Sanjait:

What I'm saying is that hetero and homo are not just social constructs. . . . In modern times, the gay/straight dichotomy seems to be very useful, and encompasses most of the population very neatly. That's why using it as part of a statistical model isn't an assumption of universality nor the presumption that a category of "other" doesn't exist. . . .

I seriously question the assertion that the dichotomy "encompasses most of the population very neatly." By that, do you simply mean that it encompasses most of the population as self-identified? If so, I'd agree with you -- but that does nothing to undercut the idea that rigid orientation classifications are largely a social construct.

The problem is that while most people do identify themselves into neat categories, their actual behavior tends to defy clean demarcation. A good example is the prison population, a point that was raised by another poster a couple of weeks ago. Many inmates would have been classified as 100% heterosexual before they entered prison, and yet, during their time there, they often adapt same-sex interests. Likewise, some studies have revealed that even the most self-professedly heterosexual males exhibit some degree of sexual response to same-sex visual stimuli.

(And, anecdotally, it is often observed that boundaries become more flexible once alcohol is introduced. On the flip side, I've personally known 100% self-professed gay men who admitted they had some degree of sexual interest in women, but didn't want to pursue the idea because they felt it threatened their social identity.)

My point is that if one pries into behaviors and responses on an individual level, the clear notions of gay and straight begin to crumble. One could postulate some complex reason why a straight person sometimes likes the same sex and vice versa -- but wouldn't the simpler explanation be that sexuality is basically fluid and that people tend to conform to social expectations?

Now, if you like, you can categorize most people farily easy by their majority inclination. Sure, that's easy -- most are not 50/50. But I would question the utility of such an exercise, since it's just a matter of tweaking the definitions. You know, all numbers are whole if you round to the nearest integer. :-)