DO BETTER-PREPARED STUDENTS LEARN MORE FROM COLLABORATIVE TESTS?
In most cases, over the course of a semester, the average group-quiz score is higher than the average individual-quiz score of any member of the group. However, from 2007-2013, the best predictor of the group-quiz average was the highest individual-quiz average of the group members (r2 = 72%). The student who had that highest average (the superstar) often led the discussion during the group quiz, and the others usually deferred to him or her. This behavior frequently extended beyond the quizzes, to in-class projects and discussions. The students with lower averages were generally contributing, but not equally in terms of quality or quantity of explanations. I became concerned that they were not learning as much as the student with the highest individual-quiz average.
From 2013 onwards, I administered a pre-test to all students at the start of class and an identical post-test at the end. This test (the CiG3) consists mostly of open-ended questions scored using a rubric. I measured the normalized score gain <g> between the pre- and post-test to determine roughly how much each students had learned during the course without the quiz scores. When all of the students are considered together, the pre-test scores are only weakly correlated with the individual-quiz averages. However, the average pre-test and post-test scores of the superstars are significantly higher than those of the other students. So are the normalized gains between those scores. Pre-service teachers at my institution are used to cooperative learning, and they may collectively agree that the most efficient strategy is to put their best-prepared members in charge. Alternatively, the individual students with the strongest Earth science backgrounds developed their knowledge because they are the most motivated to learn, and that’s why they learn more than their colleagues.