Abstract
Empirical relationhips between classical test theory and item response theory statistics. The purpose of this paper was examine empirically: (1) the relationships between the item and person statistics from Item Response Theory (IRT) and Classical Test Theory (CTT) and, (2) the invariance of parameters in both measurement frameworks. In order to achieve these objectives, a simulation study was carry out where 30 samples with 500 subjects and 50 items were generated from the 3-PL model; 10 of these samples with low level of ability, 10 with medium and 10 with high level of ability. The items were calibrated by the 1-, 2-, and 3-PL models. Likewise, the person and item statistics in CTT was calculated in each sample. The findings indicate that the relationships between the statistics from both, IRT and CTT, were very strongs, with very high correlation indices, except in the discrimination parameters. The degree of invariance of parameters across the samples wasn't very high in both measurement frameworks.
