CHAPTER 3. 



COMPARISON OF FORMS OF EXAMINATION ALPHA. 



Five forms of examination alpha, numbered consecutively from 5 to 9, were used. The 

 tests in each form were of the same types and arranged in the same order. The items differed, 

 but their difficulty had been carefully adjusted in an attempt to secure equivalence of each 

 form of each test with any other form of that test. These forms are reproduced in Part I, 

 pp. 220 to 234. The data have not been analyzed test by test to see whether this ideal was 

 reached, but a comparison based on total scores is available. The principal Hollerith sample 

 of alpha scores of the white draft (Group I) given alpha included 31,967 cases, in which the 

 form used had been recorded; 2,448 of these had been given beta also. These two groups 

 were each sorted according to form of alpha used, and comparisons made between the dis- 

 tributions of alpha scores on each form. The distributions in both actual numbers of cases 

 and percentages appear in tables 174 and 175. Similar procedure was attempted with the 

 principal Hollerith sample of the negro draft (Group IV), but the numbers proved too small 

 to make comparison valid. 



For the white draft the comparitive means, standard deviations, and probable errors 

 were computed, and are shown in table 176. 



Table 174. — Variables: Alpha score X alpha form. Group I: White draft. 

 For men who took alpha only. 



659 



