In a co-authored WonkHe article, UCL School of Management Professor Susan Smith and Associate Professor Neil Sutherland delve into the issue of whether the algorithms used to determine degree classifications in UK universities contribute to grade inflation.
In the realm of higher education, the journey toward a degree is often perceived as a standardised path, culminating in a final classification that ostensibly reflects a student’s achievements. However, behind this apparent uniformity are the intricate algorithms wielded by institutions to determine these classifications. In their exploration of this process, Susan and Neil raise a crucial question: do degree classification algorithms fuel grade inflation?
Essentially, these algorithms take a student’s module-level results and calculate their final grade, determining whether they graduate with a first class, upper second, lower second, third, or fail.
The problem arises from the fact that different universities use different algorithms, leading to potential disparities in the final grades awarded to students. This raises concerns about fairness and equity across institutions, especially considering that the UK’s framework for higher education aims for equivalence between degrees from different universities.
They highlight that there’s little empirical evidence on whether these algorithms adjust student outcomes uniformly across the sector or whether they exacerbate grade inflation. They note that a significant portion of institutions have tweaked their algorithms over the years to ensure fairness, but the impact of these adjustments on students’ awareness and behaviour remains unclear.
The focus on final degree outcomes overlooks the role of these algorithms in potentially inflating grades at the modular level. They propose a closer examination of how these algorithms contribute to grade inflation and suggest that reevaluating the modular outcomes may be necessary.
While grade point average (GPA) systems are common internationally, the UK has not adopted them widely due to variations in calculation methods. They argue that addressing the limitations of degree classification algorithms requires gathering more data and considering options ranging from adopting a single algorithm for all institutions to reducing variance in existing algorithms.
Their paper, ‘Opening the black box of degree classification algorithms: Towards a research agenda,’ aims to spark further research into these issues to better understand and address concerns surrounding degree classification algorithms.
Read the full WonkHe article.
