Thursday, June 14, 2012

Schelling & Simpson

It occurs to me that there must be a way of using Schelling’s segregation game to explain how, in some cases, Simpson’s paradox crops up. In fact the outcome of the Schelling game - marked racial or social segregation between neighbourhoods despite the existence of only mild racial or social preferences in the individuals involved & the fact that no, or only very small numbers of, individuals wish or actively plan for such a result – is itself a kind of Simpson’s paradox.

This thought came to me when playing around with my own idea of how to represent forces or processes which do not necessarily involve spatial relocation – by using the application of colour, for example yellow to a grid of squares which are either red or blue to turn some of them green or orange; the problem is the same – to have a simple, visual method of showing why you cannot predict an outcome, or deduce the process(es) which brought about the particular distribution which is evident at a point in time from purely contemporaneous data, no matter how strong the correlations with other contemporaneous factors.

It also seems to me that this kind of approach could be fruitful when applied to questions of disease, especially those which we attribute to ‘lifestyle’ rather than attacks by, for example, external insults or  infectious agents.

If we could understand the way the process works so that the population ends up segregated into groups segregated by ‘obesity’ for example, when there are at best only a mild individual preferences for such a state of being, then we might at least stop blaming the fatties themselves & look for smarter ways of fixing the process, and being kinder to all ourselves as a result..