A comparison of methods for measuring event-related potentials.

Visual event-related potentials (ERPs) were recorded from 21 hyperactive children under 2 attention conditions and 4 doses of stimulant drug (methylphenidate). This data base was used to evaluate several methods of EP component measurement. These methods were (1) conventional visual peak and trough selection; (2) automatic feature extraction based on peaks; (3) automatic features extraction based on segments; (4) gross amplitude measures; (5) principal components analysis on normalized data and latency-adjusted data. No one method emerged as the best overall. Rather it is the case that different methods are best suited to different purposes, and criteria for choosing methods are outlined.