Statistical Evaluation of Test Sets Using Mutation Analysis

Evaluation of the ability of test sets for fault detection, and indirectly also evaluation of the quality of test techniques that generate those test sets, have become more of an issue in software testing. Based on mutation analysis, this paper evaluates and compares fault detection ability of test sets using statistical techniques. In this process also different mutant types (and indirectly different fault types) are considered. A case study, drawn from a large commercial web-based system, validates the approach and analyzes its characteristics.