How do we know when an observed effect is real or spurious?
Some people say, "A p-value=0.05 is not very much evidence against the null hypothesis." Well then, how much evidence is it?
pr(B|A) ≠ 1 – pr(A|B). Why do we act like it ?!
For too long statisticians have been peddling pr(data|hypothesis) when scientists [indeed all of us] want pr(hypothesis|data).
What if AI in healthcare is the next asbestos?
It is easy to get an answer; it is very difficult to assess the quality of that answer.
Protect yourself from polio ... don't eat ice cream!