Ambiguous dual-test labels permit either sign of the direct-wiring response contrast in 15383 selected worm pairs

We linked two adult C. elegans wiring maps to a pooled wild-type signal propagation atlas. We tested whether a positive wiring contrast survives an operational response-label rule. Detection-only pairs were fixed at one and equivalence-only pairs at zero. Pairs with both results or neither result were free. Of 15383 selected directed pairs, 6602 had free labels. The q-detection contrast was positive: 8.30 percent for direct chemical wiring and 5.15 percent without it. Exact completion bounds for the contrast were -40.25 to 46.64 percentage points. Changing at least 171 free labels from the detection rule could remove its positive contrast. Equal source and observation-count stratum weights gave bounds of -43.65 to 51.53 points and a minimum of 70 edits. All five sensitivity checks permitted both signs. These are finite-sample bounds on an analyst-defined binary label, not confidence intervals or bounds on true biological effects. The method preserves the distinction between non-detection and equivalence. Different animals, selected coverage, pooled tests and an arbitrary equivalence margin limit biological use. Original code, a proof, aggregate evidence, hashes and a figure support reproduction. External records remain outside the bundle.

C. elegans · connectomics · equivalence testing · partial identification · response labels · sensitivity analysis