402 MEMOIRS NATIONAL ACADEMY OF SCIENCES. [VoC xv. 



The correlation of total performance scale scores with examination a scores by the same 

 method was 0.48; but the subjects were chiefly D and E men, and the agreement was higher 

 in the case of the few A, B, and C men than for those who scored low in examination a. This is 

 what we should hope to find if examination a is unfair to some of these subjects. 



The selection of tests for a short performance scale was based partly upon time required 

 for giving, satisfactoriness of scoring, etc. Tests 2, 4, 6, 7, and 8 were recommended as being 

 the best combination, all things considered, but in order to preserve the adaptability of the 

 scale examiners were left free to use such tests as seemed best for the particular case. 



The method of equal weighting made it possible, when less than the complete scale was 

 used, to obtain a probable total score by finding the average score for the tests actually used 

 and multiplying by 10. Hence, only one table of norms was necessary. These norms were 

 given in terms of percentile rank, but because of the selection of subjects they were only rough 

 indications. The dividing line between men qualified for regular duty and those to be 

 recommended for service organizations or development battalions, and the line between the 

 latter group and those to be recommended for discharge, were determined by examining the 

 performance scale distribution of men who had made D or E on the group examination and 

 by estimating that about 10 per cent of these should be recommended for service organizations 

 or development battalions and 5 per cent for discharge. This placed the dividing lines at 

 100 and 70 points, respectively. The latter value, however, proved to be considerably too high. 



In practice the scale had obvious limitations. While it afforded a reasonably satisfactory 

 basis for recommendation regarding a subject, the results could not be stated in mental ages. 

 As these were almost always desired by the psychiatrists to whom inferior men were regularly 

 reported, it was decided to secure data from which mental-age norms could be obtained. The 

 examiners at each camp were accordingly requested to send in records of 5 to 20 men who had 

 been given both the Stanford-Binet and the Performance Scale. The data sent in and used 

 in the revision were chiefly from examinations of men who had failed in alpha, beta, or both, 

 and included the following records: 



1. Complete performance scale: 



(a) American-born subjects 134 



(6) Foreign-born subjects and negroes , 22 



2. The recommended abbreviation: 



(a) American-born subjects 126 



(5) Foreign-born subjects and negroes 39 



All complete scale records were scored also for the short scale and for an eight-test scale 

 (in which form board and picture completion were omitted). The complete scale records of 

 foreigners were neglected and the short scale records of foreigners (61 in number, including 22 

 who had complete scale also) were used only for correlation with Stanford-Binet. There were 

 not enough cases to establish foreign norms, but this proved to be unnecessary, for the scale 

 appeared to be as fair to foreign as to American-born subjects. The data used in the revision 

 were then as follows: 



1. Complete performance scale records of American-born 134 



2. Short performance scale records of American-born 260 



3. Short performance scale records of foreign-born and negroes 61 



Score distributions. — Distributions of performance scale scores and of Stanford-Binet 

 mental ages for these subjects, and of performance scale scores for the 227 Camp Lee subjects 

 tested in the earlier experiment, are shown in table 59. 



