RELIABILITY ESTIMATE IN THE MEASUREMENT OF TWO ITEMS: ANGOFF-FELDT COEFFICIENT

• Autores: Domínguez Lara, Sergio Alexis, Merino Soto, César Ayax, Navarro Loli, Jhonatan S.
• Formato: artículo
• Fecha de Publicación: 2016
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: Revistas - Universidad Peruana de Ciencias Aplicadas
• Lenguaje: español; inglés
• OAI Identifier: oai:ojs.revistas.upc.edu.pe:article/463
• Enlace del recurso: https://revistas.upc.edu.pe/index.php/docencia/article/view/463
• Nivel de acceso: acceso abierto
• Materia: confiabilidad; Angoff-Feldt; medidas congenéricas; reliability; congeneric measures; confiabilidade

Nowadays, the use of brief measurements in psychological evaluations is extending rapidly, but many times the reliability analysis of the scores is limited by the conditions which must be overcome to use certain coefficients (e.g. α coefficient). The compliance of the tau equivalent measurement model is one of the most important the α coefficient demands, and as it is not achieved, the estimate may be biased. The aim of this paper is to present the Angoff-Feldt (rAF) coefficient for reliability estimate, to be used in congeneric measures, that is, those that do not comply with the demands of the α coefficient (e.g. tau equivalent model). Although rAF was initially thought for two congeneric halves, this paper presents the application for the measurement of two items, assuming each is a half. An example of its usefulness is provided and the possibility of implementing its use discussed.