<![CDATA[This study aims to evaluate item fit through the application of the Rasch model. The research involved 40 English as a foreign language student who took part in an essay writing class as part of a TOEFL iBT course, with writing samples gathered in partnership with the language center offering the TOEFL iBT classes. The research methodology utilizes the Rasch model with Winsteps software for quantitative analysis, employing three key statistical outputs: a "statistical summary output" to provide overall data and figures, item statistics to assess item validity, and participant statistics to evaluate participant validity. Based on the interpretation of misfit order Rasch measurement for item fit, four items ("Content," "Structure," "Mechanic," and "Diction") demonstrated excellent alignment with the Rasch model. All items exhibited Outfit Mean Square (MNSQ) values below the expected 1.0 yet within the acceptable range (0.5 - 1.5), indicating better-than-expected fit, with "Diction" showing the highest conformity. The Outfit Z-Standard (ZSTD) values for all items were near zero, signifying no significant deviations from the model. High Point Measure Correlations (PT Measure Corr) of 0.91 for "Content" and "Diction" and 0.88 for "Structure" and "Mechanic" suggest strong consistency with the overall measured ability, affirming the significant and valid contributions of these items to the scale's measurement objectives. Thus, all four items are deemed excellent within the applied measurement scale, particularly "Diction," which exhibits the highest model fit. Keywords: Item Fit, Academic Writing, Rasch Measurement]]>
Copyrights © 2024
