College admissions teams are ranking students on a 0-10 scale based on factors they can't control

Standardized test scores, grade point averages, and teacher recommendations are well-known components of college applications. But schools use other less-discussed factors in admissions decisions as well: where and to whom you were born.

Students' demographic information is used for "predictive analytics," a little-known x-factor that colleges often use for enrollment management.

The process pulls a multitude of data points into a model that predicts the probability a particular student will apply to a school, choose to attend after they've been accepted, or perform well once enrolled.

"That's one of the biggest concerns - that students might not be aware of the many ways their data is being used," Manuela Ekowo, a co-author of a paper on predictive analytics in higher education, told Business Insider.

The report, "The Promise and Peril of Predictive Analytics in Higher Education," explained how colleges rank students based on this data. Admissions teams individually score students' likelihood of becoming an applicant, being admitted, and deciding to enroll, usually on a scale of 0-10 based on factors like: race and ethnicity, zip code, high school, and anticipated major, according to the authors.

While colleges find the information helpful, predictive analytics raises questions about discrimination.

While the implicit connection between race, income, and college attendance is widely known, the predictive analytic models that Ekowo studied offers a linear determination of how these factors affect success, and thus, admission to college.

Students of color, for example, are more likely to attend a school with a high percentage of students living below the poverty line. The schools tend to have fewer resources, which can affect the quality of education.

Predictive analytics, however, take statistics like that a step further by assigning a numeric value to predict success based on uncontrollable demographics factors - unlike scoring well on exams of having an impressive GPA.

"There's nothing they can do about being born in a low income neighborhood or zip code, but those are typically standard questions on any application," Ekowo said.

But predictive analytics has its upside too. To some extent, understanding what types of student will succeed at a college is not only necessary, its laudable. Riddling students who may drop out with student loans isn't good practice for colleges - or prospective students.

"Colleges are balancing so many different priorities," Ekowo said. "They are committed to accessing diversity. They're also committed to ensuring that they can run an institution, and that requires enrolling a number of students who can pay tuition in order to keep the place up and running."

Still, its important to understand inherent biases in predictive models.

"Colleges that use predictive analytics in the enrollment management process run a serious risk of disfavoring low income and minority students, no matter how qualified these individuals are for enrollment," the paper states.