The 'Fishing License' method for analysing the time course of effects in repeated measurements

Inference on the time of onset and the duration of a treatment effect is a challenging problem in biomedical research. These studies often generate repeated measurements from subjects in a treated group and a control group during the same time period. We propose a simple approach, called the 'Fishing License' method, to test for a treatment effect occurring during an unspecified time interval. The method is based on a statistic of the largest absolute value of the t statistic between two groups obtained by considering every possible time interval in the observed time period. A bootstrapped null distribution of the test statistic is used to determine the critical value. The method also provides estimates of onset and ending times, when the null hypothesis of no effect is rejected. Simulated and real experimental data sets were generated for assessing the performance of the method.