…How’d I Get Here? College Admissions and Big Data

 

When I was an education consultant in Silicon Valley five years ago, Palo Alto was experiencing a rare phenomenon: two suicide clusters within the span of a few years. To put things in perspective: the U.S. usually experiences five clusters per year, which are defined as multiple suicides in a short time frame. What’s more, in Palo Alto nearly all of the victims were teenagers; two of the highest performing schools in the area, for example, Gunn and Palo Alto High, have 10-year suicide rates between four and five times the national average. The Caltrain (the Bay Area’s commuter rail) became synonymous with suicide, so much so that it garnered the attention of the Centers for Disease Control and Prevention, which launched an epidemiological study in the area in 2016.

Meanwhile, I was consulting students ages 12-18 at and around these schools, for a reputable contender in the rapidly growing industry of college admissions private consulting. The San Francisco Bay Area, where I was working, has the highest concentration per capita of these consulting agencies, contributing to an industry that has grown exponentially in the past few years (more than a quarter of college applicants in the top 70th SAT percentile used a college consultant in 2015). The industry is worth over $400 million, and tuition for these private programs can rival that of a community college (plus, personal consultants can freelance for up to $400 or more per hour).

stevenma-3_3b18009226408ec1bab178a37d4ce248.focal-320x417

Steven Ma, founder of ThinkTank Learning, promises students Ivy League acceptances.

Why do these families insist on using consultants? Well, perhaps what I should be asking is why there have been so many teenage suicides in Silicon Valley. The college admissions world is high intensity and high pressure; consultants are trained to be college admissions experts in standardized testing, summer internship programs (many that have lower acceptance rates than Stanford itself), course selection, college major requirements, application essays…the list goes on.

This industry has emerged from the reality that getting into college is more cutthroat than it has ever been before. While overall application rates are decreasing, “brand name” schools are at a loss from their own popularity; Stanford, for instance, received over 44,000 applications for just over 1,700 spots last year. Plus, strapped for cash, more colleges and universities are seeking out-of-state or international students to pay full tuition, particularly at state schools. With seemingly high stakes, many students report anxiety and depression due to academic pressures, and students, parents, and administrators alike argue that the admissions process is broken.

lead_960

California guidance counselors are outnumbered 1000 students to 1 counselor. (photo cred: The Atlantic)

In response, colleges have been turning to technology to get a more holistic view of applicants and social media is just the start. One of my favorite parts of consulting was Googling and Facebook-stalking my students to make sure their profiles were college-proof (trying to politely and professionally tell them that they should remove the “FADED” and “DGAF” captions from their photos was another story), but it’s not just me. About 40% of college admissions officers admit to researching applicants’ social media profiles, a number that has quadrupled since 2008. Most are doing so not only to verify details around application questions, but also to better understand the applicant’s personality or leadership style; many report finding details that actually enhance a student’s application. Students, too, are appealing to universities with multimedia applications, created through tools such as the Coalition for Access, Affordability, and Success’ digital locker, or ZeeMee’s online profile creator, which allow counselors to get to know applicants in different ways. Students can even “tour” schools using virtual reality.

ithaca

An outline of Ithaca College’s predictive model (cred)

In addition, institutions such as Ithaca College are using big data programs designed by IBM to identify the kinds of students most likely to graduate, and schools such as Wichita State use algorithms to quantify student interest, as well as potential return, then allot recruiting money based on probability of enrollment. And after those students are admitted, the schools are using predictive analytics to determine which individuals may need more academic support to succeed. In 2011, for instance, Georgia State was one of the first major universities to use ten years worth of academic and attendance data to create a model that flagged students at risk of making decisions that could lead to dropoutan approach that has saved 2016 graduates $15 million total compared to the Class of 2012.

The use of data as it relates to student interest, ability to succeed, and fit may seem invasive or even highly error-prone, but schools view it as a smart use of resources in a high stakes scenario. After all, student retention rates hover around 60% overall, and almost half of full-time, four-year students graduate in six years (rates that are even lower for black and hispanic students). Universities are constantly striving to improve these statistics, as high attrition rates can mean disrupted learning and living experiences for students. What’s moreand arguably most important with institutions of higher education inextricably tied to financial interests, a situation that could become more dire with recent tax proposals—is attrition means lost revenue for universities. At Georgia State, for example, a 1 percent increase in retention rate means an additional $3 million in return on investment.

However, there are many arguments against the use of big data to impact admissions decisions; as CTO of Dell EMC’s big data Bill Schmarzo argues, universities could easily use big data to tailor their admissions practices, as well as curriculum and services, to students more likely to become big donors after graduation (even if some of the nation’s most selective institutions already enroll more students from the top 1 percent of the income ladder than from the bottom 60 percent). While he claims that this would allow schools to recommend “better fitting” schools to students denied admission, this thinking could reinforce discriminatory stereotypes, as recent data suggests that students technically “under-qualified” but with above average test scores will perform better at more selective colleges.

Also, as with any data, there are of course privacy concerns, and what’s more, the implementation and maintenance of these programs can be costly, as well as difficult to execute flawlessly. How, for example, do you measure a student’s potential for growth in an unfamiliar environment, given limited information about the context of their existing data? How do you judge students with little to no social media presence? Furthermore, beyond admissions, communicating to “at risk” students, if not done well, can actually end up alienating or demotivating students further, so schools must be careful about who gets access to this data.

Transparency around colleges’ use of data analytics is also an interesting predicament. On the one hand, it can help students who know how to “play the game” with schoolsand this, in full transparency, was the focus of much of our consulting. If students can increase their chances of acceptance just by watching a few YouTube videos or webinars, or even talking to a chatbot, why wouldn’t they? Researchers, however, argue that the use of technology could tilt the odds in favor of students who already have increased access to college, contributing to the racial and socioeconomic divide in college admissions.

1-Fz7vi5tg4_craQEB2VwT6wPlus, many worry that colleges are putting so much pressure on applicants that students arrive on campus underprepared for their college experiences. Experts argue that students spend more time focused on college applications than actual, impactful learning in high school. And of course, as seen in the Palo Alto tragedies, there’s the stress students report as a result of the intense pressure they feel to “compete” in college admissionspressure coming from their families, their peers, their schoolsand even, the colleges themselves. Therefore, while it’s undeniable that technology has the power to upend the admissions process and its potentially harmful effects, the onus is on the schools to define what technology they need, how they should use it (ie, knowing both its strengths and failings), and most importantly, how it supports their missions as institutions of higher learning.

 

10 comments

  1. This is so interesting– I never heard of the behind the scenes technological aspect of college admissions, and it sounds like it’s changing rapidly. The ability of students to create a multimedia profile so they’re not just numbers on a page and can stand out for reasons other than grades and SAT scores could help schools use a more holistic approach to reviewing applications. I never considered that colleges stalking students’ social media accounts could actually be beneficial to them. Students don’t seem to consider this, especially given the trend of changing their Facebook names Senior year so that they can’t be found. I wonder how schools will be able to use AI to analyze these more nuanced aspects of applications like social media and multimedia profiles. It is sad to think about how the financial pressures on educational institutions could lead to distorted incentives in the admissions process. This could even lead to consequences for the economy and socioeconomic classes if it systematically prevents certain groups from getting access to a college education.

  2. Really interesting to learn about the complexity of the system that we all have gone through. I wonder if students in the application process are aware of the extent social media is used. I remember when I was going through the process four years ago, my teachers and advisors warned of social media, and encouraged avoiding or hiding use, instead of utilizing it as an asset to grow your brand. While colleges seem to be putting effort into diversifying their student populations in terms of ethnicity and socioeconomic backgrounds, this system may inadvertently disadvantage those it is looking to benefit.

  3. Great post! I think it’s completely understandable that college admissions officers are turning to social media and other digital technologies to get a holistic view of the applicant. I also agree that this technology will cause a greater socioeconomic rift in the college admission process simply because those who have more resources before college would’ve had plenty of opportunities to build their social media presence. In high school, I saw the amount of content my wealthier friends put out on social media simply because his family was able to afford all of those fancy trips during spring break and in the summer. He created albums of his trips and poetically used a punny caption for each picture in the album. I grew up in a low income family and did not think much about it then but in hindsight, that album he created was a huge asset for the college admission process that I wouldn’t have had the opportunity to replicate. Hopefully Ithaca College and other colleges that use predictive models understand this and factor it in as a predictor in the future.

  4. I found your post really informative and interesting to read! The use of data in college admissions relates to our class discussion about the use of data and AI in the job recruitment process. As we discussed in class, there are many complications with using data. There are many benefits, and I do agree that universities could use data to achieve cost savings or to get a more holistic view of a student’s application. However, I think using data for the college admission process could really disadvantage many high school students. Just like standardized tests like SAT/ACT label students with a “number,” the use of data analytics to pick which students to admit seems really standardized. Also, there are so many factors that go into college admission that can’t be quantified using data. While data analytics could bring many additional biases into the college admission process, the age of data and AI is here. More emphasis should be placed on making sure colleges stay transparent with their use!

  5. Really great post! I actually have a really good friend who went to Palo Alto High and she still speaks about the traumatic cluster of suicides, four years after attending the school. It’s really interesting to see that colleges are leveraging this sort of data to gauge interest, but I almost wonder at the fairness of it. I definitely did not do any of those sort of technological engagements, and only expressed my interest by doing in-person visits. I wonder if as the prevalence of technology has increased, if this sort of engagement is deemed necessary by admission officers. I would assume that not all students have an equal ability to engage with colleges online, given different socioeconomic backgrounds and time constraints by already juggling academics and extra-curricular activities.

  6. This is such a relevant topic! I am from the Bay Area and remember when those suicides in Palo Alto began to spike. My friend who goes to school nearby in Los Altos told me that they increased the number of counselors at her high school to try to combat the problem. My town is exactly as you described, I don’t know a single person who did not have both an SAT or ACT tutor and a private college admissions councilor. The idea of using Big Data to screen applicants however makes me nervous, as one mis-step in an algorithm can deny a very qualified student and can recommend to admit one who might not be the best fit. If colleges do start to turn to big data more, which could help smooth the process out, I agree they need to be much more transparent in their tactics. It is the uncertainty and seemingly unfair ‘picking out of a hat’ aura surrounding admissions decisions that drives kids crazy.

  7. I remember reading a great article about Palo Alto and the suicide clusters. The college admissions process has changed so much due to technology and it has become so much more competitive causing this pressure on the students. Social media is so interesting– as someone who was being recruited while applying to schools it was something I was very aware of. Many students chose to temporarily change their names so that “no one could find them”. I chose to keep my own name because I felt I had nothing to hide. However, I remember a story about a kid at Tufts who tweeted something negative about the school while at an info session. This emphasizes the point that what you put on the internet is permanent. As someone who volunteers with admissions at BC as a tour guide and panelist, we are told to encourage prospective students to follow BC on social media to try to get a “feel” for the school. As we learned from the BC social media presentation, social media has become vital to showing students the school on a more personal level. With things like virtual reality it’s crazy to think if people will choose to continue to spend money to visit schools or if they will just get a “feel” through technology.

  8. Great post, Emma – you raise a very interesting point. With so many different data points, it must be difficult to optimize for the best possible student. And what makes a good student? I’d argue that colleges should put more effort into supporting those students (and generalized demographics that they have) that don’t graduate in an allotted timeframe, as opposed to simply cutting out people who don’t have the resources or skillset to be prepared for college on their own. If they’re qualified up front, figure out how to make it work! It also is a slippery slope if something that a college controls for has a disproportionate effect on a protected class. Especially in light of Harvard’s current lawsuit (http://www.thecrimson.com/article/2017/11/26/DOJ-lawsuit-harvard/), it’s important to be careful around stuff like this…

  9. What a great, thorough post. I expect the use of analytics in admissions only to increase from here, though. As you said, it’s a high stakes game and schools need all the help they can get. Of course, it will be followed by software that analyzes the admissions decisions to help the potential applicants too.

  10. What a fascinating and informative post! When I was applying to colleges, I was always warned of keeping social media profiles clean; although I don’t think I fit the bill for a rowdy student, I knew many in my high school (and even in college) who ignored this warning, thinking it was just a rumor. It is really interesting to hear how counselors not only review social media profiles to find students who might be bad news, but also as a means for enhancing the student’s overall persona. I also remember thinking the college applications I submitted never really told the whole story of who I was, so I commend the universities exploring digital tools (multimedia applications, VR tours, big data, and so on) to not only improve the college application process, but also help students maximize their potential once they attend their school and identify those who might be struggling.

Leave a Reply

Please log in using one of these methods to post your comment:

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

%d bloggers like this: