It seems like a long time since I worked on this paper, but I am excited that it will be coming out soon in the Annals (of the Association of American Geographers for you non-geography types). My contribution was to use microsimulation to estimate the likely effects of future population change on neighborhood segregation.
We were looking at the role played by mixed-race households in shaping segregation outcomes at the neighborhood scale. Mark, Richard, and Steven have an ongoing project that looks at these issues in every way imaginable using individual level census data. For this paper though, we wanted to project impacts into the future, specifically a future in which we suspect the prevalence of mixed-race households will have increased substantially. This poses two problems: 1) we don't have the data since the changes haven't taken place yet, and 2) we don't know what role mixed-race households are likely to play in neighborhood change. This is where microsimulation comes in.
Generating a simulated population and grouping it into households across a wide range of population and mixing distributions is relatively straightforward. Rather than assuming a specific value for the rate of mixed-race household formation, I ran thousands of simulations using every conceivable parameter value to generate a sense of the solution space for our model. This is one of the real advantages of microsimulation--it allows us to generate a point estimate (the simulation mean) and also a sense of how the model behaves under extreme cases.
The second problem is a matter of understanding the ways that mixed-race households perceive, and are in turn, perceived by their neighbors. There is an important line of research on neighborhood segregation initiated by Thomas Schelling that examines the capacity of a preference for not being a racial minority in your immediate neighborhood to lead to segregation in the aggregate. But what happens in mixed-race households? Are individuals within such households separated out into their distinct racial categories or are they treated differently by different populations? Furthermore, how are the preferences of mixed-race households different from single-race households? Do they seek diversity in their surroundings? Do they identify with the minority population? With the majority population?
The animated .gif above shows snapshots of a single simulation run as the households sort themselves based on preferences. In this simulation run, mixed-race (blue) households seek diversity in their local neighborhoods and thus end up filling a sort of buffer role between the single-race households.
The larger findings of the paper indicate that an increased presence of mixed-race households is likely to drive down global measures of segregation while leaving the experience of single-race households essentially unchanged. This is a disturbing outcome since it suggests that we may see gains (in terms of reduced segregation) that do not result in any increased level of contact or association for households of different races.