Randomized trial shows improved error control in online multiple testing, indicating enhanced statistical power.
Key Points
This work aims to develop a framework for online multiple test procedures that control the familywise error rate despite dependence among test statistics.
Developed new test procedures ensuring asymptotic control of familywise error rate.
Provided concrete examples of procedures and discussed their properties.
Conducted simulations to verify type I error control for finite sample sizes.
The proposed procedures maintained control of the familywise error rate under dependence.
Simulation results indicated improved power compared to existing methods.