Ecological and landscape designations (Ancient Woodland, Green Belt, Sites of Special Scientific Interest and others) are more common around sites in less deprived areas. The evidence for a gap across all six designations together is modest; Ancient Woodland carries most of it.

What we measured

For each site we counted six protective designations nearby: Green Belt, Area of Outstanding Natural Beauty, National Park, Ancient Woodland within 500 m, Conservation Area, and a Site of Special Scientific Interest (SSSI) within 1 km. We set them against hazard layers and combined them as a log ratio: L = ln(designations + 1) − ln(hazard layers + 1). In the main run the only hazard layer is coal mining risk. Flood zone is handled separately below.

What it shows

Mean protective designations per site rise from 0.57 in the most deprived band (n = 87) to 0.75 in the middle (n = 321) and 0.87 in the most affluent (n = 242). Deprived against affluent, with all six designations: Mann–Whitney U = 8,739.5, p = 0.012, r = 0.14 (a small effect).

• After Benjamini–Hochberg correction across the thesis the adjusted p is 0.021. After Bonferroni correction it is 0.23, which is not significant.
• The result depends on how “within 1 km” of an SSSI is drawn: p = 0.010 for 0.5 km, 0.012 for 1.414 km and 0.028 for a 2 km square.
• Adding Heritage at Risk removes the gap. Five designations plus heritage: p = 0.099. Six plus heritage: p = 0.20 (see entry 5). This holds for a 500 m radius around each site; at 250 m the five-designation gap remains (p = 0.013).
• Counting flood as a hazard widens the gap between the most deprived and most affluent bands, from 0.17 to 0.24 in mean L (five designations), because deprived sites more often sit in flood zone 2 or 3 (see entry 4).

Ancient Woodland buffers carry most of the pattern: 9.2% of deprived-band sites against 33.1% of affluent ones, odds ratio 4.88 (95% CI 2.25–10.58), p = 8.4 × 10⁻⁶. Green Belt (OR 2.01, p = 0.11) and Conservation Area (p = 0.35) are inconclusive.

Correction after re-audit

An earlier version of this entry used five designations, because the SSSI check had never run successfully: its query named fields that do not exist on the Natural England layer, so it returned “not detected” for every site. With five designations the result was U = 8,370.5, p = 0.0017 (mean 0.34, 0.50, 0.64). With the check repaired (158 of 652 sites have an SSSI within 1 km, measured as the distance to the polygon), the six-designation result above is the one to use.

Why it matters

Under NPPF paragraph 11(d) as it stood in the 2023/2024 editions, the tilted balance applies with fewer legal obstacles where Footnote 7 protections such as Ancient Woodland are absent. The thesis doesn’t dispute that today’s pattern reflects history: affluent buyers long paid a premium to live near amenity. The claim is that the inherited distribution still has present-day legal consequences through that mechanism. (The August 2026 Framework restructures this test.)

What it doesn’t show

The effect is small, and this is an association, not a cause. Each site is placed at the postcode centroid given on the campaign form, so the 500 m and 1 km distances carry a position error that we have not measured.

Revised 12 November 2026: The main result now uses all six designations (SSSI check repaired): p = 0.012, adjusted 0.021 and 0.23. The five-designation result is kept as a correction. Added the effect of Heritage at Risk and of flood. The title is narrower.

Corrected again 12 November 2026 after a check of the data pipeline: added the radius qualification for the heritage result and the postcode-centroid limitation.