This study examined how four different constructed response formats of reading comprehension tests (cloze, recall, summary, and translation), developed based on the two texts with different reading difficulty, exert effects on EFL students’ reading test performance; and also investigated how those effects, if any, are differentiated with respect to the participants’ language proficiency levels. The total of 437 college students (220 male and 217 female students) participated in the current study. The test data was submitted to statistical treatment. The findings were that the four different response formats assessed different aspects of reading comprehension; the two texts with different text difficulty elicited different reading comprehension performance; and there were complicated interactions between the two variables of response format and text difficulty across the participants’ language proficiency levels. It is suggested that classroom teachers have to take it into consideration when planning to develop a reading comprehension test for EFL learners: how to choose reading texts that will be included in the test, considering an appropriate level of text difficulty and test formats to be engaged in order to assess students’ reading comprehension ability more accurately.

Ⅰ. INTRODUCTION
    Ⅱ. ASSESSMENT FORMATS
    Ⅲ. RESEARCH DESIGN
    Ⅳ. RESULTS AND DISCUSSION
    Ⅴ. CONCLUSION
    REFERENCES
    APPENDIX

• Jeong-Won Lee(Chungnam National University)