Datafication of education: why values need to be at the centre of reviews of the Ofqual algorithm debacle

Friday, August 21, 2020

Category: Blog

By Dr Rebecca Eynon (Associate Professor between the Department of Education and the Oxford Internet Institute) & Professor Jo-Anne Baird (Director of the Department of Education)

Now that the infamous Ofqual algorithm for deciding the high-stake exam results for hundreds of thousands of students has been resoundingly rejected, the focus turns to the importance of investigating what went wrong. Indeed, the office for statistics regulation has already committed to a review of the models used for exam adjustment within well specified terms, and other reviews are likely to follow shortly.

A central focus from now, as students, their families, educational institutions and workplaces try to work out next steps, is to interrogate the unspoken and implicit values that guided the creation, use and implementation of this particular statistical model.

As part of the avalanche of critique aimed at Ofqual and the government, the question of values come in to play. Why, many have asked, was Ofqual tasked, as they are every year, with avoiding grade inflation as their overarching objective? Checks were made on the inequalities in the model and they were consistent with the inequalities seen in examinations at a national level.  This, though, begs the question of why these inequalities are accepted in a normal year.

These and other important arguments raised over the past week or so highlight questions about values. Specifically, they raise the fundamental question of why, aside from the debates in academia and some parts of the press, we have stopped discussing the purposes of education. Instead, a meritocratic view of education, promoted since the 1980s by governments on the right and left of the spectrum has become a given. In place of discussions about values, there has been an ever increasing focus on the collection and use of data to hold schools accountable for ‘delivering’ an efficient and effective education, to measure student’s ‘worth’ in ways that can easily be traded in the economy, and to water down ideas of social justice and draw attention away from wider inequalities in society.

Once debates about values are removed from our education and assessment systems, we are left with situations like the one now. The focus on creating a model that makes the data look like past years – with little debate over whether the aims should have been different this year is a central example of this. Given the significant (and unequal) challenges young people have faced during this year, should we not, as a society have wanted to reduce inequalities in our society in any way possible?

The question of values also carries through into other discussions of the datafication of education, where the collection and analysis of digital trace data, i.e. data collected from the technologies that young people engage with for learning and education, is growing exponentially. Yet unlike other areas of the public sector like health and policing, schools rarely have a central feature in policy discussions and reports of algorithmic fairness. The question is why?  There are highly significant ethical and social implications of extensive data use in education that significantly shape young people’s futures. These include issues of privacy, informed consent and data ownership (particularly due to the significant role of the commercial sector); the validity and integrity of the models produced; the nature of the decisions promoted by such systems; and questions of governance and accountability. This relative lack of policy interest in the implications of datafication for schooling is, we suggest, because governments take for granted the need for data of all kinds in education to support their meritocratic aims, and indeed see it as a central way to make education ‘fair’.

The Ofqual algorithm has brought to our attention the ethics of the datafication of education and the risk that poses of compounding social inequalities. Every year there is not only injustice from the unequal starting points and the unequal opportunities young people have within our schools and in their everyday lives, but there is also injustice in the pretence that extensive use of data is somehow a neutral process.

In the important reflections and investigations that should now take place over the coming weeks and months there needs to be a review that explicitly places values and ethical frameworks front and centre, that encourages a focus on the purposes of education, particularly in times of a (post-) pandemic.