Item analysis of English paper 1 (EPI) of 2014 UPSR trial examination using Rasch measurement model

This study was aimed to identify the quality of English Paper 1 items of 2014 UPSR trial examination in terms of reliability, validity and items characteristics based on the analysis using Rasch measurement model. It also sought to determine the difficulty levels of 40 multiple-choice items consisting five constructs of vocabulary, language and social expression, grammar, cloze-comprehension and reading comprehension by examining the distribution of items in the item-person map. A number of 525 primary school pupils in Kuala Selangor was selected using proportionate stratified random sampling method. The validity evidences are shown through the Principle Component Analysis (PCA), fit statistics and item distractor analysis. Even though the raw variance explained by measure in PCA analysis is rather weak, it achieves the minimum uniformity and it clearly shows the absent of second dimension in the test. The fit statistics analyses have shown seven misfit items that are beyond the acceptable range (0.7 - 1.3 logit) and two of them have negative PTMEA Corr values. Item distractor analysis has identified five problematic items whereby three of them are also misfit items. Summary statistics of items and persons show the reliability indices of Cronbach’s Alpha are greater than 0.80 and separation indices greater than 2. Itemperson map has demonstrated good distribution pattern where most of the items fall within the range of person ability. It is expected that this study will benefit teachers in improving their assessment practice. This would also help the establishment of the item banks with validated items and enhance teachers’ ability in analysing and interpreting items accurately using modern test theory such as Rasch model. This study contributes to the field of educational measurement and evaluation in terms of expanding the use of modern test theory in language testing in Malaysia.