicated this. For the most part, DMing currently seems to be a male occupation.

Lastly, what about the sex make-up of blogs on niche topics? Here is a recent survey by Bloggasm (N = 302 -- pretty big sample for niche bloggers!). Overall, niche bloggers are 69% male. Males exceed this value for the following categories (some of these excessive values may not be significant, including sci-fi): comic books, gadget / tech, music, gay / homosexual, movie / film / TV, sports, law / legal, sci-fi / fantasy / horror, conservative, military, poker, science, and cars. Though males were below 69% for the other categories, females were only clearly overrepresented in the following categories: sex, gossip / fashion, regional interest, feminist, and food. I'll leave it to the attentive reader to discern the patterns.

Thus, investigating the problem from multiple angles converged on a single, obvious conclusion: males are much more likely than females to be interested in geeky hobbies like sci-fi. Anyone for whom this is a novel insight is simply not a good observer of the world around them. Considering that the face that launched a thousand rants was top 2% in good looks only makes the case more surprising: even if hot people, controlling for sex, did not shy away from sci-fi (a dubious assumption), we would still be left with one heck of a statistical rarity.