Sign In

Promises and pitfalls of predictive analytics

Predictive analytics has long been seen as the next big thing in education. Several different learning-management systems have the option of analyzing student progress included in them; there are also standalone systems (see “read more” for some of these options).

The allure of being able to pinpoint which students need further assistance is obvious, although technological solutions are not always necessary. Seasoned teachers would argue that identifying students who need assistance is part of the job. But technology can help by speeding the process and providing greater contextualization.

Let’s take a new class of about 30 students as an example. For a teacher to be able to identify the students who are struggling, s/he must first assess the class. Depending on the subject, level, and teaching methods of this classroom, such assessment may not occur until a month or so into instruction, at the first exam. In some subjects, particularly fast-paced advanced classes, this may be too late.

Predictive analytics use a student’s past performance and sometimes demographic data to help identify the students who are likely to need additional help. The instructor can access this information from the beginning of term. These students who may need additional help (or, conversely, those who are above class level) can receive more tailored guidance from day one. In K-12, analytics are also touted as an ‘objective’ measure of student performance and teacher quality that can be used both in parent-teacher meetings (to back up a teacher’s point about a given student’s performance) or at the administrative level (to track a teacher’s efficacy in the classroom).

There are, obviously, a number of concerns with this approach as well. The first and most obvious is that of privacy and discrimination: should demographic data of any sort be used to identify students in need?

But there are other potential problems as well. One is the assumption that historical performance can be used to predict future performance. Sometimes a student has a bad year. This doesn’t mean that this student will always be a poor student, but that stigma may remain as long as the student stays within the system. Future instruction may then center around this student being ‘at-risk’ when that is no longer the case.

Despite these pitfalls, many districts are moving forward with investments in these systems, as are institutions of higher education. There is some preparatory work and overhead involved — not just software, but also data models, training, and related tasks. Is it worth the expense? We’d like to hear from you!

Read more:

About the Author: Jaclyn Neel is a visiting Assistant Professor in Ancient History at York University in Toronto, Ontario.

Lovers, dreamers, studentsAll ArticlesRag in A-Major