As a step in trying to determine what factors are important causal determinants in producing false recall and false recognition in the DRM paradigm, the authors tested seven possible contributing variables. These variables included three attributes of the critical items (word length, frequency, and concreteness), and four related to the list (FAS, BAS, IIAS, and veridical recall). Results of the simultaneous multiple-regression analysis indicated that BAS, (the strength of the connection from the list items to the critical word), and veridical recall (the average probability of recall for the items in the list) accounted for about 84% of the explainable variance in false recall. They then discussed the implications of their results for various theories postulated to explain the false recall and recognition effects.
Procedure
The authors used information from various norms and data from previous studies as the raw data for this investigation. They obtained data for 55 lists all fitting the characteristics of the DRM paradigm. The variables they used were:
Critical Item Variables
word length Simply the number of letters in each critical item.
concreteness Concreteness ratings for the item from the Paivio, Yuille, & Madigan (1968) norms.
BAS Backward associative strength. The probability that the critical word was elicited when given a list word.
IIAS Interitem associative strength. The average IIAS for all words in the list. An average of the number of words which elicited each other word in the list. (The average of a set of dichotomous variables.)
Results
Two multiple regression analyses were conducted in which the seven independent variables were entered simultaneously to predict either probability of false recall, or probability of false recognition.
The table below includes simple correlations, beta weights, and semipartial correlations between probability of false recall and selected critical item and list variables.
length BAS veridical recall
Pearson r -.37* .73* -.43*
b .07 .70* -.40*
sr .06 .60* -.34*
Other important correlations include nonsignificant correlations between false recall and the remaining variables, because these relate to the theoretical explanations of false recall in the DRM paradigm.
Theories Discussed
|
|
|
|
|
|
|
|