A Measure of Similarity between Separate Categorizations of To-be-Remembered Words
Paul N. Russell · Perceptual and Motor Skills · 1974
The relationship between the recall and organization of to-be-remembered words has been often examined. The potential organization of lists of words to be remembered has generally been defined in terms of norms of category membership (Battig & Montague, 1969) or determined by E. Mandler (1967) saw the need to have a subjective measure, as distinct from one determined by E and normative organization. He used a task where each word to be remembered was printed on a card. Ss sorted the cards into piles which represented subjective categories. They then attempted recall of all words sorted. Two questions arise. How similar are subjective and normative (or E-determined) modes of organization? Secondly, to what extent does S use the same organization on successive sorts? Of relevance is the measure, C, outlined below. Consider a pair of words which are members of a to-be-sorted list. On cwo separare occasions (such as two sorts, or normative organizacion and one sort by S) these two words can either ( a ) share a common category (pile) on both occasions, ( b ) not share a common category (pile) on either occasion, or (c) share a common category (pile) on one occasion but be placed in different categories on the other. Two sorts or modes of categorization are in agreement in ( a ) and ( 6 ) but not ( c ) . The measure C merely counts the total number of pairs where categorization on two separate occasions is in agreement. With a list of R words to be sorted, R ( R 1) /2 different pairs of words exist. Maximum C is therefore R ( R 1 ) /2. Larger C denotes greater agreement or similarity between the two modes of categorization of the same words. The value of C expected when two categorizations are independent and randomly determined and its relation to the number of categories (piles) or number per category has not been worked out.