What is the appropriate number of events to collect for multiparameter population comparison?
The more events, the better will be your ability to discriminate different distributions. It doesn’t matter how many events you collect; the statistic is computed based on the number of events. In other words, if you collect fewer events, then you will have a much lower T(x) for two distributions than if you had collected more events–indicating that the probability that these two distributions are different is not as high.