Psychometric properties of the Frontal Assessment Battery (FAB) and its reduced version FAB-15

This study evaluated the psychometric properties of the Frontal Assessment Battery (FAB) and its abbreviated version (FAB-15) in a sample of 1,071 Chilean adults. The participants comprised both a general population group and a clinical group diagnosed with major neurocognitive disorder. By utilizing confirmatory factor analysis and item response theory, the research assessed the structural validity and item characteristics of the tool. The results demonstrated that the FAB-15, which excludes the "Prehension behavior" item, exhibited a superior statistical fit and a more stable unidimensional structure compared to the original battery. The analysis indicated that the omitted item contributed minimally to the scale's discriminative capacity. Furthermore, IRT analysis identified significant differential item functioning (DIF), indicating that the test performance is not measurement invariant. The scores were found to be influenced by sociodemographic variables, specifically educational level and age. Notably, the "Similarities" item displayed a marked nonuniform bias related to education, and significant nonuniform DIF was also detected for age. In contrast, "Lexical Fluency" was identified as the item with the highest discriminative power for executive function. Although the instrument showed adequate reliability, the presence of these biases necessitates caution in interpreting raw scores. To address this, the study developed normative data stratified by age and education. These norms are intended to correct for the identified metric biases, allowing for a more accurate and objective assessment of executive functions in clinical settings.