From f32bfb14338fd8b6d571cc00f8215a9fcb7bbfb8 Mon Sep 17 00:00:00 2001 From: Sylpheed RE agent Date: Wed, 26 Aug 2026 04:53:13 +0000 Subject: [PATCH] re: withdraw the DisableInterval / delayed-route association -- it is a clustering artefact The pooled cross-tab gave p ~= 0.02 for 'no Yes-flagged squadron has a delayed route', recorded at a cautious yellow. Breaking it down by stage removes it: all 55 Yes routes live in 2 stages, and 45 are in S14, which has NO delayed route at all -- those 45 cannot be evidence either way. Only S04 contains both a Yes squadron and a delayed route. Within S04 the exact test is P = C(47,5)/C(57,5) = 0.37, expected 0.88 observed 0. Unremarkable. The pooled p treated 1023 routes as independent draws when they are two clusters. The doc had already flagged this as the kind of near-fit that misled the corpus before; that caution was right and did not go far enough. --- docs/re/mission-wave-arrivals.md | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/docs/re/mission-wave-arrivals.md b/docs/re/mission-wave-arrivals.md index bddbcb91..a1b8b295 100644 --- a/docs/re/mission-wave-arrivals.md +++ b/docs/re/mission-wave-arrivals.md @@ -282,6 +282,34 @@ statistic is exactly the kind of near-fit that has already misled this corpus once (the "109 of 111" squadron-name base in [structures/unit-group-table.md](structures/unit-group-table.md)). +### ❌ WITHDRAWN (2026-08-26) — the p ≈ 0.02 does not survive the clustering + +The caution above was right and did not go far enough. Breaking the same +cross-tab down **by stage** destroys the result: + +| stage | No, t=0 | No, t>0 | **Yes, t=0** | **Yes, t>0** | +|---|---|---|---|---| +| S04 | 42 | 5 | **10** | 0 | +| S14 | 22 | 0 | **45** | 0 | +| the other 9 stages with delayed routes | 836 | 63 | 0 | 0 | + +All 55 `DisableInterval = Yes` routes live in **two** stages, and **45 of them +are in S14, which has no delayed route at all** — no squadron there is delayed +whether it sets the flag or not, so those 45 cannot be evidence of anything. +Only **one** stage, S04, contains both a `Yes` squadron and a delayed route. + +The pooled p ≈ 0.02 treated 1 023 routes as independent draws when they are +really two clusters. The test that actually bears on the hypothesis is within +S04, and there it is: + + P(none of S04's 5 delayed routes belongs to one of its 10 Yes squadrons) + = C(47,5) / C(57,5) = 0.37 + +Expected count under independence: **0.88**, observed **0**. That is entirely +unremarkable. **The association is not supported by the data**, and the +"`DisableInterval` finally has a plausible referent" reading above rests on the +name alone, not on this table. + Name matching for this table is sound: for Stage 02, 108 of 111 squadrons have routes and 113 route names resolve, so the cross-tab is not an artefact of failed joins.