Enhancing Learning Using Adaptive Computerized Tutoring in K-12 Settings
Carol O’Donnell, Robin Harwood, Barry Gholson, Art C. Graesser, Scotty D. Craig, Wayne Ward, Ronald A. Cole, Gautam Biswas, Daniel L. Schwartz, Kefyn M. Catley, Stephanie Ann Siler · eScholarship (California Digital Library) · 2008
Enhancing Learning Using Adaptive Computerized Tutoring in K-12 Settings Gautam Biswas 1 ([email protected]) Daniel Schwartz 2 & Kefyn M. Catley 3 Carol O’Donnell (Carol.O’[email protected]) & Robin Harwood Institute of Education Sciences, U.S. Dept. of Education, Washington, DC 20208 Department of Computer Science Vanderbilt University, Nashville, TN 37240 School of Education, Stanford University Department of Biology, Western Carolina University Barry Gholson ([email protected]) Art Graesser & Scotty D. Craig Stephanie Siler ([email protected]) Department of Psychology The University of Memphis, Memphis, TN 38152 Department of Psychology Carnegie Mellon University Pittsburgh, PA 15213 Wayne Ward 1 ([email protected]) & Ronald Cole 2 Center for Computational Language & Ed. Research, University of Colorado, Boulder, CO 80309 Boulder Language Technologies Keywords: Adaptive computerized tutoring; dialog; self- regulated learning; deep-level reasoning; science AutoTutor to a learning environment called iDRIVE (Instruction with Deep-level Reasoning questions in Vicarious Environments). The iDRIVE system provided vicarious learning in which students listened to and observed the AutoTutor agent presenting the same course content, but students did not adaptively interact with the materials. Results from the first study indicate that 342 8 th -11 th grade students showed greater learning gains in Newtonian physics and computer literacy when assigned to the iDRIVE condition (in which content sentences were each preceded by a deep-level reasoning question) than those learners who were presented with the same content offered in a monologue condition (no deep-level reasoning questions). This finding in favor of iDRIVE has implications for return on investment because it is more costly to develop an interactive AutoTutor than to script an exchange with iDRIVE. Our first classroom comparison involved six classrooms of 8 th graders (n = 160) randomly assigned to an iDRIVE condition (computerized version), monologue condition (computerized version), or standard pedagogy in which students received standard instruction given by their teachers. The iDRIVE software produced learning pretest to posttest learning gains equal to or greater than those produced by master classroom teachers on a variety of measures. A second randomized intervention paired various conditions with classroom instruction. iDRIVE software taught the conceptual component of two units of high school physics, used three vicarious conditions: standard monologue condition, standard iDRIVE condition, and an iDRIVE condition with an additional explanation. Findings indicate that students in the two iDRIVE conditions performed better than students in the monologue condition. Current findings from these two studies appear to confirm results from our laboratory experiments. How do we individualize instruction and develop cognitive skills to enhance learning? Computer tutors can provide scalable interventions that tailor instruction to each student. This symposium brings together cognitive scientists, computer engineers, content specialists, and education researchers to address adaptive computerized tutoring—a key topic in both cognitive science and education. The papers represent innovations in cognitive science research drawing from discourse comprehension theory and theories of metacognition. These theories are tested across several cognitive skills (deep-level reasoning, self-regulated learning), within multiple content domains (science, math, computer literacy), and across a range of ages in authentic classroom settings (elementary school to college). In her role as discussant, Stephanie Siler will address the common theme of these projects: finding ways to use intelligent tutoring systems to improve learning through thinking and reasoning, consistent with these theories. Carol O’Donnell, Program Officer for the Cognition and Student Learning Research Program at the Institute of Education Sciences, and Robin Harwood will moderate this panel and facilitate discussion throughout the symposium. An Implementation of Vicarious Learning with Deep-Level Reasoning Questions in Middle School and High School Classrooms The overarching goal of our research (Gholson, Graesser, & Craig) is to expose deep-level reasoning questions in the areas of computer literacy and Newtonian physics to middle school and high school students and to show how they support knowledge construction during vicarious learning. We compared an interactive intelligent tutoring system called