542 



MEMOIRS NATIONAL ACADEMY OF SCIENCES. 



[Vol. XV, 



It is not to be expected that these several groups should give the same results, since data 

 were not obtained under constant conditions. There was variation in the time intervening 

 between examinations, in the forms of examination a which were used, and in the general intel- 

 lectual level of the groups examined. Such of the conditions as are known regarding the groups 

 are given for their bearing on interpretation of the results. 



The group of enlisted men from Devens and Dix was examined as part of an investigation 

 undertaken in response to instructions issued October 13, 1917. These instructions specified 

 that the two examinations be given on succeeding days, that forms A and C be the two forms 

 used, and that A precede C for half the group and follow for the other half. This procedure 

 would obviously tend to neutralize the effect of differences in difficulty of the forms. The 

 group showed a wide distribution over the whole scale, the mean faffing toward the lower limit 

 of grade C. 



In case of the two officers' training camps the reexamination was incidental to the exami- 

 nation of candidate officer groups. The cases here considered had been examined previously 

 in the general camp survey. The periods intervening between the two examinations were 

 therefore variable, ranging for the most part from one to three months. Both groups were 

 highly selected, their means falling in grade B and their distributions covering a range from C to 

 A, inclusive, with a few scattering cases in C — . For the Taylor group no information is given 

 concerning forms used. With the Devens group all five forms had been used for the first exami- 

 nation, but form C was used exclusively in the second. The following data show the mean gain 

 in score, summarized with reference to form used in first examination: 



These data indicate that the amount of gain is influenced to an appreciable degree by the 

 form used. Taking the above figures at their face value the indications are that forms A and B * 

 are easier than forms D and E. (The case of form C is ambiguous because of repetition of the 

 identical form of examination.) In any case figures based on such small numbers of cases 

 should not be stressed unduly. Their importance consists especially in the indication of differ- 

 ence in form of examination as an important variable in this connection. 



The Devens officer group consists of cases who were tested in the course of examination of 

 officers as a group, but who had previously taken the examination under varying conditions 

 largely for the purpose of familiarizing themselves with the methods. The group was even more 

 highly selected than the officers' training camp groups, having its mean in A and with a dis- 

 tribution ranging from C+ to A, inclusive, with four scattering cases in C. Variable periods 

 had intervened between examinations, for the most part probably from one to three months. 

 No information was at hand regarding forms used. 



The Oakland school children were fourth to eighth grade pupils, and accordingly varied 

 widely in actual age. This makes significance of results very uncertain as applied to adults. 

 In the first trial forms B, C, D, and E were used; in the second trial form A exclusively. Since 

 present evidence indicates that form A is easier than the other forms used, the degree of gain 

 was probably exaggerated to some extent for this group. Examinations were given on suc- 

 cessive days. Cases fell mainly in the lower range of the scale, probably because of the fact that 

 these were not adults, so that the various grades do not retain their usual significance. The range 

 of distribution for the group was from E to C + , inclusive, with the mean falling in C — . 



Table 199 presents comparative data for the various groups on the two examinations: 



1 Scores in form B had been given the regular addition of 13 points. See p. 318. 



