Friday, 4 April 2014

Some thoughts on mapping spatial patterns of deprivation

In my research into the geography of deprivation across the UK, I frequently use maps to illustrate the spatial patterns associated with the areas identified as 'least' or 'most' deprived according to official indices such as the Scottish Index of Multiple Deprivation or the English Indices of Deprivation. Lots of other people do similar kinds of things, including mapping gurus such as Oliver O'Brien from UCL (his work is much nicer). A recent example is shown below, which I also tweeted this week. It's difficult to know exactly how people will interpret such maps, particularly when they are only seeing them on twitter without much in the way of context being provided, so this short blog fills some of the gaps and discusses some wider issues.

In previous academic papers (e.g. Urban Studies, 2009; Regional Studies, 2012; Local Economy, 2012) I've written about deprivation quite a bit, and on the need for the debate to centre not just on 'deprived' areas but more widely upon the wider dynamic of socio-spatial inequality. It's a shame that the focus is still very much on 'poor' or 'deprived' areas so in an attempt to draw attention to the urban inequalities which exist across the UK I attempt to illustrate the socio-spatial disparities within different cities. I also did this in a report on Sheffield from 2011 where I tried to draw attention to the socio-spatial inequalities within English cities, as shown below. It's not that concentrated deprivation isn't a problem (far from it) but rather that it's part of something much bigger.

These kinds of maps do draw attention to the general issue but of course they can lead to all sorts of other conclusions and claims because as we know, maps are an abstraction from reality and they do not represent an absolute 'truth'. These maps simply colour small areas within cities according to how they are classified by a government metric which attempts to say how 'deprived' places are. This may be a dubious practice in many respects, but it is woven into the fabric of how places are understood in a policy context and how problems are defined. It's important that we understand what this kind of mapping allows us to say and what it does not. Some of this is covered on the 'What does it all mean' tab of my Scottish Index of Multiple Deprivation 2012 map site, but I want to make a few more points here...

1. Colours. They are not intended to match up to any political party but some people inevitably make such inferences. The maps say nothing about the causes of the patterns, or who is responsible for them. But it doesn't stop people from talking about it and that's no bad thing. There is a lot more that could be said about colour choice but I'm going to leave it there for now.

2. The trouble with choropleth maps. Maps shaded according to some value (such as deprivation rank) present a misleading picture in a number of ways but two important ones stand out here: a) not all people in the area are 'deprived' or 'non-deprived'. This is the classic 'ecological fallacy' issue at work - the third paragraph here says more about that; and b) the shaded zoned themselves cover much wider areas than people actually live in so a big blue or red area gives the impression of a lot of something, when in fact the population of larger spatial units is similar to the smaller ones (as it often is with LSOAs or Data Zones in the examples above). 

3. The sometimes arbitrary nature of local authority boundaries. Places like Leeds are often said to be 'overbounded', whereas Manchester is 'underbounded'. This means that the local authority boundary either extends beyond the core urban area or it doesn't include much of it at all. So, in the cases of Manchester and Liverpool above if you were to extend the boundary of the map you would see more areas that are not so deprived. However, the point here is that local authorities have to deal with the financial, social and spatial implications of these patterns. What happens beyond the boundary is not part of their remit - even if it does impact upon them. The boundaries may be arbitrary in some respects but they have very real implications.

4. Why not take a different approach? A good idea, and one that Oliver O'Brien in particular has been very successful with. If we only look at where people actually live then we get a more realistic (but still not 100% accurate) picture of the spatial patterns associated with deprivation. This can be done using dasymetric mapping, where we assign the attributes of areas to individual features. This isn't a perfect definition, and the technique itself can lead people to assume a higher level of accuracy than can actually be obtained from the underlying data but it has advantages over standard choropleth maps in relation to depicting the places where people actually live or work. See also Neal Hudson's London tenure map in this style. The new OS Open Data VectorMap District buildings layer for Great Britain allows us to do this, so I've produced an example map for Glasgow based on the one in the first image above. This time you can only see the areas where there are buildings (though many are not residential properties).

Is the map above more 'truthful' than the normal choropleth? Probably not. However, this is all irrelevant if we aren't concentrating on the underlying patterns we're trying to draw attention to in the first place. The point is that in cities like Liverpool, Manchester and Glasgow the high levels of deprivation/poverty/disadvantage sit in stark contrast to areas at the other end of the scale. Also, places that we think of as 'deprived' are often far from it - as Peter Matthews might also argue. It's this kind of inequality which I'm attempting to highlight with my mapping - though I do of course like a nice looking map (I've also produced more than a few stinkers in my time). The point of all this? I hope that these maps can start a conversation about the underlying issue. I'll end with an extract from my 2012 Regional Studies paper on the issue...