236 president's address — section j. 



School examinations are a form of scholastic test which are 

 urgently in need of scientific investigation. Much criticism is 

 directed in particular against external examinations. These 

 criticisms may or may not be justified. We need a study of the 

 methods of testing scholastic ability in current use, and must 

 apply to these more familiar scholastic tests the exact methods 

 now being employed in the application of non-scholastic tests. At 

 present, for instance, no attempt is made to estimate the relia- 

 bility of an external examination. It is currently asserted that 

 the pupils are not normal when they sit for an external examina- 

 tion, and that in consequence the order of merit as given by the 

 examination results does not correspond with the true order of 

 merit. It may be so. Speculation is unproifitable, and exact know- 

 ledge could readily he secured ; first, by determining the reliability 

 oc the examination test itself; and second, by calculating the 

 oo-efficient of correlation between the examination results and the 

 estimates of competent teachers. 



It is further contended that the marking is unreliable because 

 examiners vary in their standards, and the same examiner does 

 ?iot maintain an even standard. Recent American investigation 

 appears £o lend support to this contention, but methods could 

 be adopted to counteract the effect of such variations, once they 

 are known. Again, the standard and range of marking may vary 

 from subject to subject. But if scholastic ability follows the 

 normal distribution, such variations are probably due to methods 

 and standards of marking. " The degree," says Burt, " to which 

 individuals are scattered above and below the average should, 

 apart from special reasons to the contrary, be about the same in 

 different subjects." 



" According to the usual methods of marking, the standard 

 deviation found in subjects like arithmetic appears higher, as a 

 rule, than that found in subjects like composition. Yet there is 

 no reason to suppose that literary ability is less varied than 

 arithmetical . " * 



If the standard of marking differs for different subjects, injury 

 may be done to candidates where choice is permitted and the 

 same group of subjects is not taken by all. At present, public 

 examinations determine very often the future career of candidates, 

 and for this reason alone it is desirable to secure that the order 

 of merit given by the examination is, as nearly as may be, the 

 true crdeir, and to know, at all events, how far the^ actual order 

 departs fromi the true order. 



There are many other problems bearing on the conduct of 

 examinations that are urgently in need of investigation, but only 

 a brief mention can be made of seme of them. 



» Burt, ibid., i>. 48. 



