Students take a two-phase exam twice: once individually, and a second time working in teams. Proponents hope that during the team phase, students will discuss, debate, and resolve questions by sharing their reasoning, challenging each other, and reaching consensus. Potential adopters fear that students might uncritically follow the majority answer or mimic one dominant team member. To explore this empirically, I data-mined students' solo- and team-phase responses from the final exams of three different introductory physics courses to construct multiple measures of team dynamics. My results substantiate prior findings that teams do engage in meaningful debate and explore the virtues of various possible answers. The two-phase exam implementation used does not force teams to submit a common answer, allows students to "hedging their bets" for partial credit, and incentivizes helping teammates.