S. Friel and A. H. Johnstone University of Glasgow Glasgow, G128QQ United Kingdom
Scoring Systems which for Partial Knowledge
It is clear that when a n examinee fails t o select the correct option in a multiple-choice test, his rejection of the correct ootion a n d his choice of an incorrect ontion have a sienificance Ghich should h e recognized i n t h e m k k i n g of a test. Failure t o d o t h i s has been sineled o u t a s a weakness of t h e multinlechoice item. ~ t t l t i p l e - r h o i c itemscan e probably h e m a d e mure el'frctive ii they c a n h r modifit:d t o give - t h e examinee c r r d i t for partial knowledge. The related problems of euessina a n d partial knowledee have stimulated quite a lot ofconsid&ation.hy examiners w i o a r e dissatisfied b y t h e limited a m o u n t s of information conveyed b y responses t o multiple choice items. T w o m e t h o d s have been proposed a s ways of assessing partial knowledge: (1) Differential weighting of response alternatives, a n d (2) Confidence testing. 1)
The effects of assigning different weights to different options on multiple choice items have been investigated both theoretically and empirically by Davis and Fifer ( I ) and Aitken (2). One result of allowing partial credit for options which are not absolutely correct is an increase in the total score variance and, consequently, test reliability. However, Davis and Fifer ( 1 ) have found that this procedure does not appear to alter the test validity significantly. Willey (3) claims several advantages for a special kind of five-option item which he labelled as a "three-decision multiple choice item." The examinee's task on this type of item is t o indicate which one of the five options is definitely correct and which two options are definitely incorrect. Effectively, the three-decision item requires the examinee to sort five options into three categories, an item being scored 3,2, or 0 marks depending on the category into which the examinee places the correct response. Three marks are given to an answer if the option designated as the correct one is in fact the correct one. Two marks are awarded if the correct answer is not put into the "definitely correct" or "definitely wrong" categories. No marks are given if the correct answer is placed in the "definitely wrong"category. Willey claimed that this method appeared to favor the conscientious examinee over the one who was superficial and impulsive. An added benefit was that standard multiple choice items written according to accepted principles of item construction were used.
distribution of oreferenee for ontions as wellas their certaintv
scoring procedures or use intrinsic items. Intrinsic items require a distribution of belief over the options on a multiple choice test and do not have unique correct responses. Intrinsic items reflect, realistically, situations which require choices among a finite number of responses, none of which is uniauelv correct. Such situations are common and characterize (he kunddaries of fields of knowledee. and cnndi-
to make use of his knowledge of his uncertainty, and it might be argued that, a t times, a knowledge of our uncertainty may have considerable value. At least a modified testing procedure might teach a student t o "play a different kind of game" to that which he has been learning to play under the usual multiple choice test situation. He could have agreater choice in expressing how he feels ahout each item. He could be penalized more for making a dogmatic choice than honestly assessing his uncertainty. Intrinsic items seem appropriate in exploring strategies of students confronted with problematic situations or incomplete information. They require that attention be given to the aereement of criterion .. erauo resoonses. thus almost forcine test r;;n%trurturiints, the qurstim