The Advantages of Comparative LISA Techniques in Spatial Inequality Research: Evidence from Poverty Change in the United States

Abstract

Although scholarship regarding spatial inequality has grown in recent years, past research has seen limited use of spatial statistics—let alone comparison between spatial statistical techniques. Comparing and contrasting the application and use of spatial statistics is valuable in research because it allows for more precise identification of spatial patterns, and highlights results that may be hidden when only using a single method. This study serves as a demonstration on how the use of multiple LISA statistics can benefit inequality related research. Analyzing changes in county level poverty in the rural United States from 1990 to 2015 serves as a tool to demonstrate these techniques and this study examined how the geographic distribution of poverty has changed, and well as if there is evidence of diffusion effects. The three featured techniques utilized Local Indicators of Spatial Association (LISA) statistics. The techniques are Bivariate LISA, LISA Cluster Transitions, and LISA Diffusion Transitions, with the last technique specifically designed for this study. Each technique varies in how it reports the changes in the spatial structure of poverty. Bivariate LISA and LISA Cluster Transitions are complementary to each other—with the former technique providing a single global statistic while the latter is more easily interpretable. Diffusion Transitions show how the highest and lowest values of a variable may be spreading over time. The study also produces new findings regarding rural poverty, with poverty in Mountain-West and rural Sun Belt counties on the rise. Analysis shows a diffusion effect for poverty in Southeastern metropolitan fringe counties.