rSimulations {reproducer}

rSimulations

Description

This function simulates many datasets from the same bivariate distribution to investigate the distribution of correlations for specific sample sizes.

Usage

rSimulations(
  mean,
  var,
  diff,
  r,
  N,
  reps,
  VarAdj = 0,
  seed = 123,
  returntSignificant = F,
  returndata = F,
  plothist = F
)

Arguments

mean

The mean used for one of bivariate distributions - assumed to be the control condition in an experiment.

var

The variance used for both treatment groups. It must be a real value greater than 0.

diff

This value is added to the parameter mean to specify the mean for the other bivariate distribution - assumed to be the treatment condition in an experiment.

r

This specifies the correlation coefficient to be used for the bivariate normal distribution it must be a value in the range [-1,1].

N

The number of observations in each simulated bivariate normal data set.

reps

The number of bivariate data sets that will be simulated.

VarAdj

This value will be added to the variance of the treatment condition.

seed

This specifies the seed value for the simulations and allows the experiment to be repeated.

returntSignificant

If set to true the percentage of times the t-test delivered a value significant at the 0.05 level is reported (default returntSignificant=F).

returndata

If set to FALSE, the function returns the summary information across all the replications (default returndata=F). If set to TRUE the function outputs the r and variance ratio, and variance accuracy values generated in each replication.

plothist

If set to T, the function outputs a histogram of the r-values, the varprop values and the accuracy values (default plothist=F).

Value

output If returndata=F, the output returns summary information about the average of r and the variance properties across the replicated data sets. If returndata=T, the function returns the r-values obtained for each of the simulated data sets to gather with the variance ratio, the variance accuracy measure and a dummy variable indicating whether a test of significance between the mean values was significant (which is indicated by the dummy variable being set to 1) or not (which is indicated by the dummy variable being set to 0)

Author(s)

Barbara Kitchenham and Lech Madeyski

Examples

# output=rSimulations(mean=0,var=1,diff=0,r=0.25,N=4,reps=10000)
# reduced reps to pass CRAN time limits
output <- rSimulations(mean = 0, var = 1, diff = 0, r = 0.25, N = 4, reps = 1000)
output <- signif(output, 4)
output
#  r.Mean r.Median  Var.r PercentNegative Mean.VarProp Variance.VarProp ...
# 1 0.2132   0.3128 0.3126           34.21       0.5036          0.06046 ...
# output=rSimulations(mean=0,var=1,diff=0.8,r=0.25,N=60,reps=10000,returntSignificant=TRUE)
# reduced reps to pass CRAN time limits
output <- rSimulations(mean = 0, var = 1, diff = 0.8, r = 0.25, N = 60,
  reps = 1000, returntSignificant = TRUE)
output <- signif(output, 4)
output
#   r.Mean r.Median   Var.r PercentNegative Mean.VarProp Variance.VarProp ...
# 1 0.2492   0.2534 0.01529            2.62       0.5009         0.003897 ...
output <- rSimulations(mean = 0, var = 1, diff = 0, r = 0.25, N = 30, reps = 10, returndata = TRUE)
output
#     rvalues   VarProp VarAccuracy VarDiffAccuracy tSig
# 1  0.3981111 0.4276398   0.8630528       0.6974386    0
# 2  0.2104742 0.4994285   0.7812448       0.8224174    0
# 3  0.4252424 0.4933579   1.1568545       0.8866058    0
# 4  0.3502651 0.6004373   0.8710482       0.7628923    0
# 5  0.3845145 0.6029086   0.9618363       0.7998859    0
# 6  0.1397217 0.4201069   1.1817022       1.3582855    0
# 7  0.2311455 0.3894894   0.8322239       0.8594886    0
# 8  0.3725047 0.5985897   1.1742117       0.9938662    0
# 9  0.4881618 0.2712268   0.7585261       0.5723671    0
# 10 0.1568071 0.3936400   0.9869924       1.1143561    0