Yes, I see you did use the word “proven”. That does lesson the force of my comment but not by very much. I’m curious actually what you mean by proven outside a mathematical context. In this context, the correlation is extremely strong, and there’s no obvious alternate hypothesis. I’m not completely sure what you consider in this context to be an acceptable amount of evidence, but the medical consensus seems clear, and the links Gwern gave show that the correlation exists even when one tries to control for other variables.
I think he wants randomized controlled trials. Which of course no ethics board would ever approve because avoiding aspirin in kids is not that hard or expensive, and the cost of confirming it would be too high.
(“You want to test whether slamming your face into the wall causes pain, and doesn’t just correlate with it? Why on earth?”)
We have a plausible mechanism (confirmed by massive amounts of science) by which slamming your face into the wall causes pain; for the link between Aspirin and Reye’s syndrome there is no such plausible mechanism.
Well, at least in the drug world it is more convincing if you have a rational mechanism by which a drug could cause some effect.
A plausible alternative hypothesis is that children who get sick sometimes are developing Reye’s syndrome and take Aspirin. The problem with epidemiological studies is that it is IMPOSSIBLE to control for other variables. This is why randomized controlled trials are essential and give you a kind of information epidemiological studies never can.
My point is not that Aspirin doesn’t cause Reye’s syndrome, but that it is impossible to say one way or the other. Maybe it does, maybe it doesn’t. The data is unavailable and likely never will be. So if you are an honest person who values the truth, you can’t say that it does.
I asked if it was proven that aspirin causes Reye’s syndrome. The correct answer is, “No, it is not.”
Yes, I see you did use the word “proven”. That does lesson the force of my comment but not by very much. I’m curious actually what you mean by proven outside a mathematical context. In this context, the correlation is extremely strong, and there’s no obvious alternate hypothesis. I’m not completely sure what you consider in this context to be an acceptable amount of evidence, but the medical consensus seems clear, and the links Gwern gave show that the correlation exists even when one tries to control for other variables.
I think he wants randomized controlled trials. Which of course no ethics board would ever approve because avoiding aspirin in kids is not that hard or expensive, and the cost of confirming it would be too high.
(“You want to test whether slamming your face into the wall causes pain, and doesn’t just correlate with it? Why on earth?”)
We have a plausible mechanism (confirmed by massive amounts of science) by which slamming your face into the wall causes pain; for the link between Aspirin and Reye’s syndrome there is no such plausible mechanism.
Well, at least in the drug world it is more convincing if you have a rational mechanism by which a drug could cause some effect.
A plausible alternative hypothesis is that children who get sick sometimes are developing Reye’s syndrome and take Aspirin. The problem with epidemiological studies is that it is IMPOSSIBLE to control for other variables. This is why randomized controlled trials are essential and give you a kind of information epidemiological studies never can.
My point is not that Aspirin doesn’t cause Reye’s syndrome, but that it is impossible to say one way or the other. Maybe it does, maybe it doesn’t. The data is unavailable and likely never will be. So if you are an honest person who values the truth, you can’t say that it does.