well for a quantitive study you'd need to track data about age and diagnosis of diagnosed and undiagnosed autistic as well as other fathers in quiet a large amount, assuming that about 3% of the population is autistic. i'd be very sceptical about the accuracy of such data. you could probably well show the causality in qualitative studies. but then the results are rather anecdotical and certainly somehow biased to draw conclusions about Homo sapiens s. at large.

given that i live at a place where two generations back people were mass-murdered based on their descendence, also related to psychiatric diagnosis, and that i'm trying to oppose Instrumentelle Vernunft, i'm quiet happy that i assess there's hardly a proof for that thesis.