Group sequential test (GST) calculator
Plan your interim analyses before launch and stop the test as soon as an efficacy boundary is crossed.
Calculation method
Group Sequential Calculator (GST)
Stopping boundaries calculated using the Armitage-McPherson-Rowe recursion on the B-value scale (cumulative score, with independent increments: B_k = Z_k√t_k). Alpha spending across analyses via a Lan-DeMets spending function, approximated as O'Brien-Fleming (2(1−Φ(z_{α/2}/√t))) or Pocock (α·ln(1+(e−1)t)). Boundaries solved by bisection analysis by analysis, then drift solved by bisection to reach target power, yielding maximum information and sample size.
Implementation note: boundaries derived from a spending function differ slightly from the exact tabulated O'Brien-Fleming and Pocock designs (Jennison & Turnbull). The difference is most visible on the first analysis, where the boundary is exactly z_{α/2}/√t₁ by construction.
FAQ
Frequently asked questions about group sequential testing
- A Group Sequential Test plans, before launch, a fixed number of interim analyses. At each analysis, the result is compared to a boundary calculated so that the overall risk of a false positive remains equal to the chosen threshold. If the boundary is crossed, the test stops earlier.
Tell us more about your project
We know how to increase the performance of your digital channels.
- Experimentation
- UX & Experience
- Personalization & CRM
- Data & Analytics
- Training
- Agentic commerce
