Experiments with known answers

Earlier this week I read through a student’s work placement report. Our engineering students all go out on two work placements over the course of their study with us, and need to provide reports on these. I was slightly amused to read about the student’s views on the novel experience of doing experiments where you don’t know the outcome before you start. What he meant was, for the first time in his training, he needed to undertake an experiment where he couldn’t go and look up the outcome in a textbook.

That I think is something worth thinking about. Nearly all the experiments that students do in class are things that have been done many, many times before. There’s good reasons for that – they are (mostly) reliable, the equipment is available at sensible cost, the experiment can easily be discussed in terms of underlying theory, it’s not dependent on the expertise of a single teacher that will be lost if the teacher leaves, it saves planning new experiments from year-to-year, and so on.

But it also means that students can look-up what the answers should be. That’s got a couple of clear problems: 1. The student loses their ability to critically judge the credibility of their own work and 2. there is the temptation to ‘cook’ the results so you can get a better mark. The first of these is a real skill that a scientist needs to develop – if you can’t look critically at what you’ve done and decide whether you trust the results then your science is on shaky ground. In real science, there is no answer to look up in a book – so how would you know if you’ve made a mistake?

The second is an interesting point, too. I would say if students find that they are getting better marks by making up results to conform to what the textbook says there is something going wrong with the marking process. Assessment should always be done with a view to promote learning. Copying things out of a book is not learning. I’d rather have a student try and fail to produce valid results, but, at the end of the experiment, understand what they were doing wrongly and why, and be able to correct it next time, than just to produce a graph that looks nice. Any experimentalist knows that most of his or her lab time is taken trying to troubleshoot the experiment. So why do we set experimental labs for our students that teach them otherwise?

This isn’t an easy problem to get around, especially when you have a huge class of students. I tackle it with the second and third year classes by marking a student’s work with the student present (in fact, getting them to mark it themselves first, then tell me why they deserve the mark they think they do). That way we can talk through the difficult issues and they can see that I’m not just after graphs that look good.  But in a first year class, where there are 100 or more students to shuttle through in a week, that’s not practical.

The placement student was very happy to have the opportunity to tackle a real experiment. Perhaps the answer lies in bringing in our research more into our teaching – getting the students to do some real research things.


Leave a Reply