Sunday, September 27, 2015

Cheat-Proofing Exams II: Pros and Cons of Group Work

In the past week, I have finished grading my tablet students' individual and group exams. As in previous semesters, I returned annotated PDFs back to the students individually by e-mail - with comments left in the PDFs to show the points earned only on those questions where a student did not earn all of the points. Here is one benefit of group exams: although it adds a document to return to each student (I attached the graded individual exam plus the graded group exam to each e-mail), grading is faster because I grade the group exam and assign the same point values to each member of the group.

Student Attitudes and Perceptions

Although I have yet to fully analyze the post-exam attitude survey data, there are overall positive (or, at worst, neutral) findings about the effect of group exams on students. Although almost two-thirds of the class reported feeling more stressed than normal at the start of the exam and during the exam, the bulk of the class reported feeling normal (or more relaxed than normal) following the exam!

As expected, the distribution of how well students self-assessed their performance on the exam (before receiving scores from me) was normal (very bell-curve). However, the responses that most surprised me were to my inquiry about how frequently students accessed notes and other digital resources (notes, textbook, the web, etc.) during the exam. Less than ten percent of students reported using digital resources more frequently than they had imagined they would; two-thirds of the class used digital resources less frequently than they thought they would. Informal discussions with a couple of students seems to suggest that the reason for this was that they felt rushed during the exam.

At this point, I feel good: although I don't like having students more stressed than normal at the beginning of and during the exam, this might be entangled with this being their first exam with me. I might expect to see this effect dissipate as we complete more of this style of test during the semester. Further, if students are not accessing digital notes and content frequently during the exam, then perhaps this approach is, indeed, at least cheat-resistant. In fact, I was even quite frustrated to find that one student had solved one problem (on the digital annotation of a PDF) using long division, instead of using the calculator built in to the tablet!

Student Performance

However, how did the students actually perform on the exam? Individual exams were heavily skewed towards A and B grades, which is fine with me: I like to interpret this as an indication that I was successful at helping the students understand the expectations and to study and practice the appropriate content for the exam. The class average on the group exam was exactly the same as the average for the individual exam: 75%. Most importantly, although the addition of the group exam did negatively impact the overall (individual + group) letter grade on the test of four students, it also improved the letter grade on the test of three students.

Although there are additional post-test survey data to analyze, my broad impression of the quick look I took at the data were that, generally, given the option to write a dissenting opinion on whether they agreed with the group answer to each group exam question, students generally supported the group answer that was submitted.


My overall impression is that the group exams have broadly accomplished my initial goals:
1) group exams allow me to assess material and conduct analyses that would be difficult (or unfair) to ask individual students to perform during a one-hour exam, including higher-level Bloom's questions
2) although the novelty of the exam format might have been more stressful to students, the ability to collaborate with peers and/or access digital resources during the exam might have improved their post-exam attitudes
3) group work did not appear to cause a net increase in grades, but also did not have a net negative effect. Longer-term, if the groups maintain a stable membership (which I plan to assess), having established groups early in the term might have a long-term benefit to student performance if those groups study and strategize together.
4) group work, and post-test surveys on peer evaluation of group member contributions to the exam, provide me with additional data that will help me assess students' performance. These could, for example, be particularly useful data to return to when students ask me for letters of recommendation for medical school two years hence. I will have quantitative data to suggest to admissions committees whether students were generally found by peers to be engaged in group discussions, able to help resolve discrepancies, to help distill common themes in potentially disparate individual responses, and so on.

No comments:

Post a Comment

Have an insightful comment, best practice, or concern to share? Please do!