A Critical Perspective on the Analysis of Trial-Averaged Event-Related Potentials

Abstract

The study compares single-trial multilevel analysis and averaged-trial repeated measures anova methods for analyzing event-related potentials by means of simulation procedures. Sample size, number of stimulus repetitions, effect size, and presence or absence of random slopes were systematically manipulated. Five thousand iterations were used in each simulation condition. The first and second files contain code for obtaining comparative Power and Type I error rates, as well as other indicators of comparative inferencial evidence like coverage and bias for experimental effects. The third and fourth files contain results for the individual runs in the multilevel and repeated measures anova simulations. The last file contains summary results.