Giving students feedback on their work is a Good Thing. Students certainly think so: lack of feedback is the commonest complaint we receive about our courses (frequently from students who have never supplied us with anything to provide feedback on). The National Student Survey, which devotes 5 of its 22 questions to “assessment and feedback”, certainly thinks so. Institutions, which come up with policies called things like “Feedback Is A Dialogue”, certainly think so. They aren’t, in fact, all wrong: sensibly used, regular formative feedback from teachers is an extremely valuable part of students’ learning and it’s probably right that it attracts attention. Something I’ve not seen much discussed, though, is whether there can be such a thing as too much feedback.
To illustrate what I mean, consider a department — perhaps less hypothetical than I’d like — which is concerned that its students may fall into the trap of plagiarism through ignorance when they write their assessed essays. This department has solved its problem in the following way. Plagiarism is identified using the Turnitin software. Students are given access to this software and encouraged to run their essays through it before submission. If Turnitin registers a warning, the students amend the essay and try again. This process is repeated until the essay no longer sets off alarm bells, and is therefore no longer plagiarised. By a rather behaviourist process of repeated feedback, the student has “learned” how to “avoid plagiarism” and everyone is happy. The only slight problem is that what the student has actually learned is how to sail as close to the wind as possible, and nothing whatsoever about the fundamental issues that make us concerned about plagiarism.
Lest this be thought prejudiced against the humanities, here’s an example from the IEEE Transactions on Education. A study by Pérez et al. investigated the use of a web-based self-assessment tool in an engineering course. The tool generated random tests from a database of 500 questions; Pérez et al. found that a substantial number of students simply sat the test repeatedly until they had encountered all 500 questions and memorised the answers. (When questions drawn from this database were used in the final exam, the average mark was significantly higher than when comparable questions not in the database were used.) The sheer effort involved in this strategy still amazes me.
Computers, with their unending patience, might be more susceptible to this sort of abuse than human teachers, but it occurs. Ask anyone who’s supervised a weak but amiable project student. Without very strict rationing of feedback, of the kind that gets one accused of bullying or inaccessibility, a pattern tends to develop. The student produces some work, which is mostly wrong. The supervisor corrects this work and suggests some further extensions. The student makes the corrections and produces some further work, which is mostly wrong. The process repeats, limited only by the number of weekly meetings before the deadline. By this means, it is possible for a student to turn in an entire project of which every single correct statement has in effect been dictated by the supervisor. If Stockholm syndrome has developed far enough, or if the supervisor feels that the student has good reasons to be poor at English, it’s quite possible for the entire dissertation to end up being written by the supervisor through a painful process of inaccurate transcription and iteration. (I’ve heard the process of writing a paper with a student in this manner, sentence by clumsy sentence, described as being “like doing embroidery wearing boxing gloves”; there is also a cynical maxim, attributed to Adolf Hurwitz, that “a PhD dissertation is a paper of the professor written under aggravating circumstances”.)
What this continual detailed feedback means is that students never learn, because they are never forced, to provide their own feedback to themselves. They don’t learn to check their working, and develop the nose for errors that’s an essential part of a mathematician’s equipment. They certainly don’t learn to read their own sentences — why should they, when there’s a free proofreading service available for every draft? If they’re sufficiently gifted blaggers, they may of course be able to continue this well into adult life, like the academic authors who don’t seem to know the difference between a peer reviewer and a telepathic copy-editor, or the professors who have managed the supreme trick of delegating all concerns of accuracy to their juniors. On the whole, though, these students aren’t acquiring one of the fundamental traits of an educated person, which is to care, for its own sake, whether you’re right or wrong.
Frustratingly, I suspect several techniques to improve this situation already exist, buried within the policies and guidelines on supplying feedback. Phases of peer marking or self-marking might help, for example. However, I suspect they’re unlikely to while the message that we perpetually reinforce, though questionnaires and student liaison committees and policy statements and the whole institutional propaganda machine, is that feedback is an unmitigated good and that the more of it there is and the more detailed it is, the better.
One of the most helpful pieces of feedback I recall from my undergraduate days was written — as I recall — in red pen across the top of the front page of a rather scrappy piece of homework, and it read simply “Bollocks!” In my own institution today, it would have got the writer into all sorts of trouble, not just for inappropriate behaviour but for undermining my self-esteem and failing to specify how I could improve. (An aside: the policy I mentioned earlier states that good feedback will “encourage positive motivational beliefs and self-esteem”. Am I alone in thinking that there are some situations — as when teenage boys are learning to drive — when what is required is not to encourage self-esteem but to reduce it to realistic levels?) The effect this feedback had on me at the time was not what orthodoxy might predict: it made it very clear that it was my job to hand in something worth marking, and that if I failed to do so this was nobody’s problem but my own. I don’t resent the lesson: those nine characters probably had the highest information density I encountered in three years. I just wish I could see a gentler and more ostensibly professional way to pass the message on.