ABSTRACT
The structural approaches to people–environment interaction described in Chapters 2 and 3 and the behavioural approaches outlined in Chapters 4 and 5 have a number of shortcomings. For example, spatial interaction models such as the gravity model and similar formulations may provide a reasonable description of the empirical regularity that is to be seen in the behaviour of aggregates of individuals but they offer little by way of explanation. They therefore give little insight into why individuals come to act in a given way. Aggregate statistics, in other words, frequently mask the variability in behaviour that exists between different individuals and between different groups and they thereby cloud the causes of such behaviour. Likewise, structural Marxist approaches adopt a holistic rather than an individualistic stance. They place great faith in transcendental structures as the motivating force in the functioning of society and they subjugate individuals to a passive role in a system dominated by the relationship between the mode of production and the associated superstructure. Spatial structural approaches, again reviewed in Chapter 2, fare little better. An emphasis on social space as a macro-scale phenomenon, and on the identification of social areas, runs the risk of directing inquiry at an artificial target in that many of the areas about which information is gathered (notably census districts of one sort or another) are merely administrative units designated for the convenience of census organizers. Many of them have little significance in terms of either behaviour or social belonging. Worse still, to focus on units such as census districts can encourage geographers to commit the ecological fallacy. This involves two things:
Attributing the characteristics of an area to all the individuals living within that area even where there is ample evidence to show that such ‘diagnostic’ characteristics are carried by only a minority of residents (Hamnett 1979);
Proceeding with a style of analysis whereby areal chacteristics are correlated with each other, often creating fallacious impressions of causal relationships (Openshaw 1984).
