montecarlo: performs Monte Carlo simulations for empirical cutoff thresholds
Description
Performs Monte Carlo simulations using multiple datasets without DIF.
Usage
montecarlo(obj, alpha = 0.01, nr = 100)
Arguments
obj
an object returned from lordif
alpha
desired significance level (e.g., .01)
nr
number of replications
Value
Returns a data frame with the following components:
chi12prob associated with the LR Chi-square test comparing Model 1 vs. 2
chi13prob associated with the LR Chi-square test comparing Model 1 vs. 3
chi23prob associated with the LR Chi-square test comparing Model 2 vs. 3
pseudo12.CoxSnellCox & Snell pseudo R-square change from Model 1 to 2
pseudo13.CoxSnellCox & Snell pseudo R-square change from Model 1 to 3
pseudo23.CoxSnellCox & Snell pseudo R-square change from Model 2 to 3
pseudo12.NagelkerkeNagelkerke pseudo R-square change from Model 1 to 2
pseudo13.NagelkerkeNagelkerke pseudo R-square change from Model 1 to 3
pseudo23.NagelkerkeNagelkerke pseudo R-square change from Model 2 to 3
pseudo12.McFaddenMcFadden pseudo R-square change from Model 1 to 2
pseudo13.McFaddenMcFadden pseudo R-square change from Model 1 to 3
pseudo23.McFaddenMcFadden pseudo R-square change from Model 2 to 3
beta12proportional beta change from Model 1 to 2
alphasignificance level
nrnumber of replications
cutoffthresholds for the statistics
Details
The simulated datasets have the same dimensions as the empirical data.
Group differences (impact) in theta between groups are preserved in simulated datasets.
Returns empirical thresholds for various statistics and effect size measures.
References
Choi, S. W., Gibbons, L. E., & Crane, P. K. (April, 2009). Development of freeware for an iterative hybrid ordinal logistic regression/IRT DIF. Paper Presented at the National Council of Measurement in Education. San Diego, CA.
##run lordif firstage.dif <- lordif(Anxiety[paste("R",1:29,sep="")],Anxiety$age)
##the following takes a long timemc1 <- montecarlo(age.dif,alpha=0.05,nr=500)