# Advancing Pediatric Cognitive Health: Psychometric Evaluation and IRT- and Regression-Based Norms for Two Neuropsychological Measures in Colombian Children and Adolescents

## TL;DR

This study evaluates two neuropsychological tests in Colombian children and adolescents, confirming their reliability and how factors like age and education affect performance.

## Contribution

The study provides new normative data and psychometric validation for two neuropsychological measures in Colombian children using IRT and regression models.

## Key findings

- The Token Test and Rey–Osterrieth Complex Figure showed strong reliability and no sex-related bias.
- Age and parental education significantly influenced test performance.
- Digital tools were integrated to generate normative data based on ability scores.

## Abstract

Objective: To evaluate the psychometric properties of the short version of the Token Test (SVTT) and the Rey–Osterrieth Complex Figure (ROCF) using an item response theory (IRT) framework and to establish normative data for Colombian children and adolescents based on ability scores. Methods: A total of 668 healthy participants aged 6–17 years took part in this study. Factorial structure was assessed through confirmatory factor analysis (CFA). Item parameters were estimated using a two-parameter logistic (2PL) model for the SVTT, which accounts for both item difficulty and discrimination in dichotomous responses, and a graded response model (GRM) for the ROCF, suitable for items scored on ordered categories reflecting increasing levels of performance accuracy and Differential Item Functioning (DIF) analysis was conducted to assess potential bias related to sex. Reliability was examined using the Test Information Function (TIF), internal consistency throughout Cronbach’s alpha, and the influence of sociodemographic variables was analyzed through regression models. Results: CFA confirmed unidimensionality for all measures. For most items, moderate-to-low ability was sufficient to achieve the highest scores in the ROCF, and low ability in the SVTT. DIF analysis indicated no meaningful sex-related bias in any of the subtests. Both tests showed excellent reliability and internal consistency. Copy scores were influenced by polynomial age and parents’ mean years of education (MPE), while both immediate recall in the ROCF and SVTT were affected by MPE and the interaction of logarithmic age. Conclusions: This study provides strong psychometric evidence and, together with the integration of digital tools for generating normative data, represents a meaningful advancement in neuropsychological assessment.