sim.experiment: Simulate data of an experiment with a given linear design

In ccamp83/mu: R package for general statistical analysis of univariate and multivariate data

Simulate data of an experiment with a given linear design

Description

Simulate data of an experiment with a given linear design

Usage

sim.experiment(design, subj, repetitions, random_seed = F, return_data = T)

Arguments

design	an object of the class mu.exp.design produced with the function design.experiment
subj	number of subjects
repetitions	number of repetitions per subject
random_seed	logical. When FALSE (default) the data is simulated using the seed passed via the design object. Set to TRUE to generate a random seed (ignores the seed parameter passed by the design object)
return_data	default to TRUE. Use FALSE to produce just a trials sequence

Value

An object of the class mu.exp.design with the following values

seed the seed that was used during the simulation

exp the simulated data

coef.matrix the subj-by-subj coefficient matrix

Examples

libraries()

# set a seed
trial.seed <- 9000

# initialize the factors list
lIV <- list(
  Var1 = factor(1:2, labels = c("a","b"))
  ,
  Var2 = factor(1:2)
)

tIV <- c("w","b")

# describe the model
exp.formula <- ~ Var1*Var2

# initialize the coefficients
cus.beta <- c(3, 6, 9, 12) # means of the fixed effects
cus.theta <- c(1, 2, 3, 4)  # standard errors of the random components
cus.sigma <- 3 # sigma of the model

# few examples
(r <- design.experiment(lIV, tIV, betas = cus.beta, thetas = cus.theta, s = cus.sigma, exp.formula, seed=trial.seed))

# start the simulation
exp.info <- sim.experiment(design = r, 
                           subj = 10, 
                           repetitions = 10)

# extract the simulated data
exp <- exp.info$exp

# plot the results
ggplot(aes(Var1, y, color=Var2, group=Var2), data=exp) + facet_wrap(~subjName) + 
  geom_point(size=3, alpha=.4) +                                                  # draw raw scores
  stat_smooth(method=lm, level=.68, se=F, linewidth=1, formula = y ~ x) +         # draw fit line
  stat_summary(fun.data = mean_se, geom='point', size=3, color='black') +         # draw mean points
  stat_summary(fun.data = mean_se, geom='errorbar', width=.2, linewidth=1.2) +    # draw errorbars
  theme_bw()

# check a fit of the data (with more subjects)
exp.info <- sim.experiment(design = r, 
                           subj = 100, 
                           repetitions = 10)
exp <- exp.info$exp
(lmod1 <- lme4::lmer(y ~ Var1*Var2 + (1 + Var1*Var2|subjName), data=exp,
                     control = lmerControl(optimizer = "bobyqa")))

ccamp83/mu documentation built on Sept. 10, 2023, 4:22 p.m.