The proposed work will evaluate the use of the four Cognitive Tutor middle-level mathematics courses in a range of schools, including urban and suburban schools, comparing the effectiveness of these courses to that of courses not involving educational software. Methods will be developed with which the tutoring software can detect ineffective behaviors associated with poor learning outcomes. A series of studies will address the hypothesis that instructional software can be a vehicle for creating a more learning-oriented classroom, with better learning outcomes, if it engages students in "learning-oriented dialogues," designed to help students focus and reflect on learning-oriented behaviors. The following approaches will be evaluated: Providing interactive declarative instruction prior to and during problem-solving activities, Having students set learning goals at the beginning of tutoring sessions, and Engaging students in discussions, with other students or with an automated learning companion agent, in which they evaluate their own or each others' learning behaviors. The pre- and post-assessments used in these studies will be designed to detect shallow learning, probe the depth of students' understanding, assess whether they can quickly absorb new material (i.e., whether students have become better future learners), and evaluate students' goal orientation and motivation.
Intellectual merit The proposed research will generate new knowledge about the causes and extent of students' shallow learning and ineffective learning behaviors with technology. It will also produce new knowledge about how instructional software can help students overcome those problems and be better learners. It will provide insight into the differences between students' goal orientation and learning results in urban and suburban schools. Steps will be taken to provide better integration between educational psychology theory (particularly Cognitive Load theory) and cognitive theory (ACT-R theory) and, correspondingly, between the rigor of laboratory educational psychology experiments (cf., Clark & Mayer, 2003) and the realism of classroom experiments (cf., Bransford, Brown, & Cocking, 2000).
Broad impact The proposed research, if successful, will improve American K-12 math education. Results from the proposed work will have a direct impact on middle-school math instruction across the country, since the four courses on which the research focuses will be distributed through the network of the Pittsburgh-based company Carnegie Learning, Inc. The creation of courses that work well in urban schools will be especially helpful to minority and low-SES students, who make up a large proportion of the student population of those schools. It should also benefit disadvantaged students in rural and suburban schools.
