Prep scripts quietly drop a language-preference field, and the model code treats a blank as a default majority value. Which bias-test approach finds those anti-patterns?

Prep scripts that quietly drop a language-preference field, and model code that treats a blank as the majority default, are sensitive-attribute mishandling. Static analysis of the preparation programs and the model implementation is the approach that finds those anti-patterns. A known-fair dynamic set, a Chapter 6 overfitting procedure, and skipping the scripts leave the code unexamined.