Performance measures and input uncertainty for pedestrian crossing exposure estimates

► We investigate uncertainty in annual pedestrian crossing exposure estimates. ► We test estimates made by expanding short-term counts using temporal factors. ► We use a database from video review involving over 350,000 observations. ► Using local vehicle factors, typical errors are −53% to 50% (mean −2%). ► Using external pedestrian factors, typical errors are −62% to 170% (mean 27%).