ABSTRACT
When gathering or selecting data for analysis, a choice has to be made between total enumeration, i.e. counting or measuring all the occurrences of the particular phenomenon, and sampling, i.e. counting or measuring a selected sub-set of these occurrences. In the former case, our analyses will provide an absolute statement on this totality of data, and–within the accuracy of the observations–our statement will be ‘correct’. If the total enumeration is really of all the occurrences that exist, then this is satisfactory. If, on the other hand, we have totally enumerated all the employed population in just one town (recording their occupations, income, etc.), then our results are relevant to that town alone. They do not provide us with results from which we can generalize about conditions in, for example, all other towns of the same size in the same country. Yet it is this very generalization which is necessary if geographers are to attempt to understand and explain relationships in the real world, and which has been characteristic of developments in the subject since the 1960s.
