Gaming the College Admissions System, Big Data Style

The current issue of Bloomberg Businessweek has a provocative article titled, “How to Get Into an Ivy League College—Guaranteed.”  No, this is not some commercial on late-night cable TV, paired with an ad for the slice it/dice it knife.  Instead, it’s about entrepreneur Steven Ma, who runs a booming business on how to scientifically game the system for admission to the nation’s top universities.

A few posts ago, when I wrote about Joe Green, president of the ruthless lobbying group FWD.us, I confessed that I had already had a pre-existing bias against Green back when he was in high school. He and some classmates had been the subject of a CNN documentary on the pressure on kids applying to elite colleges. To me, the students came across as cynically desiring the prestige these schools bring, rather than a wish to experience the intellectual stimulation imparted by world-class leading professors.

I am certainly not implying that most students in prestigious universities are like the ones in the CNN show, and I think the admissions officers usually manage to select students who genuinely add something to the academic, social and cultural atmosphere of their institutions.  But clearly these gatekeepers are having to scramble, what with the likes of Mr. Ma on their heels.

Ma simply applies statistical principles (call it “machine learning” if you insist) to data on applicants and their success or failure in getting into the school of their desires.  The more data he has, the more powerful his predictions are, so he’s constantly improving an already-strong track record.

Skeptical?  Surely the admissions officers don’t make decisions in such a formulaic way, you say?  Let me tell you a little story.

Way back when I was in grad school, I was employed as a Teaching Assistant, and part of my duties was to help grade exams.  One day I was grading papers, and an undergraduate happened to be in the office I shared with a fellow grad student.  The undergrad watched me grade a particular problem, say Problem 3, for a while, and after a few minutes he got to the point at which he could predict with remarkable accuracy what score I would give on Problem 3 to each student.  I was quite taken aback to learn that I had been grading on the basis of some formula that even I myself had not been aware of. Thus, in reading the BW article now, it doesn’t surprise me to learn that admissions officers at these selective schools are also unconsciously using formulas, all while thinking that they are evaluating each applicant individually.

And maybe some of it IS conscious.  I remember a friend of mine in the South San Francisco Bay Area telling me about 10 years ago that word had been circulating among his social set that Stanford was placing a major premium on applicants who had done well in a debate team.  Supposedly someone in the admissions office had leaked the word.  I’ll never know whether that rumor was accurate or not, but based on the successes my friend cited of kids acting on that tip, it may well have been legit.

One can hardly blame Mr. Ma, who is simply applying his quantitative skills to  a very lucrative market in the Asian-immigrant community.  There are many such companies, such as IvyMax, one that I pass by all the time in Fremont. (The Chinese name 飛達 means “fly to achieve”.)  Ma also is engaged in admirable philanthropic work.  However, I do blame the Tiger Mom type among his clients, as I have written here before.  Which brings me to one more story:

One Saturday in March a few years ago, when my daughter was a senior in high school, she participated in the local Science Olympiad.  One event consisted of building a catapult.  As I was watching, a pleasant fellow parent from another school  struck up a conversation with me.  “Which one is your child?”, he asked.  I pointed to my daughter, and mentioned that she and her teammates were high school seniors.  The other parent was dumbfounded that they were seniors, asking me “Then why are they here?”  Joe Green would know immediately what that parent meant:  It was March, way past the deadline for submitting admissions applications, so participating in that competition was “useless” from that parent’s point of view. To him, participating in the contest was simply a cynical act to build up a re’sume’ that college admissions committees would find attractive.  But I answered simply, though probably with an edge in my voice, “They’re here because they love science,”  The other parent recovered from his faux pas, and said, “Yes, that’s a good reason.”

This gaming of the system encourages cynicism among our young people, leading to even deeper cynicism when they become adults, Joe Green being a case in point. And I’m sure this article about Mr. Ma is causing much hand wringing among admissions officers; I don’t envy them.

Advertisements

15 thoughts on “Gaming the College Admissions System, Big Data Style

  1. From what I know about certain admission processes first-hand and second-hand (though I’m sure there is a lot of variation), most admissions committees do have explicit guidelines for 50% to 90% of an applicant’s “score”, but of course they like to keep most of that secret, and many fairly hard rules may still require some interpretation. Then there are typically scores of readers and raters who may vary in their interpretations. And finally any school worth its salt has a quota on “specials” where they look explicitly for something, almost anything, out of the ordinary. And then we get legacy admissions, and such.

    Sooo, doing a big-data reduction is always going to find a fair amount of noise, and most of what it can deduce is going to be obvious, like good grades from a good school. Will they find *anything* useful? And will it still be in force a year later? Eh. A ouija board, good grammar, and a $100,000 contribution to the building fund are going to be more useful, I think.

    Like

  2. I’m doing a big data project now for a $20B company involving Twitter, measuring sentiment. We are using Hadoop in the cloud. Interesting stuff, also requiring a data scientist early on. I see a growing need for that skill set.

    I digress.

    This big data project has been an eye opener. I think that big data is misunderstood, often misapplied, and over marketed. That being said, when applied correctly there are ethical and social implications that will come out of it. I have thought of many, but I suspect that is just the tip of the iceberg.

    I can’t resist wanting to use big data to give my business a competitive advantage. Predicting who will do what next is powerful stuff. Currently I am limited in time and money to invest very much in big data. The resources (servers) are expensive when doing it on a mass scale. You can find yourself going down aimless paths and wild goose chases. But occasionally you start to see patterns and find answers to complex questions.

    I will bet that the winner of the next election could be predicted based solely on twitter feeds. It may change as sentiment changes in any horse race, but I bet it is a good predictor of that and many other things.

    Society needs to recognize when we cross ethical boundaries, and reign us in. We probably wont know we’ve crossed a line until well after we crossed it. By then, the greed will be uncontrollable and only the law can turn things back.

    That being said, blue pill or red pill? LOL

    If you know any grad students interested in big data willing to work with a programmer and go down some interesting rabbit holes, send them my information. Right now it is just for fun (weekend stuff) but I can see it becoming a money maker. Especially for those who get past the superficial marketing of big data and learn how to apply it, and when to apply it.

    I found that pairing a developer with someone strong in statistics plus functional programming language (R) can lead to interesting things. Aside from the one Twitter project, it is more of a hobby.

    Like

    • R. Lawson: “I can’t resist wanting to use big data to give my business a competitive advantage. Predicting who will do what next is powerful stuff.”

      Let’s not get mystical about “big data”! First, let’s be clear (though you probably understand this) about what is usually meant by the term: *statistical* analysis of large datasets. The problem is, statistical analysis usually only gets one so far: either short spatiotemporal (ST) scale predictive skill WRT complex systems, or large-scale predictive skill WRT simple systems. In the latter case, this typically occurs through the derivation of the underlying dynamics of the simple system … which is what one *really* wants.

      I work in environmental modeling, which IMHO offers a useful lesson. Weather modeling (often referred to as NWP, “numerical weather prediction”) began surprisingly early on, with the work of LF Richardson in the 1920s. The fundamental physics of the atmosphere (e.g., the non-path-dependent “equations of state”) were known, but computing power was scarce (“computer” was a human job title), so weather prediction continued to be done mostly statistically: when we observe X, we can have Y confidence that Z will occur tomorrow. Unfortunately, that only enabled ST-local predictions (and therefore was useless for climate predictions, which were strictly extrapolative), and not very robust ones at that; but it *was* computationally feasible. Now that we have much more computing power, we can do dynamical modeling, and statistical modeling (or what is, “in the biz,” oddly referred to as “empirical” modeling) is denigrated.

      That being said, there are good reasons to assume that weather prediction will never be accurate at even small spatial scale for more than 2 weeks, due to non-linearities of the system. (Note that “weather” is usually defined as instantaneous observations at specific locations (i.e., small ST scale), while “climate” is *statistics about weather*, and thus is more feasibly/robustly predictable.) So it seems reasonable to assume that, barring discovery of similarly fundamental dynamics for NPP (numerical psychological prediction) and NSP (numerical social prediction), or what IIRC Asimov called “psychohistory”[1], similar advances in modeling human systems will *not* be forthcoming, and statistical predictive skill will be limited WRT to complexity of the target system (FWIW, I suspect the current US system of elite reproduction is increasingly simple and homogenous, enabling businesses like Ma’s[2]) or ST-scale (IIUC, Silver[3] limits his electoral predictions to 6 months).

      Amazing things are being done through statistical analysis of large datasets, but IMHO there is also megahype about “big data” currently.

      [1]: https://en.wikipedia.org/wiki/Psychohistory_%28fictional%29
      [2]: https://en.wikipedia.org/wiki/ThinkTank_Learning
      [3]: https://en.wikipedia.org/wiki/Nate_Silver

      Like

  3. I am Indian. My son was a Teach For America volunteer for 2 years. [I tried to talk him out of it but couldn’t. 🙂 ]

    While taking to a fellow Indian, I mentioned that son was in TFA. He nodded approvingly and mentioned (something like) “Law school?”.

    Like

  4. OTOH, there are millions of gifted students, unimpressed with the stupid games in HS and university, who don’t know the rubrics (that’s the word I’ve seen used for the grading patterns, admissions patterns, etc.) required to get in the best place for them to be able to make the progress in the direction they want to make (rather than the directions others want to steer them).

    And the result is wasted talent, lost productivity, lost innovation as their enthusiasm and opportunities are destroyed by those trying to force them into the round hole. E.g. recruiters and execs trying to force them into moving forward their own privacy violation schemes.

    Like

      • Where is the waste? Well, that’s easy. The waste lies in America’s ever-increasing loss of desirable values — values you yourself called out “A love of science”. An increasing cynicism and an inevitable turn to “for me and mine only” self-centeredness. A loss of belief in the value of a people working together to achieve common goals — worthy goals.

        Like

  5. This is from a book review of ‘Excellent Sheep,’ a new book by William Deresiewicz that’s drawing a a lot of attention, about the most elite schools and their students. I posted quite a bit about him at the bottom of Norm’s Tiger Moms post.

    https://normsaysno.wordpress.com/2014/07/08/tiger-moms-innovation-and-the-economic-and-social-good/

    ‘The trouble starts at admission. Top universities woo thousands of teenagers to apply, but seek one defined type: the student who has taken every Advanced Placement class and aced every exam, made varsity in a sport, played an instrument in the state youth orchestra and trekked across Nepal. This demanding system looks meritocratic. In practice, though, it aims directly at the children of the upper middle class, groomed since birth by parents, tutors and teachers to leap every hurdle. (The very rich can gain admission without leaping much of anything, as Deresiewicz also points out.)’

    http://www.nytimes.com/2014/08/24/books/review/excellent-sheep-by-william-deresiewicz.html?ref=books

    Liked by 2 people

  6. http://sinosphere.blogs.nytimes.com/2014/09/14/q-and-a-yong-zhao-on-education-and-authoritarianism-in-china/

    Education and Authoritarianism in China
    By DIDI KIRSTEN TATLOW

    Yong Zhao, a professor of education at the University of Oregon, has come far. Born in what he calls “one of the most ordinary villages in China,” he is now an authority on Chinese and American education and the author of “Who’s Afraid of the Big Bad Dragon: Why China Has the Best (and Worst) Education System in the World,” being published this week.

    There, Mr. Zhao examines how China’s contemporary examination-driven system emerged from an authoritarian, imperial culture, and how it has become an object of admiration among some policy makers in the West after Shanghai students ranked at the top in the Program for International Student Assessment, or PISA, test twice in a row. That throws up a puzzle that he unpicks: Chinese educators, parents and students believe their system is broken and have been trying to change it for decades. At best it produces a narrow kind of intelligence. At worst it replicates a rigid culture in which everyone competes for a few elite jobs that are dispensed, and controlled, by the state. So why is the West trying to “catch up” with China?

    Following are excerpts from an interview with Mr. Zhao:

    Q.
    You have said that traditional Chinese education actively “harms” children. How?

    A.
    It basically ignores children’s uniqueness, interests and passion, which results in homogenization. It forces them to spend almost all the time preparing for tests, leaving little time for social and physical activities. It also places them under tremendous stress through intense competition, which can damage their confidence and lowers their self-esteem.

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s