Science develops an algorithm for college selection—but does it work?

As seen in: eCampusNews: Technology News for Today's Higher-Ed Leader

Companies turn to scientific modeling and algorithms to help narrow down college selection, with the goal of dramatically decreasing dropout rate.

A new algorithm is using data and predictive analytics to determine, with noteworthy accuracy (90 percent), which institution is the best match for students. What makes it Jetson’s-worthy is that the algorithm doesn’t just match based on the student’s personality and goals now, but provides the best match for the student in the future and over time.

The algorithm was jump-started in the face of aimless college selection and high dropout rates. Selecting a new college or a transfer college is often an arbitrary process that’s often based on geographic location, availability of specific majors, a family attachment to a particular institution, or even a favorite collegiate sports team. And while transfer students already have a “taste” of what college is like – and are equipped with more information than the typical high school senior – unless they are proactive about seeking help, most receive very little support in this area.

As a result, students are largely left to their own devices and expected to “figure things out” on their own. The problem is that while 75 percent of students are consistently accepted into their first-choice university, only one-in-two college freshman ultimately graduate, according to vibeffect’s College Optimizer Index. Even more alarming is the fact that of the 80 percent of students who leave high school and enter 2-year colleges (intending to earn a 4-year degree) just 12 percent ever achieve that goal.

“America’s college dropout rate is pretty high, particularly in the online education space,” says Dave Hawn, president and CEO at ECMC Group in Oakdale, Minn., and interim CEO at Zenith Education Group (which is currently working with vibeffect to develop a higher education survey). “Students are flocking to both online and ‘ground’ programs, but a high percentage of them either don’t find the right match on the first college choice or wind up dropping out due to sheer frustration and the pressures of everyday life.”

Putting Science and Technology to Work

A couple of companies are taking a stab at the problem and using technology to help students make better choices for both selecting a college and transferring to a new one. In Washington, D.C., vibeffect has developed an algorithm-based, college-decision framework platform that students and families can use to scientifically narrow down their choices. This “unbiased, fact-based lens” costs $96 (per report) and uses a list of 66 different variables associated with the individual and measures those variables against the features of over 1,000 colleges.

vibeffect’s individual variables include things like whether someone has held a job, whether he or she likes working independently or on a team, and if the person is apt to ask for help (or not). On the college side, vibeffect factors in a school’s use of innovative teaching techniques, transportation options, and social opportunities. “Through that, we’re able to create correlations between an individual and the campus features that will help them thrive,” says Elena Maria Cox, co-founder and CEO.

For current college students, Cox says vibeffect can help the 30 to 40 percent of individuals who transfer to other institutions and – even more ambitiously – help reduce the 50 percent college dropout rate. “For the freshman who is unhappy and considering a transfer, we have individualized, self-assessments to help make those decisions,” says Cox, who calls the typical institution’s student retention efforts very “one-dimensional” in nature.

“A university and its professors are basically standing behind a one-way mirror, looking at the student’s behavior from yesterday and judging the future based on that information,” she explains. “Using data and predictive analytics, the same schools and instructors can determine which campus resources and features will put students on the path to thriving and graduating.”

So how does a student determine his or her personal traits, and how does the algorithm match those with the right college ecosystem?

Using a background steeped in research, vibeffect set a baseline to find out how many college students in the U.S. are currently “thriving” at a high level. Thriving was broken out into four separate levels: low, medium-low, medium-high, and high. According to the company, those thriving at a high level experience the maximum benefits from a specific college ecosystem, and demonstrate this through heightened academic and social integration and a deeper sense of happiness.

Using a team of experts from different fields, a national survey was developed that gathered 263 variables about how college students at four-year campuses are experiencing their college on several dimensions aligned to the definition of thriving. According to the company, this was the first research effort to define, on a national scale, what qualitative aspects of an entire college ecosystem can be identified and quantified.

The inaugural data set, established in 2013, provided the first consumer-based Index on how many students are thriving at a high level in four-year colleges. The data is updated annually, and the current data set includes more than 1.2 Million points on a representative set of students in over 1000 campuses across the country. The Index aims to be a snapshot of what is happening at a moment in time with real students in real colleges nationwide.

The initial Index found that only 1-in-5 college students are thriving at a high level and may be the root cause of the nation’s current 50 percent dropout rate, says the company.

With the Index in place, an algorithm was created using predictive sciences. The algorithm has the ability to leverage the 1.2 million data points to run predictive simulations regarding challenges in today’s higher education system.

For instance, the algorithm found that college students that are first generation Americans from low-income families have a higher probability of thriving than their wealthy peers. It also found that individuals that start in a two-year community college and transfer to a four-year university are more likely to thrive at a high level. In other words, thriving is not based on financial background, race or other homogeneous measures; but rather an individual’s confidence that what they are getting from the college experience will improve their lives and future. That desire, along with the pairing of an individual’s traits to a specific college ecosystem, creates the perfect environment for “high thriving.”

In order to help prospective students understand what their traits are and what to look for in a college ecosystem, a survey instrument was created that contains 66 individual traits and 100 campus features. The survey instrument, coupled with the algorithm, creates a predictive Model. Once a student interacts with the survey instrument, the Model provides a set of resources to predict the ideal campus environment for that individual.

The algorithm used in the Model is tested and re-tested using artificial intelligence methods and run through thousands of simulations until it produces “an acceptable probability accuracy,” noted the company. Depending on how many individual variables are captured, the algorithm has up to a 90 percent accuracy rating.

The Model does not say which college to attend by name, since the ecosystem of a college or university often changes based on the students in attendance and the faculty that are currently on staff. Campus clubs, among other things, may also come and go, changing the ecosystem in a unique way.

The Selling Point for Colleges

Cox says colleges also benefit when students use more thorough selection methods.

Consider, for example, the volume of high school graduates who get accepted to an institution but never show up. “Institutions go through all of the effort of creating a space for someone,” she explains, “and in the months between enrollment and the first day of school, the student becomes distracted and/or loses confidence in the value proposition. That’s a threat to the entire higher education system.”

By introducing scientific modeling to the selection process, Cox says vibeffect digs down into the individual motivations and behaviors of prospective students, helping them better grasp the “potential to thrive and start their higher education experience from a position of strength and the actual probability of thriving.” That, she adds, increases the probability of a positive match between an individual student and a specific campus. “Our goal is to have more students thriving and graduating and we will allow the higher education system to hold us accountable to that [mission],” Cox says.

So far, Cox says colleges and universities have responded positively to her company’s use of algorithms and scientific modeling to help students make better choices. Now, working with Zenith Education Group, vibeffect is “looking to prove that there’s a return on investment (ROI) in this for the university,” says Cox, “by getting as much data and as many examples of student success from different kinds of universities (i.e., career colleges, community colleges, online hybrid colleges).”

The company has rolled out a number of pilot projects and is looking at key measures like how many students showed up for the first day of school and how many returned for a second term more engaged (versus a control group). “Once we have the results (in February or March 2016),” Cox continues, “we’ll create a sustainable model for universities.”

Across the Education Continuum…and beyond SATs

Another company making inroads in the college and career-planning arena is Hobsons, the Arlington, Va.-based developer of Naviance. Dan Obregon, vice president of marketing, says the company takes a “holistic approach to college planning that incorporates self-awareness (on the student’s part) and then aligns that with academic plans.” At the Naviance platform’s core is the “College Power Score,” which Obregon says “helps students plan which courses they want to take in middle and high school.”

Then, based on the rigor of that course plan, the platform makes recommendations like, “Hey, it looks like you’re on track to go to a highly-selective college or university,” or, “Based on your course record, it looks like you have a fair shot at a selective college but if you step it up in a couple of key areas to increase your academic rigor you’ll be on a path to a more selective institution, if so desired.”

Along with offering guidance to individual students, the platform also helps facilitate conversations between pupils and school counselors. “With the high student-to-counselor ratio in many schools, these conversations don’t always happen,” says Obregon. “Using technology, we’ve automated that very manual, paper-based process for schools and districts, giving them early/actionable insights that they can use to handle counselors’ overwhelming caseloads.”

According to Obregon, that dialogue continues right through to the point of college enrollment, where students get advising and academic planning to ensure that they persist and ultimately graduate from college. “We’re viewing this through a continuum that spans not only what students are doing in high school, but also what steps they need to take to be successful in a post-secondary environment,” Obregon says. “Ultimately, it’s about helping students make college decisions based on a variety of factors as opposed to the standard variables like SAT scores and GPAs.”

A Groundbreaking Approach

Hawn sees real potential in a technology platform that goes beyond traditional college selection criteria that can help determine a student’s ability to thrive and succeed at a particular school. “We’re in the career school market – where everything from money to scheduling to life can get in the way,” says Hawn. “The idea that you can use technology to determine what makes one student thrive versus another is pretty groundbreaking.”

Cox says vibeffect is working on several new initiatives, all of them centered on helping students “get into the right place and the right mindset” to succeed in school. “If we can do more of this, and if we can use scientific models to reinforce our efforts – rather than the current, somewhat nebulous system that finds people making this very important decision from a position of weakness,” Cox says, “then we’re achieving our goal.”