The panellists were first asked about their data delivery priorities. Owen Beer, business transformation officer at Nottingham Trent University, described the opportunity for data to create a joined-up view of an institution’s performance.
“We want to have the assurance that everything is OK right across the institution – and to be able to target and focus on the right issues. We need to use data as evidence to challenge where things aren’t going so well,” he said. This capability would entail a shift away from “the current line of questioning around data accuracy and cleanliness to using data to begin a conversation”.
“Once we’ve cracked [digital] trust, we can challenge people’s reliance on gut feeling and become more evidence driven,” said Kevin Pickess, academic registrar at the University of Worcester, of future decisions by management and academic staff.
For students, the promise is to use predictive data to make relevant interventions in their individual learning journeys. “These would be far more targeted than previously was possible,” he added.
The conversation then turned to student well-being, a theme that surfaced repeatedly throughout. “Data will help us predict which students will perform well and identify the underperformers, not only academically but in terms of their well-being,” said Sander Kristel, executive director of UCAS Media.
“At present, as a sector, we don’t have the answers, but data would help us to formulate interventions.”
At present, there is regulatory requirement to collect well-being data, despite many universities already having rich data sets, such as attendance records. With the arrival of artificial intelligence, machine learning and data mining of free text, the capability will become more advanced, he said. “We have to be careful not to do intervention in a way that is creepy. We need permissions.”
Obtaining consent to improve student well-being as a data deliverable was a concern raised by Graham Nicholson, director of student services at the University of Dundee. “The focus is on understanding and intervening at the appropriate moment – and with consent.” In Scotland, some institutions get consent and others don’t, he said, and asked if there was a middle way.
One trend that is informing the discussion about student interventions is the way in which millennial students share their data and are keen for institutions to use them. This was confirmed by Kevin Mayles, head of quality enhancement and learning analytics at the Open University, which surveyed staff and students on data sharing.
“Students were open and keen for institution to use data, while academics were more cautious about the use of predictive data and behavioural data,” he said, of his research about context for data use.
“The Open University has adult learners and this changes the conversation with the student body, as intervention may entail encouraging students to take a managed break from study. Context is everything.”
The other aspect of using data ethically and in the students’ best interests entails data protection and privacy. Here, the General Data Protection Regulation has provided a major piece of work for generating trust, said Paul Colbran, chief information and infrastructure officer at Solent University.
“If students are more willing to share data across social media, then they’ll share with institutions who use them in their best interests.”
Panellists agreed that GDPR is helpful in progressing data-driven higher education, introducing transparency and requiring organisations to show that they are using data fairly and legally. More debate was devoted to discussing the role of other regulations, such as the teaching excellence framework, the National Student Survey and Data Futures, in advancing data capabilities and outcomes.
For students, data used in context are helpful in making a decision – and benchmarks such as the THE University Rankings and the NSS offer institutions a nudge, said Darren Tysoe, chief information officer at Regent’s University London. But he was wary about how benchmarks are used because “you can’t squash all the unique parameters of an institution into two or three measures to deem the quality of a university”.
Other bespoke metrics are more useful to Regent University’s unique proposition, he explained. “We’re located in a royal park and use metrics to allocate physical space on the campus.”
All academics want a view of the park and less glamorous rooms are less well used. Students record their physical whereabouts on campus and this is tracked. “Timeliness of data is everything. If it’s not in real time, we lose the opportunity for interventions,” he said.
Other participants voiced concerns about using historic data for data analytics conducted with current-day and evolving purposes in mind. For Mr Mayles, this “data lag” poses a particular problem.
“We’re an open-entry institution and need a broad, participative agenda. There’s a danger, if we’re not careful, that decisions are based on evidence and data according to the way we’ve done things and recruited in the past.”
In conclusion, Mr Baker asked participants for a wish list that would make data-driven higher education doable.
Requests included more investment in skills to interpret data – and a plea for Data Futures to recognise that universities deliver their curriculum in different ways and report in different ways.
A consensus emerged that much work was still to be done in order to bridge the cultural chasm between universities’ business and their academics. If the goal is a real-time, cross-institution system, a priority should be to engage academics and teach them the value of data.
First published in Times Higher Education