Using study dates - better good than haste

Study data help students and lecturers alike

On the subject of study data, Martijn de Hamer is on his guard from his position. As data protection officer, he oversees the privacy and data protection policy of the University of Applied Sciences (HvA). And study data involves processing quite a lot of data. "That's why a thorough approach is badly needed," Martijn explains. "As an institution, we very clearly see the potential of study data. But precisely because there are so many risks involved, we are deliberately going slow".

The purpose of using study data is clear for the HvA: to improve education, to guide students better through their studies, in such a way that they are optimally prepared for society. In doing so, study data can have a signalling function, for example. Suppose a student is never present and does not complete the assignments in the digital learning environment (DLO), then this is flagged up and you can start a conversation with that student. In addition, lecturers can use study data to improve their teaching. For instance, they can use the data to investigate whether a certain approach works or not. In this way, they can work more and more efficiently.

First a strategy

"So study data is going to help us, and anyway we have to keep innovating," Martijn says. "But because of the risks, we chose to take it slow and move steadily forward." Martijn came to the HvA in 2018. By then, initiatives were already underway and plans were in place, but in consultation with the Chief Information Security Officer (CISO), Martijn made the recommendation to start writing a strategy first. This had to spell out exactly what the HvA did and did not want to do with study data. For that strategy, quite some time was spent and many people were interviewed.

Students must be able to make mistakes

"We wanted to proceed carefully and create support," Martijn explains, "Because it is a lot to process with study data: in the DLO, you could see everything students, but also lecturers, do online. It is therefore very important to look closely at aspects such as privacy, protection of personal data, but also self-determination: these are fundamental rights. If you want to excel as a student, you need to be free and able to make mistakes without being caught off guard. Otherwise, the so-called chilling effect occurs: you start correcting your own behaviour when you know you are being watched. Think of the student who prefers to study between 12 and 3 in the morning; he might stop doing that if he knows a lecturer can see that and might think something about it. So: as an organisation, we want to educate students efficiently and well, but it has to be done in a safe environment. We therefore consciously choose to make as little data as possible visible, only the ones that are really necessary to properly guide the student. Incidentally, this is also in line with what the GDPR prescribes."

Algorithms are opaque

There are more challenges when it comes to student data. Martijn: "One such challenge is posed by algorithms. More and more information will become available from so many systems, about so many students and over an increasingly long period of time, that eventually you won't be able to process it manually. So I expect that algorithms will make their appearance in student data. The problem with that is that the explainability of decisions will disappear. Algorithms are often opaque, so you cannot explain to a student how the organisation would arrive at a certain advice based on the system, for example. We need to think about that right now."

First step towards practice taken by corona

The strategy is now ready and approved. It includes guidelines such as: keep an eye on the original goals and do not process more data than you need for that purpose, do not use analyses from this data for assessments and avoid excessive control of the student or lecturer.

The HvA then started to design an infrastructure for study data. To do so, they are working with short experiments in which they explore what you can do with study data analysis. One such experiment has already moved beyond the experiment stage. "Influenced by the corona pandemic, the HvA turned on the class progress module in Brightspace," Martijn explains. "The module shows how far the student is in processing the class material, but only as a percentage. The lecturer does not see what the student has done or not done exactly. This was an interesting, concrete case to really use study data in practice. By turning that module on, we started processing more of the students' personal data, and for that you have to have a good reason for the GDPR. By corona, we had that reason, namely that we need to be able to properly supervise students despite homeschooling. By the way, we have set up the module very conservatively: lecturers can only see the percentage of learning that the student has processed."

Guide makes GDPR concretely applicable to student data

When setting up the infrastructure for study data, the HvA makes a lot of use of the guide on study data and the GDPR written by SURF. Martijn: "We like the step-by-step plan, because it describes in concrete terms the steps you need to take to process personal data correctly according to the GDPR. It's also nice that the step-by-step plan focuses on study data. It covers the relevant GDPR articles and is written in clear language: our colleagues read in the guide exactly what they need to do in their own context to apply the GDPR properly."

Suddenly it's there

So there is no doubt that the HvA will really deploy study data. "But better good than hasty," says Martijn. "I see it this way: developments in the field of study data are slow, nothing much seems to happen for a long time, but there comes a moment when it is suddenly there. And at that moment, we as HvA must also be ready for it, so that we can use the study data ánd ensure that everyone continues to feel the freedom to study and work in their own way."

More information

• Getting started with student data yourself? Read Learning analytics in 5 steps: a guide to the GDPR
• Read more about the study data acceleration zone initiatives of the Education and IT Acceleration Plan