178 Irving Lorge 



person may give a correct (from the point of view of the key) 

 response — but the child may have given a response on 

 the basis of reasoning and the adult on the basis of rote 

 memory. The credit is arbitrary but the significance of the 

 credits depends upon the criterion and the degree to which the 

 process is affiliated with that criterion. 



It may be asserted that the nature of the criteria for adult 

 intelligence, or intelligence in any age range, conditions the 

 nature of the sampling of the tasks to be included in a test. 

 Indeed, there may be some point in demonstrating that 

 measures of intelligence for children are highly predictive of 

 measures of intelligence in adults. To the degree that the 

 relationship is high, the inference would be that intelligence is 

 similar in the two groups, or that the criteria for it are the 

 same. Nevertheless, the very variety of tasks in intelligence 

 tests may be such as to reduce the relationship of measures of 

 intelligence over time. For instance, Thorndike (1933), 

 using the apparently homogeneous measure derived from the 

 Stanford Revision of the Binet-Simon, shows that the relation 

 between retest coefficients varies with the amount of time 

 between retests. For repeated Binet examinations given 

 within a single month the correlation is approximately 0*89, 

 and for repeated examinations five years apart the correlation 

 is approximately 0-70. Part of the difference in these relia- 

 bility coefficients is undoubtedly due to errors of measurement. 

 It is more likely, however, that a considerable part of the 

 difference is attributable to inconsistency of what the test 

 measures at different times. 



Anderson (1939) has emphasized the apparent inconsistency 

 of measures of intelligence derived from infant tests and from 

 the Binet. He concluded "Infant tests, as at present con- 

 stituted, measure very little, if at all, the function that is 

 called 'intelligence' at later ages." Balinsky (1941), in his 

 study of the factorial composition of the Wechsler-Bellevue 

 Intelligence Scale for age groups 9, 12, 15, 25-29, 35-44, and 

 50-59 years, concluded that mental traits change and under- 

 go reorganization over a span of years. This inconsistency 



