no. 3] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 



781 



Correlation of each of the eight alpha tests with reported schooling has been made in two 

 of the groups in table 326, the combined Hancock alpha group, and the unselected English- 

 speaking group of 653 (Group X). The coefficients are as follows: 



Table 326. — Correlations between alpha tests and schooling. 



Test. 



Name of test. 



Oral directions , 



Arithmetic 



Common sense 



Opposites 



Disarranged sentences 



Number series 



Analogies 



Information 



Hancock, 



Croup X. 



Inspection of the scatter diagrams shows that the discrepancies between these two sets of 

 figures are due chiefly to the inclusion in the unselected group of 653 of a number of illiterate 

 men. The extreme skewness due to the large number of low scores on tests 4, 5, and 7 tends 

 to lower the correlation coefficients on these tests in the unselected group, while in tests 1 and 8 

 the unusual number of low scores makes the distributions more nearly normal, and thus helps 

 to raise the correlation coefficients. The differences in tests 3 and 6, and especially in test 2, 

 have not been explained. 



Most correlations of beta and individual examinations with schooling are of an entirely 

 different significance than those of alpha, because the groups are so selected that the schooling 

 is restricted. Schooling (4, 5, or 6 years) is indeed frequently the sole basis of the separation 

 into alpha and beta groups. Intelligence scores are low in these groups for the most part, and 

 under these limitations the correlation coefficient is bound to be low. This relation holds for 

 the correlations of the principal samplings shown in table 327. These coefficients should be 

 compared with the figures for the unselected group from Camp Meade (includes foreigners) 

 and for Group X (excludes foreigners), which conclude the table. The correlation plot for 

 Group X (beta raw total with years schooling) is shown in table 328. 



Table 327. — Correlations between examination beta and schooling . 



Camp. 



Group. 



Sixteen camps (beta group)... 



Do. 

 Do. 



Do 



Meade (unselected) 



Nine camps (unselected), 

 Group X. 

 Do 



Principal Hollerith sample, native 

 born. 



Principal Hollerith sample, foreign . . 



Principal Hollerith sample, northern 

 negroes. 



Principal Hollerith sample, southern 

 negroes. 



Three hundred and fourth ammuni- 

 tion train (includes foreign). 



Draft group (excludes foreign) 



.do. 



Range. 



None to college 4; beta raw, to 115. . 



None to college 2; beta raw, to 105. . 

 None to high school 4; beta raw, 



0to90. 

 None to high school 4; beta raw, 



to 100. 

 None to college 4; beta weighted, 



to 200. 

 None to college 4; beta weighted, 



to 210. 

 None to college 4; beta raw, to 110. 



Number 

 of cases. 



11,879 



5,803 

 1,737 



3,438 



262 



653 



653 



0.45 



.50 

 .36 



.64 



.67 



A report from Camp Lee on the results of giving examination beta to school children of 

 Petersburg, Va., gives the median scores of table 329 for each of the beta tests and for the 

 weighted total. 



