Minh’s Notes

Human-readable chicken scratch

Minh Nguyễn
May 11th, 2005


Limit of major weekly publication as logic approaches zero

This is just beautiful: Newsweek released its list of the top 1,000 public school systems in the nation this year, and it relies on a farce of a formula. Here’s the explanation that Newsweek gives us:

Public schools are ranked according to a ratio devised by Jay Mathews: the number of Advanced Placement or International Baccalaureate tests taken by all students at a school in 2004 divided by the number of graduating seniors.

… Although that doesn’t tell the whole story about a school, it’s one of the best measures available to compare a wide range of students’ readiness for higher-level work, which is more crucial than ever in the postindustrial age. …

Edward Felten points out that this system may actually help poor schools out. Here’s why:

  1. The numerator of this ratio, which counts “the number of Advanced Placement or International Baccalaureate tests taken by all students at a school in 2004,” doesn’t actually measure how well they do as a group, or how many were even adequately prepared to take the exam. As such, it especially rewards schools that have AP/IB exams as a requirement, and don’t force the students to pay $82 for each exam as our school does.
  2. The denominator, which measures “the number of graduating seniors” for the year, would actually give the school a higher score if it had fewer graduating students. In a worse-case scenario, it would plausibly benefit poor schools that have a high dropout rate. Best-case scenario, it benefits schools that have such high demands that even the best and brightest flunk out of the school.

But at least Indian Hill made the top 100 regardless. :^)

There is a reason why statistics these days tend to factor in such incredibly difficult concepts like percentages, per capita, and purchasing power parity, to name a few.

Nevertheless, Jay Mathews, the formula’s developer, has his reasons for using that particular formula. Prof. Felten covers this much better than I could.

Judging from article titles from Jay Matthews’ column at The Washington Post, it looks like he’s trying to defend himself. Fair enough, since I’ve had to do that on occasion as well. But, unfortunately, the Post requires a registration. Registration there is free, but I make a point of not reading sites that require such arcane methods of demographics research. I refuse to have to use BugMeNot or the BugMeNot extension for Firefox, which nonetheless are very useful tools that I would recommend.

Thanks to Prof. Felten for the scoop.