The intuition says housebuilders favour land where the ecological record is thinnest. The data say the reverse.

What it shows

28.5% of large residential sites (97 of 340; the residual category in our classification; see entry 8) sit in low-record areas, against 42.1% (126 of 299) of all other sites. Odds ratio 0.55 (95% CI 0.39–0.76), Fisher p = 0.00035, robust under whole-thesis Bonferroni (adjusted p = 0.0063). Low-record areas concentrate in grid-scale energy (63.6%) and logistics and freight (47.1%, 8 of 17, a small number).

Testing whether developer-owned large residential sites sit in low-record areas more than other such sites gave OR 0.42 (95% CI 0.09–1.95), p = 0.35. That test is underpowered, with a minimum detectable odds ratio of about 4.2, and it is provisional until the ownership classification covers the sites new in v6. Read it as inconclusive, not as evidence of no link.

Reframing

For large residential sites, siting in low-record areas is not the pattern; urban infill, the other housing type, sits at 36.6% (83 of 227), close to the overall 34.9%. The data don’t show why housing sites differ. Low-record areas are more common among energy and logistics sites, but most low-record sites are housing or infill sites (180 of 223, against 43 for energy and logistics).

Revised 12 November 2026: Title and wording changed. A sentence attributing the pattern to the 21-day consultation window is removed, because that window was not measured.

Corrected again 12 November 2026 after a check of the data pipeline: the claim that low-record areas are mainly an energy and logistics feature is replaced, because most low-record sites are housing or infill.