Multilevel Modeling

11. Multiple Spatial Contexts

Ascertaining the relative importance of different spatial scales can, after taking into account (individual) compositional effects, provide important clues to the level ‘at which the action lies.’ A multilevel framework is ideally and readily suited to this task. Thus, underlying Figure 8 is a multilevel model based on a three-level structure of individuals (level-1) nested within neighborhoods (level-2) nested within regions (level-3). The micro model can be written as:

 13) yijk = β0jk + β1x1ijk+e0ijk

with an additional subscript to represent the regions. In addition, there would be a macro model at the neighborhood level (level-2):

14) β0jk= β0k+u0jk

where, β0k, is the poor health proportion for region k; and u0jk is the differential for the jth neighborhood in the region. There would also be a macro model at the region level (level-3):

15) β0k= β00k

where, β0 is the average poor health score; and υ0k is the differential for the kth region, to form an overall three-level ‘random-intercepts’ model:

16) yijk = β0+ β1x1ijk+(υ0k+u0jk+e0ijk)

Depending on the relative size of the neighborhood and region level variance terms (σ2uo) and (σ2υo), respectively, that summarizes the place-specific differentials at each level, this model would produce one of the patterns shown in Figure 8.

<a href="/longdesc/Subu_Sec11_Exer5.html">Flash Description</a>

Flash is not available on mobile devices. Please view the Flash Description.