findSize: Determine a sample size

In dkahle/bayesRates: Two-Sample Tests and Sample Size Determination from a Bayesian Perspective

Description

Determine a sample size for a given power and hyperparameter configuration

Usage

findSize(power, a1, b1, a2, b2, a = a1, b = b1, pi0 = 0.5, pi1 = 1
  - pi0, c = 1, family = c("binomial", "poisson"), method = c("ones",
  "halves"), quiet = TRUE)

Arguments

power	desired power
a1	alpha, the hyperparameter of the prior distribution of the first poisson rate
b1	beta, the hyperparameter of the prior distribution of the first poisson rate
a2	alpha, the hyperparameter of the prior distribution of the second poisson rate
b2	beta, the hyperparameter of the prior distribution of the second poisson rate
a	alpha, the hyperparameter of the prior distribution under the null
b	beta, the hyperparameter of the prior distribution under the null
pi0	the prior probability of the null hypothesis
pi1	the prior probability of the alternative hypothesis
c	relative loss constant (loss due to type II error divided by loss due to type I error)
family	"binomial" or "poisson", depending on test
method	"ones" or "halves"
quiet	message the user with calls to samplePower

See Also

samplePower

Examples

## Not run: 

	
a1 <- 2; b1 <- 6
a2 <- 3; b2 <- 3
plotGamma(c(a1, a2), c(b1, b2))
findSize(power = .8, a1, b1, a2, b2, family = "poisson")

a1 <- 5; b1 <- .2 # E = 1
a2 <- 30; b2 <- .167 # E = 5
findSize(power = .8, a1, b1, a2, b2, family = "poisson")

a1 <- 4; b1 <- 2
a2 <- 6; b2 <- 2
findSize(power = .5, a1, b1, a2, b2)
findSize(power = .6, a1, b1, a2, b2)
findSize(power = .7, a1, b1, a2, b2)
findSize(power = .8, a1, b1, a2, b2) # takes a while
findSize(power = .8, a1, b1, a2, b2, method = "halves")

findSize(power = .9, a1, b1, a2, b2, method = "halves")



## End(Not run)

dkahle/bayesRates documentation built on May 15, 2019, 9:07 a.m.