no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 417 



Table 76 shows for every one of the tests the coefficients of correlation between scores 

 obtained under different times and officers' ratings. It will be noted that the correlations 

 between double and single time for all tests except 3 are above 0.90. This means, again, that 

 double time has no great general effect. Arranging the tests in order of the size of these co- 

 efficients we have from greatest to least 6, 4, 2, 7, 8, 5, 3 — the same general order as above. 

 Tests 2, 4, and 6 occupy the three first positions — that is, they are affected least. 



Table 7G. — Correlations between the separate tests, double and single time, with a variety of different measures. 



Correlation between— 



1. Double and single time 



2. Officers' ratings and double time 



3. Officers' rating and single time 



4. Officers' rating and gain l 



5. Total score ( single time) and gain * 



6. Total score (single time) and single time 



7. Total score (singlo time) and double time 



8. Single time and total score of remaining six tests (singlo timo). . 



9. Double time and total score on remaining six tests (single time) 



Tests. 



0.937 

 .384 

 .320 

 .331 

 .540 

 .732 

 .800 

 .706 

 .784 



0.879 

 .432 

 .450 

 .188 

 .271 

 .784 

 .736 

 .743 

 .704 



0.940 

 .422 

 .433 

 .206 

 .347 

 .868 

 .867 

 .785 

 .802 



0.902 

 .405 

 .382 

 .232 

 .270 

 .805 

 .745 

 .740 

 .690 



0.960 

 .350 

 .308 

 .313 

 .316 

 .753 

 .711 

 .701 

 .662 



0.920 

 .409 

 .352 

 .329 

 .395 

 .824 

 .786 

 .721 

 .700 



0.910 

 .490 

 .473 

 .120 

 .155 

 .848 

 .821 

 .761 

 .754 



1 Correlations only approximately representative, since distributions are greatly skewed. 

 Table 77. — Correlation for alpha total score single time with alpha total score double time. i/=0.967.) 



We have seen that doubling the time does not result in any demonstrable improvement 

 in alpha as a whole. We have further to ask, however, whether some single test may not be 

 appreciably benefited by an increase of period. If we take officers' ratings as a measure of 

 excellence in the test, then the change in correlation with officers' ratings from single to double 

 time will show whether the test is improved, or the converse. In other words, we may tabu- 

 late the differences between the second and third rows of table 76. These differences are: 



It will be noted that in tests 3 and 4 the difference is negative, which means that the cor- 

 relations of these tests, single time, with ratings are greater than their correlations, double 

 time, with ratings. The differences in tests 2, 7, and 6 are probably significant, but not enough 

 to justify recommending double time. 



