Universities must ensure inclusive analytics

Data analytics and machine learning “quietly power” decision making and predictive process in universities. Which can be good for administrators but not all students, “if adopted uncritically they can also amplify social inequalities and historical injustice, often by stealth,” Bret Stephenson (La Trobe U), Andrew Harvey, (Griffith U), and Qing Huang (LT U) warn in a new report for the National Centre for Student Equity in Higher Education.

Data as oft argued, can be used for good, discovering discrimination in admissions and identifying students as individuals, not as part of an equity group but, the authors argue, “uncritical adoption, and a failure to maintain effective oversight, can result in a dramatic undermining of equity goals.” And they warn, “analytics-driven interventions may be either ineffective or even counter-productive, in some cases leading to self-fulfilling prophecies of failure.”

They recommend nine measures needed to build “inclusive analytics, including;

explicit equity protections be built into institutional data: “the ‘peril’ posed to equity interests in universities is not limited to teaching and learning activities but extends out beyond the university through pre-enrolment marketing and post-graduation employability projects and digital tracking of alumni”

diversify analytical expertise: “domain experts in non-ICT related fields – librarians, student advisers, recruitment officers, teachers/lecturers, etc. – are indispensable” as are ethics and equity practitioners

governance oversight: for example, “standing committees to oversee analytics, similar to ethics committees”

don’t build assumptions about groups into predictions and monitor projects in process for fairness

And above all there is culture, “In sum, we argue that Australian universities, and equity advocates within these institutions, should advocate for a fully democratic, and decidedly human and “in-house”, process of machine learning and artificial intelligence.”