Three-day ozone runs had no common excess over a fixed-pair reference near selected US bird surveys in 2005-2019

We tested three-day high-ozone runs in 1,120 complete windows before selected bird surveys in four US states in 2005-2019. An exact reference kept each window high-day count and adjacent high-high pair count. Observed run totals were 0.90-1.13 times reference means across two regions and two periods. Excess counts had both signs. Only 73 windows allowed run counts to vary under the reference. Second sites and three selection checks gave no common positive excess. A composition proof and exhaustive binary checks support the calculation. These are finite-sample reference contrasts, not tests of bird effects or weather causes. Selected monitors, shared episodes, sparse variable windows and reference assumptions limit inference. The package gives daily evidence, code, source hashes, all results and offline reproduction instructions.

ozone · exposure timing · bird surveys · observation bias · adjacent pairs · runs · conditional reference