Some of my old lab mates are getting some well-deserved press for their education research. Their message that students should be challenged is good and very worth reading and taking to heart.
But I would argue that their study doesn't support their claims that errors are not harmful. Specifically, I think if you delayed the test phase then the initially generated error would trump the later-learned correct answer. (As would be predicted by the NToD dynamics, and even Jost: the guessed-answer gets a bigger initial RS and SS boost from the potency of guessing, and it came first.) So, as per the UCPD training I used to do, you give the trainee a chance to guess but if they are about to produce an error, you step in with the correct information.
Also, their baseline comparison is off - sure generation produces better recall than passively provided answers, but how would it fare against the generation of correct vs incorrect information?
Note to self, what would Catherine Fritz think?
|comments: talk to me|