Teen cracks the code to higher SAT scores

And he's got an MIT professor to back him up

Published November 6, 2010 12:01AM (EDT)

Do you remember the night before you took the SATs? The butterflies, the insomnia, the nightmares, the last-minute cramming. About 1.6 million American high schoolers are experiencing that same pre-exam anxiety right now as they await tomorrow morning and the most important test of their lives.

Fourteen-year-old Milo Beckman, dubbed a prodigy and a student at arguably the best specialized math and science secondary school in the country, will be one of those test takers, but with a trick up his sleeve. He's cracked the code to getting a higher score on the essay portion of the infamous SATs.

Milo wanted to test the research of Les Perelman, the director of the undergraduate writing program at MIT, who found that the longer the SAT essay, the better the grade. According to CollegeBoard.com, the essays, added to the test in 2005, are graded not for their individual qualities but for the general impression they give the reader. Perelman deduced that "general impression" translated into "length," and was right 90 percent of the time. However, Perelman only studied CollegeBoard's sample essays, so Milo set out to look at the actual test and see if Perelman's theory still held water.

And it did. Using Facebook, Milo gathered data on the line length and scores of 115 essays and found that there was a direct correlation between a higher score and a longer essay. The probability of that correlation being mere chance is basically zero. He is using his findings to advocate for more time to be allowed to write the essays and more time for teachers to grade them. He even made a really impressive YouTube video to demonstrate his research.

It's doubtful Spicoli did all that well on the SATs, or even bothered to show up for them: 

Movie Videos & Movie Scenes at MOVIECLIPS.com


By Michelle Fitzsimmons

Michelle Fitzsimmons is an editorial fellow at Salon.com.

MORE FROM Michelle Fitzsimmons


Related Topics ------------------------------------------