I tried a new way to give a test on my students last week. Instead of a test that lasted an hour, I divided the test into two halves. After finishing the first half an hour, I collected the papers and let the students study for an hour, then I gave them the second half of the test. The students were informed of this arrangement a week before the test.
The students did not know before hand, but part of the questions (S) in the second half of the test had appeared in the first half, and the rest (M) were modified slightly. I had expected that the students would study the material that they had difficulties with. Because of this, and that they had more time to consider the questions, they should be able to do better on S. But M might actually confuse them, therefore they might actually do worse.
The results were pretty much as expected. The average mark on the questions that were the same (S) improved by about 7.5%. Among the 64 students, 47% did better on S, 38% got the same marks, and only 14% did worse. Among the students whose marks were below average in the first half, 55% improved on S, and only 10% did worse. Among those whose marks were above average, 38% improved on S, and 15% did worse. So it can be argued that the weaker students benefited from this experiment more than the stronger ones.
But the drop in the average mark for the modified questions (M) - a drop of 17% - was quite a bit larger than I expected. Perhaps, when I tried to make the questions different, I actually made them harder.
During the break, I observed that many students were studying hard, as expected. Several actually came up to asked me about some of the questions in the test. I answered the questions openly and used the white board, so that all those who wanted to, could follow our discussions. This was partly to try to be fair to everyone, and partly because the purpose of the experiment was to help them to study in the first place.
I also noticed that some of the students (more than 10% of the class) were just sitting there, or taking a nap. Some of them told me that they felt it might be better to rest, so that they could be fresher for the second half - which may not be a bad idea. What I could not determine was whether there was any correlations between what they did during the break and their performance on the second half of the test. I would love to find out next time.
It was a fascinating experiment, for me anyway.