ABSTRACT
In general, ecological studies are characterized by being based on grouped data, with the groups often corresponding to geographical areas, so that spatial aggregation has been carried out. Such studies have a long history in many disciplines including political science (King, 1997), geography (Openshaw, 1984), sociology (Robinson, 1950), and epidemiology and public health (Morgenstern, 1998). Our terminology will reflect the latter application; however, the ideas generalize across disciplines. Ecological studies are prone to unique drawbacks, in particular the potential for ecological bias, which describes the difference between estimated associations based on ecological-and individual-level data. Ecological data are a special case of spatially misaligned data, a discussion of which is the subject of Chapter 29.
