My doctoral thesis originated out of the Belfast Social Malaise Project, on which I was employed for two years as a research assistant. As the project progressed, I became more concerned about the scientific ramifications of the complex multivariate quantitative techniques that we were using. I therefore decided to explore the implications in more detail for my doctorate thesis. Working part-time, this took a total of 7 years to complete (1971-1978).
The two most probable reasons for analysing multiple social deprivation are (a) to objectively identify which areas are most deprived possibly with a view to designating areas for additional resources in pursuance of a policy of positive territorial discrimination, or (b) to gain a better understanding of the processes resulting in areas of multiple deprivation. The basic argument of the thesis was that the use of complex multivariate techniques in such studies tends to create an impression of scientific rigour and objectivity, whereas in reality they actually disguise a high degree of subjectivity and some fundamental technical problems with regard to both objectives.
There are at least three possible ways to conceptualise multiple deprivation. The first is as a collective term for a variety of social problems; the second is as the common underlying cause of different types of social problem; and the third is as a holistic concept which incorporates a complex web of causes and effects. The conceptualisation employed should be reflected in the choice of method. The first conceptualisation lends itself to principal components analysis and various other methods which may be used to combine individual measures into a weighted composite index; the second conceptualisation is more appropriately modelled using factor analysis; whilst the third conceptualisation requires some form of causal modelling. The Simon Blalock approach was favoured in the thesis. In many studies using these techniques the conceptual implications of the statistical method are not explicitly explored. The thesis explored some of the limitations implicit in the methods used to implement each of the three conceptualisations.
The thesis also explored a wide ranges of more technical problems associated with spatial correlation (the basic building block for most multivariate techniques) such as data quality, the methods used to quantify variable attributes, the implications of unfulfilled statistical assumptions, problems of ecological inference, and a wide range of other issues related to the spatial framework used for the analysis (e.g. the modifiable areal unit problem).
The thesis concluded with a number of suggestions for improved practice, but unfortunately there are no simple solutions for most of the problems raised.
The thesis ran to over 700 pages. I doubt if anyone other than the examiners ever read it, but it makes a very effective door-stop.
Return to the main Social Geography page