Resumo

CHOCOTEA POCA, Omar  e  VILLA CABERO, Dafne Maritza. Modelo de ojiva normal de dos parámetros: una alternativa para el análisis de instrumentos de medición. Revista Varianza [online]. 2019, n.16, pp. 10-19. ISSN 9876-6789.

Abstract Models of item response theory, IRT (Item Response Theory, in English) are more efficient compared to the Classical Theory Test, TC or TCT, for the analysis of the technical quality of measuring instruments alternative. Compared with CT, TRI allows more information ontest items and on the construct or latent variable of interest measured in examinees; however, there are still unresolved problems estimate, causing sometimes impossible to adjust certain models in specific datasets. This article focuses on presenting the theory model warhead normal two parameters, where the parameters and the latent trait also play an important role. Given the advantages of the model, the problem lies in choosing the best method of parameter estimation, being a nonlinear model, you must use iterative simulation processes. The parameter estimates were obtained by the method of Monte Carlo simulation via Markov Chain (Markov Chain Monte Carlo: MCMC) in WinBUGS, where the corresponding runs were made in Brugs, which is a library of R. Once obtained values of the parameter estimates can be obtained probabilities item response given student latent trait, the respective Item Characteristic Curves, Test Information Function and Standard Error Information. For this application the examination of the summer course 2005 was used

Palavras-chave : Measurement; Psychometrics; Item response theory; Item analysis; Model estimation.