Seven-day survey date shifts changed ozone episode labels in 1-7% of complete four-state bird survey windows in 2005-2019

We tested whether a prior-month ozone episode label stayed the same when a bird survey date moved seven days earlier or later. We used public USGS survey dates and EPA ozone records in California, Oregon, New York, and Pennsylvania. An episode was at least three consecutive days above 70 ppb. We held the monitor and year fixed. In four region and period samples, 1.1-7.2% of complete windows changed label. Among windows positive at any of the three tested dates, 31.1-100% changed label. The 100% estimate used only four windows. Complete records covered 951 of 1,592 eligible date windows. The second monitor did not repeat the later eastern pattern. Missing data, repeated monitor use, rare episodes, and selected roadside surveys limit the result. This is a test of exposure definition, not a test of bird damage or climate effects. The package has daily evidence, raw source hashes, code, missing-data bounds, and calendar checks.

ozone · bird surveys · survey timing · exposure sensitivity · observation bias · missing data