Strategic Interface Design Can Improve Learning Efficiency in an Intelligent Tutoring System

EasyChair Preprint no. 13879

Abstract

Educational data mining can help us discover patterns in data that can be used to help improve learning environments and assess those improvements. Here, we discovered that students were spending a significant amount of time in a tutoring system typing in newly derived problem-solving statements (about 12% of students’ total time in the tutor). We designed a new problem-solving interface to allow students to select from among auto-generated multiple-choice statements and performed a between-groups study to compare students' problem-solving time and learning with and without access to auto-generated multiple-choice statements, between two different semesters. Our results showed significant time savings, with students in the control group taking 4.3 hours in total and the intervention taking 2.07 hours. These substantial time savings did not significantly impact the efficiency of student solutions (both groups had similar solution lengths) and only marginally reduced rule application accuracy on the posttest. This led to a significant difference in learning efficiency between conditions since students learned a similar amount in both groups but saved substantial time in the intervention.

Keyphrases: cognitive load, Human Computer Interaction, Intelligent Tutoring System, learning efficiency

@Booklet{EasyChair:13879,
  author = {Sutapa Dey Tithi and Behrooz Mostafavi and Arun Kumar Ramesh and Tiffany Barnes},
  title = {Strategic Interface Design Can Improve Learning Efficiency in an Intelligent Tutoring System},
  howpublished = {EasyChair Preprint no. 13879},

  year = {EasyChair, 2024}}