Publication: Reformulación del coeficiente k para juicio de personas expertas aplicada a las competencias digitales docentes
Loading...
Date
2026
Authors
Elías Aspee, Juan ; Silva Quiroz, Juan ; Matar-Khalil, Shadye ; González Campos, José
item.page.secondaryauthor
item.page.director
Publisher
Universidad de Murcia. Servicio de Publicaciones
publication.page.editor
publication.page.department
DOI
https://doi.org/10.6018/rie.679191
item.page.type
info:eu-repo/semantics/article
Description
Abstract
El juicio de personas expertas es una técnica ampliamente utilizada para validar instrumentos de evaluación, autoevaluación, autopercepción, mediante la evaluación informada de ítems. El coeficiente de competencia experta (K) combina la autoevaluación del conocimiento y la argumentación objetiva, ha sido adoptado extensamente en educación y salud; pero su aplicación como promedio simple y equiponderado presenta limitaciones epistemológicas y estadísticas por su vulnerabilidad al sesgo de autopercepción. Se propone una generalización del coeficiente K que incorpora un tercer componente, referido a la proximidad temática del experto con ponderaciones diferenciadas para cada dimensión. Se integra un modelo inferen-cial basado en la distribución beta y la aproximación normal asintótica, permitiendo estimar intervalos de confianza y realizar pruebas de significación sobre la competencia experta. Se aplicó en dos contextos: un caso simulado y un estudio real de validación de un instrumento sobre competencias digitales en educación parvularia. Los análisis muestran que el coeficiente K tradicional sobrestima la competencia en comparación con la versión generalizada. En el caso real, la muestra inicialmente clasificada como válida bajo el umbral fijo de 0.8 se reduce al aplicar los criterios inferenciales propuestos, aumentando la precisión en la identificación de personas expertas. Una prueba para muestras pareadas confirma diferencias significativas entre ambas metodologías. La reformulación del coeficiente K mejora la rigurosidad, transparencia y validez del juicio experto al considerar tanto criterios objetivos como contextuales y al respaldar las decisiones mediante inferencia estadística. La propuesta ofrece un marco sólido y adaptable para futuros procesos de validación en diversas disciplinas.
Expert judgment is a widely used technique to validate assessment, self-assessment, and self-perception instruments through informed item evaluation. The expert competence coefficient (K) combines self-assessment of knowledge and objective argumentation. It has been widely adopted in education and healthcare; however, its application as a simple, equally weighted average presents epistemological and statistical limitations due to its vulnerability to self-perception bias. A generalization of the K coefficient is proposed that incorporates a third component, referring to the expert’s thematic proximity with differentiated weights for each dimension. An inferential model based on the beta distribution and the asymptotic normal approximation is integrated, allowing for the estimation of confidence intervals and significance tests on expert competence. It was applied in two contexts: a simulated case and a real-life validation study of an instrument on digital competencies in early childhood education. The analyses show that the traditional K coefficient overestimates competence compared to the generalized version. In the real-world case, the sample initially classified as valid under the fixed threshold of 0.8 is reduced by applying the proposed inferential criteria, increasing the accuracy of expert identification. A test for paired samples confirms significant differences between the two methodologies. The reformulation of the K coefficient improves the rigor, transparency, and validity of expert judgment by considering both objective and contextual criteria and by supporting decisions through statistical inference. The proposal offers a robust and adaptable framework for future validation processes in various disciplines.
Expert judgment is a widely used technique to validate assessment, self-assessment, and self-perception instruments through informed item evaluation. The expert competence coefficient (K) combines self-assessment of knowledge and objective argumentation. It has been widely adopted in education and healthcare; however, its application as a simple, equally weighted average presents epistemological and statistical limitations due to its vulnerability to self-perception bias. A generalization of the K coefficient is proposed that incorporates a third component, referring to the expert’s thematic proximity with differentiated weights for each dimension. An inferential model based on the beta distribution and the asymptotic normal approximation is integrated, allowing for the estimation of confidence intervals and significance tests on expert competence. It was applied in two contexts: a simulated case and a real-life validation study of an instrument on digital competencies in early childhood education. The analyses show that the traditional K coefficient overestimates competence compared to the generalized version. In the real-world case, the sample initially classified as valid under the fixed threshold of 0.8 is reduced by applying the proposed inferential criteria, increasing the accuracy of expert identification. A test for paired samples confirms significant differences between the two methodologies. The reformulation of the K coefficient improves the rigor, transparency, and validity of expert judgment by considering both objective and contextual criteria and by supporting decisions through statistical inference. The proposal offers a robust and adaptable framework for future validation processes in various disciplines.
Citation
Revista de Investigación Educativa, RIE (2026), núm. 44, pp. 532-550
item.page.embargo
Ir a Estadísticas
Este ítem está sujeto a una licencia Creative Commons. http://creativecommons.org/licenses/by-nc-nd/4.0/