466 



MEMOIRS NATIONAL ACADEMY OF SCIENCES. 



[Vol. XV, 



It is clearly necessary to define exactly "military efficiency" or "value to the service" 

 before any attempt is made to study correlations of test performance with it. (The alpha 

 tests, although often claimed to be alike, are nevertheless different enough to reflect the quali- 

 tative differences of selected groups.) All of the tests show a rather satisfactory degree of 

 differentiation between groups of considerable differences in military efficiency. It is possible 

 that greater refinement in the definition of standards of military efficiency and in the applica- 

 tion of them would make possible a better selection of alpha tests than exists at present. So 

 long as such standards remain as vague and undefined as the present study indicates them to 

 be there is little ground for regarding one alpha test as considerably more or considerably less 

 efficient than another, always excepting test 5, which seems consistently from every point of 



view the least effective. 



Section 2. — Examination beta. 



In general the beta tests show the same tendency to vary in sharpness of differentiation 

 in the different groups compared. Data on three groups only have been available and these 

 groups are rather small. There are but 242 cases in all. These facts make it almost im- 

 possible to draw any certain conclusion as to the relative values of the tests. Tables 138 to 

 140 give the comparisons. 



Unfortunately the distributions of scores for the "poorest" Field Artillery, group of test 

 4 could not be used owing to clerical errors in transferring scores to individual record cards. 



In the comparison of "poorest" Infantry v. "poorest" Field Artillery, test 3 ranks first 

 in sharpness of differentiation, and in the differentiation of "poorest" Field Artillery v. standard 

 it is second. In the comparison of "poorest" Infantry v. standard and men of "low military 

 value" v. standard, it ranks sixth. This is probably the only case of considerable shift in rank 

 for which an explanation can be offered. A study of the distribution of scores of the standard 

 group indicates that test 3 is of the " all-or-none " variety, since its frequency curve is U-shaped. 

 Such a test will give a very clear distinction between two groups which he on opposite sides of 

 the point in the scale of ability which is crucial for the test. 



Table 138. — Means and standard deviations of the mean for scores in the seven beta tests made by various military groups 



A. 984"UNSELECTED" (ENGLISH-SPEAKING) CASES. 



C 92 CASES AMONG "10 POOREST," NINETEENTH INFANTRY, CAMP TRAVIS. 



D. 78 CASES OF "LOW MILITARY VALUE," CAMP CUSTER. 



' Several cases obviously scored erroneously. 



