no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 413 



Table 72. — Number of subjects making each score in each test. 



i Tests selected for short scale. 



In several oamps the point scale was abbreviated by using the odd numbered tests and 

 multiplying the score thus covered by 1.7. For 168 oases at Camp Sherman, mostly beta 

 failures, the correlation between the complete scale and the odd-numbered tests was 0.88. 

 For 167 cases at Camp Greenleaf the correlation was 0.96. The median deviation of score 

 on the odd numbered from score in the complete soale was stated as approximately six months 

 in the case of 168 Camp Sherman records. It was found at Greenleaf that total score on tests 

 4, 9, 10, 12, 15, 16, 17, and 20 correlated 0.97 with score on the complete scale. The regression 

 equation for this combination of tests was found to be as follows: Total score equals partial 

 score multiplied by 2, plus 6. The probable error of estimate by this formula was stated as 

 approximately 2 points. In the same study, tests 4, 9, 16, and 17 gave a score which correlated 

 0.S6 with score on the complete scale. The regression equation was : Total score equals 3.5 times 

 partial score, plus 15. This gave a probable error of estimate of approximately 4 points, which 

 corresponds on an average to about eight months in terms of mental age. A short point scale 

 arranged at Camp Sherman consisted of tests 2, 4, 9, 15, and 17, total score of which correlated 

 0.S9 with total score of complete soale. These five tests yield a total score which approximates 

 one-third of score by entire scale, but with a probable error of estimate of somewhat more 

 than a half year. It is evident that the scale thus reduced must be used with caution. 



Table 73. — Correlations of separate tests of point scale with total score of point scale. 



