The Comparative Study of IRT and CTT in the Assessment of English Standardized Examination
Chen Jin
Abstract
Chen Jin
Abstract
The assessment of examination quality is of great importance to policy making in teaching administration.With the emergence of large-scale standardized examinations,more importance and stringent requirements have been placed on fair examination and scientific critical evaluation of examinations.The present paper addresses the basic principle and computational methods of IRT theory,makes an empirical comparison between the IRT theory and CTT and contrasts between the test items parameter estimation and the students' ability estimation on the basis of the data collected in the standardized examination at GUCAS.It is found that the test item parameter estimation of IRT is independent of the quality changes of the sample,showing better stability over CTT;at the same time,the scoring system of the IRT theory takes into full consideration the functional difference among the test items,test scores being more accurate with fewer identical scores.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
The assessment of examination quality is of great importance to policy making in teaching administration.With the emergence of large-scale standardized examinations,more importance and stringent requirements have been placed on fair examination and scientific critical evaluation of examinations.The present paper addresses the basic principle and computational methods of IRT theory,makes an empirical comparison between the IRT theory and CTT and contrasts between the test items parameter estimation and the students' ability estimation on the basis of the data collected in the standardized examination at GUCAS.It is found that the test item parameter estimation of IRT is independent of the quality changes of the sample,showing better stability over CTT;at the same time,the scoring system of the IRT theory takes into full consideration the functional difference among the test items,test scores being more accurate with fewer identical scores.
Key concepts: Item response theory, Classical test theory, Test (biology), Estimation, Quality (philosophy), Econometrics, Computer science, Sample (material)