Psychometric Validation of the ``General Attitudes toward Artificial Intelligence Scale (GAAIS)'' in Costa Rican Engineering Students: Evidence of Reliability and Validity Validación psicométrica de la ``Escala General de Actitudes hacia la Inteligencia Artificial (GAAIS)'' en estudiantes costarricenses de ingeniería.

Abstract

This study examines the reliability and structural validity of the General Attitudes toward Artificial Intelligence Scale (GAAIS) in a sample of engineering students from Costa Rica. Two versions of the instrument were evaluated: the original 32-item version and a shortened 20-item version, increasingly used in recent literature. Analyses included Cronbach’s alpha and McDonald’s omega coefficients, sampling adequacy (KMO index and Bartlett’s test of sphericity), and an exploratory factor analysis complemented by Horn’s parallel analysis. Results showed satisfactory internal consistency in both versions and a factorial structure consistent with the positive–negative bidimensional organization. The parallel analysis suggested the possible existence of additional subdimensions; however, the bidimensional model proved more parsimonious and theoretically coherent. Overall, the findings support the use of both the complete and the abbreviated scale, highlighting that the former provides broader conceptual coverage, while the latter constitutes an efficient and parsimonious alternative for specific research contexts.