NOVEMBEE 23, 1906.] 



SCIENCE. 



663 



TABLE IV. MEASTJBEMENTS OF THE ACCUBACT OF JUDGMENT OF TEN OBSEBVEKS. 



validity of the judgments varies to a measured 

 degree, the arrangements made by the indi- 

 viduals could be weighted. I have not under- 

 taken the somewhat tedious calculations neces- 

 saiy; they would not considerably alter the 

 order, but would make it somewhat more exact, 

 at the same time decreasing the probable 

 errors. 



There is here measured for the first time, 

 I think, the accuracy or reliability of judg- 

 ment. This is obviously a complex and im- 

 perfectly analyzed trait, depending on a large 

 number of varying conditions. A man's judg- 

 ment may be good in some directions or from 

 certain points of view, and bad in other ways. 

 Still we understand vaguely what is meant by 

 good judgment and value the trait highly in 

 ourselves and in others. Thus most people 

 complain that they have a bad memory, but 

 I have never heard any one acknowledge that 

 he had a bad judgment. It appears that the 

 measurement of the reliability of judgment 

 of individuals may have wide-reaching appli- 

 cations in civil service examinations and in 

 all cases where individuals are selected for 

 special purposes, a balanced judgment being 

 nearly always more important than the kind 

 of information that can be tested by a written 

 examination. I have measured the accuracy 

 of observation and memory'^ and Dr. F. B. 

 Sumner has measured the validity of beliefs." 

 When we learn to look upon our observations, 

 recollections, beliefs and judgments object- 

 ively, stating in numbers the probability of 

 their correctness and assigning probable errors 



^ ' Measurements of the Accuracy of Recollec- 

 tion/ Science, N. S., Z: 761-6, 1895. 



« ' A Statistical Study of Belief,' Psychol. Rev., 

 5: 616-31, 1898. 



to them, there will be an extraordinary change 

 in our attitude in religion, politics, business 

 and all the affairs of life. 



There are two cases in which these judg- 

 ments were subject to special conditions which 

 it may be worth the while to notice — that in 

 which a man of science gave his own position 

 and that in which he gave the positions of his 

 immediate colleagues. In sending out the 

 slips, nothing was said as to whether it was 

 expected that a man should include his own 

 name. Of the 120 who made the arrange- 

 ment, 34 gave positions to themselves; 20 as- 

 signed positions to themselves lower than 

 that resulting from the average judgment, 

 twelve higher positions and two the same posi- 

 tions. On the other hand, 22 gave themselves 

 positions higher than the average grade (which 

 is lower than the position, being related to it 

 somewhat as the average is to the median), 

 ten lower and two the same. The judg- 

 ments were somewhat more accurate than 

 the average judgments. In 21 cases the 

 departures from the mean were less than the 

 average departures and in 13 cases they were 

 larger. It thus appears that there is on the 

 average no constant error in judging our- 

 selves — we are about as likely to overestimate 

 as to underestimate ourselves, and we can 

 judge ourselves slightly more accurately than 

 we are likely to be judged by one of our col- 

 leagues. We can only know ourselves from 

 the reflected opinions of others, but it seems 

 that we are able to estimate these more cor- 

 rectly than can those who are less interested. 

 There are, however, wide individual differ- 

 ences; several observers overestimate them- 

 selves decidedly, while others underestimate 

 themselves to an equal degree. 



