Description Usage Arguments Details Value Examples
This function allows to run a simulation study of mpt2irt models. Data are
generated either from the Boeckenholt Model (genModel = "2012"
) or
from the Acquiescence Model (genModel = "ext"
). Subsequently, one or
both of these models are fit to the generated data using either JAGS or Stan.
The results are saved in an RData file in dir
.
1 2 3 4 5 6 7 8 9 | recovery_irtree(rrr = NULL, N = NULL, J = NULL, prop.rev = 0.5,
genModel = c("ext", "2012"), fitModel = c("ext", "2012", "pcm",
"steps", "shift", "ext2"), fitMethod = c("stan", "jags"),
theta_vcov = NULL, betas = NULL, beta_ARS_extreme = NULL,
df = NULL, V = NULL, M = 500, n.chains = 3, thin = 1,
warmup = 500, method = "simple", outFormat = NULL,
startSmall = FALSE, df_vcov = 50, dir = NULL, keep_mcmc = FALSE,
savext_all = FALSE, savext_mcmc = TRUE, add2varlist = c("deviance",
"pd", "popt", "dic"), ...)
|
rrr |
Sequence of integers (e.g., |
N |
number of persons |
J |
number of items. Can be a vector for multiple traits (e.g., J=c(10,10,10)). |
prop.rev |
number of reversed items. Can be a vector for multiple traits(e.g., prop.rev=c(5,3,5)/10) |
genModel |
Character. The data generating model (either "2012" or "ext"). |
fitModel |
Character. The model for data analysis ("2012", "ext", or both as vector c("2012", "ext")). |
fitMethod |
Character. Whether to use "stan" or "jags". |
theta_vcov |
true covariance matrix of response processes (order: middle, extreme, (acquiescence), trait). standard is diag(3) / diag(4). Can be a vector of variances (not SDs). |
betas |
Optional list. May have entries |
beta_ARS_extreme |
Numeric. Only for |
df |
degrees of freedom for wishart prior on covariance of traits (default: number of processes + 1) |
V |
prior for wishart distribution (default: diagonal matrix) |
M |
number of MCMC samples (after warmup) |
n.chains |
number of MCMC chains (and number of CPUs used) |
thin |
thinning of MCMC samples |
warmup |
number of samples for warmup (in JAGS: 1/5 for adaptation, 4/5 for burnin) |
method |
Passed to |
outFormat |
either "mcmc.list" (can be analyzed with coda package) or "stan" or "runjags" |
startSmall |
Whether to use random starting values for beta sampled from "wide" (FALSE) or "narrow" priors (TRUE; beta and theta closer to 0; might solve problems with slow convergence of some chains for extreme starting values). |
df_vcov |
Numeric. Degrees of freedom for wishart distribution from which the variance-covariance matrix for generating the data is drawn. |
dir |
Path to directory where results should be stored, |
keep_mcmc |
Logical indicating wheter to retain, besides a summary of the parameters, the raw mcmc samples. |
savext_all |
Logical indicating wheter to save the output from Stan/JAGS in an external RData file. |
savext_mcmc |
Logical indicating wheter to save the mcmc samples in an external RData file. |
add2varlist |
Additional variables to monitor (e.g., |
... |
further arguments passed to |
Note that a text file "progress.txt" is written (and updated) to dir
informing you about the progress of the simulation.
Function does not directly return anything but saves an external
RData file to dir
. This object is a list containing the generated
parameters in sim-results$param.sum$gen
, fitted parameters and other model fit
information in sim-results$param.sum$foo
, as well as a summary of the setup.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 | ## Not run:
recovery_irtree(rrr = 1:2, N = 20, J = 10, genModel = "ext", fitModel = "ext",
fitMethod = "stan", M = 200, n.chains = 2, warmup = 200,
dir = "~/")
# run multiple simulations in parallel using the 'parallel' package
no_cores <- parallel::detectCores() - 1
cl <- parallel::makeCluster(no_cores)
parallel::clusterApplyLB(cl, x = 11:13, fun = recovery_irtree, cores = 1,
N = 20, J = 10, genModel = "ext", fitModel = "ext",
fitMethod = "stan", M = 200, n.chains = 2, warmup = 200,
dir = "~/")
parallel::stopCluster(cl = cl)
## End(Not run)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.