Enoch O. Oladunmoye, PhD

The Two-Parameter Logistic IRT Model: Understanding Item Discrimination, Difficulty, and Differential Item Functioning

The two-parameter logistic model (2PL) occupies a central position within Item Response Theory (IRT) as the standard framework for analysing dichotomously scored items whose discriminating power is permitted to vary from item to item. Unlike the one-parameter logistic model and the Rasch model, which constrain every item to a common discrimination value, the 2PL model estimates a discrimination parameter and a difficulty (or location) parameter for each item individually. This additional flexibility allows the model to capture genuine differences in how sharply individual items distinguish among respondents situated at different points along a latent trait continuum. This research note provides a comprehensive treatment of the 2PL model, covering its mathematical formulation, the interpretation of item difficulty and discrimination, the item characteristic curve, item and test information functions, person-parameter estimation, model assumptions, identification and scaling, model-data fit, Differential Item Functioning (DIF), applications to scale development, and practical implementation within the PsychtrixWeb platform. Particular attention is given to the distinction between statistical discrimination and substantive construct relevance, since a high discrimination value is often mistaken for evidence of validity, and to the relationship between the 2PL and Rasch models and its consequences for invariant measurement. The note closes with a proposed PsychtrixWeb 2PL workflow that integrates item calibration, information functions, DIF analysis, visual diagnostics, and publication-ready reporting, with the aim of moving researchers from simple questionnaire scoring towards genuinely information-based psychological measurement.