Skip to contents

Get simulation based cutoff values for Yen's Q3 residual correlations

Source: R/easyRasch.R
RIgetResidCor.Rd

Based on Christensen et al. (2017, DOI: 10.1177/0146621616677520).

Usage

RIgetResidCor(data, iterations, cpu = 4)

Arguments

data

Dataframe with response data

iterations

Number of simulation iterations (needed)

cpu

Number of CPU cores to use

Details

Uses a dataframe with response data to simulate residual correlation values across n simulations based on estimated item & person locations.

Results include mean, max and difference between the mean and max for each iteration. Also, 95th and 99th percentile values are reported, and the latter is recommended for use with RIresidcorr() as cutoff value, since the max value seems spurious and reliant on number of iterations.

Uses multi-core processing. To find how many cores you have on your computer, use parallel::detectCores(). Remember to keep 1-2 cores free.