no. 3] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 833 



Mr. Carl R. Brown utilized Pearson's method of comparing distributions in order to deter- 

 mine the probability that the observed differences between them are due to the operation of 

 different types of selection (or significantly different degrees of the same type of selection). He 

 reports as follows: 



Using letter-grade frequencies, the quantity x 2 was calculated for pairs of distribution as follows: 



Comparison. 



Engineers v. accountants 242. 97 



Accountants v. telegraphers 100. 7 



Telegraphers v. locomotive firemen 100. 17 



Locomotive firemen v. teamsters 124. 44 



These values of r 2 are too large to allow the use of Elderton's tables. By means of the original formula the value 

 of P for the first comparison (Engineers v. accountants) was found to be —47, (P=25.10). Since the probability that 

 the difference between these two distributions of letter grades is the result of chance is so small, we may conclude (using 

 the converse statement) that the two populations represented by our samples are clearly differentiated as regards Army 

 intelligence ratings. 



Although the x 2 values obtained from the other comparisons are only about half that for engineers v. accountants, 

 they are outside the limits of Elderton's tables, and indicate such an infinitesimal probability that chance is the sole 

 factor that all of the groups under consideration may be regarded as distinct from the point of view of Army intelligence 

 ratings. 



The significance of Group I is brought out not only by its subdivision into four or five 

 occupational levels, but also by the fact that similar occupations are "bunched" on the intelli- 

 gence scale. That is, similar occupations have similar distributions of intelligence with medians 

 approximately the same. For example: General auto repairmen, auto-engine mechanics, and 

 auto assemblers are all in the upper part of the C group of occupations. Then in the upper part 

 of the C+ group we have railroad clerks, filing clerks, and general clerks "bunched" together 

 with practically the same median scores. It is also interesting to note that telephone operators 

 fall in the uppermost part of the C group, while telegraphers are well in the middle of the C + 

 group. Bookkeepers bear the same relation to accountants with the exception that each is in 

 a higher letter grade. 



In view of the possibility that the table of occupational intelligence standards is unreliable 

 in part because of inaccuracies of the personnel methods, data were requested for soldiers who 

 had been occupationally classified by trade tests. Sufficient returns were obtained on eight 

 occupations to warrant the construction of table 379, which gives a percentage distribution 

 of intelligence ratings of men not trade tested (data previously secured and reported in table 

 378), and of men who passed the trade test, i. e., qualified as apprentices or better in that trade. 



Table 379. — Intelligence and occupation. Percentage distribution of letter grades of men not trade tested and of men who 



passed the trade tests. Study of Surgeon General's Office. 



It will be noted that in every case except that of truck drivers the per cent of beta men 

 qualifying on the basis of trade tests is much lower than on the basis of personnel interview. 



