A noticeable feature in the second presentation of Series II and III was the numerous memory illusions. These occurred with every subject but with some more than others. The women were generally more subject to them than the men. In each case the subject was asked whether he rememberd having made the decision before, and if so whether he decided in the same way. In many cases the subject would vigorously affirm that he actually remembered arranging the cards in the same way before, when the previous arrangement was really quite different. There were also some fewer cases when the subject remembered that his decision was different, when it was really the same. CONFIDENCE Three of the subjects—J, K, and L-were asked to state the degree of their confidence in each decision. This was reported by the subjects as meaning either "The subjective feeling of satisfaction in the arrangement," or "The assurance of constancy." The same subject would alternate between these two attitudes from time to time. The subjective feeling is different in each case; but for our purposes, the correlation of constancy and confidence, we assume that they need not be distinguished; for the former always implies the latter, though the opposite is not always true. The degree of confidence was given as A, B, or C,A being perfect confidence, C very little or none, and B medium. It is clear that the subjects might differ not only in the degree of their confidence but also in their way of describing it. The same degree of confidence might be described as A by one and as B or C by another. This, however, is a criticism that can be raised as well against the introspective method as a whole. The results of this confidence test are shown in Tables VI-A and VI-B. The subjects are compared with one another; and the confidence of the decisions that were changed with that of the whole. The "numerical value" is obtained rather arbitrarily by valuing A at 2, B at 1, and C at o; but perhaps it is not so arbitrary after all, for A is introspectively perfect confidence (1), B medium (1⁄2), and C none (0); and to avoid fractions 2, 1, and o are used instead. The average confidence is obtained by dividing the numerical value by the number of decisions. The following points are noteworthy: (1) The more confident subjects are also the more constant. (2) The confidence is slightly less with the decisions that were changed. (See Tables VI-A and VI-B under “Average Confidence.") (3) One subject (J) is regularly less confident on the second presentation; the others are equally or more confident. (4) The decisions changed are divided fairly equally among A, B, and C; but the amount of the change is as a rule greater in C decisions than in B's or A's. There are, however, exceptions. (See tables VI-A and VI-B, last section.) Average Amount of Change for A, B, or C Confidence N. V. of Changes Ave. for A, B, C 6A J 7A 1% A IC I В BC ABC ABC P.' and P. First and second presentations. = N. V. and N. V. Numerical Value of confidence in first and second Decisions = No. D. Number of Decisions. Average Amount of Change for A, B, or C Confidence N.V. of Changes Ave. for A, B, C The degree of difficulty of the decisions was also obtained from the same three subjects, but only for the first presentation. The results are given in Tables VII-A and VII-B. D means difficult, M medium, and E easy. An arbitrary "numerical value" is obtained by valuing the difficulty of D at 2, of M at 1, and of E at O; and the average difficulty by dividing the numerical value by the number of decisions. It will be noticed that: (1) The subjects differ considerably in the amount of difficulty, and the least confident subject has the greatest difficulty in the threes, but the least in the fives. (2) The easy as well as the difficult decisions are changed, but the average difficulty for those changed is slightly greater than for the total. Two experiments were carried out to ascertain the association time of the subjects. There were nine subjects, six of whom were the same in each case and three different. In the first a list of fifty words were used. These were repeated to each subject who was directed to reply as quickly as possible with the first word that came to his mind. The time was taken with a stop watch visually; that is, the watch was started and position of the hand noted when the word was given, and when the reply was heard. This method was found difficult and inaccurate; and in the second experiment, which consisted of forty words, the watch was started when the word was given and stopped when the reply was heard. The results were very much the same except in the case of C, whose time in the second experiment is much shorter than in the first. Full results of both experiments are given in Table VIII. The amount of the correlation between association time and decision and constancy time may be brought out by the "Method of Unlike Signs." This method is recommended by Whipple only for preliminary survey and not for "final determinations of important correlations, because the probable error is too large." It seems, however, sufficient for our immediate purpose. The central tendency (median) of the association times of the twelve subjects is obtained, using the times of column 4 except for the subjects not represented therein. In Table VIII-A the subjects are arranged in the same order as in Table VIII, and their deviations above (+) or below (-) the central tendency (1.15) recorded. A similar procedure is followed for the decision time with threes (column 6) and with fives (column 8). By comparing each subject's deviation from the central tendency in decision time for threes with his deviation from the central tendency in association time (see Table VIII-A) it will be found that there are 50% cases of unlike signs; but a similar comparison in the case of decision time for fives gives 66 23% cases of unlike signs. Now if we refer to Whipple's Manual, Table 9, we see that this means a correlation coefficient of o and —.509. The 'G. M. Whipple, Manual of Physical and Mental tests, p. 40. |