A spatial conditional approach to modelling unemployment and poverty in the counties of Missouri


Session:

Accounting for Dependence in Small Area Estimation

Authors:
Noel Cressie (National Institute for Applied Statistics Research Australia (NIASRA), University of Wollongong, Australia) (Speaker)
Rhys McDonald (National Institute for Applied Statistics Research Australia (NIASRA), University of Wollongong, Australia)
Andrew Zammit Mangion (National Institute for Applied Statistics Research Australia (NIASRA), University of Wollongong, Australia)
Abstract:

Efficient multivariate estimation for small areas requires that both the multivariate and the spatial nature of the dependence be recognised. However, building a dependence model for all possible combinations of two or more variables and their locations in a discretely indexed domain is not easy, since any covariance matrix that is derived from such a model has to be nonnegative-definite. In this talk, a conditional approach for multivariate-spatial-statistical-model construction is given that captures interactions between variables in a natural way and allows non-negative definiteness to be established easily. Starting with bivariate spatial models, we give its connection to multivariate models defined by spatial networks. Small-area studies often have a strong spatial component, and they should recognise the interaction between different variables that describe the socio-demographic phenomenon under study. Our conditional approach is demonstrated on a bivariate dataset from the American Community Survey of unemployment and poverty in the counties of Missouri, where the two variables are seen to interact asymmetrically.