So you're saying that when looking for a correlation between age of the father and prevalence of autism of the child, you need to account for the confounding factor of heritability since among a population of fathers who are older, there will be a higher prevalence of autism than in a population of fathers of an even distribution of ages (the general population) due to the social factor.

Sounds like a reasonable and logical take. Are you aware of any studies that show this or is it your own hypothesis?