Jan 28 2020

6 Best Practices for Using Student Data for Student Success

Colleges must be conscientious in matters of access, disclosure and bias when developing data-driven interventions.

The 2020 list of EDUCAUSE’s Top 10 IT Issues provides evidence that the higher education IT community is increasingly focused on using technology to better understand students and rethink systems, culture and process to improve student success. This focus is especially critical when it comes to using student data to improve retention and completion rates.

To develop the capabilities and systems that can provide students with personalized, timely support, institutions need to understand how data technology, such as predictive analytics, can address the factors that lead to student success. They also must know how to use information from institutional data to introduce changes that promote sustainable, effective and efficient practices, while considering students’ experiences and needs as their starting point.

These best practices can help campuses balance the imperatives of serving a more diverse population and enabling success for all students.

1. Adopt Predictive Analytics and Other Advising Tools

Institutions have increasingly become more access-focused while recognizing the limits of their resources to serve more students well. Applying predictive analytics to advising processes could enable institutions to pinpoint the students who need a particular kind of help, without expanding staffing levels, thereby increasing their reach in a financially sustainable way.

Technology-enabled advising tools also have the potential to increase the quality of academic or counseling interactions. Advisers would gain capacity to have transformational conversations with students who have been identified as needing that critical support and guidance because more transactional activities, such as degree mapping and transfer articulation, can be completed with technology.

2. Take Care to Protect Against Implicit Bias in Algorithms

It’s no secret that implicit bias is present in all predictive analytics models, arising from the fact that human beings are the programmers behind those algorithms and the resulting decisions. As a result, algorithms may obscure the needs or accomplishments of particular groups of students and, at worst, extend the impacts of structural racism.

For example, some predictive algorithms use factors outside of the student’s control, such as home ZIP code, race or ethnicity, and high school. If an institution uses these factors for admissions or to identify priority students for advising or other support services, students who are already underserved may find themselves further excluded because these data points are using existing structural inequities. This is why Georgia State University, a leader in the use of predictive analytics, excludes any nonbehavioral or nonchangeable dimensions from their predictive models, such as living in a certain ZIP code.

It is important for institutions to be vigilant about these algorithm inputs to reduce the amount of human bias that is present. This can be achieved by working with an internal analytics team or in partnership with technology providers to continually refine algorithms to ensure equity in access for all students.

MORE ON EDTECH: See how Georgia State University used data analytics to close achievement gaps based on race, ethnicity and income level.

3. Train Employees to Use Data Insights Effectively

Campuses making the shift from an institutional mindset to a student-centered mindset are noticing ways they can become student-ready. The effective use of data can help with this endeavor. But to use data effectively, advisers, faculty and support staff must be trained to have deeper — and potentially more difficult — conversations with students based on the data insights.

These difficult conversations may cover topics such as nearing financial aid limits; challenging life situations, such as food or housing insecurity; or academic issues, such as not accessing the support and resources needed for success. In addition, advisers and administrators need to be trained to create opportunities for collaborative decision-making with students, rather than making decisions on their behalf.

For example, if a student has been flagged as at risk for maxing out his financial aid, it is important for the adviser to raise this issue and assist in finding alternative solutions. This is likely to be a sensitive issue that will require an empathetic and collaborative conversation between adviser and student. Subtle changes, such as being mindful of the language we use to refer to students, can make a difference in our student-centered culture as well.

For example, California just approved a bill to remove references to “at-risk youth” in K–12 schools and refer instead to students as “at-promise.” Universities are also beginning to adopt this terminology.

4. Ensure Students Have Agency Over Their Own Data

It is critically important to give students a choice to opt in or out of providing data to institutions and to educate students about how campuses are using their data. There are some pieces of information that students may not want disclosed without permission, including mental health concerns or tracking of the locations a student visits on campus.

Sacramento State University is implementing a pilot program to collect data from students who opt in to track what campus services freshman students are using. This data is used to improve the first-year experience, but it has no personally identifiable information. Rather, the university is looking at the class collectively to make improvements around campus, while giving students agency to decide whether or not they feel comfortable sharing their data.

DISCOVER: Find out why top college bodies are saying data analytics could be a budget booster for universities. 

5. Allow Selective Access to Student Data

Even on campuses that promote a holistic view of the student, very few staff or faculty members actually need access to all of a student’s data. Having data governance and custodial practices in place allows the right people to see the right data at the right time. Faculty might have access to one slice of information, such as the student’s attendance or participation in class. A residence hall adviser might notice the student’s behavior in the dorm. Maybe the office of accessibility and accommodations has still another slice.

Instructors, administrators, students and third-party vendors all contribute to the process of data production, whether it’s through a technology tool or observation and professional expertise. All of these parties should therefore have a shared understanding of the basic purposes and limits of data collection and review, so that campuses can balance the proactive identification of students who might need support with a sustainable and ethical approach to student privacy.

6. Create a Process for Comprehensive Data Reviews

There are issues or situations where institutions do need to bring all the data slices together to create a holistic student view. In these cases, there should be a process in place to bring individual staff or faculty members together as a committee to collectively make a decision. But these situations are relatively uncommon.

For example, in K–12, schools may have 504 plans that require teachers, support teams, administrators and families to come together with comprehensive data to better support a student’s learning. Cross-campus partnerships between academic affairs, student affairs, advising and IT can also support care for the student as a whole person. The University of South Florida, for example, created a Persistence Committee that uses data from its student information system and learning management system, as well as departmental expertise, to make real-time decisions to impact student success.

READ MORE: Check out our coverage of data analytics trends in higher education from the 2019 EDUCAUSE conference.

We may not have all the answers to complex questions regarding who owns student data, who gets to see the data or the best ways to ensure students get to make decisions related to their data. There aren’t yet standard frameworks or industrywide structures to guide us in answering these questions.

Yet, EDUCAUSE’s 2020 Top 10 IT Issues show we’re beginning to focus on the right questions. Paying attention to how student data is collected, how it is being used to make decisions and how we can build in student agency throughout the process should drive important conversations in the coming year.

These vital discussions will, in turn, help us find innovative ways to approach the transformation of our culture, workforce and approach to data technology for ensuring student success in the coming decade.

anyaberkut/Getty Images