Simulate Longitudinal Binomial Data — sim_dat

TODO: Build in check if the correlations are too high! Where is that code? Note that the variances aren't used if passed, Requires MASS R package

sim_dat_binom(
  N = 100,
  number.groups = 2,
  number.timepoints = 4,
  reg.formula = NULL,
  Beta = 0,
  corr = "ar1",
  cor.value = NULL,
  cond.mcar = F,
  Covariate = F
)

Arguments

reg.formula	this is a regression formula to pass to the data generation; null is formula(~ Group + Time + Time*Group),
Beta	this is the Beta for the regression equation; numeric matrix of Beta values OR a scalar value != 0 that will be the final value of the interaction parameters, default is all(Beta == 0) for Type I error simulations
corr	options for correlation structure c('ind', 'ar1', 'cs'), default is 'ar1'
cor.value	numeric, the first corr in ar1, the corr in cs option
cond.mcar	logical; do you want a conditional MCAR data generation
Covariate	logical; do you want a random simulated covariate?

Value

returns a dataframe containing the simulated data