Psychometric Toolbox


Function Item Response Theory-Basic analysis -> item calibration and test scoring based on the one and two parameter logistic models.

The IRT-B function performs item calibration and individual scoring on the basis of the one-parameter and the two-parameter logistic models. Unlike the remaining CTT macros, we recommend that IRT-B be used only for teaching purposes. For clinical assessment or for research a more sophisticated IRT program such as BILOG or MULTILOG is recommended. As a teaching tool, however, IRT-B has interesting features and provides useful information.

The IRT-B function only contains one computing button.

The output is organized in two parts, which generally coincide with the calibration and scoring stages. The first part first provides the CTT-based item analysis results (including the alpha reliability estimate). Next, the IRT item estimates: discriminations (slopes), and locations are displayed. This first part also provides: (a) the test information values, (b) the descriptive statistics of the raw and IRT scores, and (c) a mean-squared residual statistic for each item in order to assess the appropriateness of the model chosen. Two graphs are available in this part: the item characteristic curve (for the selected items), and the test information function curve.

The second part of the output provides: (a) the raw scores and (b) the individual trait estimates for each respondent together with the corresponding standard errors obtained using Bayes EAP estimation. The test characteristic curve, which is the graph that relates both types of scores, is also included in this part of the output.

Download example data file: