IRTPRO

What is IRTPRO?

THE IRTPRO (Item Response Theory for Patient-Reported Results) is an entirely new application for item calibration and test scores using IRT.

Item response theory (IRT) models for which item calibration and score are implemented in IRTPRO are based on one-dimensional and multidimensional versions [confirmatory factor analysis (CFA) or exploratory factor analysis (EFA)] of the following widely used response functions:

• Two-parameter logistics (2PL) with which equality constraints include one-parameter logistics (1PL)
• Three-parameter logistics (3PL)
• Graded 
• Generalized partial credit
• Nominal 

These item response models can be combined into any combination in a test or scale, and any user-specified equality constraints between parameters or fixed values for parameters can be specified.

IRTPRO implements the Maximum Likelihood (ML) method for item parameter estimation (item calibration) or calculates Maximum a posteriori (MAP) estimates if previous (optional) distributions are specified for the item parameters. That said, alternative computational methods can be used, each of which offers better performance for some combinations of dimensionality and model structure

The calculation of irt scale scores in IRTPRO can be done using any of the following methods:

• Maximum a posteriori (PAM) for response patterns
• Expected a posteriori (PAE) for response patterns
• A posteriori (EAP) is expected for the scores summed up

Data structures in IRTPRO can categorize respondents items into groups, and latent population variables and covariance and variance matrices can be estimated for multiple groups. Most often, if there is only one group, the variable (s) mean(s) and variance(s) of the latent population are fixed (usually at 0 and 1) to specify the scale; for multiple groups, a group is usually referred to as a "reference group" with standardized latent values.]

To detect the differentiated functioning of items (DIF), IRTPRO uses Wald tests, modeled after a Lord proposal, but with precise item error-covariance matrices calculated using the supplemented EM algorithm (SEM).

Depending on the number of items, response categories, and respondents, IRTPRO reports several varieties of adjustment and diagnostic statistics after item calibration. The probability values of -2 log, Information Criterion and Bayesian Information Criterion (BIC) are always reported. If the sample size sufficiently exceeds the number of cells in the complete cross-classification of respondents based on item response patterns, the general likelihood ratio test in relation to the general multinominal alternative will be reported.

System Requirements

Operating system: Windows 7, 8, 10

Min. CPU: Processor 486 or higher

Minus. RAM: 16 MB the RAM

Disk space: 10 MB