# Data audit (P1)

File: data/design.csv — SHA-256 b1a17a529973e7994bbedad5ec1bc73fe968c777301ca8745657ea300e3c74f0 (matches bundle.json).

• 30 rows x 3 columns: i (int64, 1..30 consecutive), x (float64, dimensionless, range -1.3942..1.9919, mean 0.2823, sd 0.8212), sigma (float64, dimensionless, range 0.6034..1.7903, mean 0.8195).
• Missing values: 0. Duplicate i: 0. Duplicate (x,sigma) rows: 0. All sigma > 0. x sorted ascending.
• No date columns (synthetic fixed design). sigma rises with |x - ~0| (approx. quadratic pattern), i.e. largest variances at extreme x — the high-leverage points — which is the case where conventional SEs are expected to understate the variance of beta_hat.
• Minor observation: rows 21/22 have nearly identical x (0.7123, 0.7126); not an error.
• Information-availability: the only input is the permitted design file; no violation. No exclusions applied.