no. 2.i PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 



375 



Test 15. Memory for designs. — Much the same may be said of test 15 as of test 14. It was 

 excellent, but took more time than test 10 (also a visual memory test), which was preferred to 

 test 15 on this basis. It would make an excellent alternative for test 8 (picture sequence). 



The revised beta was therefore composed of the following tests: (1) Maze, (2) cube analysis, 

 (3) X-0 series, (4) digit symbol, (5) number checking, (6) pictorial completion, (7) geometrical 

 construction, and (8) spot pattern. 



Section 6. — Scoring the tests. 



It was found that if the tests in beta were not weighted, two tests (number checking 

 and cube analysis) would contribute almost half of the variability of the total. Of these, 

 cube analysis was not considered one of the best tests. Moreover, geometrical construction, 

 which it was desired to weight heavily, would be making the least contribution. It was 

 therefore decided that the tests in beta should be weighted. Weighting was done on the basis 

 of the interquartile range in each test (table 46). 



Table 46. — Interquartile range and weights given in each test of revised beta. 



Test. 



Old No. 



New No. 



Interquartile range. 



Lee. 



Meade. 



Weight 

 given. 



Weighted range. 



Lee. 



Meade. 



Maze 



Number cheeking 



Pictorial completion 



Cube counting 



Digit symbol 



Spot pattern 



Geometric construction. 

 X-0 series 



5.2 

 11.5 

 5.1 

 8.6 

 9.0 

 3.6 

 3.6 

 6.0 



3.5 

 13.7 

 6.2 

 5.8 

 17.0 

 6.0 

 5.4 

 6.2 



2 

 1 

 2 

 1 

 »1 

 2 

 4 

 2 



10.4 

 11.5 

 10.2 

 8.6 

 9.0 

 7.2 

 14.4 

 12.0 



7.0 

 13.7 

 12.4 



5.8 

 17.0 

 12.0 

 21.6 

 12.4 



1 The raw score of the digit-symbol test is one-third of the number of correct items. 



The weighting adopted gave a total score of 219. 



As a matter of fact, the correlation of weighted and unweighted beta (as revised) was 0.987 

 for the Meade unselected group, and their separate correlations with alpha were practically 

 identical (0.793 and 0.790). The wisdom of the decision to weight the scores is therefore open 

 to question. 



The grades assigned to scores in beta were adjusted according to the distribution of 

 ihe scores (weighted totals) in the three groups of data. By drawing the line between C and 

 D at SO, all the cases in the institutional group (except one) would be graded D or E, and 

 thus it was expected all such cases would be recalled for individual examination. The line 

 between D and E was set at 40, placing all the 6 and 7 year cases in the E group. This ruling 

 gave just 5 per cent of the Meade unselected group a grade of E, and 15.3 per cent D, which 

 seemed satisfactory. For the upper part of the distribution only the Meade group was avail- 

 able. Since this was mistakenly thought at the time to be a group rather below the average, 

 A was made to include only 2.1 per cent (200-219) and B, 16.8 per cent (170-199). Letter 

 grades were tentatively defined in terms of score as follows: A = 200-2 19; B = 170-199; 

 C+ =140-169; C=100-139;C- =80-99; D = 40-79; E = 0-39. 



Table 47 gives the score distributions (expressed as per cents) found for the three 

 groups: (1) 298 unselected men, Meade; (2) 479 Lee men who fell below 100 in alpha; (3) 77 

 feeble-minded at Yineland. 



Table 47. — Score distributions for the separate tests of the preliminary form of beta. 



TEST 1. MAZE. 



i Test 1 was given, but older scoring rule was used. 



