Publisher's Synopsis
MULTILOG 6 employs item response theory (IRT) to perform analysis and test scoring for multiple category items. It provides item parameter estimation and subject scoring under four different multiple category logistic models:
* the SAMEJIMA logistic model for graded responses;
* the BOCK multinomial logit model for multiple nominal categories;
* the BOCK-SAMEJIMA-THISSEN model for multiple-choice items with guessing; and
* Master's partial-credit model.
Each model may be fit to a latent ability continuum by marginal maximum likelihood, or to a manifest ability criterion by maximum likelihood.
MULTILOG 6 is able to impose equality constraints on selected subsets of item parameters and permits continuous (or quasi-continuous) measured variables to be mixed with categorical item responses. Truly unique, MULTILOG 6 differs from other programs in that it can estimate a continuous latent trait from both qualitative data and fallible quantitative measurements.
MULTILOG 6's minimum hardware requirements: an IBM-compatible personal computer with 384K RAM, and two disk drives (any combination of floppy and hard). The use of a math coprocessor is recommended, but not required.
