Fig. 3.

Posterior probability of benefit (hazard ratio $\le 0.75$) and harm (hazard ratio > 1.0) for the mixture prior (blue lines) and for monitoring based on the partial likelihood (black lines), which is equivalent to using a flat (improper) prior. Monitoring was at calendar dates (15 Jan 1991, 31 Jul 1991, 31 Dec 1991, 30 Mar 1992) with the X-axis indicating the cumulative number of toxoplasmosis or death events from the combined arms.