no. 2.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES AKMY. 391 



This shows no serious drop from the correlation of 0.731 ±0.012 between Stanford and 

 weighted score. 



Table 54 presents the scatter table for Stanford with beta according to the method suggested. 

 Inspection shows that this form handles the group at the lower end of the scale at least as well as 

 does the weighted score. (This point will be discussed more fully in connection with the estab- 

 lishing of the lower dividing line for this method.) 



Table 54. — Shot/ring correlation of beta raw score (suggested form) with Stanford-Binet mental age — English-speaking 



whites from nine camps. 



r=.727±.012. 



Beta raw (suggested): Mean=f.2.8 a=25.7. 



Stanford-Binet: Mean=13.42»= 2.85. 



1 In tables from which computations were made intervals of half-year mental ages were used. 

 5 to 5.9, 6 years includes 6 to 6.9, etc. 



In the present tables 5 years mental age includes 



As further evidence that this is a safe change and that correlations are not determined 

 mainly by some erratic quality of the total group, correlations were computed separately for 

 each of four camps — Dbc, Jackson, Lee, and Meade. Table 55 shows the correlations obtained 

 with Stanford by the method under consideration and those obtained by use of the former 

 method. In no case is there an important drop in the amount of correlation though there is for 

 each a slight reduction. 



Table 55. — Correlations of Stanford until beta weighted score and with beta raw score (suggested form) for four different 



camps, English-speaking cases only. 



As an additional check on the proposed change the correlation of alpha with the new form 

 was computed. (Since it had been decided, on evidence from other sources, to use raw instead of 

 weighted scores on alpha, raw alpha scores wefe used for this correlation. Since the correlation 

 of raw with weighted alpha scores for this group is 0.994, change from weighted to raw scores 



