research upon the vaHdity of indices 

 of prediction, it is recommended that: 



The test score and rank-in-class in 

 school should be combined into one 

 index of academic promise which 

 should be used as the basis of screen- 

 ing. For each State a critical score on 

 the index should be set at a point 

 which would yield twice the State 

 quota, provided that the State critical 

 score were above the national mini- 

 mum score. 



For the candidates above the State 

 critical score, additional information 

 should be collected so that in the 

 final selection it will be possible to 

 take into account certain important 

 qualities such as originality, creative 

 ability, motivation, emotional sta- 

 bility, and qualities of leadership. 



For the convenience of the State 

 committees of selection, a summary 

 sheet would be prepared giving the 

 essential data from the four sources 

 of information indicated above and 

 this summary sheet would be at- 

 tached to the front of each candidate's 

 dossier when it is sent to the State 

 committee. 



The State committees should be 

 provided with directions to assist 

 them in interpreting the various items 

 of information about each candidate. 

 The State committees, however, 

 should be entirely free to use and 

 evaluate the information in accord- 

 ance with their best judgments and 

 should be encouraged to collect addi- 

 tional information, •such as interview 

 reports, concerning the applicants, to 

 provide the broadest possible bases 

 for the process of selection. 



]. Test of Scientific Promise 



(a) Length. — The test should be 

 of sufficient length for efficient selec- 

 tion, perhaps of 5 or 6 hours dura- 

 tion. It should not be a speed test. 



(h) Type of questions. — The test 

 should be of the objective or con- 

 trolled-answer type. The unreliability 

 of free answer questions as well as 

 the limitation in sampling imposed 

 by such questions restricts their use- 

 fulness for the present purpose. 



(c) Content. — There should be 

 several sections in the test. The ma- 

 terials throughout should be such as 

 to involve a complex of aptitude and 

 achievement as the most satisfactorv 

 measure. The subject matter should 

 be related particularly to scientific 

 ability instead of to general academic 

 promise. 



(d) Level of difficulty. — It is im- 

 portant that the test be of maximum 

 selectivity at and above the critical 

 score. Studies of the results of the 

 best mathematics and science tests 

 now in use indicate that students 

 who make very high scores on such 

 tests can be expected to succeed in 

 scientific courses during the first year 

 of college with a high degree of 

 certainty. 



(e) Preparation of the tests. — (I) 

 The test should be prepared after 

 consideration of the specifications rec- 

 ommended by the advisory committee 

 which would include both scientists 

 and testing specialists. (2) The test 

 material should be pretested on a 

 suitable population and the final test 

 made up of the most successful items. 



2. School Record 



The school record is important be- 

 cause it is a measure not only of 

 ability but of the application of that 

 ability in academic work over a 

 period of several years. It has been 

 found to be as useful in predicting 

 college success as an aptitude test 

 and when combined with the test 

 score, the combination provides an 

 index that is superior to either item 



182 



