Sounds like a good title for a paper ... and it is! Atila Abdulkadiroglu, Parag Pathak, and I published this study a few years ago; here's a copy: .
"The Elite Illusion" asks whether high schools that seem extraordinarily successful, specifically, the highly selective public schools in Boston and New York known as exam schools, actually improve student outcomes. The Boston Latin School (BLS) is the oldest high school in the country. Its graduates indeed do well: they have amazing test scores and go to top colleges. A good school, right? Most Boston parents would give a kidney to get their kid into BLS; many New York parents would give a liver to get a child into Stuyvesant!
What's the "illusion" here? Schools like BLS and Stuy are indeed super selective - they admit only top students, that is, applicants with high middle school grades and top scores on the schools' tough admissions tests. Of course, these students tend to do well - because only those predicted to do well get to go in the first place. Sounds like another likely case of selection bias, and it is! The Elite Illusion tackles selection bias by examining outcomes for applicants with application scores very close to admissions cutoffs. This approach is what econometricians call a "regression discontinuity design," and it's one of our most powerful tools. Looking at outcomes for students on either side of an admissions cutoff is a compelling natural experiment. The students on either side are essentially indistinguishable: those who clear the cutoff simply had better luck the day they took the test.
Say you lucked out and got into Stuy. This means you go to school with lots of other smart kids. In fact, your Stuy peers are so smart, they have middle school test scores 2 standard deviations above those of the classmates you would have had if your luck had soured and you'd failed the admissions test. How does being around such high performers affect, say, your New York State Regents scores or the type of college you go to? The answer is ... it doesn't! Kids close to exam school admissions cutoffs who fail to get an offer learn just as much and attend colleges no less desirable than do kids who received the coveted elite admissions offer.
Elite illusions are pervasive. Even sophisticated observers are inclined to mistake peer quality--good outcomes--for causal effects. I sometimes observe this among my academic colleagues, a reasonably sophisticated bunch. Metro Boston is composed of medium-sized towns like Cambridge, Brookline, Lexington, and Newton, that run separate school systems. New faculty of childrearing age pick their towns in part by surveying old-timers like me on school quality. Overheard in our department lunchroom are comments like this: "Newton," avows one eminence grise. "But Lexington is better" someone even older older pipes up. "We live in Brookline, but I wasn't impressed by their math and science," says a recently tenured high-flyer. "Our kids go private."
Alas, these distinguished scientists' observations on the quality of Boston's suburban school systems are without scientific merit. Superstar economists live in Newton. Because the Newton schools are frequented by the children of such luminaries, they can't help but have spectacular outcomes. Hard evidence on school quality will have to come from elsewhere.
This question originally appeared on Quora - the place to gain and share knowledge, empowering people to learn from others and better understand the world. You can follow Quora on Twitter and Facebook. More questions:
- Jobs and Careers: How did you deal with brutal referee reports in your early career?
- Econometrics: Who is regarded as the father (or mother) of econometrics?
- Data Science: With respect to careers in data science and analytics, what advantage or unique value is there in graduate studies in econometrics when compared with statistics, engineering, or computer science?