rB: Data generation: Log-risk-ration of a binomial-Gaussian model

In metagen: Inference in Meta Analysis and Meta Regression

Description

Random draws of log risk ratios from a hierarchical binomial Gaussian model.

Usage

  rB(n, h, s, a, r, x, b)

Arguments

n	number of draws.
h	heterogeneity.
s	study sizes.
a	balance of group assignments.
r	fixed risk in the treatment group.
x	design matrix.
b	regression coefficients.

Details

It is always assumed that at least one response in a study has happend, i.e., a response of 0 in a treatment or control group is rounded up to 1. Note that this may lead to an overestimation of small risks. If possible, make sure your sample sizes are large enough to compensate for this effect.

Value

A (2k,n) matrix. Each column is an independent draw.

Examples

h_test <- .03
x_test <- cbind(1,1:13)
b_test <- c(0.02, 0.03)
s_test <- rep(2000, 13)
a_test <- rep(.3, 13)
rB(n=10, h=h_test, s=s_test, a=a_test, r=.3, x=x_test, b=b_test)

metagen documentation built on May 2, 2019, 6:08 a.m.