There has been quite a bit of recent activity around the release of the 2010 Indices of Deprivation (for England) as we all try to get to grips with what it says about the state of neighbourhood and communities throughout the country. (My colleague Alasdair Rae has posted a load of fascinating stuff on his blog.) A lot is, of course, at stake because the IMD (as it’s known) inevitably winds up being used in all sorts of ways, including as a basis for prioritisation of policy and funding attention.
But it has its limitations, like anything that essentially aims to reduce complex social phenomena and constructions into a score or rank. We’re all tempted, as well, to use the IMD to look at trends and to compare places over time — despite the warnings that that is not what the IMD was designed for, or even suitable for. One of the good things about the more recent editions of the IMD is that it includes much more up to date information on a variety of different indicators, grouped into reasonably useful ‘domains’. Getting at the individual scores for these domains means we can begin to do more with the data as a snapshot (and resist the temptation to just look at the overall score and how it has changed over time).
As someone interested in GIS and mapping techniques I’ve always been attracted to the small-area data contained in the IMD. Yet, the number of different variables it contains can sometimes frustrate the possibilities of looking at subtle differences in the qualitative nature of deprivation between places without recourse to more complex statistical techniques (which become less immediate and less intuitive).
So using multivariate mapping techniques offers some way forward. The attached example, which I’ve called a ‘continuous multivariate map’, was produced by separating three of the IMD’s domain scores (income, living environment, crime) into different colours in the RGB (red-green-blue) colour model. Each place is therefore given an overall colour that is dependent on the mixing of the three colours (or the three domain scores). All three domains are equally weighted. A place that is ‘red’, for example, is deprived mainly in terms of income but not the other two domains. Places coloured white score highly on all three domains. Darker colours score lower and are comparatively less deprived. And so on.
The results are mildly interesting because they show how the mix of facets of deprivation can vary within a city. The example map shows Sheffield, a notoriously divided city in social and economic terms. The affluent west end comes out quite clearly as not being particular deprived – if it is, it is in terms of the living environment of those suburbs closer to the main employers (universities and hospitals). A band of blue, again near to the universities, shows neighbourhoods where poor living environment and problems with crime appear to coexist, but where in general low incomes are not so much of a problem. Contrast this with the east end of the city where higher levels of deprivation manifest themselves in quite different ways. The north-eastern suburbs score highly on all three domains. Some of the south-eastern suburbs, which have quite attractive built environments and generous built-form standards, are nevertheless afflicted by high crime and low incomes.
Of course none of this is more than an interesting exercise in visualisation. It tells us nothing about the complex links between these domains, nor of causality. But moving away from a “one number” approach to measuring and visualising deprivation can only be a good thing.
The original inspiration for this type of map came from Stan Openshaw’s Census Users’ Handbook, 1995, Geoinformation International, Cambridge. See in particular Danny Dorling’s chapter, ‘Visualising the 1991 Census’.