184 Discussion 



another issue, the estimation of health in individuals. We have tried to 

 study a representative community in New York State. We wanted to get 

 a relation between responses to questions about physical and mental 

 symptoms and psychological behaviours and health status. We enlisted 

 the co-operation of some 30 physicians. All agreed that they can recog- 

 nize ill-health but none was willing to report a measure of the adequacy 

 of the health status of the individual being appraised. Your point, 

 however, is well made. Wetzel, in the United States, demonstrated 

 that certain measures of health are related fundamentally to the school 

 success of children, and that the child's health can be estimated, in part, 

 by the fact that he is not achieving adequately in school ; this is recog- 

 nized as ill-health. On medical treatment, which may be nutritional 

 or corrective in another way, the school performance does, in many 

 instances, improve and the averages definitely do go up. The number of 

 variables that must be kept under control in the appraisal of performance 

 or of intelligence or of emotion are so complex that sometimes it is very 

 courageous of a psychologist to attempt to do anything. After Prof. 

 Danielli's contribution, I have a feeling that you people are in the 

 same spot. Perhaps by the compounding of assays in the penumbra of 

 shadowy ignorances, we may learn something. 



Welford: I entirely agree, but I do think there is one point which is 

 very often missed and that is that the type of sample required and what 

 we need to control vary greatly according to the purpose of our study. 

 If we have simply an applied aim, for instance, to further the employment 

 of older people, we need to study the older people as we find them in the 

 existing population; we need then a truly representative sample. We 

 might perhaps complicate this by saying we need a representative sample 

 of the population as it will be ten years hence. The sample would still 

 be fundamentally of the same type. For a study of ageing, as such, 

 we do not need a representative sample; we need one in which people 

 of different ages are comparable. In a representative sample, death, 

 disease and so forth all take their toll ; but this does not affect the validity 

 of the sample. In the comparable sample such factors are important and 

 their effects must be controlled. 



Friedman: I wonder whether there would be any validity in one 

 additional approach to the two which Dr. Lorge has mentioned in his 

 combined approach. The problem, it seems to me, is in part complicated 

 by the fact that the testing method itself is a variable in the experimental 

 arrangement. Could that be smoothed out by the ordinary biological 

 procedure of selecting groups which were even on the test, i.e. in which 

 the test performance was the same, and then working from that to a study, 

 say, of ageing or other factors ? 



Lorge: That method has been utilized in some testing of the intellect- 

 ual ability of adults at age 34, for whom scores on mental ability at age 14 

 were available. Individuals were equated for age- 14 scores and measured 

 again at age 34. The test results at age 34 were different depending 

 upon the amount of education intervening between the fourteenth and 

 thirty-fourth birthday. The greater the amount of schooling in the 

 twenty-year interval, the greater the difference in the age-34 perform- 



