Group Sequential test Procedures included in version 3.0.

In a group sequential design there is at least one analysis (test of a hypothesis or confidence interval) performed before the planned final analysis. Data is collected until a pre-defined information level is achieved (for example a certain amount of data has been collected) and a test of the null-hypothesis is performed. If it turns out that the null-hypothesis is rejected at a pre-determined level of significance at the interim analysis, the testing stops, Otherwise more data is collected and a new analysis at the next interim analysis is performed. This process continues until either the null-hypothesis is rejected at an interim analysis or until one reaches the planned final analysis. Since there is more than one analysis performed of the same null-hypothesis, the p-values at the interim analyses and the final analysis have to be adjusted to control the pre-determined type I error by choosing one of the design option.



Test Procedures
  • Normal Distribution Means
    • Design
    • Conditional Power
    • Predictive Power
  • Test of Proportions
    • Design
    • Conditional Power
    • Predictive Power
  • Logrank test
    • Design
    • Conditional Power
    • Predictive Power

Design Options
  • Pocock
  • O'Brien and Fleming
  • Wang and Tsiatis
  • Haybittle and Peto
  • Lan-DeMets (Pocock)
  • Lan-DeMets (O'Brien-Fleming)
  • Gamma family (Hwang-Shih-DeCani)
  • Rho family (Kim-DeMets)
  • User defined
  • Power family

 

New in Version 3.0 .

CREOSTAT HB, Enbarsvagen 11, 42655 V.Frolunda, Sweden, email: mail@studysize.com