SELECTION 



503 



306-309). Poor criteria have an equally 

 serious effect on training, since a poor 

 method of evaluation will pass incompetent 

 men to duty and perhaps fail good ones. 

 Men may pass tests supposed to show pro- 

 ficiency, but if the tests themselves are in- 

 valid, months may pass before it is realized 

 that the training program lacks effectiveness. 

 Any research that can be pursued now to 

 develop good criteria in training schools or in 

 the Fleet is desirable ; even more desirable is 

 research to develop good general methods for 

 the design of criteria as nev/ tasks become 

 important. 



Objective Criteria 



One line of approach is to objectify cri- 

 teria. A general opinion from a supervisor 

 or observer is notoriously inadequate. Bias, 

 "halo effect," differences between observers, 

 and other errors, make such estimates crude 

 at best. JMethods of eliminating obsei'ver 

 errors are many, but it is not certain which 

 will be most practical. 



One attack is to require superior officers 

 to collect specific records of good and poor 

 job performance, instead of merely rating 

 from memory at the end of a six-month 

 period. Such a procedure is now being de- 

 veloped for the Air Force by the American 

 Institute for Research. The checkride in 

 aviation is a technique based on specific ob- 

 servation rather than opinion; this method 

 has been greatly improved by defining care- 

 fully the test maneuvers and the facts to be 

 noted (47). Similar field tests of undersea 

 warfare personnel can be used effectively, as 

 demonstrated by the USS SYLPH, during 

 the recent war, A few graduates from each 

 class of the two fleet sound schools were sent 

 to New London and tested at sea, and weak- 

 nesses which were detected led to revision of 

 faulty programs in the schools. Perhaps all 

 that can be said at present is that the weak- 

 ness of subjective criteria has been overcome 

 very little by research to date, but that 

 studies of the type made during wartime 

 should be pursued (44, pp. 357-379). 



Combat Criteria 



No research during peacetime can really 

 evaluate a militaiy selection or training pro- 

 gram. The real test of a bombardier is 

 whether his bombs hit the target even when 

 he is under tension and threatened by flak. 

 The test of a submarine officer's leadership 

 is whether he can maintain the morale of a 

 crew on a long patrol during which they 

 are depth charged repeatedly. No method 

 of making such tests is feasible in peacetime. 

 And it is probably not true that the man who 

 does best in simple peacetime demonstra- 

 tions and dr>' nms will also have the reserve 

 capacity which makes an outstanding com- 

 bat performer. Some of the best men in 

 wartime are those who were undistinguished 

 in peacetime but were brought to a peak by 

 a real challenge. 



The problem, then, is to establish methods 

 for evaluating how v\^ell men do in combat. 

 These methods, if thought out in advance, 

 should greatly increase the dependability 

 of selection methods. 



Multiple Criteria 



Past workers have tried to predict a single 

 criterion: "goodness" on the job. But per- 

 formance is not so easily described; any 

 job has many aspects, and most men are 

 better at some aspects than others. Any 

 research on prediction of complex duties 

 should be so designed that a criterion will 

 show how well the man does each aspect of 

 his job. For example, the lookout might 

 be characterized thus: "He is superior in 

 locating small targets; he frequently mis- 

 classifies a cargo ship as a warship; he makes 

 some mistakes in estimating heading, though 

 he is generally accurate; he is one of the best 

 men I've ever had for maintaining alertness 

 on a long watch; and so on." Such a break- 

 down of the job should peiTnit far more 

 exact prediction of job performance than the 

 usual single criterion. In aviation selection, 

 personality tests predicted very poorly who 

 would graduate from school (18); this is 



