554 MEMOIRS NATIONAL ACADEMY OF SCIENCES. tvoL.xv, 



of individual examinations that belonged with a given set of group examinations. It is evident 

 that the selection of a representative sample from such data could be done only by the scrutiny 

 of every card and by a careful checking of the numbers of the various types of examinations 

 against such records of these numbers as happened to be available. Very often it occurred 

 that the relative numbers of the different types of examination taken from a file that gave 

 every evidence of being complete would not tally very closely with the corresponding figures 

 in the weekly reports from the camp. In such cases the cards for the sample were taken in 

 the frequencies in which they were found (or in the case of a small sample in the frequencies 

 in which they occurred in a larger group) . Such discrepancies are quite possibly to be accounted 

 for by the fact that the weekly reports could never be exactly identified with a group of undated 

 cards. 



In order to fix certain of the conditions of sampling and to meet in part the difficulties 

 arising from the varying constitution of the groups reported by the different camps, it was 

 decided to separate the principal sample into the four main groups mentioned above, viz, the 

 white draft (representative of the greater part of examining), the negro draft, the selected 

 group of white officers, and the more or less selected group of men already established in training 

 in permanent divisional or other military organizations. Hence, in selecting cards, attention 

 had always to be paid to the group in which the card belonged, although, fortunately, in many 

 camps separate files had been kept for these four divisions. 



The principal sample for the white draft, as well as that for the negro draft, is intended to 

 provide a group that shall be representative of the draft from the country at large. Since it 

 was necessary to select particular camps from which the cards should be drawn and since these 

 camps had a somewhat abnormal geographical distribution which was in turn cut across by some 

 unusual assignments of draft quotas, it was decided to pull from the records a large number 

 of cards in accordance with the State from which the men had been drafted. From May to 

 August, 1918, psychological examiners had been instructed to place on the back of every psycho- 

 logical record card the home town and State of the recruit, and it was in accordance with these 

 entries that the geographical selection by State was made. The discontinuance of the rule for 

 entering the recruit's residence necessitated the selection of all cards in these geographical 

 groups from the period of examining prior to the middle of August, 1918. In other groups the 

 cards may, and frequently do, come from later periods. It was not possible to select all the 

 cards in the same period of time, because in some cases sufficient numbers were not available. 



The general rule for selection of records from the files in the case of an alphabetical arrange- 

 ment was to pull cards from different parts of the alphabet, most usually by taking an equal 

 number from the front of every file box (a file box contains about 2,000 cards). After the 

 requisite number of cards had been obtained in a particular case the numbers taking examina- 

 tion alpha only, examination beta only, alpha and beta together, and any form of individual 

 examination were counted separately and compared with such statistical or general data as 

 were available for that camp. For example, a failure to find alpha-beta records should have 

 meant that the camp at some time or other was unable to recall men from alpha to beta. 

 Absence of records of individual examination most frequently meant that the cards had been 

 filed elsewhere. In such a case> the additional file was located, cards of the same group were 

 isolated when feasible, and a sufficient number of individual records added to the group already 

 pulled to make a fair sample. The number of individual records was determined either by the 

 weekly reports from the camps or, when the total group could be isolated both for group exami- 

 nations and individual examinations, by estimating the number of cards in each part and pro- 

 rating. When this course was necessary, small batches of cards were counted by hand, large 

 batches were counted by placing the cards under compression and measuring — there are almost 

 exactly 100 cards to the half inch. 



When discrepancies between actual findings in the files and office records were great an 

 effort was made to secure other samples and thus avoid the possibility of error. For example, 

 the records for Camp Sherman were abandoned entirely in the case of draft quotas because no 

 individual examinations could be found to go with the group examinations. On the other hand, 



