Ph. Harasym et al., NEGATION IN STEMS OF SINGLE-RESPONSE MULTIPLE-CHOICE ITEMS - AN OVERESTIMATION OF STUDENT ABILITY, Evaluation & the health professions, 16(3), 1993, pp. 342-357
There is accumulating evidence that negation (eg., not, except) should
be used sparingly or avoided in the stems of multiple-choice items. I
n these items, examinees are required to select the one alternative th
at is false from among those that are true. The same items can also be
given as positively worded multiple-response items in which the stude
nt is directed to identify those alternatives that are true. The prese
nt study compared the responses of examinees on single-response, negat
ively worded (SRNW) items and on multiple-response, positively worded
(MRPW) items. The experimental design allowed students to respond to b
oth SRNW and MRPW formats that employed identical item alternatives. T
he findings of this study provide further evidence that the use of neg
ation in stems should be limited and that the SRNW item should often b
e converted to an MRPW item. This recommendation is based on the inher
ent cuing (i.e., the artificial elevation of student scores) observed
within SRNW items that significantly lowers item reliability and great
ly distorts information about examinee performance.