Missing assistant status gave no clear common count association in four US bird survey samples in 1990-2025

We tested whether unreported assistant status was linked to recorded bird taxa in 55,627 consecutive-year pairs on the same route with the same observer. Separate models covered western and eastern US routes in 1990-2006 and 2007-2025. Controls included weather, timing and state-year effects. Adjusted associations ranged from -0.30 to +0.39 taxa per change from reported to missing status. All four primary intervals included zero after correction for four tests. Some sensitivity results differed. Missing status rose from about 6 percent to 22-23 percent of weather-complete runs between periods. The evidence gives no clear common count association. It does not establish that complete-case analysis is safe or that missingness is random. Selected roadside surveys, unmeasured assistant use, detection and shared shocks limit inference. The package includes code, source hashes, model inputs, numerical checks and offline reproduction instructions.

biodiversity · observation bias · missing data · assistant status · Breeding Bird Survey · weather · first differences · null result