calculateLargeSampleRandomizedBlockDesignEffectSizes: calculateLargeSampleRandomizedBlockDesignEffectSizes

In reproducer: Reproduce Statistical Analyses and Meta-Analyses

calculateLargeSampleRandomizedBlockDesignEffectSizes

Description

The function uses a simulates a large experiment to estimate the asymptotic values of the probability of superiority, Cliff's d and the standardized mean difference data for a four group randomized blocks experiment for four different distributions: Normal (i.e. type='n'), log-normal (i.e. type='l'), gama (i.e. type='g') and Laplace (i.e., type='lap').

Usage

calculateLargeSampleRandomizedBlockDesignEffectSizes(
  meanC = 0,
  sdC = 1,
  diff,
  N = 5e+06,
  type = "n",
  Blockmean = 0,
  StdAdj = 0
)

Arguments

meanC	to act as the mean of the distribution (default 0) used to generate the control group data (note for the gamma distribution this is the rate parameter and must not be zero)
sdC	the variance/spread of the distribution (default 1) used to generate the control group data.
diff	a value added to meanC to generate the treatment group data (default 0).
N	the size of each group (default 5000000)
type	the distribution of the data to be generated. One of: 'n' for normal (default), 'l' for log-normal, 'g' for gamma, and 'lap' for Laplace.
Blockmean,	a value that can be added one of the blocks to represent a fixed block effect (default 0).
StdAdj	a value that can be added to sdC to introduce heterogeneity into the treatment group (default 0).

Value

A tibble identifying the sample statistics and the values of the probability of superiority, Cliff's d and StdMD (labelled StdES)

Author(s)

Barbara Kitchenham and Lech Madeyski

Examples

set.seed=400
calculateLargeSampleRandomizedBlockDesignEffectSizes(
  meanC=0, sdC=1, diff=.5, N=100000, type='n',Blockmean=0.5,StdAdj = 0)
#  MeanC   SdC MeanT   SdT    BE  Phat Cliffd   UES   Var StdES
#  <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>  <dbl> <dbl> <dbl> <dbl>
#1     0     1   0.5     1   0.5 0.638  0.277 0.501 0.998 0.502

reproducer documentation built on Oct. 18, 2023, 5:10 p.m.