Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

[Phys-L] college admissions minus SAT plus MOOC ... or not



There's an amusing but spectacularly flawed article in today's
Gomorrah Post:
Kevin Carey
"Goodbye, SAT: How online courses will change college admissions"
http://www.washingtonpost.com/opinions/goodbye-sat-how-online-courses-will-change-college-admissions/2015/03/19/85c4d6c0-cc0c-11e4-a2a7-9517a3a70506_story.html

It is an object lesson in faulty reasoning. The article contains
a great many interesting observations. Most of what the article
says is true ... except for the conclusion!

Massive open online courses, or MOOCs, offered by dozens of elite
colleges give students a chance to prove that they’re ready for a
university — and in turn, the institution gets an accurate measure of
whether a student is prepared for its academics,

There is a proverb to the effect that if something seems like
a reliable proxy for something else, as soon as you start
putting weight on it, it stops being reliable. People will
find a way to game the system.

I think this proverb even has a name, such-and-such law,
but I haven't managed to remember it or google it. Can
anybody lend me a clue?

Carey points out that when college admissions depend on a
"personal essay" applicants will hire ghost writers. I
am reminded of the celebrated Peter Steiner cartoon in
the New Yorker:
http://michaelmaslin.com/inkspill/wp-content/uploads/2013/07/dog-on-the-internet-by-peter-steiner.jpg

It's disappointing that Carey didn't realize the same problem
affects MOOC grades -- or it would, if anybody ever started
relying on them for any serious purpose. It's a catch-22:
The grades are reliable just so long as they aren't important
enough to be worth fudging.

On the upside, if you ignore Carey's conclusion, the rest
of the article is a good read. It offers lots of anecdotes
about how screwed up the existing admissions process is.
He cites the study by Jesse Rothstein from 10 years ago
showing that "after controlling for students’ background
characteristics, SAT scores predict only 2.7 percent of
the variation in students’ college grades."
http://eml.berkeley.edu/~jrothst/publications/rothstein_sat_jmetrics2004_policybrief.pdf

I've spent enough thousands of hours looking at test scores
and transcripts etc. to have some opinions on the matter.
I agree with Carey that relying on SAT scores is ridiculous.
Doing better is certainly possible ... but it's not as easy
as Carey suggests.