This chapter examines differences between delinquents and nondelinquents, between good and bad boys, between middle- and lower-class boys. Theory involves conjectures about relations among variables for a given set of units, usually individuals. For example, if lower-class areas are known to have higher crime rates than upper-class areas, the theorist may infer that individuals with low incomes are more likely to commit crimes than individuals with high incomes. Goodman and others have shown that it is sometimes possible to derive estimates of the individual relation from group data, and Herbert Menzel has pointed out that ecological correlations are often necessary in the study of groups and other aggregates. The investigator wishing to compare on a variety of variables boys nominated by teachers as good boys with boys nominated as potentially delinquent cannot be expected to describe what may be a peculiar combination of traits for each boy.