Bayesian Decision Analysis and Patient Preferences
The interactive online graph shows the expected “cost” to patients for an RCT design with a particular statistical significance level. It also shows the trade-off between the false positive rate (i.e., Type 1 error) and the true positive rate (i.e., 1–Type 2 error) for the given clinical trial design. Can you set the significance level so that it minimizes the expected cost to patients? What happens when the cost of a Type 1 (false positive) or Type 2 (false negative) error changes?