no. a.] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 569 



time was available for checking the entire group. Samples of every clerk's work were taken 

 at various times; errors were indicated on less than 1 per cent of the cards. Some of these 

 errors were found later when the records were transferred to code slips. If an adequate force 

 of 50 clerks could have been available for two weeks all records could have been done promptly 

 and checked and work of the analysis could have proceeded without the numerous delays that 

 occurred subsequently. 



The transfer of the information to code slips, the punching of cards, the mechanical sorting, 

 and the tabulation of the records was accomplished by the assistance of the Actuary of the 

 Bureau of War Risk Insurance. At one time over 100 clerks were employed on the work. 

 Had it not been for gratuitous assistance from outside the War Department it would not be 

 possible to put forth the present tables. 



Sorting. — It would have been desirable to prepare Hollerith tables with respect to 

 every pair of relevant variables both for every one of the combinations of examination (see code) 

 separately and for every kind of score. For example, in a comparison of schooling with intelli- 

 gence it would be desirable to prepare scatter tables of schooling against the alpha scores of men 

 who took alpha only, against the beta scores of men who took beta only, against the alpha scores 

 of men who took alpha and beta together but no individual examination, and against the beta 

 scores of men who took alpha and beta together. When the three types of individual examina- 

 tion are added to such a scheme we find, however, that 28 tables are necessary for every 

 comparison, for we need to have the alpha scores of men who took alpha, beta, and performance, 

 and the beta scores of men who took alpha, beta, and performance, and the performance scores 

 of men who took alpha, beta, and performance, and the alpha scores of men who took alpha, 

 beta, and point scale, and so on. To multiply the number of desired comparisons by 28 is, 

 however, impracticable. Accordingly the number of tables for every comparison had, in the 

 mechanical sorting, to be reduced to seven, as follows: 



1. Alpha scores of alpha only. 



2. Beta scores of beta only. 



3. Alpha scores of alpha-beta only. 1 



4. Beta scores of alpha-beta only. 



5. Performance scale scores of all men who took the performance scale whether, with 



alpha or beta or without. 



6. Point scale scores of all men who took point scale whether with alpha or beta or 



without. 



7. Stanford-Binet mental ages of all men who took the Stanford-Bine t examination 



whether with alpha or beta or without. 



Ideally it would also be desirable to have available in the treatment of these data separate 

 tables for every camp. The results are always a function of camp procedure and frequently 

 it is impossible to tell whether we are dealing with a real difference or with an artifact that 

 grows out of differences between camps which contribute, it may be, unequally to two different 

 samples. In almost no case, however, are less than 15 camps involved, and in many cases the 

 number of camps is over 20. To multiply the desired number of comparisons by 15 is just as 

 impracticable with the available faculties for analysis as it would have been to have increased 

 the comparisons for all the various combinations of examination. Accordingly it has been 

 necessary to be satisfied with sorting various main divisions of the principal sanmle by camps 

 and to trust that the sampling has so far distributed the camps as to render camp differences 

 ineffective in the particular comparisons. 



It may be noted in this connection that an early plan for sorting, in which the minimum 

 number of divisions by combination of examination were made and in which it was planned 

 to sort separately for camps in all the more important sortings, called for the preparation of 



1 We have referred to the alpha examination briefly as "alpha" and to the beta examination as "beta." We shall use the term "alpha-beta" 

 for the combination of examination in which alpha and beta were taken but no individual examination. Similarly, we shall speak of "alpha-beta- 

 individual" when all three examinations were taken. Similarly, the terms "alpha-individual" and "beta-individual" indicate an individual 

 examination with one or the other of the two group examinations. 



121435°— 21 37 



