Figure 3

Bayesian tri-plots of the efficacy of thymectomy using different prior probabilities in the matched and unmatched datasets. In Figure 3A and B, the blue line depicts the skeptical prior, assuming prior belief of no efficacy of thymectomy, with a mean HR = 1 and 95% CI:0.6 -1.4. The green line depicts the likelihood, which is the probability of thymectomy efficacy for each model, using the observed data only (uninformative prior). The red line reflects the posterior probability, which incorporates the prior belief to the likelihood. Figure 3A shows the probabilities for the matched dataset. The likelihood has a 96% probability of thymectomy efficacy (HR >1), with a posterior probability of 79% after incorporating the skeptical prior. Figure 3B reflects the unmatched dataset. The likelihood has a 98% probability of thymectomy efficacy, with a posterior probability of 86% after incorporating the skeptical prior.