Evaluation of computerized tests applying the classical theory of the tests and the theory of response to the item

• Autores: Menacho Chiok, César Higinio, Alva Trinidad, Jesús María Cano
• Formato: artículo
• Fecha de Publicación: 2020
• Institución: Universidad Nacional Agraria La Molina
• Repositorio: Revistas - Universidad Nacional Agraria La Molina
• Lenguaje: español
• OAI Identifier: oai:revistas.lamolina.edu.pe:article/1638
• Enlace del recurso: https://revistas.lamolina.edu.pe/index.php/acu/article/view/1638
• Nivel de acceso: acceso abierto
• Materia: Pruebas informatizadas; teoría clásica de los test; teoría de respuesta al ítem; modelos logísticos binarios; calibración de la prueba.; Computerized tests; classic test theory; item response theory; binary logistic models; test calibration.

The objective of this study was to evaluate the reliability and validity of computerized tests via the web through the measurement of their psychometric and statistical properties by applying the Classical Test Theory (TCT) and the Item Response Theory (TRI). The TCT methodology was applied to assess the difficulty and discrimination of the test and the items. The data was adjusted to the TRI binary logistic models of one, two and three parameters. A computerized test of 30 questions was applied to 775 students enrolled in the Basic Statistics course in the 2016 II semester. The results indicated a good reliability of the test with a Cronbach’s alpha of 0,833 and was corroborated with a correlation of 0,815. For the TCT the difficulty index identified three very easy questions (V7, V8 and V12) and the discrimination index did not find any questions to withdraw it. The assumption of unidimensionality with factor analysis was tested with an explained variance of the first factor of 24,7%. The binary logistic model of the three parameter TRI (3PL) was better adjusted to the data. For the calibration process with the 3PL model, questions V28 (discrimination index greater 0,65) were withdrawn; V8, V12, V16 and V18 (chance index greater than 0.4) and none with the difficulty index.