> Because the uncontrolled nature of the observations means evidence cannot be reliably collected in a way that separates one cause from any number of unintended causes and side effects. The only way to get around these obstacles would be to feed one group of people doses of lead and compare them to another in a controlled setting, or isolate two groups of people, give one group the right to birth control and abortions, deny the other group those rights, and keep them from finding out about each other.

Uh, no, statistical controls are a real tool of real science. While laboratory conditions are in some ways (ethics aside) ideal, they aren't a prerequisite for real science.