After literally centuries of experience and observed outcomes based on millions of students, colleges and universities thought they knew which key indicators ultimately could predict a student’s success. Grades in core courses, test scores, and attendance in the first year were identified as the most accurate predictors of whether a student would graduate.
But as the four-year national graduation rate fell below 50 percent, and student loan debts began to reach crisis proportions, new external pressures have led some universities to consider a new approach: data mining. In collaboration with data analysis companies such as Civitas Learning and EAB, or else relying on tracking systems they have developed on their own, universities have uncovered some surprising patterns about which metrics and performance indicator actually correspond with students’ eventual scholastic success.
These big data mining software programs for use in higher education analyze hundreds of millions of past and present student records, searching for patterns. The results are often surprising. For example, one large nursing school anticipated that students’ eventual success in its programs would be tied to their performance in the school’s core course, “Conceptual Foundations of Nursing.” But a sophisticated data analysis instead revealed that a student’s performance in an introductory math course was a much more reliable indicator of the likelihood that he or she eventually would graduate from the program. As a result of the new information, the college redirected its efforts to include new math tutoring offerings, designed to help struggling students grasp the concepts and prevent them from dropping out of school.
The data mining software also can be customized to track a range of variables. By examining everything from new student applications to where and when students scan their university ID cards, the software can make predictions about how students’ background or social habits might contribute to their eventual success or failure.
However, as is so often the case, the new tool also has some drawbacks. The cost for a university to contract with a data analysis company and license the software is steep. A standard three-year contract typically costs over half a million dollars. Many university faculty and administrators are also concerned about the impact of using (or perhaps overusing) predicative analytics. Universities seeking to increase their rankings or ratings may refuse to accept students whose backgrounds feature certain “warning signs,” and such choices may have disproportionally negative effects on historically underrepresented students (e.g., those from lower socioeconomic households, international students).
Moreover, the extra interventions that schools adopt to get struggling students back on a particular track may prevent students from switching majors, even if doing so would be more appropriate or appealing for the students in the long term. Finally, many critics cite potential privacy concerns: Student data could easily become compromised if a hacker accessed the system. With potentially millions of records containing sensitive personal information, the data files would be an attractive target for criminals seeking to commit identity theft.
Ultimately, proponents of data mining believe that the software’s use in a college or university setting will be beneficial to students. As long as human advisors remain in the equation, offering advice and recommendations in face-to-face encounters with students, the big data and personal information can be used as one of many tools designed to help students succeed. Thus far, only a few colleges have adopted the technology, and it has not been in use long enough to show a measurable increase in graduation rates. Time will tell whether data mining by colleges and universities ultimately is worth the (actual and potential) price.
- If the results prove successful in a college setting, should data mining programs be expanded into high schools to predict student success as well? Why or why not?
Source: Joseph B. Treaster, “Will You Graduate? Ask Big Data,” The New York Times, February 2, 2017