Key points are not available for this paper at this time.
Language reading examinations in French and Spanish were administered to students in order to compare the behavior of “natural” four‐choice items with “natural” five‐choice items rescored as four‐choice items after removing the least popular incorrect alternative. No significant differences in the regression systems of these items were found. However, “natural” four‐choice items were significantly less reliable than “natural” five‐choice items.
Ramos et al. (1973) studied this question.