One thing I was thinking about which could affect the correlation they found between father's ages and the risk of autism is that there tends to be a higher rate of parents with advanced educations (Masters, PhDs, MDs) than in the general population. And aside from these advanced educations taking more time to complete and then get started in a career, many couples may simply be more married to their careers, especially early off. So by the time the clock starts running down, they only have time for one child (decreasing the chances they'll have a multiple autism incidence family).
Also, in couples, males are often older than females (not always, but often). So by the time they have children, the male of the couple may have reached that 40+ mark while the female may not have.
Plus, if there is an increased risk of Polycystic Ovarian Syndrome (as proffered by Baron-Cohen's team), then this could affect fertility and number of children (PCOS is the #1 cause of infertility in women). For my mother, it wasn't until 14 years after they married that they finally had me and she had to have some fertility treatment.
If she hadn't needed treatment, who knows, I might've had a whole slew of siblings and they'd've had children younger.
And finally, with our poorer social skills, it may simply take us longer to find a mate and settle down (if we ever do).
Those were a few of the possibilities I thought of which could be affecting the correlation.