At some point while considering equitable grading practices, I found myself searching the archives of TPT looking for some ideas regarding retakes. While I appreciate the idea of an honest retake, my experience has been that it is simply more time and effort on my part, and minimal effort and a hope to just “do better this time” on the part of my students. I ran across Jeff McManus’ article regarding the “box score” (“Retests”: A better method of test corrections)
In short, when the students turn in their exam, they receive a blank copy of the exam and they get to redo it, using any resources. If the redo is perfect, their old score gets a bump on the square root curve. I liked this notion, but had a dilemma—my exams in AP Physics are taken from secure college board documents which are not to leave my classroom. Additionally, I knew that certain groups of kids would work together, while others would not take the initiative to join a group, attempt to work on their own, and not reap as much of the benefits. Not wanting to lose the integrity or security of the exams I needed to make a modification on the assignment.
I informed students of the opportunity to do a retake. Since they needed time to really work the exam, I offered them a “collaboration day” during lunch (our students have a shared lunch hour). The retake would be the following day at lunch as well. (Collaboration day came and I was enthralled. Two thirds of my students came (this has increased to as high as 80%) , received a blank copy of the test, and started talking and working together. Large groups of students formed around white boards to tackle problems, the energy was palatable and the camaraderie was invigorating. Since the students had no number to form an idea how they had actually done on the exam, there was a wide range of abilities in the room.
One of my best students commented to me after collaboration day, “I thought I did really well, but I realize there was a lot I didn’t know” The need to score a perfect in order to obtain an increase in points also motivated students to grill each other for explanations until they understood and could reproduce the work themselves.
Retake day arrived and I had a full house. Students were able to finish their previously 40 minute exam in 20 or less because they knew how to attack the problems and most students were able to perfectly answer the problems.
I struggled, however, with the notion that students might memorize steps to a solution, rather than it being truly valid. I added a reflection component to the retake. Students needed to explain to me what they had previously misunderstood that now they comprehended. The reflections were telling. Students who had obtained 100% on the exam could clearly indicate their faults in either concept or problem-solving approaches. Students who were unable to obtain 100% were unable to adequately reflect on what they misunderstood.
I have continued this practice, in particular with the energy exam, for the last 5 years since I first came across the article. It is not my first or only method for re-assessments, but it is certainly a powerful one. A few changes and observations I’ve made over the years:
- To avoid the memorization piece, instead of testing next day, we test 5-7 days after the collaboration day in order for students to “forget”
- I had a really hard time not bumping a student who earned a 60% and then got all of the FRQ right and missed one MC. So I do a half-bump… so if the full bump is 10*sqrt(60) = 77, the half bump is 77-60/2 = 8.5 60+8.5 = 68.5, which I’ll likely round to 70 out of generosity.
- I’ve had one instance where a student with extreme anxiety and perfectionism this was problematic. I made alternative arrangements for that student ahead of the retake (they got 100 anyway).
Question about test versions. Is this all the same test version? Or do you do something like original test is version A, collaboration day test version B, re test day version B?
Sounds very intriguing! I might do this for my senior chemistry students.
LikeLike
It’s the exact same test
LikeLike