Introducing the problem
Asexuality is defined as the lack of sexual attraction. From a bayesian perspective, this makes it harder for a person to conclude that they are asexual. To demonstrate that you are a person who has a stable pattern of attraction towards one or more genders, all you have to do is experience attraction. To be really sure that it's "stable", you could experience it two or three times. But to demonstrate that you do not stably experience attraction, the best you can do is sit around and wait.
Asexuals are not the only people with this problem. Straight people have the same problem demonstrating to themselves that they are not bisexual. Gay and lesbian people do too. This can be illustrated in the Storms' Model diagram (published in 1979).
In this diagram, the purple arrows illustrate "hard" Bayesian problems. That is, if you're in the group at the base of an arrow, it may be "difficult" to demonstrate that you are not actually part of the group where the arrow is pointing, because you'd have to sit around and wait a while.
The orange percentages show, roughly, the number of people in each group. I know the estimates of homo- and bisexual people are pessimistic, but the estimate of asexuals is in accordance with the literature. What puts asexuals in a unique situation compared to the other three orientations, is that there are just so few asexuals as compared to heterosexuals. That means that any Bayesian estimate of probabilities will initially favor heterosexuality.
So what does the Bayesian calculation look like? (If you aren't familiar with Bayes' theorem, you should look it up. Alternatively, skip to "Result".)
Technical Section on Bayes' Theorem
Let A be the proposition that a person is asexual, while H is the proposition that they are heterosexual. E is the proposition that so far they haven't experienced sexual attraction.Result: One may conclude that they are more likely to be asexual than heterosexual if they are such an age that more than 99% of heterosexuals would have experienced sexual attraction.
P(A|E)/P(H|E) = P(A)/P(H) * P(E|H)/P(E|A)
P(A)/P(H) is the prior odds ratio between being asexual and heterosexual. It's about 1/97, because there are about 97 times as many heterosexuals as asexuals. P(A|E)/P(H|E) is the new odds ratio, in light of the fact that a person has not yet experienced sexual attraction.
P(E|A) is the probability that an asexual would not have experienced sexual attraction yet. I think it is possible for asexuals to experience sexual attraction without having a stable pattern of sexual attraction, so I think P(E|A) is less than one. But let's approximate it as one.
P(E|H) is the probability that a heterosexual person would not have experienced any sexual attraction yet. When P(E|H) is less than 1/97, one may conclude that they are more likely to be asexual than heterosexual.
Applying Literature on Sexual Attraction
We can estimate when heterosexual people experience attraction with research! Sciatrix recommended a 1996 review by McClintock & Herdt, which says that the average age of first sexual attraction is age 10. According to one of the references, the standard deviation is 3.6 years for gay and lesbian people. I glanced at another reference, and it reported a standard deviation of 4.7 years for lesbian women and 3 years for heterosexual women. Since heterosexuals are the relevant group, I will go with the 3-year estimate.
Let's start by assuming that the ages are normally distributed. There's a simple expression for the probability that a heterosexual at age y will not have experienced sexual attraction yet:
0.5 * erfc( (y-10) * 0.2357 )
The tricky thing is that it is not safe to assume a normal distribution! A normal distribution is the result when there are lots of random factors, each of which contributes a small delay or a small advancement. But lots of real phenomena do not occur in normal distributions. And the differences in the distributions are very important here, because we're interested in the tails of the distributions. It's important to know, what is the skew of the distribution? What is the kurtosis? We may never know.
If I were to build a toy model (as I've been trained to do as a physicist), I'd say that at some age, sexual attraction "turns on" for a person (though this age may be different for different people). After this point, they have a certain probability per unit time of experiencing sexual attraction. This model predicts an "exponentially modified Gaussian distribution", and it would only increase the age necessary for a person to conclude they are asexual. But perhaps I'm getting too much into sketchy territory. Let's take a step back.
Problems with this analysis
You may be saying "Wait, 17? But aren't there lots of people who know they're asexual before that? How can that be right?" Indeed, based on a community survey, about 10% of the online community is under 17.
Those 10% aren't wrong. Actually, it's my analysis which is wrong. In so many ways.
The first problem is that people could have poor recollections of when they experienced sexual attraction. They're being asked many years after the fact. Poor recollection would probably inflate the variation between individuals.
The second problem is that I believe that there are gradations of sexual attraction. It's likely that it ramps up over a long period of time. People just recall the one event that felt most like a breakthrough, but this could occur at different times during the ramp-up for different people. This would also inflate the variation between individuals.
The third problem is that "so far no sexual attraction" is not the only possible evidence that a person can be asexual. If, as I believe, there is a ramp-up of sexual-attraction-like experience, then asexuals might be hyperaware of this ramp (since asexuals notice they're different from other people). If a person isn't even experiencing the beginning of that ramp, that's additional evidence for being asexual.
Oh, and don't forget that we don't really know the distribution of first sexual attraction experiences.
Feel free to point out further problems in the comments.
Conclusions
I think the Bayesian analysis I presented above is what a lot of questioning people want to exist. They'd like an objective tool, something to tell them what they really are. Preferably something involving MATH and SCIENCE. But this is the best you can get, and it still won't tell you much. There's a broader lesson in here about relying on math and science to describe one's experience. It's really hard!
But I'll at least say this: asexuals really are in a unique epistemic position compared to other orientations. It is harder to demonstrate asexuality than any of the others. It is unclear just how hard that is. For some people it may be harder than for others. Some people might know fairly early on, while others may be unable to know until they're older.
I am of the mind that we should accept that asexuality involves enhanced uncertainty. It is okay for people to be uncertain. It is okay for people to identify as one thing, and later change their mind. In fact, we should advocate a better etiquette of uncertainty. Don't assume that an identity is a commitment. Don't treat identity shifts as betrayals or marks of shame. Don't condescend to people as if you know better than them who they are. (Pretty much this is all stuff asexuals already do.)
