UTILIZATION OF BRAIN POWER IN ARMY 199 



various measures of the validity of the army mental tests 

 are presented. 



The probable error of an Alpha score is about five 

 points. This is approximately one-eighth of the stand- 

 ard deviation of the scores for unselected soldiers. The 

 reliability coefficient of examination Alpha approximates 

 .95. This group examination correlates with other meas- 

 ures of mental ability as follows: (i) With officers' rat- 

 ings of their men, .50 to .70 for the total Alpha score 

 and .30 to .54 for the separate tests; (2) with Stanford- 

 Binet measures of intelligence, .80 to .90 for the total 

 Alpha score and .31 to .85 for the separate tests; (3) with 

 the Trabue B and C Completion tests combined, .72 for 

 the total score and .39 to .76 for the separate tests ; (4) 

 with Examination Beta, .80; (5) with the composite re- 

 sult of Alpha, Beta and Stanford-Binet examinations, 

 .94; (6) in the case of school children results of Alpha 

 examination correlate (a) with teachers' ratings .67 to 

 .82, (b) with school marks .50 to .60, (c) with school 

 grade location of thirteen and fourteen-year-old children 

 .75 to .91, (d) with age of children .83 (for soldiers the 

 correlation of Alpha score with age is practically zero) . 



The Alpha examination given with double the usual 

 time allowance correlates approximately .97 with the reg- 

 ular time examination. 



The following data indicate the reliability of Exami- 

 nation Beta : It correlates with Alpha, .80 ; with Stanford- 

 Binet, .73 ; with the composite of Alpha, Beta and 

 Stanford-Binet, .915. The correlation of the separate 

 Beta tests with the Stanford-Binet ranges from .47 to 

 .63 (average .58). Results of Beta given with double 

 time allowance correlate with those obtained with the 

 regular time allowance .95. 



For the several forms of individual examination used 

 in the army the principal correlations at present avail- 

 able are as follows : 



Results obtained by repetition of Stanford-Binet ex- 



