Now Playing: Conditional Power for Two Proportions (2:42)
Suppose a study has been planned to detect a 0.1 change in proportions, from P1 = 0.6 to P2 = 0.7, at an alpha of 0.025, using a one-sided z-test. The target sample size is 114 per group. An interim analysis is planned after half the data have been collected. The data monitoring board would like to have the conditional power calculated for z values of 0, 0.5, 1, 1.5, 2, and 2.5.
The conditional and predictive power values corresponding to each z value are shown in the numeric results section.
For a z value of 2.5, the conditional power is just over 0.8, and the predictive power is over 0.94.
Now, suppose that an interim analysis is conducted after half of the data has been collected. The analysis yields a z-test value of 2.12. The value of P1, the control group proportion, is found to be 0.643.
The data monitoring board would like to recalculate the sample size assuming a P2 value of 0.743 (keeping the difference at 0.1), and a conditional power of 0.8.
Notice that the target sample size has increased from 114 per group, to 197 per group.