Abstract

A Monte Carlo experiment with reaction times (RTs) was conducted to evaluate the performance of four P-value adjustment procedures in the context of multiple comparisons. The Ex-Gaussian model was taken as reference for RT generation. We manipulated sample size and the exponential component of the Ex-Gaussian distribution. Multiple comparisons were then conducted by adjusting P-values according to the Bonferroni method and three procedures based on False Discovery Rate. Both type I error and power rates were examined for each of the four different methods. The results showed that after decomposing RT, power rates were higher for methods based on False Discovery Rate than for Bonferroni.