Fig. 4

Type I error rate when not adhering to the stopping rule and always proceeding to the second stage \(\alpha ^d_\text {no stop}\) for \(p_a-p_0 = 0.15\), \(\alpha = 0.05\) and \(\beta = 0.2\). The introduced optimized design always keeps the type I error rate even if the stopping rules are not adhered to