But (to illustrate the tension) and RA Bjork ran an empirical study on the effects of students' choices to drop flashcards and found "small but consistently negative effects on learning." web.williams.edu/Psychology/Fac
Conversation
Of course, it's important to note that the broad term "learning" here is used to refer to "accuracy." Trading off accuracy for not-being-bored may actually be net positive for learning writ large!
… though, given the reported behavior, it's not clear that's what was happening.
2
5
It's always hard to know how to interpret attitudinal survey responses in studies like this. The participants were learning Swahili vocabulary provided by the investigators. How would their behavior differ in a personally-motivated setting outside formal education?
1
3
That's one reason why I'm excited about learning from Quantum Country readers: it's a serious context of use, and participants are (so far!) people who've chosen to study quantum computation.
2
1
4
Another (2020) paper on this tension from Daphne Whitmer and colleagues, with a more authentic scenario: andymatuschak.org/files/papers/W
In-training US Marines used SRS to learn to identify armored vehicles. Group who could drop cards remembered much less well.
1
9
I wonder why No Drop did so much worse than Mastery Drop in the delayed post-test. Any intuition you on this from Quantum Country?
1
The relevant diff is between the delayed samples, which are within each others’ error bars.
1
1
Well, Whitmer and all think the difference is relevant:
“We presented preliminary evidence that using mastery criteria as an AT strategy to drop cards promotes long-term retention compared to not dropping cards during study.”
2
I don’t think that conclusion is supported by their data. They find no stat sig difference in delayed retention, delayed transfer test performance, or # of mastered objectives between the two groups. N=47 tho so ¯\_(ツ)_/¯
1
1
Hm, you're right. Given that you have looked at data with QC that's much larger N, what's your intuition - would this bear out or is it unlikely that there's a difference in the real world?
1
Replying to
Hard to say with confidence, since we don’t quite test that. A related observation: there are a few prompts which ~20-30% of readers have trouble remembering beyond a few weeks, even w many repetitions. Long intervals sort of simulate mastery drop—for the worse in these cases.
But it’s not really a fair comparison. A better observation is just that the SRS curve should be shallower for these prompts, or that these prompts should be refactored. Also, I suspect readers *know* these are their “tough” prompts.
1