ran.CBData {CorrBin}

Generate random correlated binary data

Description

ran.mc.CBData generates a random CBData object from a given two-parameter distribution.

Usage

ran.CBData(
  sample.sizes,
  p.gen.fun = function(g) 0.3,
  rho.gen.fun = function(g) 0.2,
  pdf.fun = qpower.pdf
)

Arguments

sample.sizes

a dataset with variables Trt, ClusterSize and Freq giving the number of clusters to be generated for each Trt/ClusterSize combination.

p.gen.fun

a function of one parameter that generates the value of the first parameter of pdf.fun (p) given the group number.

rho.gen.fun

a function of one parameter that generates the value of the second parameter of pdf.fun (rho) given the group number.

pdf.fun

a function of three parameters (p, rho, n) giving the PDF of the number of responses in a cluster given the two parameters (p, rho), and the cluster size (n). Functions implementing two common distributions: the beta-binomial (betabin.pdf) and q-power (qpower.pdf) are provided in the package.

Details

p.gen.fun and rho.gen.fun are functions that generate the parameter values for each treatment group; pdf.fun is a function generating the pdf of the number of responses given the two parameters p and rho, and the cluster size n.

p.gen.fun and rho.gen.fun expect the parameter value of 1 to represent the first group, 2 - the second group, etc.

Value

a CBData object with randomly generated number of responses with sample sizes specified in the call.

Author(s)

Aniko Szabo, [e-mail]

See Also

betabin.pdf and qpower.pdf

Examples

set.seed(3486)
ss <- expand.grid(Trt=0:3, ClusterSize=5, Freq=4)
#Trt is converted to a factor
rd <- ran.CBData(ss, p.gen.fun=function(g)0.2+0.1*g)
rd