simcreate: Generate simulations for comparison

In mrzdcmps/changeofevidence: Change of Evidence

Generate simulations for comparison

Description

This function generates a dataframe containing multiple simulations of random data.

Usage

simcreate(
  trials,
  n.sims = 1000,
  mean.scores = NULL,
  method = c("pseudo", "files", "quantis"),
  filespath = "RandomFiles/",
  parallel = TRUE,
  nstart = 5,
  alternative = c("two.sided", "less", "greater"),
  prior.loc = 0,
  prior.r = 0.1,
  p = 0.5,
  use.files = NULL,
  use.quantis = NULL
)

Arguments

trials	The amount of trials in a single experiment (this includes all subjects).
n.sims	The amount of simulations to be generated. 10,000 is recommended, this might take several hours or days, though, depending on the sample size and number of trials.
mean.scores	Should bits be summed up to represent a normal distribution centered about a mean? If yes, indicate the desired mean score here. If you are comparing to binomial data, set to NULL.
method	The method used to generate the random data. Options are "pseudo" (software-based pseudo-RNG), "files" (text-files containing random bit sequences) and "quantis" (trueRNG device). Default is "pseudo".
filespath	If random files should be used indicate the path to those files.
parallel	If set to TRUE, multiple cores are being used in parallel to generate the simulations (recommenden).
nstart	Number of data points that are considered before calculating the first BF (min = 2)
alternative	Set parameter for Bayesian testing (t-Test).
prior.loc	Set parameter for Bayesian testing (t-Test).
prior.r	Set parameter for Bayesian testing.
p	Set parameter for Bayesian testing (Binomial) or data generation for probabilities of success other than 0.5.

Details

The tests of this package rely on comparisons of the experimental data to simulated random data sets. This function can provide a dataframe containing multiple simulated runs of the experiment with completely random data. To generate the random data a software-based pseudo-RNG can be used or - better - the package provides text-files containing random bit sequences previously generated by a quantum-based trueRNG (Quantis). Simulations can be created for binomial datasets and bits can be summed up to represent normally distributed mean scores.

Value

A dataframe with trials*nsims rows containing the columns "simid","index","rw","density.rw","bf" and "density.bf".

Examples

sims.pseudo <- simcreate(trials = 100, n.sims = 1000, method = "pseudo")
sims.files <- simcreate(trials = 100*20, n.sims = 1000, mean.scores = 10, method = "files", filespath = "RandomFiles/")
sims.quantis <- simcreate(trials = 1000, method = "quantis")
sims.parallel <- simcreate(trials = 376*100, n.sims = 1000, mean.scores = 50, method = "pseudo", parallel = TRUE) # 376 participants with 100 summed up trials each

mrzdcmps/changeofevidence documentation built on Feb. 27, 2025, 3:10 a.m.