no. 3] PSYCHOLOGICAL EXAMINING IN THE UNITED STATES ARMY. 555 



it was necessary in some cases to admit cards from complete alphabetical files in which the 

 percentage of individual examinations varied considerably from the average reported from the 

 camp. The errors arising from such discrepancies are probably not great, since the individual 

 examinations are few in comparison with the total group. Errors of this order do exist, how- 

 ever, and must be kept in mind as bearing upon the precision of the data. 



In counting the cards to make up a single sample and in determining the relative numbers 

 of the different types of examination, the counts were made rapidly by hand with the aid of 

 rubber finger tips. All the counting and selection of cards was done by a single officer and an 

 assistant. It is no small task thus to select 160,000 cards. Only rarely can most of the consecu- 

 tive cards in a file be taken for a single group. Very often it is necessary to leaf over several 

 thousand cards in order to find a bare fifty or a hundred. Usually every card had to be handled 

 several times; once, say, to make the selection by States (written on the back), again perhaps 

 to exclude negroes or noncommissioned officers or privates who were not recruits, and then 

 finally to make the count of different types of examination. This counting was so time-con- 

 suming that it proved impracticable to check the numbers, and many differences between desired 

 totals and the totals actually appearing in the tables of sortings are due undoubtedly to errors 

 in the original selection. These errors are as a matter of fact of no serious moment. They are 

 always small because a rough check was exercised by measuring the piles or by placing successive 

 piles of 100 cards alongside of one another and noting an equivalence of height. 



An additional source of error lies in the uncertainty of the meaning of the State-name 

 written on the back of the card. It was intended that this entry of State should represent the 

 State from which the recruit was drafted, but in one camp at least it is known that State of 

 birth was written instead. In some other files disparity of procedure seemed to exist, for in 

 some cases the State written on the back of the card would differ from the State entered under 

 birthplace on the front of the card, while in other cases in the same file the entry on the back 

 of the card would be the name of some foreign country, obviously an entry of nativity and not 

 of residence. It is thought that this error is perhaps also not large when compared with other 

 forced errors of selection that occurred later; for example, the failure to find any cards at all 

 for Arizona and but few for Florida. 



A description of groups is given below in section 2. A map, showing the geographical 

 significance of the various camps at which psychological examining was done, is presented 

 herewith. It must be remembered, however, that the camps did not draw entirely in the draft 

 from adjacent territory, although such had been the original plan. 



Section 2. — Description of groups. 



The principal sample for Hollerith analysis was divided, as indicated above, into eight 

 main groups. The description of these groups follows: 



Group I: white draft, pro-rated, by States. — This group is intended to represent the draft of 

 the United States at large. It was selected from 1 Scamps which were, with one exception, 

 the National Army camps in which examining was firmly established at an early date and to 

 which the greater part of the draft was assigned. These camps are: Custer, Devens, Dix, 

 Dodge, Funston, Gordon, Grant, Lee, Lewis, Meade, Pike, Taylor, Travis, Upton, and Wads- 

 worth. The camps in themselves show a wide geographical scatter and it was thought at first 

 that this distribution would render a sample, selected with equal increments from every one of 

 the camps, representative of the country. It was found, however, that there were many of 

 unexpected assignments (e. g., Camp Wadsworth had mostly Pennsylvania men) which would 

 leave some section of the country totally unrepresented and some other section unduly weighted. 

 For this reason it was decided to pro-rate in proportion to the male population of the States as 

 given in the census of 1910. Inspection of the records indicated that it might be difficult or 

 impossible to obtain a sample on this method of selection that was larger than 40,000 or 50,000 

 cards. 



