an econometrics prof i was TA'ing for brought this up in front of undergrads and made them feel very uncomfortablehttps://twitter.com/ThatsMauvelous/status/1445211581592260613 …
-
-
Replying to @eigenrobot
I made the mistake of looking into fetal alcohol syndrome and acquired the extremely bad opinion that it’s as much about genetic luck as alcohol
9 replies 1 retweet 55 likes -
Replying to @literalbanana @eigenrobot
BMJ: Hazardous Journeys Parachute use to prevent death and major trauma related to gravitational challenge: systematic review of randomised controlled trials https://www.bmj.com/content/327/7429/1459 …pic.twitter.com/S2w9VQO8aU
2 replies 0 retweets 10 likes -
The problem is that this is false and parachutes are all subject to rigorous trials. The trials are not *randomly* controlled because *we already know the control outcome*.
3 replies 0 retweets 11 likes -
Replying to @AlexGodofsky @BenLiddicott and
grumpgrumpgrump I just hate the parachute thing because vaguely rationalist-adjacent people use it as an excuse for bad epistemology but don't even get the details right
1 reply 0 retweets 11 likes -
Replying to @AlexGodofsky @BenLiddicott and
The other thing is I in fact would not be willing to jump out of an airplane wearing a parachute if all the manufacturer offered me was a theory of how it should work.
1 reply 0 retweets 11 likes -
Replying to @AlexGodofsky @BenLiddicott and
isn't this the WHOLE point of the parachute paper? you don't need to be all that randomized, all that controlled, or all that tried, if you have other evidence --- RCTs are when you have weak priors.
1 reply 0 retweets 5 likes -
Replying to @fakegeekguy @AlexGodofsky and
no RCT exists for if I'll die if I fly a rocketship into the sun, but I don't need a trial to know, let alone a randomized or controlled one
1 reply 0 retweets 7 likes -
Replying to @fakegeekguy @BenLiddicott and
You do need trials. You don't need (randomized) controls when you have an extremely strong prior *about the control*. The satirical paper doesn't even contemplate the idea of nonrandomized unctrolled trials; it thinks the dichotomy is between RCTs and observational data.
3 replies 0 retweets 9 likes
what if i disagree with the prior of the person designing the trial 
-
-
Replying to @eigenrobot @fakegeekguy and
If there is substantial disagreement about the prior that's often a good indication that nonrandomized uncontrolled trials are a weak source of evidence :)
1 reply 0 retweets 9 likes - Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.