With a $2 million Department of Education award, researchers at Worcester Polytechnic Institute and Carnegie Mellon University will continue development of an intelligent tutoring system for middle school mathematical education, transforming into an unparalleled tool for educating students and tracking their progress. The system will give schools long-term data on student performance and teachers and parents immediate feedback to help them work individually with struggling students. The system does this while also tutoring students
WORCESTER, Mass. – Researchers at Worcester Polytechnic Institute (WPI) and Carnegie Mellon University have received a four-year, $2 million award from the federal Department of Education to continue development of a powerful computerized tool designed to help middle school students master mathematical skills.
With the award, researchers will enhance an intelligent tutoring system called ASSISTment, giving it new capabilities and transforming it into an unparalleled tool for both educating students and tracking their progress. The system will give school systems the long-term data on student performance they must report under the federal No Child Left Behind Act. More important, it will provide teachers—and parents—immediate, day-to-day feedback on what students have and have not learned, making it easier to offer individualized instruction to help students master concepts they may be struggling with.
Remarkably, the system does all this at the same time it tutors students. In fact, ASSISTments is the only system that can provide longitudinal data and a benchmark assessment of student skills without taking time out from classroom instruction, says Neil Heffernan, associate professor of computer science at WPI and leader of the ASSISTments research team.
"The No Child Left Behind Act is putting pressure on states to find out what areas student need to improve in, which is leading to a rush to do more testing," Heffernan says. "Unfortunately, this testing cuts into classroom time, and the tests don't provide the kind of immediate feedback that teachers need to do a better job in the classroom. Our system can do that, and can also help students master concepts they're struggling with, without sacrificing instructional time."
"Students shouldn't need to stop learning while they are taking a test—especially a practice test," says Kenneth R. Koedinger, professor in the Human Computer Interaction Institute at Carnegie Mellon University and a co-principal investigator on the grant. "Students keep learning while they are using the ASSISTment system, and we are showing that we get just as good if not a better idea of what they know and do not know than we can from high pressure, one-shot tests."
Heffernan, whose expertise is in artificial intelligence and intelligent tutoring system design, leads a team that includes experts in cognitive psychology, psychometrics, and Web-based educational technology. In addition to Koedinger, the principal investigators are Brian W. Junker, professor of statistics at Carnegie Mellon, George T. Heineman, associate professor of computer science at WPI, and Murali Mani, assistant professor of computer science at WPI.
With the DOE award, the team will add significant new capabilities to the system, which was developed over the past four years with support from the DOE, the Office of Naval Research, the National Science Foundation, and other agencies and tested extensively by teachers and students in the Worcester, Mass., Public Schools. The school system has adopted ASSISTment for use in all of its eighth grade math classes.
With the changes, the system, which is currently built around more than 900 test items from the MCAS (Massachusetts Comprehensive Assessment System) 8th grade math exam, will be able to dovetail with the mathematics curricula used by individual school systems, so students can receive tutoring tied directly to what they are learning in class. It will be expanded to include content from sixth and seventh grade mathematics curricula, and will be able to generate user-friendly reports that show teachers and parents how individual students are progressing and what topics they are struggling with.
Providing regular feedback to parents will help them give their children the individualized attention they need, something teachers don't always have the time to do. "Even with the best diagnostics in the world, teachers are generally better at adapting to trends that involve the whole class then they are to the idiosyncratic needs of each individual student," Heffernan says.
Finally, the system will incorporate new features designed to help students move toward mastery of math topics. The system will keep track of each student's progress and record which skills they have not yet mastered. It will also develop and implement an individualized mastery program for each student and tailor the tutoring it delivers accordingly. The system will let students elect to take tests when they feel ready to demonstrate their mastery. The mastery component of ASSISTment will differ significantly from the approach taken by most intelligent tutors, Heffernan says.
Rather than requiring students to master a topic before they move on, the system will keep track of what they need extra help with and, as they move on to other topics, circle back from time to time to provide additional instruction, including videos showing a teacher explaining how to solve the problem and a Web site showing worked examples.
Along with expanding the capabilities of ASSISTment, Heffernan and his team will continue to evaluate the success of the program by conducting research with the 4,500 students enrolled in grades 6-8 in Worcester's public schools and their more than 60 teachers. Research on the system has already resulted in dozens of peer-reviewed papers and presentations at scientific conferences. "One of the most important findings of that research," Heffernan says, "is that we have shown, without a doubt, that students are learning while they are taking our test."
