The group work test is an assessment strategy that promotes higher-order thinking skills for solving context-rich problems. With this format, teachers are able to pose challenging, nuanced questions on a test, while providing the support weaker students need to get started and show their understanding. The test begins with a group discussion phase, when students are given a “number-free” version of the problem. This phase allows students to digest the story-like problem, explore solution ideas and alleviate some test anxiety. After ten to fifteen minutes of discussion, students inform the instructor of their readiness for the individual part of the test. What follows next is a pedagogical phase change from lively group discussion to quiet individual work1. The group work test is a natural continuation of the group work in our daily physics classes and helps reinforce the importance of collaboration. This method has met with success at York Mills Collegiate Institute, in Toronto, Ontario, where it has been used consistently for unit tests and the final exam of the grade 12 university preparation physics course.
Context-rich problems and tests Our students regularly work in groups and practise solving context-rich problems using a process based on Pat and Ken Heller’s Cooperative Group Problem Solving2 and Randall Knight’s homework solution process3. To begin this process, each group receives one copy of the problem statement. Context-rich problems are story-like: they normally avoid technical physics vocabulary and often lack a well-defined question. In other words, they don’t ask students to “solve for x”. Rather, they require students to interpret and focus the problem; construct a precise physics question to answer, identify and describe the relevant physics; plan and execute a solution; and evaluate the results. This process helps train students in a systematic way to unpack more challenging problems and to think deeply about them, instead of jumping for the first equation that comes to mind. Due to the greater demands of context-rich problems, students need the support of working in cooperative groups to generate ideas, find solution strategies and check one another’s work. As a result, a typical problem takes a full 70-minute class for the group to complete. In the past, context-rich problems didn’t work on our traditional tests: they took too much time, limiting the coverage of material; and working alone, weaker students were often unable to start the problems, or they significantly misunderstood the situations, preventing them from demonstrating their understanding. To make context-rich problems work in a test situation, I decided to introduce an element of group work to the test and changed my philosophical approach to testing. Rather than designing a test to exhaustively survey students’ understanding of content, I decided that the test should focus on the skills of unpacking challenging problems and carefully explaining one’s understanding. These changes helped to create an assessment that reinforces our program’s learning goals for students (higher-order thinking, conceptual understanding, explanation), and the students’ mode of learning (cooperative group work).
Rethinking how we assess The inspiration for group work tests came in equal parts from Carl Wieman’s Physics Teacher article “Physics Exams that Promote Collaborative Learning”4 and Eric Mazur’s online lecture “Assessment: the Silent Killer of Learning”.5 Wieman describes two-stage exams that begin with individual work and finish with a collaborative portion that helps provide immediate feedback from the exam experience. We used this approach a number of times in our classes and were pleased with our students’ level of engagement. Mazur, in his lecture, reminds us that our educational system creates risk-averse students who can excel on traditional tests, but are woefully unsuited for academia or industry. If our goal in STEM education is, as Wieman suggests in an Issues in Science and Technology article, “to maximize the extent to which the learners develop expertise in the relevant subject, where expertise is defined by what scientists and engineers do” ,6 our assessments should promote the behaviours and skills of experts. The insights from Mazur and Wieman led me to rethink our approach to testing. When an expert has a challenging problem to solve, one of the first things she might do is confer with colleagues to help define the problem and generate ideas for its solution. With a clearer understanding of the problem, our expert begins working out the solution with access to a wide range of information and tools to assist the solution process. I realized that Wieman’s two-stage strategy could be reversed to model this expert approach, so I decided to begin the test with group work. During the individual portion of our tests, students keep the notes they made on their copy of the problem statement. This page also includes a list of equations from the course, which helps to de-emphasize memorization and encourage a focus on problem solving.
Starting the group work test For the group work portion of the test, students sit in their regular three-person groups from their daily class work. I design these groups based on the advice of Heller and Heller2: each group has a strong, average and weak member whose personalities I try to match to encourage tutoring and discussion. The first groupings of the course are almost random, only avoiding the combination of a male, male, and female student until I know their personalities better. A typical group work test consists of one or two context-rich problems. Each student is given a page with the problem statement: a description of the scenario with all the numbers replaced by descriptive phrases. In the sample below, the problem is to decide whether a stunt can be performed safely. The problem statement does not ask students to solve for a particular physics quantity; instead, it describes the desired outcome for the scenario. In addition to the problem statement, the group work sheet provides instructions to guide students’ discussions. In the sample provided below, we see typical notations on the problem statement sheet by a student who scored average on this test. Note how the student’s simple notes help to define the problem. Students do not try to exhaustively record all of the group’s discussions. Instead, they highlight the main features and clarify a few ideas: just enough to prepare themselves for the individual portion of the test. 
The individual portion of the test When a group has finished its discussions, its members put up their hands, indicating their readiness to move on, and adjust their seating to space themselves out. On the individual part of the test, the problem is repeated almost verbatim, but with numbers replacing some of the descriptive words. In a conventional test without the group work phase, one long question could lead to a risky all-or-nothing situation where the student either does or does not “get” the problem. The group discussions, combined with problems that focus on the core content of the unit, help to reduce that risk. Furthermore, we use a solution process that requires students to describe the problem using multiple representations including pictures, technical diagrams, word descriptions, and mathematics7. This process emphasizes conceptual understanding and provides many different ways in which a student can demonstrate his understanding, even if he cannot completely solve the problem. Our students have been trained in this process since grade 11, making it very familiar. Only the individual part of the test is marked, which keeps the evaluation process simple and focuses on what the individual can accomplish. Below is shown the individual portion of the test with my solution. The solution format is highly structured, with subheadings under each representation that serve as a helpful checklist. Note the variety of ways in which students are required to show an understanding of accelerated motion: a curving dot pattern, a sloping velocity graph, a velocity vector diagram, and a description of the vertical motion. Consistency between the representations is a good challenge for stronger students. The second page of the individual part of the test has the mathematical representation of the problem. We teach our students to organize their work carefully and provide descriptions before each section of math. This helps to check whether students understand the purpose of the math and whether they can interpret a mathematical quantity’s physical meaning. It also makes complex solutions much more readable.

Results of group work tests Our school has found many positive results from the group work test. First, it serves as an important step toward achieving Wieman and Mazur’s goals in STEM education, and it is a valuable complement to Wieman’s two-stage exams. Second, the format reinforces the sought-after twenty-first century skills of collaboration, communication, critical-thinking and creativity8. Third, it emphasizes conceptual depth and expert problem-solving skills. It challenges strong students to demonstrate their deep understanding and ability to handle the fine nuances of the problem, while it supports weaker students through the group discussions and highly structured solution process. A concern with this format might be that students with little understanding get a “free ride”, but this does not seem to be the case: I have witnessed the complete solution process being explained to a struggling student, but without the skills and basic understanding, the student was able to do little on his own. The administration of the group work test has been straightforward and scaled upwards easily from classes of 30 students to the final exam of 110 students. Since the students were familiar with the process, the exam was easily managed by myself and two other non-physics teachers after just a couple minutes of instruction.
Our students have responded enthusiastically to the group work test format. I asked students to anonymously answer the question “how did the group discussions affect your performance on this test?” The most common two responses were that it helped them relax or reduce their anxiety, and that it helped them understand the problem or check their solution ideas. Fairness of group composition was not raised as an issue, leaving me to believe that students feel they are benefiting from the process. The group work format also eliminated most student questions during the test, the ones which actually mean, “Am I doing this correctly?” While observing their group discussions during the test, I was quite affected by the care students took in checking one another’s understanding, going back and discussing, before moving on to the individual part. The group work test is now a fixture of our senior physics course and we are beginning to introduce it in other courses.
6 Carl Wieman, "Applying New Research to Improve Science Education." Issues in Science and Technology 29, no. 1 (Fall 2012).
7 David Rosengrant, Eugenia Etkina, and Alan Van Heuvelen. "An overview of recent research on multiple representations." 2006 physics education research conference. Vol. 883. 2007.
8 Partnership for 21st Century Learning (p21.org) is coalition of businesses, educators and policy makers promoting these skills.
