I ran a cycle of in-class quizzes for the first time this past semester, substantially inspired by ideas you’ve presented. It went well! Thank you for all your advice.
The course was a mid-level math course: Intro. to Analysis with around 70 students. I split the material into seven chapters and offered three quizzes on each, and marked them as “not yet” (notated by a 0 in Gradescope), “progressing” (a 1), or “proficient” (a 2), with only the strongest performance counting to the course grade.
I didn’t see test-anxiety problems. Students reported that the approach was less stressful than the traditional prelim-prelim-final. I think it helped a lot that the quizzes followed a closely consistent style, so, while the questions varied, the students quickly learnt what sort of challenge to prepare for. It was immensely heartening to see them improve over the three attempts, often dramatically.
The two biggest problems I ran into were cheating and the logistics of mandated accommodations for some students. The cheating issue was disappointing to see, especially given that most students worked extremely hard and with absolute integrity. It was exacerbated by students being close together in class and by some of the questions being multiple-choice, where answers are visible at a glance. To combat that, I sometimes circulated two versions of a quiz. As for the accommodations, I kept the quizzes short and gave all the students as much time as they needed (within the bounds of practicalities), which substantially worked. Both of these issues added substantially to the burden of running the course.
There were students who struggled, but I didn’t perceive a big “snowball” problem. The first attempt on each chapter was not long after that material featured in class, and I think it helped them to have to grapple with it at least once in a timely manner. I think that with a “traditional” grading scheme, the same snowball problem can be present but it can be worse, because the gaps between tests mean it isn’t exposed so frequently.
Incidentally, I saw an opposite sort of effect. I mostly gave quizzes two at a time: one on the most recent chapter and one on the chapter before that. A couple of students would deliberately bomb the most recent chapter, just using it to get a sighter on that material, but they would nail the prior chapter. I could only admire this strategy.
Hacking Gradescope to mark a 0, 1, or 2 on each quiz was cumbersome. There were points in the background of each quiz, which students could see, and I felt that saying “not yet” (0) again and again to a handful students (who were perhaps attempting the course prematurely) was rubbing it in too much. So I may suppress the 0, 1, and 2 next time and replace them by stating the point-ranges which they, in effect, represented.
Anyway, it was a good experience and I think it had a positive impact. I look forward to running it again next semester and honing it further.
I ran a cycle of in-class quizzes for the first time this past semester, substantially inspired by ideas you’ve presented. It went well! Thank you for all your advice.
The course was a mid-level math course: Intro. to Analysis with around 70 students. I split the material into seven chapters and offered three quizzes on each, and marked them as “not yet” (notated by a 0 in Gradescope), “progressing” (a 1), or “proficient” (a 2), with only the strongest performance counting to the course grade.
I didn’t see test-anxiety problems. Students reported that the approach was less stressful than the traditional prelim-prelim-final. I think it helped a lot that the quizzes followed a closely consistent style, so, while the questions varied, the students quickly learnt what sort of challenge to prepare for. It was immensely heartening to see them improve over the three attempts, often dramatically.
The two biggest problems I ran into were cheating and the logistics of mandated accommodations for some students. The cheating issue was disappointing to see, especially given that most students worked extremely hard and with absolute integrity. It was exacerbated by students being close together in class and by some of the questions being multiple-choice, where answers are visible at a glance. To combat that, I sometimes circulated two versions of a quiz. As for the accommodations, I kept the quizzes short and gave all the students as much time as they needed (within the bounds of practicalities), which substantially worked. Both of these issues added substantially to the burden of running the course.
There were students who struggled, but I didn’t perceive a big “snowball” problem. The first attempt on each chapter was not long after that material featured in class, and I think it helped them to have to grapple with it at least once in a timely manner. I think that with a “traditional” grading scheme, the same snowball problem can be present but it can be worse, because the gaps between tests mean it isn’t exposed so frequently.
Incidentally, I saw an opposite sort of effect. I mostly gave quizzes two at a time: one on the most recent chapter and one on the chapter before that. A couple of students would deliberately bomb the most recent chapter, just using it to get a sighter on that material, but they would nail the prior chapter. I could only admire this strategy.
Hacking Gradescope to mark a 0, 1, or 2 on each quiz was cumbersome. There were points in the background of each quiz, which students could see, and I felt that saying “not yet” (0) again and again to a handful students (who were perhaps attempting the course prematurely) was rubbing it in too much. So I may suppress the 0, 1, and 2 next time and replace them by stating the point-ranges which they, in effect, represented.
Anyway, it was a good experience and I think it had a positive impact. I look forward to running it again next semester and honing it further.