#everybodylies — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #everybodylies, aggregated by home.social.
-
I really loved Seth Stephens-Davidowitz's book, "Everybody Lies." I'm now struggling to read his new book, "Don't Trust Your Gut," but am finding book's title SO OFF-PUTTING. Alas.
#SethStephensDavidowitz
#EverybodyLies
#DontTrustYourGut
#DataScience
#EvidenceBasedPractice
#DecisionSupportSystems
#OperationsResearch
#Counterintuitive
#Strategy
#Planning
#Intuition
#Epistemology
@epistemology -
Regression Discontinuity, or the Stuyvesant High School Effect
Do elite schools, or prison incarcerations, have an effect on students or prisoners? I'd mentioned this a few months back on HN and finally tracked down the source.
The Stuy High effect is addressed in "The Elite Illusion: Achievement Effects at Boston and New York Exam Schools",
Atila Abdulkadiroglu, Joshua D. Angrist & Parag A. Pathak (2014) https://economics.mit.edu/files/9518Upshot: students who came within two answers of the cuttoff (fallling two points above or below the threshold) of Stuy High admissions ... have very similar life outcomess: indistinguishable AP and SAP scores, and indistinguishable admissions rates to prestigious universities.
Other times effects do matter, in another study by Chen & Shapiro on federal inmate prison assignments, showing harsher incarceration translates to higher recidivisim rates.
In Seth Stephens-Davidowitz, Everybody Lies
NBN interview: https://newbooksnetwork.com/everybody-lies
My HN comment: https://news.ycombinator.com/item?id=29745533
#RegressionDiscontinuity #StuyvesantHigh #Education #BigData #SethStephensDavidowitz #EverybodyLies #Prison #MKiethChen #JesseShapiro
-
Jeff "never puts away anything, especially oven mitts" Cliff, Bringer of Nightmares 🦝🐙 🇱🇧🧯 🇨🇦 @[email protected] ·If we can be sure of anything in the 2100s it's that we're going to invent a *lot* of interesting, and disturbing types of blackmail we can't even imagine yet. I'm not sure how much Stéphane Leman-Langlois gets this.
Also I suspect he would benefit by reading #EverybodyLies - he seems to understand the first order problems of #bigdata, but #everybodylies contains a cursory overview of ways of dealing with those big problems at the first order. There's no doubt second order problems beyond that (including ones justifiable by bayesian reasoning similar to Leman-Langlois's) but he doesn't seem to present this to the reader, at least. #BigDataAndSecurityIntelligence
