374 MEMOIRS NATIONAL ACADEMY OF SCIENCES. [V6l. xv, 



Test 3. Number checking. — Test 3 was retained, being satisfactory on all counts except 

 that it does not make a very sharp separation between good and poor groups. The only change 

 is the substitution, in the demonstration, of simpler numbers for some of those originally used. 



Test 4- Dot imitation. — Of tests 4, 10, and 15, which seemed to be testing similar processes, 

 test 4 was dropped as the one of the three showing the greatest number of zero and perfect 

 scores, the poorest correlation with alpha total, and the poorest separation between good and 

 poor groups. The differences were small, however, and test 4 had the advantage that the feeble- 

 minded made very low scores in it. It might well be used in an alternative form of beta exam- 

 ination. 



Test 5. Pictorial completion. — Test 5 was retained on all counts, except that it took two pages 

 in the book and over 6 minutes to give. By better arrangement of pictures, by rejecting the 

 largest and putting in drawings with simple lines which could afford reduction, 20 items have 

 been put on one page. The test was studied item by item for relation to alpha and beta total 

 scores, and the 11 best items from the original form remain in the present form. Of the six 

 samples in the new demonstration, four are new. The time was later reduced from 5 minutes 

 to 3 minutes. 



Test 6. Cube analysis. — Test 6, while not among the most successful of the beta tests, 

 was not particularly poor, and had the advantage of brevity. It seemed likely, too, that by 

 replacing some of its items by less obvious ones it might very easily be improved. Test 6 

 was thus tentatively included in the new beta, pending improvement, with a reduction of the 

 time limit to 3 minutes. 



Test 7. Line comparison. — Test 7 was an immediate case for rejection. The very high 

 per cent of zero scores, in conjunction with the high per cent of perfect scores, showed that it 

 was much too easy once the idea was conveyed, but that far too often the demonstration failed. 

 The distribution is the reverse of normal — i. e., U-shaped. It seemed not worth while to 

 try to remedy these difficulties, which apparently are inherent in the test. 



Test 8. Picture sequence. — Test 8 was tried both with response by numbers and arrows. 

 The former was better, but even so the demonstration did not succeed very well. Since it 

 seemed closely related to tests 5 and 14, and poorer than either, test 8 was dropped. 



Test 9. Digit-symbol. — Test 9 was retained on all counts. The time limit was reduced 

 because it seemed after trial at Camp Lee that the amount covered in the first 2 minutes was 

 probably a better index than that in 3 minutes. 



Test 10. Spot pattern.— Test 10 was indicated for retention on all counts except that it 

 required much time and was closely related to test 15. Test 15, although it gave excellent 

 results, took even longer to give than 10. Accordingly 10 was retained, but with a reduction 

 of the number of items from 20 to 12. The correlation between the left-hand and the right- 

 hand halves of the page (10 items each) was 0.73, and since the items were correlated with alpha 

 and beta totals and the best 12 selected instead of the left-hand 10, the reliability of this test 

 in its revised form is probably fairly high. This reduction brings the time under 6 minutes. 



Test 11. Analogies. — Test 11 was dropped as impossible to demonstrate in its present form. 

 It was not studied in other respects. 



Test 12. Geometrical construction. — Test 12, good in all respects except for its high per cent 

 of zero scores, was particularly good in differentiating the feeble-minded. It was therefore 

 included and heavily weighted in the new beta total. The time limit was increased to 3 minutes, 

 with the idea that this might serve to distribute the scores more widely. A simpler sample 

 in the demonstration was substituted for one of the more difficult samples. 



Test 13. X-0 series. — Test 13 is rather easy and gives a high per cent of perfect scores, 

 but since it seemed to be effective in indicating the institutional feeble-minded group and in 

 separating good and poor groups, and was a brief, one-page test, it was included in the beta. 



Test 14- Picture situation. — Test 14 proved an excellent test and was omitted from the new 

 beta on practical considerations only. It seemed related to test 5 and test 8, and could less 

 easily be reduced to one page than could test 5. It was hoped, however, to revise test 14 in 

 this direction and to have it ready for an alternate form of examination beta. 



