804 TRANSACTIONS OP SECTION L. 



answered by nearly the same number, whether the method of Galton, of 

 McDougall (Mental Measurements Committee), or that followed in several other 

 investigations is employed. This remarkable conformity is shown by the diagram. 

 If the supernormal individuals who accomplish the test are added the result 

 is nearly always the same — a percentage of 77. 



5. Experimental Tests of General Intelligence. By Cyril Burt, 31. A. 



A series of experiments was carried out at Oxford two years ago, mainly 

 upon thirty elementary school children, 12^ to 13g years of age. The chief object 

 was to determine the relative value, as tests of general intelligence, of a dozen 

 brief tasks, involving mental processes at various levels, in various aspects, and 

 of various degrees of complexity. By general intelligence was understood innate, 

 unspecialised mental efficiency, as distinguished both from acquired knowledge, 

 interests, and dexterities, and from specific endowment, aptitude, or talent. To 

 form tests of general intelligence, the tasks were required, not necessarily to 

 prove a means of measuring its amount in any individual child, but merely, with 

 sample groups of children, readily and rapidly to yield results which should be 

 reliable in themselves, and correspond to a constant and definite degree with the 

 results of prolonged and careful observations of the teacher. The degree of 

 correspondence was calculated by the method of correlation, and the coefficients 

 obtained were taken as indicating the relative value of the tests. 



Views attributing to sensory discrimination, whether general or specific, an 

 intimate functional correspondence with general intelligence were not confirmed. 

 Auditory and visual tests, indeed, showed positive, though not considerable, cor- 

 relations with intelligence ; but these seem rather to be referred to the dependence 

 in the course of evolution of the progress of intelligence upon the perception of 

 space and upon the perception of spoken words, and of these respectively upon 

 delicacy of eye and ear. Tests of discrimination of touches and of weights 

 showed approximately no correlations with intelligence. Simple motor tests, such 

 as tapping and dealing, showed somewhat higher correlations than the sensory 

 tests. But of the six simpler tests, sensory and motor, none gave correlations 

 above 0'50. 



The remaining six dealt either with processes of a higher mental level — such 

 as memory, habituation, scope, and maintenance of attention — or with more 

 complex mental processes, involving co-ordination of both sensory and motor 

 activities, such as the ' alphabet ' and ' dotting ' tests devised by Mr. McDougall. 

 Each of these six yielded correlations of over 0'50, the coefficients in the case of 

 the last two being particularly high. An amalgamation of the results of the six 

 gave correlations with intelligence of 0'85 to 091 ; and these figures are distinctly 

 higher than those for the estimates of one teacher with another's, or with the 

 results of examinations. 



Further experiments have since been made in Liverpool at a mixed secondary 

 school and at a secondary school for girls. The main object of these was to 

 investigate three problems suggested by some of the limitations of the foregoing 

 investigation— viz., how far such tests are affected by difference in sex, how far 

 they can be undertaken with success by teachers untrained in a psychological 

 laboratory, and how far they can be carried out as mass-experiments with num- 

 bers of children simultaneously instead of singly upon individuals. Tests have 

 also been added to represent processes of the highest mental level — abstraction, 

 judgment, inference, perception of relations — a level untouched by the previous 

 research. The results indicate that, as compared with simple sensory or motor 

 tests, tasks involving higher and more complex processes are vitiated to a far 

 less extent by difference of sex in the subjects, absence of special training in the 

 experimenter, and the peculiar conditions of experiments upon children in class. 

 They also appear to possess the most intimate relations to intelligence. Tests, 

 therefore, of this type seem the more practicable for educational investigations 

 and sociological surveys upon a scale sufficiently extensive for statistical treatment 

 of the results. 



