Analysis of fault models for student use

Computer science educators have long been concerned over the difficulty with which some students learn to write computer programs; especially the wide disparity in students' abilities to locate and fix faults in the programs they write. We hypothesized that requiring students to categorize faults (as the faults are encountered) would help them better understand the faults and, in turn, improve their corrections and speed up the programming process. This paper analyzes five fault categorization models, with an eye towards application by novice programmers, and provides experimental evidence showing that simple fault categorization can aid students' programming abilities.