no. 3.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 779 



Table 323. — Per cent distribution of alpha scores of officers. Comparison of officers with schooling only to eighth grade or 



less with recruits of more than eighth-grade schooling. 



That education is certainly not the chief conditioning factor in scores on examination 

 alpha is very well shown in table 323 and figure 49. Here the alpha scores of native-born white 

 recruits of high school and college education are compared with those of officers of eighth grade 

 schooling or less. Every recruit in the recruit group has had more schooling than any officer in 

 the officer group; the least educated recruit in the group has had a longer education than the 

 best educated officer included. And the group of officers nevertheless makes a slightly better 

 record on examination alpha. It is evident then that the examination is measuring other 

 qualities, in which officers stand above recruits, to a greater extent than it is measuring education. 



Section 4. — Correlations hetween schooling and intelligence ratings. 



In view of the close relationship between education and intelligence ratings which have 

 appeared in the preceding sections, it is not surprising that the computed coefficients of corre- 

 lation (Pearson r) run high. In general it may be said of examination alpha that in an unse- 

 lected group (i. e., including those men who would ordinarily be considered too illiterate to take 

 alpha) the correlation with the number of years schooling reported approximates +0.75; in an 

 alpha group (i. e., excluding illiterates) the correlation coefficient approximates +0.65. If 

 alpha were to be given to an unselected group the dispersion of whose scores covered the whole 

 range of the examination, and whose reported schooling varied from none at all to seven years 

 or more of college work, the correlation coefficient would doubtless be greater than any here 

 presented. The nearest approach to this is a combination of two of the groups which appear 

 in table 324 (the Hancock group of 489 and the unselected group from nine camps, Group X) ; 

 this gives a wide range both of schooling and of alpha scores and a correlation coefficient of 

 + 0.S1 between reported schooling and alpha weighted total. 



Except for the principal Hollerith sample (Groups I, II, III), table 324 is compiled from 

 camp reports and from data already at hand before the principal Hollerith samplings were made. 

 The most heterogenous single group included (the unselected group of 653 men from nine camps, 

 Group X) contains no weighted alpha score higher than 360 and no schooling record of more 

 than four years college, thus not covering the upper part of the range in either respect. For 

 this group the correlation is 0.75, whether raw or weighted totals are considered. The same 

 relation is found for the unselected group from the Three hundred and fourth Ammunition 

 Train, Camp Meade, in which the range of scores and schooling is about the same as for the 

 unselected group of 653. 



