Changes in Human Performance with Age 165 



studies because these normally relate age and work done and 

 so control them automatically. They have, however, their 

 own problems of control which depend upon the type of 

 industrial data used. 



(a) Production records or studies of actual work done 

 provide what are at first sight the most valuable and informa- 

 tive data on performance, but are subject to serious limitations. 

 The most important of these is that a marked fall in produc- 

 tivity would be unlikely to be tolerated for long by either the 

 employer or the man concerned, so that if productivity falls 

 much as a man becomes older he is likely to move or be 

 moved to another job. Older people left on the job will thus 

 be unrepresentative of their age group, and inferences based on 

 their performance will be unduly favourable to older people. 

 Selection in the opposite direction by promotion will also 

 occur, although probably less often, and may result in age 

 trends appearing unduly unfavourable to older people. We 

 cannot without further evidence be sure in any particular case 

 that these two selective trends balance. Progressive selection 

 of this kind is not overcome by a longitudinal study of men 

 who have been on the same job for a long period of years, 

 because the fact that they have been so may imply that they 

 are a selected group whose performance has declined or risen 

 less than that of others. 



Studies of work done in industry, whether cross-sectional 

 or longitudinal, are thus valid only if labour turnover and 

 transfers to other jobs are negligibly small. Ideally they 

 require also a substantial group of people well spread over 

 the age scale and all doing exactly the same work, since what 

 are minor variations of work from the industrial point of view 

 may imply major differences in the psychological demands 

 the work makes. This requirement is very seldom satisfied 

 in industry but can be at least partly overcome by relating 

 each man's performance to the average of men in a parti- 

 cular age range — say the thirties — or in the whole group who 

 are doing the same work. Wackwitz (1946) has used such 

 standard measures to compare one operation with another in 



