jml {MiscPsycho} | R Documentation |
Estimates the Rasch model using joint maximum likelihood (JML)
jml(dat, con = 0.001, bias = FALSE)
dat |
A data frame or matrix containing item responses |
con |
Convergence criterion |
bias |
Correction for JML bias |
JML estimation only has good asymptotic properties. Marginal Maximum Likelihood (MML) is a better technology for parameter estimation. However, this option is provided for users of Winsteps who may want to explore their data in R prior to working in Winsteps. JML is known to return biased parameter estimates. If bias = TRUE, then a correction for bias is made using (L-1)/L * b where L is the total number of test items and b is the original item parameter
jml returns item parameter estimates, their asymptotic standard errors, the sample size per item, and the Infit and Outfit statistics found in Winsteps.
Harold C. Doran
Best Test Design. Wright and Stone (1979)
ltm