Difference Method: A New Research On Multiple Testing In Biostatistic

In order to deal with the high dimension biological data,we usually use multiple testing which means to test a lot of hypotheses simultaneously.The old approach to multiplicity problem calls for controlling the familywise error rate(FWER),but it is thought to be too strict.Now,the popular approach is to control the false discovery rate(FDR)(Benjamini,Y.and Hocherg,Y.1995).John D.Storey(2001)introduce a modified version of the FDR called positive false discovery rate(pFDR).His approach is more applicable and powerful than the Benjamini-Hochberg FDR method.Having researched these methods,we introduce a new one called "difference method".This method applies the difference of the two adjoining p-values which come from the sequential p-values to fix the cut off point.If the differences are always smaller than some constant,we will have sufficient reason to reject the hypothesis which are performed with the p-values with small differences.Our method has an advantage that it can find some non-true null hypotheses in the acceptance region,so it maybe more powerful. At last,we compare our method with the Benjamini-Hochberg's FDR method and the John D.Storey's method.In some instance,our method has more power.In the Section 2 of this paper,we will introduce different visions of error rate in multiple testing and the approaches with different error rates.And we will propose a new method(the difference method)and analysis the advantage of our method in Section 3.In Section 4,we mainly compare our method with the other two methods by simulations.