Measure of Similarity between Recall Attempts in Multitrial Free Recall

Abstract

Of relevance is P, a measure of the extent to which the number of items common to two recalls, exceeds the number expected if all items stood an equal chance of recall on both attempts, as a proportion of the maximum possible deviance from the expected number common to two recalls. It is -1.0 when the minimum number of items is shared, 0.0 when the expected number is common to both recalls, and 1.0 when the maximum number of items is shared. It is analogous to the adjusted ratio of clustering of Roenker, Thompson, and Brown (1971). The derivation of P is as follows: R, the number of items in the list to be remembered r., the number of items recalled on the nth trial ?+I, the number of items recalled on the n + lth trial x, the number of items common to recall trials n and lz + 1 If the probability of recall of items to be remembered is equal for all R items on every trial, the expected number of items common ro two recall trials, E(x), is given by