Well, the metric is a debatable point. The fact that the T-test is inapplicable to the data in question is not a debatable point. So naturally the critics would choose to argue about the use and abuse of T-tests.

But you're certainly right that the entire exercise is kind of silly, because the metrics we are measuring against are highly dubious in the first place. So even if there had been a signal, it would not have proven the case, only given cause for further investigation.

OTOH, that turns out not to matter, because even in the most generous possible interpretation of the data - according to the metric Vladimir proposed - there is no signal. Full stop. No deconstruction of metrics necessary.

- Jake

Friends come and go. Enemies accumulate.