Office of Technology Transfer – University of Michigan

Entropy-based Analysis of Scores on Graded Assignments

Technology #6068

Current trends in education increasingly emphasize digital education technologies and individualized learning. There is a pressing need for educational tools which allow educators to better understand the needs of students. Current grading and administrative tools used in classrooms, MOOCs, professional certification programs and distance learning courses allow teachers to implement grades and obtain basic statistics about the class distribution. However the use of less typical statistical analyses can give educators information which helps to ensure that the students are understanding course material.

Entropy-based distribution calculations give more information

By using Shannon entropy-based statistics, a better understanding of student progress can be extracted from grade distributions on assignments and coursework. While typical statistics give information about averages and standard deviation, entropy calculations give information that allows educators to understand which parts of assignments were too difficult or too easy, complex distributions of grades (like bimodal distributions) and how the scores will impact course grades. Access to this information allows educators to better understand and serve the needs of the students.

Applications

  • Autograders in college courses
  • Learning Management Systems
  • MOOCs
  • Professional certification programs

Advantages

  • Better understanding of student learning of material
  • Statistical analysis of grade distribution