Individualizing Tutoring with Learning Style Based Feedback

In open­structured domains such as object­oriented design, feedback cannot be standardized error messages or the solution to the problem. To attain the success of effective human tutors, these systems should adapt to a student's learning style. We introduce a pedagogical framework that incorporates the Felder­Silverman learning style model and their validated instrument for assessing individual learning style. The framework provides a feedback infrastructure based on the learning style model dimensions (such as visual, verbal, intuitive, sensor, etc.). It has been implemented as part of the DesignFirst­ITS, helping novices learn how to design a class in UML from a problem description. The system has been evaluated in a CS1 class with results that show that students found the learning­styles based feedback effective in helping them identify and correct errors so that all were able to complete a non­trivial problem.