Multiple hypotheses testing problems are highly relevant whenever data is evaluated statistically. In business research, for example, the topic gains more importance due to the increase in data driven approaches for efficient decision making. By now, the decision on a suitable multiple hypotheses method presupposes a two-step selection. First, the user has to decide on the multiple type I error definition. This definition guarantees distinct conservatism regarding false inferences. Second, the decision is on a procedure for the predefined error definition. We introduce and prove an error flexible Bonferroni modification for medium size hypotheses applications, called SiMaFlex procedure. The method is able to flexibly outline Familywise Error Rate (FWER), False Discovery Rate (FDR) and unadjusted conservatism at the same time. Thus, our new method transforms the two-stage decision into a one-stage decision on the type I error definition. In addition, our method is one of the few that controls the FDR by a single step concept. Extensive simulation studies show that our new method outperforms existing procedures.

每当对数据进行统计评估时，多重假设检验问题都非常相关。例如，在商业研究中，由于使用数据驱动的有效决策方法的增加，该主题变得更加重要。到目前为止，对合适的多重假设方法的决策以两步选择为前提。首先，用户必须确定I型多重错误定义。该定义保证了有关错误推断的独特保守性。其次，决定用于预定义错误定义的过程。我们介绍并证明了针对中等大小假设应用的错误错误Bonferroni修改，称为SiMaFlex过程。该方法能够同时灵活地描述家庭错误率（FWER），错误发现率（FDR）和未经调整的保守性。从而，我们的新方法将两阶段决策转换为关于I类错误定义的一阶段决策。此外，我们的方法是通过单步概念控制FDR的少数方法之一。大量的仿真研究表明，我们的新方法优于现有程序。