Item Response Theory Overview

Principles

Definition

Item response theory is a family of statistical models used primarily in psychometrics, educational measurement, and related fields. It investigates the probabilistic relationship between an individual's unobservable latent trait level (such as ability, attitude, or health status) and their observed responses to individual items on a test, questionnaire, or scale. Characterized by separately modeling parameters for both the items (e.g., difficulty, discrimination) and the persons (e.g., ability), its significance lies in providing a framework for test development, scoring, and analysis that offers advantages over classical test theory, including the potential for invariant item and person measures under model assumptions and the ability to estimate latent traits from incomplete response patterns.

Ontological type

Core Models

Test Development Applications

Parameter Estimation

Key developments

Key Figures

Foundations and Estimation era

Ronald K. Hambleton [1] is a leading figure in the Foundations and Estimation era, with affiliations at Columbia University [2] and the American College of Financial Services [3]. His 1992 work Fundamentals of item response theory [4] articulated the fundamentals of IRT and its estimation framework, contributing to the standardization of latent-trait modeling and the basis for model-based scoring that underpinned adaptive testing in this era. Through this publication, Hambleton's emphasis on separating item parameters from person abilities and on rigorous estimation laid the groundwork for the conditional and marginal likelihood approaches that defined the era's methodological core [4]. As a consequence, the era established consistent rules for test construction and comparability, with Hambleton's contributions shaping practice across institutions such as Columbia University [2] and the American College of Financial Services [3], underpinned by Fundamentals of item response theory [4].

Fairness and Accessibility era

R. Chris Fraley [1] operated within the Fairness and Accessibility era, with affiliations at the University of Illinois Urbana-Champaign [2] and the University of California, Davis [3]. His key contribution in this era is the item response theory analysis of self-report measures of adult attachment [4], demonstrating how IRT can reveal item-level properties and potential bias in attachment assessment, thereby supporting fairness and reproducible measurement. This work by Fraley [1] anchored the era's emphasis on differential item functioning and measurement invariance, illustrating how IRT-based diagnostics can flag bias in self-report instruments within applied psychology. Thus Fraley [1] helped establish item-level evaluation as a standard workflow component in measurement practice during 1994-2007.

Open Bayesian Multidimensional IRT era

Paul-Christian Bürkner [1] is a leading figure in Open Bayesian Multidimensional IRT (2008-2023), with affiliations at Heidelberg University [2] and Ghent University [3] during this era. His work, particularly the paper Bayesian Item Response Modeling in R with brms and Stan [4], advances practical Bayesian IRT modeling by providing an accessible, open-source workflow for fitting multidimensional item response models, enabling coherent uncertainty quantification and reproducible analyses that are central to this era.