498 



SELECTION AND TRAINING 



be more desirable to make one test of each 

 ability than to have the factor tested over 

 and over for each job. Yet the specific type 

 of arithmetic important may vary from job 

 to job. One task calls for skill with frac- 

 tions, one for decimals; one uses little divi- 

 sion, one uses much. By GuUiksen's 

 method (20) one could establish separate 

 keys for the same test, each key giving the 

 estimate of arithmetic ability that correlated 

 best with one single job. Research can de- 

 termine how much predictive validity is in- 

 creased by establishing multiple empirical 

 keys for a single standard aptitude test. 



In this connection, an important lesson 

 arises from a BuMed study. Aviation psy- 

 chologists concerned with predicting pilot 

 success had found a biographical inventory 

 to be a helpful selection tool. This is in 

 accord with many other studies, but it should 

 be noted that life histories are useful only 

 after specific research has shown this or that 

 item to be correlated with success. Permit- 

 ting a classification officer to assign men 

 merely on the basis of his opinion as to what 

 past background is desirable produces assign- 

 ments based more on superstition than on 

 science. In the BuMed study, the investi- 

 gators followed up a guess that some items in 

 a biography were effective predictors for one 

 group of men but not for others. Their 

 investigation showed that there were 70 

 items in the inventory which identified, 

 among men of high ability, those who would 

 pass the training and those who would fail. 

 There were 85 items which identified pro- 

 spective failures among those with low ability 

 test scores. But only seven items were pre- 

 dictive at both ability levels. As a result, 

 the psychologists developed three separate 

 keys for the Biographical Inventory, to be 

 used at different ability levels (55) . Apply- 

 ing to a man's inventory the key for his abil- 

 ity level raised the validity coefficient only 

 by a few points, it is true; but even such 

 gains cumulate to major savings in military 

 processing. The procedure in this study 

 seems worthy of further trial with other pre- 



dictors. This multiple-key technique should 

 certainly be applied to empirical personality 

 tests. 



On this and other problems in military 

 processing investigators will find the mono- j 

 graph. The Prediction of Personal Adjust- ^ 

 ment (28), informative and provocative. 

 Horst and his collaborators listed, in 1941, 

 many cmcial unsolved problems in predic- 

 tion, and suggested fundamental studies 

 needed. They frequently developed tenta- 

 tive mathematical solutions to such problems 

 as item weighting. Their suggestions were 

 rarely developed adequately for application 

 in World War II, but their monograph is an 

 excellent point of departure for many of the 

 studies proposed in the present report. 



Psychomotor Functions 



It is very important in military processing 

 to measure motor aptitudes: coordination, 

 speed of movement, steadiness, etc. Tests 

 of this sort have often been useful, especially 

 in selecting aviation personnel, but they are 

 expensive to design and administer. Un- 

 fortunately, different jobs call for different 

 motor abilities; no two or three motor tests 

 can measure all the factors essential in key 

 jobs. It would be desirable to have system- 

 atic research on the testing of psychomotor 

 functions to determine ways in which they 

 may be used effectively in mass processing. 

 One possibility is to devise pencil-paper tests 

 which will test motor abilities. For func- 

 tions not so treatable, it may be possible to 

 devise an apparatus which will quickly and 

 efficiently give many motor aptitude scores. 

 In the past, motor tests have been designed 

 individually, to test some one ability. Prob- 

 ably design could be improved so much that 

 many men could be tested at once by a 

 single examiner, and a single testing could 

 yield all the needed aptitude scores. 



Training to Overcome Low Aptitude ^ 



Aptitude testers are inclined to think of 

 abilities as fixed and not subject to training. 

 Actually, many of the deficiencies men show 



