Slashdot Mirror
Getting Into College the Old Fashioned Way: With Money
Businessweek (in a story spotted via Tyler Cowen's Marginal Revolution) profiles ThinkTank Learning, a college-admission consultancy founded by Steven Ma, and largely catering to ambitious Asian immigrants like Ma, and their offspring — kids who'd like to go to elite schools, and can afford to have Ma's firm help them navigate the path to getting in. It's a statistics driven system, and backed by a money-back guarantee, so long as the applicant meets certain requirements: ThinkTank will refund their tens of thousands of dollars in fees if they don't make it into the sort of school that the ThinkTank algorithms say they will. Basically, they've reverse engineered the admissions policies at schools, particularly elite schools like MIT, Stanford, and the Ivies, and done so well enough to know which factors in a student's portfolio can be tweaked to increase their odds of getting into the big-name schools. A slice: [Ma's] proprietary algorithm assigns varying weights to different parameters, derived from his analysis of the successes and failures of thousands of students he's coached over the years. Ma's algorithm, for example, predicts that a U.S.-born high school senior with a 3.8 GPA, an SAT score of 2,000 (out of 2,400), moderate leadership credentials, and 800 hours of extracurricular activities, has a 20.4 percent chance of admission to New York University and a 28.1 percent shot at the University of Southern California. Those odds determine the fee ThinkTank charges that student for its guaranteed consulting package: $25,931 to apply to NYU and $18,826 for USC.
Its racist to acknowledge language barriers? ITs racist to expect that people attending an American university be able to speak English to participate in group assignments? Or do you expect the American students to learn Arabic, Mandarin and Somali?
Good-bye
Gather enough anecdotes and you can start to do statistics on them.
Sure, you could. If you want to work with a complete unreliable dataset, where all your conclusions are much more likely to be invalid.
If you are scientific you will follow up with a well randomized survey.
"Well-randomized surveys" are not anecdotes. Anecdotes are individual stories, which may all have their individual bias. Since they are reported without context or regard for selection, they are more subject to cherry-picking, confirmation bias, etc.
But usually inquiry begins with anecdotes.
Agreed. You have to get interested in a topic first, and if you've never heard anything about it, you probably would never look into it. But after we've heard a couple stories then we move onto better data collection techniques if we want to draw any valid conclusions.
Or didn't you take Statistics 101?
Well, I've actually written articles in peer-reviewed professional publications and essays on the topic of the use of statistics. What are your credentials?