How Bias, Context, and Data Gaps Shape What We Know About Code Security

This section outlines the methodological limits of using code review data to study software security, addressing annotation bias, taxonomy overlap, dataset quality, and generalizability.