A reaction-time result includes display, browser, input device, operating-system scheduling, frame rate, and anticipation as well as human response. Use the same setup, complete several trials, discard obvious false starts according to the stated rule, and compare medians rather than one best attempt. Wireless or touch input can add variable latency. Rest between sets and do not treat a web test as a medical assessment. For a timing bug, include client, device, display rate, input method, trial values, and false-start behavior.
Akousa