no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 341 



Test 4- — Changes: Only minor changes were made in this test, and these had to do merely 

 with the phrasing of certain items. Test 4 showed exceptional evenness of difficulty among the 

 five forms. Became test 2 of examination alpha. 



Test 5. — Changes: (1) About 15 of the easiest items were eliminated from each form and an 

 equal number of difficult items were added at the lower end. (2) The test was altered in such 

 a way as to bring an equal number of each type of item (historical, literary, scientific, etc.) into 

 each form. (3) It was decided to print the response words in heavy-faced type. 



This test contributed too much to total score, especially in the lower ranges. Ability to 

 succeed in the first 10 or 15 items seemed to depend more on reading ability than on informa- 

 tion. Too many officers were able to complete the test (20 per cent). There was marked dif- 

 ference in the difficulty of forms. Became test 8 of examination alpha. 



Test 6. — Changes: (1) Two easier items were added at the beginning of the test and five 

 harder items toward the end in each form; enough eliminations were made to leave the total 

 number as before. (2) Attempt was made to rearrange the items in order of difficulty on the 

 basis of examination records of a random group of privates. (3) A few items were shifted from 

 one form to another in order to equalize difficulty of the forms. 



The test yielded too large a percentage of zero scores, but in all other respects was one of 

 the best tests of the scale. Became test 4 of examination alpha. 



Test?. — Changes: (1) The test was extended from 10 items to 16 items. (2) The number 

 of alternative responses was reduced from four to three for each item by eliminating in each case 

 that one which had least often been checked. (3) The scoring was changed from right^minus- 

 wrong to number right. (4) Several unsatisfactory items were eliminated or altered. (5) 

 Several more difficult items were included. (6) The items were graded in difficulty on the basis 

 of estimates made by the office force and were distributed so as to equalize the difficulty of the test 

 in the five forms. Became test 3 of examination alpha. 



As it stood, the test was not long enough and the score was too much dependent upon speed 

 of reading. The elimination of the most absurd alternative response reduced the amount of 

 reading without altering the judgment factor appreciably. The increase in the number of items 

 reduced the factor of chance error in adjustment and improved the test in the higher ranges. 



Test 8. — Changes: (1) The instructions were revised so as to include more samples. The 

 statement of instructions was also simplified with the view to having the subject understand 

 the task through concrete illustration. (2) The subject is required to write two additional 

 numbers instead of one for each item, the purpose being to insure that a correct response indi- 

 cates a complete understanding of the problem. (3) Time extended from two to three minutes. 

 (4) The number of items was changed from 15 to 20. The first items are intended to be easier 

 than those of examination a and those at the end of the test very much harder. (5) Effort was 

 made to equalize the difficulty of the test in the various forms by systematic inclusion, in each 

 form, of the same types of problems. Became test 6 of examination alpha. 



This test proved of greatest value with officers, but the distribution of scores was unsatis- 

 factory. As it stood, the test was too much of the "all or none" type. The addition of both 

 easier and harder items was made to eliminate this defect. 



Test 9. — Changes: (1) The instructions were simplified by the elimination of free associ- 

 ations in the incorrect alternative responses of the first two samples. The correct answer was 

 thus made more obvious. (2) The first items in each form were made somwehat easier in 

 order to reduce the number of zero scores. (3) Some items were shifted from one form to 

 another in order to equalize the difficulty of forms. (4) Several objectionable items were 

 altered. (5) Items were arranged in the order of difficulty according to the examination 

 records of a random group of privates. Became test 7 of examiation alpha. 



Test 10. — This test was eliminated. Like test 2 it was found to correlate very poorly 

 with Stanford-Binet mental age and with other measures. It counted too much in the make- 

 up of the total score. It was of little value in the higher ranges of intelligence and was time 

 consuming. 



