Automated auditory response detection: statistical problems with repeated testing

Sequential application of a statistical test is usually applied in an automated auditory response detection algorithm. The sequential test strategy is very time-efficient but increases the probability of a false rejection of the null-hypothesis. For this reason, it is necessary to correct the critical test value. However, the well-known Bonferroni correction leads to an over-correction when dealing with dependent or partly dependent data. The objective of the study reported here was to develop a method to determine the critical test value for the sequential testing of dependent data. Extensive Monte Carlo simulations were used to develop this method. The simulation results were reviewed and the benefit of the suggested method, in comparison to the Bonferroni correction, was shown using a large sample of real amplitude modulation following response data. The detection rate determined for these data and the ROC curve demonstrate the advantage of using the method suggested here.