suppose for a moment that this were actually true what do physical reality do you imagine might cause this outcome? what specific actions can you imagine all white doctors taking that no black doctors take that would *triple* infant mortality independent of everything else?
-
Show this thread
-
Replying to @eigenrobot
note that this is not actually what they found black infant mortality rate is triple the white rate, the effect they see drops that to (afaict) ~double the white rate
2 replies 1 retweet 12 likes -
Replying to @halvorz @eigenrobot
Were you able to see how they determined the pediatrician's ethnicity?
1 reply 0 retweets 5 likes -
Replying to @MaiqTL @eigenrobot
"Physician race is not coded by the data and is captured from publicly searchable pictures of the physician." non-white, non-black doctors were dropped from the analysis
1 reply 0 retweets 6 likes -
Replying to @halvorz @eigenrobot
That seems suspicious. Thanks for looking that up!
1 reply 0 retweets 3 likes -
it does but tbf it does not seem they had another way to do it and they do seem to have taken steps to do it accuratelypic.twitter.com/5qwatPMx8z
4 replies 0 retweets 2 likes -
lol if this is pure measurement error the "true" coefficients should be /greater/ ironically this possibility of measurement error decreases my confidence in the studys results
1 reply 0 retweets 1 like -
I get how measurement error would underestimate coefficients but why would it have different effect on white/black?
1 reply 0 retweets 1 like -
well I mean it would bias downward estimates of p(death | black, everything else) - p(death | everything else) and the other thing
1 reply 0 retweets 1 like -
yea but wouldn’t it bias down p(death|black baby&black doctor) and p(death|black baby&white doctor) more or less equally?
1 reply 0 retweets 1 like
Yeah but latter doesn't get a coefficient in their setup
-
-
then maybe I’m misunderstanding how they did this why doesn’t their model cover all the possible pairings?pic.twitter.com/YarmnUreMl
1 reply 0 retweets 1 like -
It does cover them White/black infant and white/black doctor are both dummies defined on {0,1} They have coefficients on each dummy individually, plus the interaction White baby black doctor is case where black doctor = 1
0 replies 0 retweets 2 likes
End of conversation
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.