primetime
Pro Bowl Player
- Joined
- Jun 15, 2005
- Messages
- 13,627
- Reaction score
- 15,375
We know so little about the workings of the human body that it's remarkable we've stumbled into a lot of the miraculous cures and treatments we have now. I think of general anesthesia, which shouldn't even work based on what we know - the fact you come back as you after general anesthesia is literally inexplicable based on what we know of the human brain and the self. And of course some small percentage of people don't, even perfectly healthy people, and we don't know why that happens either. (It's also a risk on a population level but the risk on an individual level is probably lower than dying in a hurricane if you live in Boston. Certainly substantially less risky than driving.)
We know even less about nutrition, and the science there is often very crappy and paradigms change year to year. Part of that is industry's meddling, part of that is that it's just hard to do those kind of studies. And the things that get trumpeted on the front page of the Times - red wine is good! - are usually small samples, etc.
Science writing is a problem. I think of the phenomenon where geriatric pregnancy is said to increase risk of autism, for example. But the studies themselves only look at first pregnancies, and women with older first pregnancies may be categorically different in some way than those with younger first pregnancies. And even a statistically meaningful and sizable increase in risk ratio may be effectively meaningless in terms of overall incidence for individuals. Say the risk ratio is 1 in 10,000 and increases to 1 in 8,000. That's a big change... on a population level. It effectively makes no difference to individuals. But science journalism will often report it as percent difference (an increase of 20%!) rather than in those magnitudes. Always be suspicious when you're reading about percent changes in probabilities, because big numbers are usually signs that something odd is at play.
We know even less about nutrition, and the science there is often very crappy and paradigms change year to year. Part of that is industry's meddling, part of that is that it's just hard to do those kind of studies. And the things that get trumpeted on the front page of the Times - red wine is good! - are usually small samples, etc.
Science writing is a problem. I think of the phenomenon where geriatric pregnancy is said to increase risk of autism, for example. But the studies themselves only look at first pregnancies, and women with older first pregnancies may be categorically different in some way than those with younger first pregnancies. And even a statistically meaningful and sizable increase in risk ratio may be effectively meaningless in terms of overall incidence for individuals. Say the risk ratio is 1 in 10,000 and increases to 1 in 8,000. That's a big change... on a population level. It effectively makes no difference to individuals. But science journalism will often report it as percent difference (an increase of 20%!) rather than in those magnitudes. Always be suspicious when you're reading about percent changes in probabilities, because big numbers are usually signs that something odd is at play.