Exactly. The graph showing that older coders' answers are not significantly better than those of younger coders is a case in point: maybe the good older coders are too busy actually coding to spend time answering questions on StackOverflow.
One of alan's points is that it is incorrect to assign behaviors of "older Stackoverflow users" to the universe of "older developers".
The population of "older Stackoverflow users" is not randomly drawn from the population of "older developers", and nothing is put forward to claim that the former is representative of the latter, so you cannot make this assumption.
