EXAMINING THE QUALITY OF ENGLISH TEST ITEMS USING PSYCHOMETRIC AND LINGUISTIC CHARACTERISTICS AMONG GRADE SIX PUPILS
Examining the Quality of Test Items Using Psychometric Properties in Classical Test Theory
AbstractPurpose - This study examined the quality of English test items using psychometric and linguistic characteristics among Grade Six pupils.
Method - Contrary to the conventional approach of relying only on statistics when investigating item quality, this study adopted a mixed-method approach by employing psychometric analysis and cognitive interviews. The former was conducted on 30 Grade Six pupils, with each item representing a different construct commonly found in English test papers. Qualitative input was obtained through cognitive interviews with five Grade Six pupils and expert judgements from three teachers.
Findings - None of the items were found to be too easy or difficult, and all items had positive discrimination indices. The item on idioms was most ideal in terms of difficulty and discrimination. Difficult items were found to be vocabulary-based. Surprisingly, the higher-order-thinking subjective items proved to be excellent in difficulty, although improvements could be made on their ability to discriminate. The qualitative expert judgements agreed with the quantitative psychometric analysis. Certain results from the item analysis, however, contradicted past findings that items with the ideal item difficulty value between 0.4 and 0.6 would have equally ideal item discrimination index.
Significance -The findings of the study can serve as a reminder on the significance of using Classical Test Theory, a non-complex psychometric approach in assisting classroom teacher practitioners during the meticulous process of test design and ensuring test item quality.
This work is licensed under a Creative Commons Attribution 4.0 International License.
The Malaysian Journal of Learning and Instruction (MJLI) has taken all reasonable measures to ensure that material contained in this website is the original work of the author(s). However, the Journal gives no warranty and accepts no responsibility for the accuracy or the completeness of the material; no reliance should be made by any user on the material. The user should check with the authors for confirmation.
Articles published in the Malaysian Journal of Learning and Instruction (MJLI) do not represent the views held by the editors and members of the editorial board. Authors are responsible for all aspects of their articles except the editorial screen design.