Results Related to the Automated Reassessment and Relearning System (ARRS)

The Automated Reassessment and Relearning System (ARRS) is a system that gives kids practice spaced out over days. To read more about how we use this for science take a look at this NSF Grant that funds some of this work. WPI has also have used ARRS as part of the National Center for Cognition and Mathematics Instruction (NCCMI) .

ARRS can used in a few different ways.

  1. ARRS can use useful as an intervention (Does it help kids learn more?)

    1. In this paper we showed ARRS lead to more student learning. This finding was just highlighted in EdWeek.

    2. More recently we have rolled out a personalized ARRS scheduling system that pays attention to the student's initial speed of learning to decide students' retention test pace.

  2. ARRS can be used as a scientific tool to randomly assign students to different delays of reassessments.

    1. In this paper, we were trying to learn what is the best schedule to optimize long-term retention.

  1. ARRS can also be used as a better measurement device (Does using a delayed reassessment lead to more precision in estimating students knowledge than immediate assessment does? )

    1. In this paper, we attempted to see how much better we could predict student knowledge 6 months later. If you use ARRS, and add to your assessment how well they do one week after the test, you get increased predictive validity on the posttest 6 months later. Why would you care? We subscribe to Koedinger, Corbett & Perfetii theory of "robust learning", that that we want to measure student learning in a robust manner. Robust Learning has three components. One is whether the student knowledge transfer to "far transfer" questions that tap that knowledge in a new way. A second way is whether that knowledge retains in the long term. The third part of robust learning is whether it prepares students for future learning. ARRS is a "longer term" measure. It turns out that these delayed reassessments from ARRS are good predictors of student knowledge.

    2. We are working right now on an unpublished paper where we are looking to see how much better ARRS tests would be as a measurement of knowledge than immediate tests. We expect ARRS tests are more sensitive measurements to show the performance difference among students,

    3. The third component of Robust Learning is preparing students for future learning. We are working on a project to see if students' speeds of learning of post-requisites can be a measure on students learning activities.