the present study investigated construct equivalence of multiple choice (mc) and constructed response (cr) item types across stem and content equivalent mc and cr items (item type ‘a’), non-stem-equivalent but content equivalent mc and cr items (item type ‘b’), and non-stem and non-content equivalent mc and cr items (item type ‘c’). one hundred seventy english-major undergraduates completed mc and cr forms of a reading comprehension test. the effect of different item formats on test takers’ performance was measured using multifaceted rasch measurement. the conclusion was that cognitive demands of mc items were higher than cr items. the findings of this study showed important pedagogical implications for test constructors as to the fact that the testing method can make a difference in tests of reading comprehension.