# Data audit (P1)

• File: data/design.csv, SHA-256 b1a17a529973e7994bbedad5ec1bc73fe968c777301ca8745657ea300e3c74f0
• 30 rows x 3 columns: i (int, 1..30, unique), x (float, dimensionless, range -1.3942..1.9919, mean 0.2823), sigma (float, dimensionless, range 0.6034..1.7903, all > 0).
• Missing values: 0. Duplicate rows: 0. No dates (not applicable).
• Anomalies: none. Note: sigma^2 is strongly correlated with x^2 (r = 0.987), i.e. error variance is largest at high-leverage points -> the setting in which conventional SEs are expected to be biased downward. This is a design feature, not a data issue.
• Information-availability: the only input is the permitted design file; no violations.
• Exclusions: none (see out/exclusions.csv).