Statistical analysis of data

The results of the comparison within matched pairs are shova in Table 2. I n these comparisons scores for a few adult subjects are missing in most tests (cf. above). The actual number of pairs from which the data hare been computed varies from 67 to ‘76 (for tests 21-22: 55 pairs). Among the children test data for one member of the matched pair are missing in n number of cases, leaving a total of 43 pairs. A negative difference always means that the test performance of P was inferior to that of C. For example: the time score of P in variable 15 (M=S4.87) is higher than that of C (M= ‘i6.S3), but since this means that the performance of P is inferior, the difference is entered with a negative sign. The estimate of significance, however, is based on a two-tail test and accordingly indicates the probability of obtaining, by chance alone, a difference as large as or larger than the one at hand, regardless of direction. The last column of Table 2 has been added in order to indicate the order of magnitude of a difference, as indicated by its absolute value. It is obtained by dividing the mean difference between groups by the S. D. of the Control group. This S .D. has been chosen, some\J-hat arbitrarily, in order to standardize the differences and make them comparable regardless of number of items in the test, the scoring scale used etc. There is a rather general trend in this table. The vast majority of differences are negative. On the whole, group P does not perform quite as well as group C. Four of the differences are significant a t the 0.03 level: those in tests 1, 4, S and 1s. A common feature of these tests is that speed plays a decisive role in performance. On the other hand, the differences are small or moderate. Even the largest difference (test 1) is moderate-it amounts to 40 70 of the S. D.