Acquisition phase. As four out of the five participants in the control group reached perfect performance by the third trial (criterion), the fourth and fifth trials were not administered, and the repeated measures comparison was carried out on only the first three trials for both groups. The assumption of homogeneity of variance-covariance matrices was not tested , however, according to Tabachnick & Fidell (1996) the assumption does not need to be assessed when samples are of equal size or not notably discrepant (Tabachnick & Fidell, 1996). The patient and control groups differed in the number of object-place associations they were able to recall over three trials F(1, 8) = 9.04, p =.02, but there was no group by trial interaction F(1, 8) = .351, p>.05, or a within-subjects trial effect F(1, 8) = 1.73, p>.05. The overall group difference accounted for 53% of the variance in the task. Figure 2 shows mean improvement across the first three trials for the two groups, and Figure 3 shows individual tra jectories for participants in both groups. As can bee seen, the control group showed a clear learning effect, with four out of the five subjects reaching asymptote by the second trial. By contrast, two patients achieved perfect performance by the third and fourth trials, while the remaining three did not reach this level by the fifth trial (Figure 3).
Delayed recall trial. Patients recalled fewer associations (M = 1.7; S.D =1.14) than did the controls (M = 4.6; S.D = .89), t(8) = 4.93, p=.001. Given the sample size, Cohen's d measure of effect size was calculated and was found to be d = 2.83, representing a very large effect (Cohen, 1992). Whereas one subject in the control group could remember only three items, the rest of the subjects remembered all five items. Conversely, among the patients, only one subject remembered as many as three items, with the others remembering between zero and two items.