The 0-100 scale is unfair

The 0-100 Scale is Unfair

A linear scale is fair and non-inflationary

A lightly edited version of this post was published in Times Higher Education (THE)

Photo by Kristen MacAdams

The 0-100 scale is used routinely in assessment in UK Higher Education. Students submit work and receive a mark. These marks accrue and determine degree outcomes. The problem is that the 'percentage' scale, such a familiar feature of assessment, is inherently unfair.

The unfairness comes at the top and the bottom end, because of the arbitrary degree classification boundaries that are a part of UK marking tradition. By convention, scores between 70-100 denote first class performance, while scores between 0 and 39 are classified as 'fails'. Between them, the first-class and fail ranges account for 70 per cent of the scale. Meanwhile most students are receiving marks that are crammed into the remaining 30 per cent. The traditional apportioning of the scale makes little sense. But the problems go deeper than that.

First, it causes inflation of grades and degree outcomes. In countering archaic practices that would see markers employing implicit ceilings for the marks that they would award, universities (urged on, rightly, by generations of external examiners) have established structures that encourage usage of the full range of available scores.

Very often, in pursuit of fairness, universities specify the marks that can be awarded for different levels of performance. The '2-5-8 system' is frequently used. This results in the award of 65 (for example) for a 'mid' upper-second piece of work, with 62 and 68 being available for minor up or down adjustments. This ensures that when two different markers make the same criteria-driven judgments about equivalent pieces of work, they will end up awarding the same score.

Translating this structure into the first-class range would set the top mark at 78, for a 'high' first. And if this was where universities landed, then the problems of inflation and unfairness would not occur. But why have a 100-point scale that stops at 78, and, for the same reasons, starts at 30? That would be fair, but stupid.

Instead, what happens is that things get stretched at the top end. Typically, we might see something like 72 being awarded for a 'low' first, 80 for a 'mid' first, '90' for a high first, and '100' for the odd exceptional piece of work. It's important that a university agrees on what these numbers should be, in order that some students do not fall foul of having their work assessed by a 'hard marker', while others, performing to the same standard, get higher marks. There's a whole tradition of unfairness, right there.

So, let's see what happens now that we have stretched the scale to make sensible use of the whole of it, and agreed on ways in which specific marks should be awarded fairly for work of the same standard. We will leave the fact that every university does this differently, such that students performing at the same standard at different universities will get different marks.

Take student A, who is, to date, averaging a high upper-second, equating to 68. For their next piece of work they improve, and achieve a high first, for which they receive 90. All well and good. We must value the improvements that students make in their work.

Meanwhile, on the same course student B is averaging a high lower second, equating to 58. For their next piece of work, they make a jump in improvement of the same magnitude, to a high upper-second. They are awarded 68. Well done them.

These two students have both improved their work in ways that an assessment system should value equivalently. But student A gets 22 extra marks to feed into their degree outcome, while student B only gets 10. The reverse effect happens at the bottom end, in fact, with students being over-penalised in ways that do not happen at the top end. There isn't a world in which this is fair.

All students who come to university, whatever their trajectory through a course, should have equal opportunities to learn and to improve. A student who is working to the best of their abilities in the middle of the scale, for example, should have access to the same rewards for improved performance as those achieving at the top end of the scale.

The inherent unfairness of the 0-100 scale is plain to see. But it is so familiar that we tend to look straight through it, leaving its structural problems hiding in plain sight. It wouldn't, in our view, survive a proper equal opportunities audit.

One could plausibly argue that improvements at the top end of the scale are 'harder' to make, and therefore worthy of greater reward. That position is not, as a rule, explicated in assessment systems, and is in truth an after the fact justification of an essentially arbitrary decision, sometime in our distant past, to make 70 the threshold for first-class honours. That's what causes this unfairness. And, indeed, that's also what makes the 0-100 scale feed grade inflation at the top end.

The answer is to change the scale to a linear one. For example, in a 0-16 ‘grade-point’ structure, all steps up and down the scale attract equal reward for improved, and equal cost for degraded performance. In the examples above, students A and B would both gain three extra points. That's fair. It also happens to be non-inflationary.

. . .