| Summary: | Asking students to reflect on course content and ask questions about that content has been shown to improve comprehension in numerous domains. More recently, tools have been developed to store multiple-choice questions created by students in an online repository where they can be shared, evaluated and discussed with their peers. Although benefits are reported from the use of such systems, multiple-choice questions are not suitable for all teaching contexts: many instructors prefer to use free-response questions to assess learning. This thesis describes the development and evaluation of an online tool designed to support pedagogies involving student generated free-response questions. Using a design-based research methodology, an online tool named StudySieve was designed, implemented and evaluated in naturalistic educational settings. StudySieve was used in three large undergraduate Computer Science courses in which students developed questions with sample solutions, answered the questions contributed by their peers and evaluated both questions and answers. The effectiveness of free-response question generation and related activities is evaluated by considering the response from students, the nature of the questions created and the relationship between activity and subsequent exam performance. Students believe that the activities of question generation, answering questions and evaluating both questions and answers do help them learn. They rarely write more questions than required for assessment, but many students use the questions generated by their peers to practice prior to exams. The questions created by students reflect all the major topics of a course covered prior to the question generation activity, but the cognitive level of student generated questions tends to be lower than those generated by instructors. Although student question-generation activity is correlated with exam performance, the correlations can be explained by other levels of activity in a course. However, students who write questions on a topic sometimes have improved performance in exams on that topic compared with students who have not authored a question on the topic. Generating free-response questions is a valuable educational activity, but the exact conditions that result in improved exam performance require further investigation by future research.
|
|---|
