no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 305 



3. For the marines each test correlated as follows with the total score of the ten tests: 



Test 1234 567 8 9 10 



Correlation 0.75 0.70 0.73 0.82 0.72 0.79 0.88 0.56 0.76 0.62 



4. The correlation of the total score with the number of school grades completed was 0.73. 



The correlations which the tests gave were therefore in the mam satisfactory. They were 

 high with outside measures of known value; they were high enough with one another to indi- 

 cate that all were reasonably good tests of general intelligence; at the same time the inter- 

 correlations of the tests were not high enough to suggest that the tests were only repetitive of 

 one another. 



The tests were scored in various ways in order to determine the best method empirically. 

 The principle adopted was that of finding by trial the method of scoring which would give the 

 highest correlation with Binet or with total score. All the tests were scored both for " number 

 right" and "right minus wrong." In addition, test 4 was also scored for "right — 2 wrong"; 

 and tests 5 and 7 for " 3 right — wrong." It has been supposed that it would be found necessary 

 to penalize, more or less extensively, for errors. However, it was found that, in general, penal- 

 izing was of little or no advantage except in those tests where it was necessary to counterbalance 

 the factor of chance, as in tests 3 and 6. Test 4, for example, gave significantly higher correla- 

 tions with Binet and with total score when scored for number right than when scored for " right — 

 wrong" or "right — 2 wrong." Only in tests 3, 6, 7, and 10 did the data justify subtraction 

 from the score in case of error. In tests 3 and 6 the subject has an even chance of making a 

 correct response by guessing, and this is offset by scoring the test "right — wrong." In tests 7 

 and 10 the advantage gained by penalizing for errors was small. 



The time allowances for the tests were checked up by noting the per cent in each group who 

 attempted all the items of a test. The following changes in time seemed to be justified: Test 1, 

 slight reduction in time for certain items and slight increase for others. Test 2, slight decrease 

 for certain items. Test 3, reduction from three to two minutes. Test 4, no change, five minutes. 

 Test 5, reduction from four to three minutes. Test 6, reduction from two t one and one-half 

 minutes. Test 7, reduction from two minutes to one minute. Test 8, reduction from four to 

 two minutes. Test 9, reduction from four to three minutes. Test 10, unchanged, two minutes. 

 The time allowances agreed upon at this revision remained unchanged until the original group 

 scale was modified into the alpha scale in January, 1918. 



Other results considered were the number of zero scores for different tests in various groups, 

 the standard deviation of the scores in different tests, and the relative difficulty of the items 

 composing the 10 tests. It will be recalled that when the group test was originally made up 

 enough items were collected for 10 different "forms" of each of the 10 tests, so that 10 different 

 record booklets could be used, all equally difficult but not duplicative. Only one of these forms — 

 namely, Form A — was printed for the preliminary try-out. Hence the rearrangement of items 

 at this time to accord with order of difficulty was made only for Form A. Even in this form no 

 serious attempt was made to secure an exact order according to difficulty. As a result of the 

 experiment, a few items in the tests were modified in content or form. The directions for the 

 separate tests were modified to some extent and several changes were made in the examiner's 

 prologue. Whipple was largely responsible for the final wording of the prologue. All the 

 members of the committee participated in the revision of the group tests after the recess except 

 Haines, who was unable to be present. 



On July 1 word was received that $2,500 had been made available for a trial of the methods 

 in various military organizations. On July 2 copy for Forms A, B, C, D, and E of the group 

 examination were placed in the hands of Bingham, who was charged with the responsibility of 

 seeing the material through press. One thousand copies of each form were printed. 



