correllation is not causation?
agggh! Read this: The study controls for teachers' reports of aggression and impulsivity at age 10, the child's gender, and parenting style.
Do you think scientists with >10 years training know less about statistics than you? They actively try to exclude other causes, which is what "controls for" means. Any other ideas for root causes that do not include those controlled for? Or were you just trying to be smart with a nice one-liner because it worked so well for others?
It is entirely possible that he *does* know more about statistics than these researchers. As someone has already observed, the statistics behind this research are essentially flawed by having too small a pool size of violent criminals. I could point out a couple of other flaws in the study, including the question of how effective their approach to controlling for aggressivity is, but that seems unnecessary.
More importantly, correlation is not causation! It doesn't matter if you've corrected for a few factors. Even if you corrected for every possible factor, the possibility would remain that both events had some shared cause. There is *never* a time when correlation alone, without other evidence, can serve to establish causation.
There are a lot of people who seem to implicitly trust that someone who calls himself a scientist knows what he's doing. The truth is, the social sciences are mostly filled with people who struggled in their basic statistics courses, and never really learned the stuff. As a math major, I often tutored people in statistics classes, and the people in the social sciences statistics classes never had the sort of grasp of the material needed to do anything useful really. Having never understood statistics, they mostly learn by mimicking their research advisors, who never learned statistics either.
There are some researchers out there doing good work, but they are vastly outnumbered, and your default presumption of competence is totally unwarranted.