There was a radiolab episode awhile ago about a study in Sweden. They had records going back centuries, of family trees and each years harvest. They found that grandchildren of people who starved at the age of 9 had significantly better life outcomes.
That sounds like a textbook case of p-value hacking and hypothesis fishing. Is there a solid statistical analysis on the space of hypotheses considered and the size of effects measured?
Yeah, this was the first thing that came to my mind, but I couldn't remember where I read this (I thought it was posted on HN recently), but I guess I heard it on Radiolab. I also found a more recent study about the same phenomenon:
Indeed, there seems a casual assumption in some pop accounts and discussions that such 'inherited stress' must be bad, a continuation of the ancestral tragedy. In fact some evidence, like that Swedish data, suggests the opposite can also be true.
Fascinating. Also entirely consistent with genetic determinism by the usual natural selection story - 9 year olds who survive starvation have bad-ass genes, etc.
Dying vs. merely damaged is a red herring. Interesting fact re: 9 yo => making eggs/sperm - good comment overall. "Effect wouldn't be as strong" why?
Re: 'regardless of age': you can't rule out genes that influence near-starvation thrive/fail differentially at young vs teen ages. Nevertheless I agree that this is suggestive evidence for "it's not just regular genetics". Good analysis if it indeed applies to this case!
http://www.radiolab.org/story/251885-you-are-what-your-grand...