- Thomas Thurston
School admissions suck! Can data science help?
Remember the college admissions process… and how much it sucked? After years of struggling to get decent grades and racking up brownie points, you throw yourself at the feet of admissions people. Getting into college has become insanely competitive, with amazing students from every corner of the world competing for relatively few slots in great schools.
This creates problems for both students and colleges themselves.
This, in turn, bombards schools with so many applications the differences between one great student and another can be negligible. They all run together. Even when good candidates are found, schools don’t know if their offers will be accepted since students are casting wider nets. The application process for both schools and students is increasingly unpredictable.
Faced with these problems, students and schools are turning to data science.
For example, Fast Company recently wrote about Wichita State University, where data scientists use algorithms to predict which students will succeed and which won’t. Working with IBM, WSU is using a recruitment model with reported 96% accuracy in identifying “high-yield” admissions candidates (compared with the 82% accuracy from professional admissions consultants).
According to IBM the use of data science is growing in academia, spanning not just admissions but also topics such as class scheduling, identifying at-risk students and tailoring student assistance to specific needs. Algorithms can include variables such as student academic results, social media activity and other out-of-school factors such as employment and a student’s family situation.
Did we say schools are evaluating candidates based on social media?!? Yep. Tell your kid to take those beer-bong photos off his Facebook page. For example, Samford University is using a Facebook algorithm (article here) dubbed “enrollment intelligence” that mines social media to help colleges figure out which applicants will likely enroll, who’s “on the fence” and who probably won’t enroll. These algorithms are based on social media activity, friends, messages and other behavioral factors. It’s even used for students who don’t participate in social media discussions themselves, so students can run from these algorithms but they may not be able to hide.
Schools aren’t the only ones using data science to their advantage. Students are doing it too. Dozens of automated “college admissions predictors” exist to help students estimate their odds of getting into schools. For examples check out MyChances.net, Cappex.com, The Princeton Review’s “School Finder” or the Facebook app Admission Splash.
Such data science is a good thing if it helps better pair students and schools in a positive way. That said, if misapplied it could lead to results that do more harm than good. Imagine for a moment your top choice of college is Princeton. You apply, but then an algorithm scans your Facebook data and tells Princeton, for some reason, you aren’t likely to enroll. Based on this input, Princeton denies your application when they would have otherwise made you an offer. As a result, you don’t get into Princeton, Princeton doesn’t get a qualified student and you have no idea why Princeton said “no” so you can’t correct them.
I’m not saying admissions algorithms are inherently good or bad – just that we need to be careful. The stakes are high so we must be diligent. On a positive note, assuming a specific algorithm is good, it also holds potential to make the admissions process a lot more transparent. Rather than over-relying on the intangible “gut feelings” of admissions experts, you can literally see how algorithmic decisions are made, step-by-step. You can also incrementally refine an algorithm over time, since it is explicit and consistent. While most schools would rather eat glass before making their admissions decisions transparent, perhaps more explicitness and openness could make the whole process more efficient and fair? If their processes are really above-board, what do admissions departments have to hide?
Algorithms are making inroads into college admissions. Like it or not, it’s happening. There’s potential for good or bad depending on how thoughtful academia is as it enters this new phase. That said, it would be awesome if college admissions were more open, fair and accurate. At the very least, perhaps someday the process won’t suck… as much.
#GrowthScience #admissionsalgorithm #collegeadmissions #admissionspredictor #gettingintocollege #thomasthurston #datascience