Does the position of correct answers in multiple-choice items influence item difficulty?

Due to economical reasons, the multiple-choice response format is often the preferred response format for test construction, especially in large-scale-assessments which claim for quick processing on the examinees side and automated scoring on test makers side. As a disadvantage, research on effects of answer-key balancing and various “response sets” indicates an unfavourable influence of the position of correct answers on item difficulty. To examine this effect, two different multiple-choice response formats (“1 out of 6” and “2 out of 5”) were administered in 3 versions each (correct answers at extreme position, middle position or random) in the course of a large-scale competence assessment in reading comprehension. Furthermore, to investigate age-dependent effects this design was applied to 4th and 8th grade students and finally analyzed using the LLTM (Fischer, 1973). Results give useful hints for further test developments, especially in the field of Large-Scale-Assessments.