Probability density difference between invasive and conservative arms in TACTICS for combined death/MI outcome . A binomial model is used for the control and intervention groups, and then, because the sample sizes were large, the posterior distribution, which is really the difference between two beta distributions, is approximated with a normal distribution. A Haldane prior (beta(0,0)), which contributes no information to the analysis, has been used. The area under the curve to the left of any given point represents the probability of the superiority of the invasive arm. For example, there is a 97% probability that the invasive arm is superior to the conservative arm (area under the curve to the left of the vertical line at 0). However, the probability that the benefit exceeds 1% or 2% is only 85% and 57%, respectively.