NotesFAQContact Us
Search Tips
ERIC Number: ED263177
Record Type: Non-Journal
Publication Date: 1985-Jul
Pages: 14
Abstractor: N/A
Reference Count: N/A
Predicting Performance in Variants of the Multiple-Choice Test.
Hutchinson, T. P.
For over 50 years, the overwhelming weight of evidence has been that subjects are able to make use of partial information when responding to multiple-choice items. The subject chooses the alternative which has given rise to the lowest mismatch, except that if this minimum mismatch is larger than some threshold, the question is left unanswered. Assuming some statistical distribution of mismatch, we may obtain the proportions of items answered correctly and answered wrongly, in terms of ability and willingness to guess. This study reanalyzed two datasets. One was an answer-until-correct test of spatial reasoning (386 examinees, 30 items, 5 alternatives). Evidence for the operation of partial knowledge was given by two findings: performance when second and subsequent choices are made (after the first choice is wrong) is above the chance level, and is positively related to first-choice performance. The second dataset was a four-alternative test of chemistry administered to 407 subjects. There were 20 genuine items plus four nonsense items. The proportions of the genuine items answered correctly, answered incorrectly, and omitted can be used to predict the proportion of the nonsense items attempted. Fairly good agreement between predicted and observed proportions was found. (Author/PN)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Audience: Researchers
Language: English
Sponsor: N/A
Authoring Institution: N/A