no. 3.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 645 



the same procedure with class-interval frequencies for mental age, although the variability of 

 these groups is practically constant so far as can be determined from our data. Such a pro- 

 cedure simplifies the addition of frequencies coming from the three original, different distribu- 

 tions at the same level of the combined scale. The result of this treatment of the mental age 

 variable is table 163, in which the standard deviation of the mental age arrays is 37.45 in 

 points of combined scale. . 



Table 163. — Showing how the frequencies of each class interval of a Stanford-Binct mental age distribution should be 



disposed on the combined scale. 



(This table is made from the regression of combined scale on mental age, which is : Combined scale=1.01 M. A. (in 

 years)-|-0.33. Standard deviation of combined Bcale arrays of any type (mental age) is 1.498 class intervals.) 



Our plan of combining all types of examination used in the experimental sample, Group X, 

 is thus complete, and we are in position to determine empirically its reliability by testing its fit 

 to the sample upon which it is based. Tables 159 and 162 have been tested for the transforma- 

 tion of alpha and beta totals scores, respectively, for four different divisions of the sample into 

 alpha and beta groups. Thus we have treated the whole sample (1) as having been measured 

 by alpha only, using table 159 only; (2) taking all cases scoring less than 25 points in alpha 

 as beta cases, remainder as alpha cases; (3) all cases below 50 alpha taken as beta cases, remain- 

 der as alpha; and (4) all cases below 75 alpha taken as beta cases, remainder as alpha. Histo- 

 grams of the resulting distributions are given in fig. 6. Mean, standard deviation, and fi t and 

 /3 3 have been calculated for each of the four frequency distributions obtained in this way. These 

 constants are: 



