322 



MEMOIRS NATIONAL ACADEMY OF SCIENCES. 



[Vol. XV, 



Camp Devens, 107 men, foreign eliminated, but largely inferior cases, r for examination a 

 and Stenquist, 0.35. Same group, r for examination a and abbreviated Stenquist (consisting 

 of alternate items), 0.32. 



Camp Lee, 76 unselected men, r for examination a and Stenquist, 0.30. For the 30 men 

 of this group whose scores in examination a were below 50 the correlation with Stenquist was 

 0.00. 



Camp Dix, 909 men of the Three hundred and third Engineers, unselected, r for exam- 

 ination a and Stenquist, 0.510 (see table 4). The same men were also given an improvised 

 group examination consisting of five tests, among which were the designs test, a digit-symbol 

 test, and a maze test. Although these required only three to six minutes each, their corre- 

 lations with examination a were, respectively, 0.54, 0.77, and 0.46. For another group of 

 69 unselected men at Camp Dix the correlation between examination a and Stenquist was 0.62. 



Camp Lee, 17 men who had fallen below 100 in examination a, correlation between Sten- 

 quist and officers' estimates, was 0.20. The correlation between Stenquist and mental age 

 for 216 men who had been given an individual examination was practically zero. (See table 5). 



Camp Dix examiners gave the Stenquist test to 40 inmates of a New Jersey State colony 

 for the feeble-minded. Five earned scores above 50. The mental ages of these ranged from 

 7.5 to 11 years. Other subjects of the mental ages 8 and 9 made to 5 points. 



Camp Devens reported tests of 48 inmates of the Massachusetts School for the Feeble- 

 minded, Waverly. The correlation between Stenquist score and mental age by the method 

 of rank differences was only 0.32. Three-fourths of these subjects earned scores as high as 48, 

 and one subject, of mental age 8.6, the score of 80. The same 48 subjects were also given 

 test 2 (memory for digits) and test 4 (arithmetical reasoning) of examination a and test Q 

 (designs) from the individual examination series. The combined score of these, given as a 

 group test, gave a correlation of 0.65 with mental age, twice as high as that for Stenquist and 

 mental age. 



The above correlations are so low that the Stenquist test can not be considered a satis- 

 factory test of general intelligence. For two groups of 179 and 107 unselected men the 

 correlation between the complete and abbreviated Stenquist (items A, B, D, E, and I) was 

 0.77 and 0.84, respectively. The correlation of one half with the other half would necessarily 

 be somewhat less than this. A reliability coefficient less than 0.90 is unsatisfactory. 



The correlation arrays of tables 4 and 5 are typical of the low correlations found for 

 Stenquist and examination a. 



Table 4. — Relation of Stenquist scores to examination a scores (r. =0.510). 



