p1.given.p0 {MChtest} | R Documentation |
Consider the SPRT for testing Ho:p=p0 vs H1:p=p1 where p1< Alpha < p0. For Monte Carlo tests, we want to reject and conclude that p<Alpha. In terms of the resampling risk at p (i.e., the probability of reaching a wrong decision at p) the minimax SPRT has a particular relationship between p0 and p1. Here we calculate p1 given p0 or vise versa to obtain that relationship.
p1.given.p0(p0, Alpha = 0.05, TOL = 10^-9) p0.given.p1(p1, Alpha = 0.05, TOL = 10^-9)
p0 |
null such that p0>Alpha |
p1 |
alternative such that p1<Alpha |
Alpha |
want tsprt associated with testing p=Alpha or not |
TOL |
tolerance level input into call to uniroot |
either p0 or p1
Michael P. Fay
Fay, M.P., Kim, H-J. and Hachey, M. (2007). Using truncated sequential probability ratio test boundaries for Monte Carlo implementation of hypothesis tests. (to appear Journal of Computational and Graphical Statistics).
p1.given.p0(.04)