Like most areas of the educational world these days, technology is forcing higher education institutions to do more with less. Institutions are under increasing pressure to admit more students, retain these students, and do their best to ensure student success. Facing this pressure, tech-savvy institutions can benefit greatly from predictive analytics and predictive models to help achieve their goals.
This article provides an overview of predictive modeling and how it can be used to assist higher education institutions in predicting certain behaviors, specifically regarding enrollment. It also discusses the common types of predictive models, and what institutions that utilize these models can expect if data is applied correctly.
What is Predictive Modeling?
At its highest level, predictive modeling is a statistical technique that is used to determine the probability of an individual performing a certain desired behavior. In terms of predictive analytics, it’s one such statistical technique that analyzes current or historical facts to make predictions about the future, along with machine learning or data mining, for example.
For higher education institutions specifically, predictive models can help make predictions around enrollment, fundraising, marketing, and student success. Take enrollment for example: institutions can analyze the behavior of previous students and identify variables that influenced their enrollment behavior. Once institutions identify the relevant predictor, independent variables can then help construct a statistical model that will predict future behavior.
Once this predictive model is developed, it’s applied to prospective or enrolled students to determine the likelihood that they will exhibit a desired result—apply, enroll, graduate, etc. Institutions can then tailor their marketing or recruitment in ways that will maximize human and financial resources.
How can it help in Higher Ed?
In addition to maximizing resources when it comes to enrollment as discussed above, predictive modeling also helps colleges and universities identify students who are likely to succeed, as well as students that are more at-risk of not doing well. Armed with this knowledge, institutions can target support accordingly and not made assumptions about behavior. They can use data to their advantage.
This was the focus of a recent New York Times article on the subject. In it, the Times points out that while a little less than half of college students graduate in four years, the number only increased to 60% after six years. This is putting more pressure on schools to bring up their graduation numbers, and predictive modeling is helping with this.
Specifically, the article points to Georgia State University, the University of Arizona (UA) and Middle Tennessee State University as examples using various methods of predictive models. Georgia State is using it to help nursing students who are at-risk of dropping out; at UA data showed that English professors needed to develop more resources to writing. And at Middle Tennessee, predictive models determined that a D in a required history course was the most common denominator of students who didn’t graduate, so the school devoted more resources to this area.
What Tools are Available?
The use of predictive models in higher education is becoming increasingly popular throughout the nation, with a number of vendors helping to build models and predictive tools. However, this doesn’t mean that there is a one-size-fits-all option for most institutions. In fact, it’s quite the opposite and institutions must be weary as to which service they decide to use to help build their models.
Although many vendors are transparent about their models and algorithms, and allow institutions to have direct involvement in the design process, not all vendors take this approach. Institutions must be very familiar and with their current practices and know exactly what they want to achieve, and should become knowledgeable about predictive models and analytics before partnering with a developer.
One tool that works very well and is widely used at many institutions is Einstein Analytics. This application automatically analyzes billions of data elements to provide a portfolio of relevant, self-service apps to unveil insights of student behavior based on previous data. Envision is another great option, as it’s widely viewed as the most accurate predictive models in the industry.