Although students tend to think that exams are merely an evaluation tool, a lot of research demonstrates that testing actually helps with the learning process. Researchers Pyc and Rawson published a short article in Science exploring why testing is beneficial to memory. They found that testing helps students develop more effective mental “mediators” – or ways of remembering the material that they study.
The researchers had students study several Swahili-English translation word pairs over the course of a study session. They were then given the same words to study again at three subsequent study sessions. At each study session, the students were asked to write down a “keyword” (a mediator) that helped them remember each translation. For example, for the Swahili-English translation pair “wingu – cloud”, many students came up with the keyword “wing”. Wing sounds like wingu, and makes one think of flying… sky… clouds.
At the last three study sessions, half of the students spent all their time studying, and half of the students were given a practice test before they were allowed to study. One week later, all the students were given a final exam. As expected, the students that had had pretests during their restudy sessions scored better than those that didn’t (40% vs 15%) without being given any hints.
When students were given the keyword that they’d written down while studying as a hint on the test, the students that had had pretests only did about 5% better. However, this hint helped the non-pretest students quite a bit (improving their score from 15% to 30%). This result indicates that testing as part of the study process helps students get better at remembering mediators – the pretest group was already pretty good at remembering the keyword “wing”, so being given that word as a hint didn’t help much. In fact, if the test explicitly asked the students to write down the keywords they’d come up with while studying, students in the pretest group were able to remember about 67% of their keywords, compared with only about 32% from the students that didn’t have any pretests.
Furthermore, the fact that non-pretest students didn’t perform as well as pretest students, even when given their keyword as a giant hint, indicates that testing helps students get better at mentally decoding their mediators (e.g., coming up with the word “cloud” when they think of “wing”, rather than drawing a blank).
This research applies specifically to keyword mediators, so I’m not sure how directly applicable it is to learning things like reaction mechanisms in chemistry or how to solve an algebra equation. Nevertheless, it’s consistent with what teachers always say – to test yourself while you study by trying to solve problems without looking at the solutions manual.