| simulateRandomizedBlockDesignEffectSizes {reproducer} | R Documentation |
simulateRandomizedBlockDesignEffectSizes
Description
This simulates a two-block and two-treatment design based on one of four distributions, and finds the values of ktau and Cliffs d and their variances. It simulates a randomised blocks experiment with two treatment groups and two control groups each of which being divided into two blocks. It assumes equal group sizes but group spread (standard deviation can be changed, see StAdj). It returns values of both parametric and non-parametric effect sizes and their variance and significance. For the logarithmic distribution it calculates effect sizes based on the log transformed data as well as the raw data.
Usage
simulateRandomizedBlockDesignEffectSizes(
mean,
sd,
diff,
N,
type = "n",
alpha = 0.05,
Blockmean = 0,
BlockStdAdj = 0,
StdAdj = 0,
AlwaysTwoSidedTests = FALSE,
ReturnData = FALSE
)
Arguments
mean |
The default value for all groups which can be changed for the two treatment groups using the parameter diff and for the two block 2 groups using the parameter Blockmean |
sd |
The default spread used for all four groups unless adjusted by the StdAdj. It must be a real value greater than 0. |
diff |
This is added to the parameter mean to obtain the required mean for treatment groups. It can be a real value and can take the value zero. |
N |
this is the number of observations in each group. It must be an integer greater than 3. |
type |
this specifies the underlying distribution used to generate the data. it takes the values 'n' for a normal distribution, 'l' for lognormal distribution,'g' for a gamma distribution, 'lap' for a Laplace distribution. |
alpha |
The level used for statistical tests (default 0.05). |
Blockmean |
if >0 an adjustment made to both group means in Block 2 |
BlockStdAdj |
if >0, an adjustment that can be made to the sd of each group in block 2 |
StdAdj |
this specifies the extent of variance instability introduced by the treatment and if >0 will be used to amend the sd parameter for both treatment groups. This value must be positive and less than 0.5 |
AlwaysTwoSidedTests |
Logical varable (default FALSE) if TRUE the function always performs two-sided tests. Otherwise if the parameter diff is not equal to zero, the function performs one-sided tests. |
ReturnData |
Logical variable, If TRUE, the function simply returns the generated data. If false (which is default value) the function returns various effect sizes and whether the effect sizes are statistically significant. |
Value
data frame incl. either the non-parametric and parametric effect sizes and whether the effect sizes are significant at the 0.05 level or the generated data depending on the value of the ReturnData parameter.
Author(s)
Barbara Kitchenham and Lech Madeyski
Examples
set.seed(123)
as.data.frame(
simulateRandomizedBlockDesignEffectSizes(
mean = 0, sd = 1, diff = .5, N = 10, type = "n", alpha = 0.05,
Blockmean = 0.5, BlockStdAdj = 0, StdAdj = 0))
# N phat phat.var phat.df phat.test phat.pvalue phat.sig phat.ci.upper phat.ci.lower d
# 1 40 0.79 0.005866667 30.15715 3.786189 0.0003403047 TRUE 1 0.6600213 0.58
# vard d.sig d.ci.lower d.ci.upper cor sqse ctvar n1 n2 sigCVt sigCVn
# 1 0.02430788 TRUE 0.2775601 1 0.3052632 0.01315789 0.006953352 20 20 TRUE TRUE
# ttest.sig ES Variance StdES BlockEffect MedianDiff
# 1 TRUE 0.9402999 0.7829385 1.06268 0.307119 1.313642
set.seed(123)
as.data.frame(
simulateRandomizedBlockDesignEffectSizes(
mean = 0, sd = 1, diff = 0.5, N = 10, type = "n", alpha = 0.05,
Blockmean = 0.5, BlockStdAdj = 0, StdAdj = 0, AlwaysTwoSidedTests = TRUE)
)
# N phat phat.var phat.df phat.test phat.pvalue phat.sig phat.ci.upper phat.ci.lower
# 1 40 0.79 0.005866667 30.15715 3.786189 0.0006806094 TRUE 0.946392 0.633608
# d vard d.sig d.ci.lower d.ci.upper cor sqse ctvar n1 n2 sigCVt
# 1 0.58 0.02430788 TRUE 0.2135334 0.8033737 0.3052632 0.01315789 0.006953352 20 20 TRUE
# ttest.sig ES Variance StdES BlockEffect MedianDiff
# 1 TRUE 0.9402999 0.7829385 1.06268 0.307119 1.313642
set.seed(123)
as.data.frame(
simulateRandomizedBlockDesignEffectSizes(
mean = 0, sd = 1, diff = .5, N = 10, type = "l", alpha = 0.05,
Blockmean = 0.5, BlockStdAdj = 0, StdAdj = 0, ReturnData = TRUE))
# BaselineData.B1 AlternativeData.B1 transBaselineData.B1 transAlternativeData.B1
# 1 0.5709374 5.6073700 -0.56047565 1.72408180
# 2 0.7943926 2.3627208 -0.23017749 0.85981383
# 3 4.7526783 2.4615013 1.55870831 0.90077145
# 4 1.0730536 1.8416883 0.07050839 0.61068272
# 5 1.1380175 0.9456894 0.12928774 -0.05584113
# 6 5.5570366 9.8445021 1.71506499 2.28691314
# 7 1.5855260 2.7124451 0.46091621 0.99785048
# 8 0.2822220 0.2307046 -1.26506123 -1.46661716
# 9 0.5031571 3.3246217 -0.68685285 1.20135590
# 10 0.6404002 1.0275821 -0.44566197 0.02720859
# BaselineData.B2 AlternativeData.B2 transBaselineData.B2 transAlternativeData.B2
# 1 0.5667575 4.163950 -0.5678237 1.4264642
# 2 1.3258120 2.023702 0.2820251 0.7049285
# 3 0.5909615 6.653384 -0.5260044 1.8951257
# 4 0.7954150 6.541284 -0.2288912 1.8781335
# 5 0.8824622 6.181624 -0.1250393 1.8215811
# 6 0.3052289 5.412117 -1.1866933 1.6886403
# 7 3.8106015 4.729964 1.3377870 1.5539177
# 8 1.9220131 2.555092 0.6533731 0.9380883
# 9 0.5282757 2.001781 -0.6381369 0.6940373
# 10 5.7765980 1.858053 1.7538149 0.6195290