174 Irving Lorge 



greatly enhance the value of data collected by any method. 

 In personality appraisal, retrospective evidence about the 

 parents' values and attitudes may be clues for the sources of 

 the underlying motivations in their offspring. The influences 

 of early learning upon later learning may have profoundly 

 affected not only the behaviour of the child as a child but also 

 that child as a maturing, and mature, adult (cf. Hebb, 1949). 

 The limitations of retrospective data must be recognized : such 

 evidence may be somewhat less reliable than data collected 

 more specifically and more objectively. Retrospective data 

 may be in error to some degree depending upon the recency 

 of the event, the frequency of its occurrence, or the tendency 

 to schematicize the evidence according to some preconception 

 about what should be reported. 



The method for accumulating evidence, therefore, intro- 

 duces some special problems in the interpretation of evidence. 

 A second sort of methodological concern relates to the 

 representativeness of the samples of the population of dif- 

 ferent ages in cross-sectional study, or of a cohort for a longi- 

 tudinal study. Wesman (1955), indeed, indicated the scope of 

 problems involved in obtaining a representative sampling of 

 adults at different ages for standardizing an intelligence test. 

 The persons to be tested must be selected to be representative 

 of the population by sex, educational attainment, occupation, 

 residence (urban-rural), geographic region, etc. In addition, 

 the person so chosen must be willing to co-operate by taking 

 the test, and the examiner must be the kind of person who can 

 keep the willing subject sufficiently motivated to make him 

 willing to continue for a fairly long time, an experience that 

 may be occasionally frustrating. 



Since the optimal resolution of these complex factors is 

 achieved but rarely, it is obvious that cross-sectional, as well 

 as longitudinal, studies suffer from some kind of bias in the 

 sampling. In appraising the intelligence of a cohort (Terman 

 et al., 1947; Owens, 1953), the bias has been toward appraising 

 the ability of either the more intelligent or the better-adjus- 

 ted, or the ability of the institutionalized old or mentally 



