Degree grading

15 February 2010

Degree classification is something that crosses my radar every now and then, most recently in the FT article PwC mounts backlash at "2:1 tyranny". time to let off steam..

The history

In the olden days the idea was that the distribution of marks was approximately a Gaussian distribution with most of its weight within the Second class region; a few people did very badly and got thirds, and a few did very well getting firsts, but between these two extremes people did not really differentiate.

In practice the spread around the 2:1-2:2 divide is lopsided, with the tailing off on the 2:2 side a lot faster than the 2:1 side. In fact a square wave with its rising/falling edges at 58% and 72% is a better approximation. There are several reasons for this:

If people are not on track to get a 2:1 there is an increased tendency for them not to finish the course at all. Certainly in places like Bristol getting a degree is much more expensive than it used to be, and students are more inclined to cut their losses. Flunk rate figures have a dubious reputation for accuracy, even before considering reporting of transfers between courses.
Standards more rigidly enforced.
Within Engineering at Bristol, minimum standards on the 4-year courses are enforced more rigidly than in the past. Basically anyone not on track for a 2:1 has to drop down to the 3-year course, and if those who do make it into the 4th year bomb that year they tend to be given a B.Eng 2:1 rather than an M.Eng 2:2. This is a big incentive, as in most engineering subjects, those with only a B.Eng cannot get UK accreditation.
Putting in that bit extra work
Those who know they are in the high-2:2 region also know that a few gradings at 12-13 (out of 20) rather than 10-11 may well be enough to push them over the 2:1 threshold. A similar effect can also be seen with students close to the 2:1-first border, where people have been known to work 14-16 hour days. This is particularly common in coursework-heavy degrees such as Computer Science where students get quantitative feedback on their progress frequently throughout the year.
Of course this does come at a very heavy personal cost, with substance abuse (everything from caffeine & tobacco to alcohol & MDMA), heart-break, and total mental breakdown (I know several people who ended up requiring medical exemption from exams) being all too common problems. Suicides and accidental deaths where stress was a contributing factor are surprisingly common.
Self-fulfilling prophecy
A lot of people go to university because a degree is a job requirement, and the common company view of the 2:1-2:2 split has filtered its way back into some undergraduate marking. Increased workload as the result of extra students has meant the decision of whether to award a 2:1 or 2:2 for a piece of work is down to whether the student appears to have put in proper effort, and then objective criteria are considered to decide on the final mark.

Why fine-grained degree classification is a bad idea

It does not solve the problem. If degree classifications was much more finely grained, companies would go for the cheap option of filling N places by taking the top N ranked candidates. Some companies have already gone to the extent of requiring firsts, let alone 'good' 2:1s. It might make people who worked flat out and got >75% feel hard done, but they are relatively few and far between (and yes, I was one of them).

Leaving aside differences between departments/courses, the biggest problem with fine-grained marking is that it makes the system sensitive to errors. The most obvious source of such errors is inherent variances when a coursework marking load is split among multiple markers. The problem gets worse because the difference between 64% and 67% becomes significant, leading to a large increase in students challenging their marks.

So what about those with 2:2s

It tends to be the companies that do not care what degree you have that make 2:1s mandatory. Many engineering companies only specify 2:1s for specific roles, because they realise that it is on-the-job training that actually matters. I've known people who have previously done internships to have 2:2s and even thirds overlooked.