ANALISIS VALIDITAS BUTIR DAN KETEPATAN TARGET UKUR INSTRUMEN BIOLOGI KELAS XII: APLIKASI PEMODELAN RASCH
Keywords:
Rasch Model, item analysis, misfit items, logit difficulty, person fit, test calibration, biotechnology assessment, measurement validity, educational evaluationAbstract
This study aimed to analyze the quality of a 20-item biotechnology test using the Rasch Model. The item analysis identified three misfitting items (S7, S14, S19) whose Outfit MNSQ values exceeded the model’s acceptable range. The difficulty levels ranged from –2.94 to +5.11 logits, indicating a wide spread of item complexity but with a tendency toward extreme categories, dominated by very easy and very difficult items. At the respondent level, 11 students (22%) showed misfitting response patterns, suggesting potential guessing or inconsistent answering behavior. Although item and person reliability were classified as good, the presence of misfit patterns indicates the need for instrument refinement. Key recommendations include revising the misfitting items, adding more items within the moderate difficulty range to fill gaps at the extremes, and restructuring the test blueprint to achieve a more balanced distribution of difficulty. The findings highlight the importance of Rasch-based calibration to produce more accurate and objective educational assessments.