A few months back I ran a survey that included questionnaires on autism, schizotypy, depression/anxiety1, mating orientation2, and sexual behaviour3. Most responses were collected via twitter, receiving a total of 147 completions (~75% male). Not a huge sample size, but would still be pretty good for a psychology study, and we might be able to squeeze some value out of it. While admittedly this sample is rather unrepresentative of the overall population, this is less critical for analyzing associations than it is for establishing baseline rates. As long as the associations hold relatively constant across different subgroups in the population, meaningful conclusions can still be drawn.
The autism quotient I employed was the AQ-Short, a 28-item measure of autistic traits across five subdomains: social skills, routine, switching, imagination, and numbers/patterns. A social behaviour factor subsumed the lower-order factors aside from numbers/patterns (which only weakly correlates with them).
I also included SPQ (Schizotypal Personality Questionnaire), but it was mostly a wash. The ‘interpersonal’ factor correlated highly with social AQ (.74 with the social skills subscale) and didn’t explain additional variance, and the others didn’t correlate with anything sexual, so we’ll just focus on the autism results here.
First, the correlations:
Most of the relationships seem to be in the expected directions. Social AQ correlated negatively with relationship status, past year and lifetime sex partners, and positively with depression and anxiety scores and autism diagnosis4. It was also negatively correlated with short- and long-term mating orientation, while an autism diagnosis predicted a lower short but not long-term orientation.
Despite this, social AQ also correlated with sexual frustration, and this relationship was strengthened when mating orientation was included in the model. It was also robust to the inclusion of depression/anxiety score.
Linear regression models revealed significant effects of social AQ on men’s number of past year sex partners (negative), age of sexual debut, sexual frustration5, and masturbation frequency. There was a sex interaction effect for sexual frustration, with AQ predicting it more strongly (and exclusively) for men. The sex interaction effect didn’t quite reach significance for age of sexual debut, but there was only a statistically significant effect when women were excluded.
A negative binomial model is more appropriate for (highly overdispersed) count data such as lifetime sex partners. This found a significant effect (IRR = 0.504, p < 0.001), with a 1 SD increase in social AQ predicting a 50% decrease in the lifetime sex partner count.
Comparing the means and proportions of various variables of interest between men by virginity and sexlessness in the past year, we find that virgins and sexless men scored significantly higher on social AQ, depression and anxiety scores, and both short- and long-term mating orientation, with mostly large effect sizes. Virgins were also more likely to report an autism diagnosis, with a medium effect size. No significant differences were observed for depression or anxiety diagnoses.
In logistic regression models controlling for mating orientation, age, and sexual orientation, the effect of social AQ remains significant and large, with a 1 SD increase in social AQ predicting an increase in the odds of virginity, sexlessness, and never having been in a relationship by 2.2–3.2 times. Given a baseline virginity rate of 50%, a 1 SD increase would increase the estimated virginity rate to 76%. Virginity and never relationship were robust to the inclusion of anxiety and depression score, but sexlessness wasn’t to depression6.
A sex interaction effect indicating the association was stronger for men emerged for virginity (β = −1.24, p = 0.024) and sexlessness (β = −1.15, p = 0.039), but not never relationship (β = −0.7, p = 0.274).
A significant effect of autism diagnosis was also observed, with an odds ratio of about 5 for virginity, but only after including depression diagnosis as a covariate.
Another way to break down the data is to compare men high and low in social AQ. The top 20% in social AQ (higher in autism traits) reported an average of 2.7 lifetime sex partners (boosted significantly by one guy hitting the cap of 50 I set, but this might be a BS response - he also said he lost his virginity at 30+ and reported 126 partners) and 0.4 sex partners in the past year. For those who were in the bottom 20% of social AQ, the averages were 10.2 and 1.9, respectively. Additionally, 82% of the top 20% were virgins and 91% had no sex partners in the past year. For the bottom 20%, these proportions were 23% and 45%, respectively.
This brief analysis further underscores the value in looking at autism not just as a dichotomous variable but a continuum of traits. Somebody who is ‘sperg-adjacent’ will tend to have substantially different outcomes to a ‘giga-NT’.
There might be other angles worth exploring in this data, but I think this has covered the main ones.
Some assume that these outcomes simply reflect lower interest among autists, but higher levels of sexual frustration and masturbation frequency among more autistic men challenge this assumption.
Some have also proposed that autism supports a reproductive strategy that prioritizes parental investment, with higher focus, attention to detail, and technical skills facilitating resource acquisition and higher status in their profession. At least when it came to social AQ—whose traits seem more relevant to mating outcomes—it predicted lower STMO, but also LTMO, which seems at odds with the theory (not that this will perfectly capture an underlying reproductive strategy). Autism diagnosis also predicted lower STMO, but had no effect on LTMO. It’s possible that a history of rejection, anticipating failure due to poor social skills, aversion to touch, etc. could reduce mating interest among more autistic people, but this question won’t be resolved here. Regardless, more autistic people were less experienced even after controlling for this lower mating interest.
Incels, as morphological reductionists, assert that any association between psychological characteristics and sexual or romantic outcomes are fully mediated by physical attractiveness, but we’ve seen how various personality traits and psychopathologies are effectively orthogonal to physical attractiveness. From what research is available in regard to autistic traits, it appears to be the same story. Ward & Scott (2018) found that facial morphs of men in the top 20% of AQ were rated as lower in mental health and higher in masculinity than those in the bottom 20% of AQ, but not as less attractive.
This article can be seen as a primer for the next one, which will delve deeper into autism and its intersection with the incel community and dating.
I made sure to use shorter scales for autism and schizotypy, but in hindsight these and the mating orientation questionnaires were still needlessly long, and it would've probably gotten more completions if I used even shorter scales or just narrowed the scope down to autism.
The effect size wasn’t large however; it seems more due to a good number of respondents high in AQ not being diagnosed more so than those low in AQ being diagnosed.
Despite social AQ negatively correlating with short- and long-term mating orientation and these correlating positively with sexual frustration, this relationship was present even without controlling for these factors.
This could potentially be due to sexlessness/singleness leading to higher depression in people higher in social AQ rather than the reverse.
"(boosted significantly by one guy hitting the cap of 50 I set, but this might be a BS response - he also said he lost his virginity at 30+ and reported 126 partners)"
Easily explained by a 30 year old Silicon Valley engineer finally plucking up the confidence to hire a prostitute, and having the financial means to feed the resulting addiction
> I made sure to use shorter scales for autism and schizotypy, but in hindsight these and the mating orientation questionnaires were still needlessly long, and it would've probably gotten more completions if I used even shorter scales or just narrowed the scope down to autism.
True autist wouldn't be deterred by a long questionnaire.