The other interesting thing they did was changing up the conditions of retrieval practice. Some students took an open-book test, where they literally had the text sitting in front of them while they answered the questions. Some took a closed-book test (no text available), but some were in a closed/open switch task. Essentially, students could either look at the text OR they could have the generative test questions, but not both at the same time. They could switch back and forth between them. Students were then given a delayed test to see the effects of the open- and closed-book testing.
The results were interesting:
1) Closed-book tests resulted in worse performance on the generative learning activities. This was likely due to the effect I described above. If students were unable to retrieve the information they needed to complete the activities then they wouldn’t perform as well on those questions.
2) Open- and closed-book tests resulted in the same performance levels. Because open-book tests had better generation and closed-book tests had better retrieval, each group got a boost for remembering information on a final test.
3) The switch group had the highest long-term learning. This group likely got both a benefit of retrieval and a benefit of generation/elaboration. While the retrieval was of shorter duration than a pure closed-book test might be, they had more successful retrievals, making them more able to complete the generation activities as well!
There are a few important takeaways from this study.