164 A. T. Welford 



of the results of longitudinal studies using material of the 

 level of an intelligence test on more than one occasion. 



With repetitive tasks such as sensory-motor skills this 

 difficulty can be at least largely overcome by training the 

 subject before each test until he has ceased to improve with 

 practice. Unfortunately, however, this may take a long time 

 and make the test unduly time-consuming. 



With problem-solving and other tasks where "insight" can 

 be gained the task cannot be used more than once and subse- 

 quent tests have to be done with similar, but different 

 problems. The problems must either be equated for difficulty 

 or must be presented to different subjects in different orders 

 in a balanced design. Even with different problems, however, 

 substantial learning effects in the way of tackling particular 

 types of problem are likely to enter, so that the longitudinal 

 method would seem seldom appropriate for the study of per- 

 formance at this kind of task. 



(b) The obvious objection to longitudinal studies that a 

 very long wait is required if the study is to span a substantial 

 part of a man's life, can be overcome in some cases by using 

 animals. Where animals are unsuitable, it may be possible to 

 resort to a mixture of the longitudinal and cross-sectional 

 methods. A group of subjects covering a substantial age range 

 is tested and the same group is tested again after a period of 

 years. For some functions this period can be as short as five 

 years, as with Weston's (1949) experiment already mentioned. 

 For most tasks, retesting subjects at intervals over a period of 

 ten to twenty years would be better. This method has been 

 little used because although it avoids the necessity of ex- 

 tremely long waits it still requires, with preparation before 

 the first test and analysis of results after the second, a term of 

 years longer than any research team on ageing has yet been 

 able to look forward to with confidence. 



Industrial Studies 



Problems of equating subjects for education, occupation and 

 other background features do not usually arise in industrial 



