powerplot: Power plots for multivariate RR methods

In danheck/RRreg: Correlation and Regression Analyses for Randomized Response Data

Power plots for multivariate RR methods

Description

Uses the function RRsimu to estimate the power of the multivariate RR methods (correlation RRcor, logistic regression RRlog, and/or linear regression RRlin.

Usage

powerplot(
  numRep,
  n = c(100, 500, 1000),
  pi,
  cor = c(0, 0.1, 0.3),
  b.log = NULL,
  model,
  p,
  method = c("RRcor", "RRlog", "RRlin"),
  complyRates = c(1, 1),
  sysBias = c(0, 0),
  groupRatio = 0.5,
  alpha = 0.05,
  nCPU = 1,
  show.messages = TRUE
)

Arguments

numRep	number of boostrap replications
n	vector of samples sizes
pi	true prevalence
cor	vector of true correlations
b.log	vector of true logistic regression coefficients
model	randomized response model
p	randomization probability
method	multivariate RR method
complyRates	probability of compliance within carriers/noncarriers of sensitive attribute
sysBias	probability of responding 'yes' in case of noncompliance
groupRatio	ratio of subgroups in two-group RR designs
alpha	type-I error used to estimate power
nCPU	either the number of CPU cores or a cluster initialized via makeCluster.
show.messages	toggle printing of progress messages

Value

a list of the class powerplot containing an array res with the power estimates and details of the simulation (e.g., model, p, pi, etc.)

See Also

RRsimu for Monte-Carlo simulation / parametric bootstrap

Examples

# Not run
# pplot <- powerplot(100, n=c(150,250), cor=c(0,.3,.5),
#                   method="RRlog", pi=.6, model="Warner", p=.3)
# plot(pplot)

danheck/RRreg documentation built on Dec. 3, 2022, 7:50 p.m.