Why segregation studies are stuck in Chicago By David Manley (Univeristy of Bristol)

The way in which we study segregation hasn’t really moved forward since the 1950s. Much of the literature that we use and the measures that we employ owe their conception to the Chicago urban school of thinking.  The thing is, these traditional ways of understanding segregation do not fit the modern urban environment. Of course, they didn’t really fit the 1950s urban landscape in the United States either but that was not the point: a measure that described the proportion of people that wold need to move to gain an equal distribution of whatever characteristic was of interest was sufficient. In the US that characteristic was racial inequality, and there wasn’t really a need to debate how the segregation had come about or what drove it: what they needed to know was how much segregation was there. Recently, scholars have noted that the city of Chicago isn’t particularly representative of US cities (no surprise), but that it also isn’t particularly representative of the other rust belt cities either.  Given that our understanding of segregation comes from a unique place at a particular time frame in history it is perhaps time to rethink how we investigated segregation more generally.
The most common measure of segregation is known as the Index of Dissimilarity. The thing is, it is biased. Studies by economists which simulated urban data have demonstrated that the Index of Dissimilarity has a tendency to record higher values of segregation than actually exist – that is to say it is upwardly bias. Actually, more recent work has suggested that it is more complicated than that and that the bias in the Index tends to move values towards 0.5 (or saying that to achieve an urban landscape with no segregation around half the population would need to move home). This is troubling because it suggests that what we thought we knew about segregation is wrong – or at least not as clear as we thought. However, even if the index was not biased there is a bigger problem that pervades the studies that have used it. I mentioned above that in the US segregation studies were, rightly, concerned with racial inequality. The modern segregation literature is much more diverse than this and not only considers segregation along racial or ethnic lines but also include multiple other factors as well: class; social; economic, and; cultural to name a few and the index has been deployed to measure all of these and more beside. But in doing so researchers trying to understand how segregation is playing out in our urban environments are forced to over simplify. In the urban environment people do choose where to live (or get forced to live) purely because of their ethnicity or social status or age. There is a complex interaction between these factors (and others) which can serve to intensify or the resulting distribution of individuals within neighbourhoods. But the index has no way of help us determine which of the factors is the most important (remember that the index is bias so a higher value may not necessarily mean that there is more segregation, just that the index was bias) and nor can it help us understand how much age segregation there is once we have accounted for the level of educational segregation that occurs. Moreover, index values calculated for units of one size (say small neighbourhoods based on a couple of streets) cannot readily be compared with units of another size (say large neighbourhoods or even regions within a city) because the index is relative to the size of the units used. Crucially, this prevents us from identify the scale at which segregation is occurring.
What should we do about this, are all segregation studies doomed to repeat the same mistakes as before? No. Recent work undertaken with colleagues at the University of Bristol has developed an approach using multilevel models harnessing the idea that segregation is about the variability in the numbers of different groups within neighbourhoods, the greater the variable in the number of a group within neighbourhoods over a city the greater the segregation. Importantly, once measured in a modelling framework we are able to include multiple different types of segregation at once and even multiple scales. In doing this we might be able to further our understanding of one facet of the complex urban environment.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s