Theories and patterns of productive group interaction in two-stage introductory physics exams

Two-stage" exam formats, in which students take an exam and then repeat portions in a small-group setting, have been studied for years as a method to bring interactivity and peer-instruction into the classroom.Despite evidence of educational benefits, they have not yet seen widespread adoption at the scale of other active-learning interventions, perhaps in part because the mechanisms behind the potential benefits have been less thoroughly explored.We propose a theory of successful group exams in which students advocate for their chosen answers and deliberate thoughtfully among the options presented, without resorting to strategies like majority voting or deference to a single seemingly-strong student.By categorizing decision patterns across a sample of 59 group exam questions we provide statistical evidence that group exams can indeed support these kinds of pedagogically valuable interactions, and we examine the ways that different exam questions can influence productive or counterproductive interaction types.