Teaching of Psychology, 35: 305���312, 2008 Copyright C Taylor & Francis Group, LLC ISSN: 0098-6283 print / 1532-8023 online DOI: 10.1080/00986280802373841 SCHOLARSHIP OF TEACHING AND LEARNING EXEMPLARS Improving Learning Through Interventions of Student-Generated Questions and Concept Maps Jack W. Berry UAB Injury Control Research Center, University of Alabama at Birmingham Stephen L. Chew Samford University Using the principles of the scholarship of teaching and learn- ing, we evaluated 2 learning strategies to determine if they could improve student exam performance in general psy- chology. After the second of 3 exams, we gave students the option of participating in a specific learning activity and assessed its impact using the third exam. In Study 1, participating students generated a minimum of 3 questions per week over course material. Lower performing students who participated improved their exam performance such that they were indistinguishable from stronger students who did not participate. In Study 2, students had the option of generating concept maps over course material. Generating concept maps significantly improved performance. Virtually every teacher has students who perform poorly on the first exam or two and who then come to the teacher for advice on how to do better. In this situation, teachers are better able to help if they can recommend specific, empirically verified courses of ac- tion to the students. We describe two studies of class- room research in which students improved their exam performance after a midsemester intervention of one of two learning strategies. In Study 1 the learning strategy was student genera- tion of questions about the course material on a weekly basis. In Study 2, the learning strategy was student generation of concept maps of course topics. We chose these two strategies because of research that suggested they would be effective in improving student academic performance (e.g., Lee & Hutchison, 1998 Nesbit & Adesope, 2006). Support for their po- tential comes largely from controlled, randomized ex- perimental studies in learning and memory. Daniel and Poole (in press) cautioned against recommendations of pedagogy based solely on controlled laboratory exper- iments. They provided compelling examples of study strategies that work in one setting but have no effect (or even have harmful effects) when applied in other con- texts. They recommended an ���ecological approach��� to pedagogical research, in which teachers base peda- gogical strategies on research conducted in real-world learning contexts. Their critique mirrors similar con- cerns raised in clinical research between clinical effi- cacy research, which typically uses randomized clinical trials, and clinical effectiveness research, which pri- marily focuses on the generalizability of results to real- world contexts (Howard, Moras, Brill, Martinovich, & Lutz, 1996). In teaching psychology, variables such as student effort, study strategies, prior knowledge, and teacher assessment strategies are all likely to form part of the ecological context. To guard against making faulty generalizations from laboratory to classroom, we employed the scholarship of teaching and learning framework (cf. Halpern et al., Vol. 35, No. 4, 2008 305

1998). Specifically, we took principles derived from predominantly laboratory-based research and tested them systematically in a teaching context. We ob- tained approval for both studies reported here from the Samford Institutional Review Board. Study 1 Study 1 investigated the effect of a voluntary ques- tion generation activity on student performance. After the second of three exams, we gave students the option to participate in a program in which they would gen- erate a minimum of three questions per week relating to course content. We based this study on research by Harper, Etkina, and Lin (2003), who examined the effect of question generation on learning physics. In their study, students in an introductory physics course generated questions for 8 weeks. Independent raters coded the questions for conceptual difficulty. Roughly half the questions were minimal or low in conceptual difficulty, meaning they related to simple facts, definitions, clarifications, or low-level comprehension. The other half were medium or high in level of conceptual difficulty, meaning they were analytical, synthesizing, or involved applications of concepts. The students completed a standard test of physics concepts both before and after the question generation phase. Harper et al. found no relation be- tween the number of questions students generated and learning. However, they did find that generating ques- tions at medium or high levels of conceptual difficulty was linked to significantly better learning of concepts. Theoretically, generating conceptual questions might improve learning in several ways. Questions in- volving meaningful analysis, synthesis, or application likely lead to deeper levels of processing (Craik, 2002) and reflection (Lee & Hutchison, 1998), as well as the development of schema (Herbert & Burt, 2004). The form of encoding produced by conceptual questions probably also leads to transfer-appropriate processing (Morris, Bransford, & Franks, 1977) for answering con- ceptual exam questions. We predicted that higher performing students would generate conceptually deeper questions. Consistent with Harper et al. (2003), we predicted that deeper questions would be associated with greater improve- ments in learning, but we did not expect to find a re- lation between the number of questions generated and learning improvement. Finally, consistent with prior studies of learning strategy interventions, we predicted that the questioning activity would have a greater effect on lower performing students than higher performing students (e.g., Schmid & Telaro, 1990). Method Participants. Participants were 102 undergrad- uate students (23 men, 79 women) enrolled in two sections of general psychology at Samford University. Most students were freshmen. The two classes met 3 days per week in the morning for approximately 1 hr per class meeting. There were 53 students in the earlier class (21% men, 79% women) and 49 (25% men, 75% women) in the later class. Procedure. Over the semester, the course in- structor (the first author) administered three examina- tions and an optional final examination (all multiple- choice). We used data only from the midterm exams. After the second exam, we informed the students that we were conducting a study about their abilities to ask questions about material covered in the course. We of- fered the students an opportunity to earn extra credit each week until the third exam by submitting ques- tions to the instructor concerning the material covered during the prior week. We distributed the following in- structions to all students: Provide us with three questions that you would like answered concerning the topics covered in your textbook readings or in lecture. These can be any questions you might have, as long as the questions are about the material or are stimulated by the material. They can be questions about concepts you are still unclear about, about further information you would like to have, or questions about how some issue applies to your own life or to other courses or concepts. When students submitted a set of three questions by e-mail, the instructor sent them an acknowledg- ment and awarded the extra credit. Students received no feedback regarding either the answers to the ques- tions or the conceptual depth of the questions they submitted. We assessed gains in learning through their performance on the third exam. Results and Discussion The two classes did not differ significantly in per- formance on any of the three examinations. For the first class, exam scores (percent correct) were 84.1%, 85.8%, and 82.4% for Exams 1, 2, and 3, respectively. For the second class, scores were 83.4%, 85.1%, and 83.2%. 306 Teaching of Psychology