Elektrine lite

← Feed

@joepolitz@types.pl

2022-12-29 17:03 UTC

I've done video exams for a few years now, and finally analyzed some (with Rachel Lim and Mia Minnes): Post: https://jpolitz.github.io/notes/2022/12/27/video-exams.html Full paper: https://jpolitz.github.io/docs/sigcse-2023-video-exams.pdf We see misconceptions in video traces that are hard to see in static artifacts because they have to do with the *order* students did things in. For example, the screenshot on this post has 3 different orders used to fill in values in traces of the same kind of loop. Strategy B matches Java's semantics best, and there are hints that submissions of students in that group did better on the exam (and no, trace ordering wasn't on the rubric!). From this analysis, I can think of loops to trace that distinguish these strategies. We can poke at if helping students pick a trace strategy helps with other learning, and I can highlight in class. I wouldn't have noticed without these videos in the loop in their assessments.

Replies (0)

No replies.