404 MEMOIRS NATIONAL ACADEMY OF SCIENCES. [vol. xv. 



The digit symbol test (7) was too difficult; but the large number of zero scores was no 

 doubt partly due to the method of weighting. Raw scores from zero to five equaled zero 

 weighted. The weigh ting was therefore changed so as to make the test one to two points 

 easier at the lower end of the scale. 



The picture arrangement test (9) had a rectangular distribution. A study of the method 

 of scoring in this test showed that there is considerable probabilit} 7 of getting credit as a result 

 of chance juxtaposition of cards. It therefore seems probable that if chance scores were 

 eliminated the lower portion would approach zero as in the picture completion test (10). The 

 test was really too difficult, but, nevertheless, it seemed advisable to reduce some of the chance 

 scores by a change in weighting (increasing the number of zero scores). This new weighting 

 still allows some credit for chance arrangement, but a more radical change might interfere too 

 much with the scale norms. 



Both tests 9 and 10 are too difficult for men who have failed in beta, and they can not be 

 improved by weighting. 



The remaining six tests of the scale — manikin and feature profile (2), cube imitation (3), 

 cube construction (4), form board (5), designs (6), and maze (8) — give good distributions of 

 scores for each group of subjects, and the original weighting is satisfactory. 



The scoring and weighting finally adopted for each test are as given in the revised Examiner's 

 Guide and in the performance scale record blank. 



Mental-age rating. — Mental-age equivalents were obtained for the long scale (134 cases), 

 for the eight test scale (134 cases), and for the short scale (260 cases). Since the norms for 

 eight tests differed very little from those for the whole scale, only the long and short scale norms 

 are given in the Examiner's Guide. The short-scale norms were derived from tests 2, 4, 6, 7, 

 and 8; but since tests 1 and 3 are very nearly equal to these in difficulty, the same norms may 

 be used with a minimum of error for any number and any combination of tests 1, 2, 3, 4, 6, 7, 

 and 8. If eight or more tests are given the long scale norms should be used. 



These mental-age equivalents were obtained by equating equal ranks in the performance 

 and Stanford-Binet scales. Thus the mental-age equivalent for a given score is not necessarily 

 the most probable mental-age for any individual making that score. To obtain this result the 

 regression equation of Stanford-Binet mental age upon performance score would have to be 

 used. However, since the relation of the different individual examination scales to army 

 achievement is unknown, it may be assumed that each scale ought to give much the same 

 result, in the sense that the number of men recommended for discharge or development battalions 

 on the basis of the two scales should be approximately equal. This result is obtained by tne 

 method used in equating ranks. 



Correlations.— -Correlation arrays were plotted and Pearson product-moment correlations 

 computed for long scale, eight test scale, and short scale scores with Stanford-Binet mental ages; 

 for short scale with long scale scores; and for scores on each test with total performance scale 

 scores and with Stanford-Binet mental ages. These calculations were based upon 134 cases, 

 except in the case of correlation of scores on the short scale and its various tests with Stanford- 

 Binet ratings. Here 260 cases were available. 



The correlations of the separate tests with mental age ratings and with total performance 

 scale scores are on the whole higher than expected. If the tests are arranged according to the 

 degree of correlation, the following orders are obtained: 



With Stanford-Binet. 



Digit symbol 0. 777 (260 cases) 



Designs 0. 735 (260 cases) 



Picture arrangement 0. 723 (134 cases) 



Manikin and feature profile 0. 676 (260 cases) 



Ship 0. 661 (134 cases) 



Maze 0. 655 (260 cases) 



Picture completion 0. 650 (134 cases) 



Cube construction 0. 633 (260 cases) 



Cube imitation 0.597 (134 cases) 



Formboard P. 480 (134 cases) 



With total performance scale. 



Digit symbol 0. 881 (260 cases) 



Designs 0. 858 (260 cases) 



Picture arrangement 0. 858 (134 cases) 



Manikin and feature profile 0. 830 (260 cases) 



Cube construction 0. 819 (260 cases) 



Picture completion 0. 815 (134 cases) 



Maze 0. 788 (260 cases) 



Ship 0. 768 (134 cases) 



Cube imitation 0. 661 (134 cases) 



Form board 0. 606 (134 ca 



