I attended the annual symposium of the University of Copenhagen’s Center for Internationalisation and Parallel Language use (CIP), which, as always, gave me a lot to think about. There was an interesting presentation about the importance of feedback, for example, which led to a no less interesting discussion. Nina Nellemann Rasmussen and Janus Mortensen had done a survey of their writing students which found, not surprisingly, that students found it generally useful to get feedback and more useful to get feedback from their teachers than from their fellow students. Someone asked the very relevant question of whether this showed in the students’ work, or whether this was just a matter of the students’ perceptions. The study had indeed only measured the students’ own perceptions of the utility of the feedback.

This led me to propose that we imagine a large-scale randomized controlled trial. Suppose we take a cohort of 600 students and divide them randomly into three groups (A, B, C). All three are given the same lectures and the same final essay exam. All three groups are given a pre-test to set a baseline against which to measure improvement, and are given deadlines to submit work throughout the semester, but while one group (A, the control) is merely given a pass/fail on the submitted text, the other two are given feedback. In group B, the students have to give each other feedback, while in group C they receive feedback from the teacher. After the final grades are given, the final test is compared with the pre-test and are evaluated for signs of improvement. The question is, in which group will we expect to see most improvement?

I think group C will marginally outperform group B, but B and C will significantly outperform group A. It’s possible that four groups would be needed to test whether giving feedback itself improves performance. (This raises the possibility that group B would outperform group C because the act of giving feedback is actually worth more than receiving “qualified” feedback from a teacher.)

In other words, I believe feedback is important. But I’m actually not so sure that it has to be very “expensive”, i.e., that it has to be provided by teachers. I do think students value teacher feedback more than peer feedback. But I guess I’m saying I think this may be an overly generous valuation. Even if it is worth a bit more to the student (in terms of actual improvement) it may not be worth the cost of the teacher’s time given to all the students. 80% of the value, perhaps, can be provided by peers.

I actually think the point goes deeper. I deliberately gave group A submission deadlines that are not strictly necessary (since they are neither getting nor receiving feedback) in order to remove the confounder that the regular writing practice implies. (If one difference between group A and the other two is that B and C wrote more during the semester, the comparison won’t tell us anything.) Another 80% (of the improvement independent of giving feedback), I suspect, comes simply from practicing. Finally, I think a substantial proportion of the improvement can also be predicted from characteristics of the students themselves (though this would ideally have been controlled for by randomization).  This leaves writing instructors with the following somewhat unhappy hierarchy of what improvement in writing depends on:

1. Character
2. Practice
3. Feedback
4. Instruction

In other words, I think we might be spending too much time trying to figure out what to tell the students about writing, not enough effort figuring out ways of getting them to write and seek feedback from their peers. Instruction should be organized around these student-based activities. The efforts of instructors are wasted if the students are not writing a great deal and devoting time to reading each other’s work. We have to remember that writing instruction is ultimately a species of coaching. It’s not what you tell the students that matters but what you can get the students to do.