Mistreatment Probabilities
This app explores the probability that an incorrect treatment decision is made as a result of measurement error in a tailoring covariate. Error-prone data of the form X* = X + U are generated, with X and U normally distributed. A treatment rule of the form 'Treat if X > t' is applied, for some treatment threshold t. The resulting graph shows the probability that an incorrect treatment decision is made if it is based on the error-prone X*. For example, if X* = 16, X = 14, and the treatment rule is 'treat if X > 15'.
Enter mean of X
Enter variance of X
Enter variance of U
Enter treatment threshold
Adjust plot range
Resample!