At Florida International University we have been experimenting with a novel exam format. We have been giving our introductory physics students a group exam followed by an individual exam that contains transfer questions related to the group exam. The group exam requires students to work together on a difficult new problem. This format reflects one of our primary learning goals for our students: to be able to learn physics on their own. Videos of the group exam reveal that students are highly collaborative and engage in productive learning activities; such as, sense-making and constructing new representations. The question addressed in this paper is: Is students' participation in the group exam related to their ability to transfer their knowledge to the embedded questions? We present analysis that shows that students' ability to transfer their knowledge is related to how much they participate and more subtly, how they participate in sense-making and representational activities.