sim_test: Simulation tests

In nguyenllpsych/thirt: Thurstonian FC-IRT Model

Simulation tests

Description

Examine parameter recovery and performance using simulated data through parallelization

Usage

sim_test(icondition, stan = TRUE, sem = TRUE, seed = 20220415)

Arguments

icondition	integer indicating the iteration of simulation condition
stan	boolean indicating whether estimation results should be compared with thurstonianIRT stan
sem	boolean indicating whether estimation results should be compared with thurstonianIRT lavaan and mplus
seed	random seed

Value

save a timestamped RData file in the current working directory

Examples

## Not run: 

# load parallel package
library(parallel)

# Simulation conditions
n_person      <- 2
n_item        <- 4
n_neg         <- 1
n_block       <- 5
n_dim         <- c(4, 6)
n_iter        <- 100
n_burnin      <- 20
step_size_sd  <- 0.1

condition_mat <- expand.grid(n_person = n_person,
                             n_item   = n_item,
                             n_neg    = n_neg,
                             n_block  = n_block,
                             n_dim    = n_dim,
                             n_iter   = n_iter,
                             n_burnin = n_burnin,
                             step_size_sd = step_size_sd)

n_condition   <- nrow(condition_mat)

# create clusters
n_core <- detectCores()
cl     <- makeCluster(n_core)
RNGkind("L'Ecuyer-CMRG")

# load libraries in clusters
clusterEvalQ(cl, library(thirt))
clusterExport(cl, "condition_mat")

# simulation tests
parLapply(cl, 1:n_condition, sim_test)

# stop cluster
stopCluster(cl)

## End(Not run)

nguyenllpsych/thirt documentation built on Feb. 14, 2024, 10:53 p.m.