no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 339 



age was then compared with the average correlation for five other Binet groups in which regular 

 time had been used. The following figures show the resulting changes in correlation coefficients : 



The correlations are all higher in the double-time group. That this was not due to the 

 more liberal time allowance is made clear by the fact that tests 1 and 2, in which time was 

 not extended, show as large increases as the others. The chief factor was the wider range of 

 mental ability in the double-time group. A comparison of the percentage of increase in the 

 correlations for the separate tests indicates that if any of the tests are improved by increase of 

 time it is probably tests 4, 8, 9, and 10, and that if any are injured it is probably tests 6 and 3. 

 On the whole the evidence does not suggest the desirability of any large changes in the time 

 allotments for the tests. This is in harmony with the results set forth in chapter 9. Accord- 

 ingly, only minor changes in timing were made. 



Section 11. — Methods of scoring and weighting. 



The chief questions in regard to scoring concerned the choice between the "number-right" 

 method and the "right-minus- wrong" method. A study was made at Camp Lee, in which 

 every test was correlated with total score by each method. In the case of the first four tests 

 this was done for a group of 191 random cases, and in the case of the last six tests for a group of 

 70 randum cases. The results were as follows (correlation of each test with total score) : 



The results indicate that tests 3 and 6 are the only ones in which the righ1>minus-wrong 

 method is better; in all the others, particularly tests 2 and 1, it is distinctly less satisfactory. 

 Similar data collected at Camp DLx indicated a slight advantage of the righ1>minus-wrong 

 method for test 3, no advantage for this method in the case of test 6, and a distinct superiority 

 of the number-right method in the case of tests 7 and 10. 



Terman tried both methods of scoring tests 7 and 10 with certain groups and found the fol- 

 lowing effects on correlations: 



Group. 



1. 13-year hoys 



2. Double-time group . 



3. University students 



4. 1,160 unselected 



Nature of correlation. 



With grade location 



With mental age 



With average mark . 

 With age 



Test 7. 



Right. 



0.72 

 .62 



Right 

 minus 

 wrong. 



0.58 

 .66 

 .15 

 .53 



Right. 



0.63 

 .22 

 .33 



Right 

 minus 

 wrong. 



0.55 

 .44 

 .24 

 .61 



Here the comparison of most significance is that for the first group (r with grade location 

 of unselected 13-year-old school boys) and it is seen that this distinctly favors the number right 

 method. Group 2 indicates that with double time test 10 is better scored by the right-minus- 

 wrong method. 



On the whole, the evidence was thought to justify the elimination of the right-minus- wrong 

 method for tests 7 and 10, and the retention of penalties for error only in the case of tests 3 and 6. 



