Challenges

Turpen C, Finkelstein ND (2009). Not all interactive engagement is the same: variations in physics professors’ implementation of peer instruction. Phys Rev ST Phys Educ Res 5, 020101. The authors investigate the implementation of peer instruction (PI) in six high-enrollment introductory physics classes, developing a system for describing and measuring classroom practices that contribute to different classroom norms. They observed four types of questions, logistical, conceptual, algorithmic, and recall, but relatively little variation in the extent to which the six instructors used each type. They also observed that student-student interaction around clicker questions was consistent in the six classrooms. They did observe significant variation, however, in instructors’ interactions with students during the question response period, with some instructors remaining in the stage area and others moving into the classroom and interacting extensively with students. Further, they saw variation in the clicker question solution discussion stage in two ways: first, some instructors always addressed incorrect responses during the discussion, while others sometimes eliminated this step; second, student contribution to the class-wide explanation varied, with the average number of students contributing ranging from 0-2.4.  In addition, the way that instructors interacted with students varied significantly, resulting in different classroom norms around discussion of reasoning. They report that differences in instructor practice produces variations in students’ opportunities to practice conceptual reasoning, talking about the subject matter, agency and scientific inquiry.

Turpen C, Finkelstein ND (2010). The construction of different class­room norms during peer instruction: students perceive differences. Phys Rev ST Phys Educ Res 6, 020123. The authors investigated whether differences in the ways that instructors in three physics classes implemented PI led to discernibly different classroom norms. The authors used classroom observations to characterize instructor actions and instructor-student and student-student interactions around clicker questions. Each type of interaction provided the instructor and the students with different possible roles (e.g., rebutting a peer’s physics ideas, presenting a question, listening to the instructor’s explanation). One instructor varied the way he implemented clicker questions but typically promoted student-student discussion. The other two instructors used a more limited range of faculty-student interactions; one of them explicitly promoted student-student discussion while the other did not. Students in the more varied class reported more peer discussion and greater comfort and frequency of collaborating with the instructor. In addition, the researchers examined whether instructors’ implementation practices impacted the value students placed on explaining their reasoning and supporting their answers; greater instructor emphasis on sense-making enhanced students’ perception that this was important. The authors conclude that instructional practices influence the norms in a course, determining to what degree students perceive peer discussion and student-instructor collaboration as valuable means for making sense of course concepts. Their description of the seven types of interactions also serve as a valuable resource for instructors, providing concrete descriptions of several ways to implement clicker questions.

DeMorgan TP, Wakefield C (2012). Who benefits from peer conversation? Examining correlations of clicker question correctness and course performance. J Coll Sci Teach 41, 51-56. This study examined the proportions of students who switched their answers after peer discussion from their initial individual vote, both in positive and negative directions.  The data set included clicker and exam performance from 214 physics students who answered a minimum of 70% of clicker questions with peer discussion and a revote.  The authors found that 23% of student answers went from incorrect to correct, and 36% were correct twice. Only 8% went from correct to incorrect, while 33% were incorrect twice. The authors observed a strong positive correlation between students with a high percent of initially correct responses and a high course grade. However, there was no correlation between a student’s overall course grade and the percentage of times a student responded correctly twice, or changed from an incorrect to correct answer. The lack of correlation could be because not all clicker question topics were explicitly tested on exams. Students may also change their answer temporarily during class as a result of peer discussion, but still retain original incorrect ideas. In addition, the in-class questions were not graded, and the instructor always explained the correct answer after voting, both of which could lead to lack of student motivation in answering clicker questions.

Relling AE, Giuliodori MJ (2015). Effect of peer instruction on the likelihood for choosing the correct response to a physiology question. Adv Physiol Educ 39, 167-171. The authors investigated learning from peers’ knowledge versus co-construction of knowledge during peer discussion. The study was conducted with 101 students in a veterinary physiology course and examined the impact of several factors, including the presence of correct answers within the group, confidence about the original answer, gender and group size, on answer correctness and answer-switching. Groups of 2-4 students answered questions individually and indicated their level of confidence, engaged in peer discussion, and then answered a final time, providing both group and individual final answers.  Unsurprisingly, students who were “not so sure” or “just guessing” about their initial answers were more likely to change answers than students who indicated they were very sure about their initial answer, and incorrect answers were more than threefold more likely to be changed after peer instruction than correct answers. Students in groups that had at least one initial correct answer were more likely to change to a correct answer than students in groups with no initial correct answers, although changes to correct answers occurred 24% of the time when no members of a group initially had the correct answer. Gender and group size had no effect on the probability of changing answers.

Miller K, Schell J, Ho A, Lukoff B, Mazur E (2015).  Response switching and self-efficacy in peer instruction classrooms. Phys Rev ST Phys Educ Res 11, 010104. This study collected response data from 91 students in an introductory electricity and magnetism course, who engaged in PI for 83 different clicker questions over the semester.  Students were grouped into categories based on whether and how their answers changed from their individual vote to their vote after discussion.  In particular, the authors were interested in “negative” switching, in which students switched from correct to incorrect, or incorrect to another incorrect answer after peer instruction. Switching data were then correlated with the difficulty of the clicker question, and with the students’ scores on a self-efficacy survey. Students switch their answers 44% of the time, usually in a positive direction (73% from wrong to right). Students with low self-efficacy were significantly more likely to negatively switch their answers than students with high self-efficacy, even when controlling for incoming knowledge with a pre-test. Students are more likely to engage in a negative switch on difficult items than on easier items. In addition, females reported lower self-efficacy than males, and were more likely to engage in negative switching. The authors conclude that the strong correlation between switching and self-efficacy beg interventions to help students have positive experiences during peer instruction, including building mastery, providing modeling, and reducing stressful in-class situations.

James MC, Willoughby S (2011). Listening to student conversations during clicker questions: what you have not heard might surprise you. Am J Phys 79, 123-132. The authors characterized peer discussions occurring as part of PI cycles in high-enrollment introductory astronomy classes. For each cycle, students discussed a multiple choice question before voting individually; the instructor displayed the histogram of student responses and discussed various responses. The authors recorded and transcribed 361 peer discussions of 45 clicker questions, identifying “standard” and nonstandard conversations. In standard conversations, individual clicker responses represented ideas articulated in the discussion, and the discussion focused on at least one MC alternative. Nonstandard conversations fell into three categories: those including unanticipated student ideas, such as misconceptions not represented in the question; those in which correct clicker responses misrepresented student understanding; and those that represented pitfalls, such as student passivity or self-evident answers. 136 of the 361 conversations analyzed were standard, but 12.5% included unanticipated student ideas, 26% represented cases where students gave the correct clicker response without understanding, and 37% were unproductive due to pitfalls. Interestingly, questions requiring factual recall and those requiring higher-order cognitive skills did not yield statistically different results in these categories. Based on these results, the authors offer suggestions to increase the percentage of productive discussions.

Jensen M, Johnson DW, Johnson RT (2002). Impact of positive interdependence during electronic quizzes on discourse and achievement. J Educ Res 95, 161–166. Students (n=151) in an anatomy and physiology course took a quiz for the first 25 minutes of their weekly lab section. Students were split into two different groups, either assigned to an “interdependence” group, or to a “no interdependence” (individual) group.  All students had the opportunity to work together in a chat room, available only to members of their group, to answer each quiz question. In the interdependence group, one student’s quiz grade was randomly selected and given to all students. Students assigned to the individual group condition were able to chat with other members of their group in the chat room during the quiz, but each person was graded individually.  Groups remained the same for the entire study. All students were then graded individually on four tests. The authors found that as the course progressed, students in the interdependence group engaged in significantly more instances of interaction and had longer interactions, while those in the individual condition engaged in less of both.  Students in the interdependence group also performed significantly better individually than the independent group on tests three and four, but not on the first two.