Abstract

The present paper is aimed at presenting an application to reaction times of an alternative data analysis method for treating unplanned comparisons in a repeated measures series. We compared such method with two different procedures. The first procedure consisted of a series of single pairwise comparisons, without controlling for type I error. The second procedure was the standard version of the Bonferroni method. It is worth noting that, as the number of comparisons gets larger, with the first method a substantial increase of the probability to make a Type I error is observed. Conversely, with the application of the method developed by Bonferroni, the increase in the number of comparisons results in a higher probability of making a Type II error. The empirical comparison, supported by a "Bootstrap" analysis, allowed to show that the alternative method, known as "False Discovery Rate" (FDR) possesses many advantages and is particularly easy to implement.