Reducing confounding bias in predicate-level statistical debugging metrics

• DOI: 10.5555/2337223.2337278
• Publication Date: 2012-06-02
• Authors: Ross Gore; Paul F. Reynolds
• Keywords: Predicate (mathematical logic); Statistical Inference; Control flow

Statistical debuggers use data collected during test case execution to automatically identify the location of faults within software. Recent work has applied causal inference eliminate or reduce control and flow dependence confounding bias in statement-level statistical debuggers. The result is improved effectiveness. This encouraging but motivates two novel questions: (1) how can be predicate-level (2) what other biases eliminated reduced. Here we address both questions by providing a model that eliminates reduces failure We present empirical results demonstrating our significantly improves effectiveness variety debuggers, including those only single source bias.