This study analyzes a nationwide English placement test in the Rasch framework Data were obtained from 297 first-year junior high school students. The assessed person (.75) and item reliability (.95) indicate that the test was fairly consistent and reproducible with other samples of examinees. About 60% of total variation was explained under the assumption of unidimensionality. The person separation index suggests that two and a half ability levels can be differentiated by means of the test. Overall, the ability distribution of students was higher than the difficulty distribution of items. The fìt statistics identified a few misfìt items, but their impact on the utility of the test appears nonsignificant.

I. INTRODUCTION
II. LITERATURE REVIEW
III. METHOD
IV. RESULTS AND DISCUSSION
V. DISCUSSION OF MISFITTING ITEMS
VI. CONCLUSION

• Munhong Choe(University of Hawaii)