Skip to content

Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization

Research output: Contribution to journalArticle

Standard

Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization. / Wolf, Levi; Anselin, Luc; Arribas-Bel, Daniel; Mobley, Lee.

In: Annals of the American Association of Geographers, 24.05.2018.

Research output: Contribution to journalArticle

Harvard

Wolf, L, Anselin, L, Arribas-Bel, D & Mobley, L 2018, 'Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization', Annals of the American Association of Geographers. https://doi.org/10.17605/OSF.IO/KS6T3

APA

Vancouver

Wolf L, Anselin L, Arribas-Bel D, Mobley L. Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization. Annals of the American Association of Geographers. 2018 May 24. https://doi.org/10.17605/OSF.IO/KS6T3

Author

Wolf, Levi ; Anselin, Luc ; Arribas-Bel, Daniel ; Mobley, Lee. / Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization. In: Annals of the American Association of Geographers. 2018.

Bibtex

@article{8075a63863844d0ca3dcc038198ea726,
title = "Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization",
abstract = "Multilevel (or variance components) models have been applied in many areas of regional science, epidemiology and polimetrics. They are most often used to model treatment nonstationarity in policy regimes, a form of spatial process heterogeneity. Multilevel models with spatially-correlated components are increasingly used to model the presence of both spatial heterogeneity and spatial dependence. Previous treatments typically focus on studying a single model specification in detail, deriving its estimators, and demonstrating its properties in a case study. Instead, a general approach is possible. In this paper, a generic Gibbs sampler for spatially-correlated multilevel models is developed and its empirical properties examined in model of cancer screening. Critically, there is tension between spatial dependence and spatial heterogeneity terms in the model. This results in apparent ``growth'' in some multilevel models where the dependence terms overpower the effects of estimate regularization. This results in locally-smoothed estimates, which, depending on the strength of spatial dependence, may be both larger and more precise than their corresponding single-level standard linear models. Thus, shrinkage always applies, but not always from the standard single-level linear model. Due to the complexity of these tradeoffs between model fit, dependence, and heterogeneity terms, this formal attention is critical to building systematic and transferable understanding.",
author = "Levi Wolf and Luc Anselin and Daniel Arribas-Bel and Lee Mobley",
year = "2018",
month = "5",
day = "24",
doi = "10.17605/OSF.IO/KS6T3",
language = "English",
journal = "Annals of the American Association of Geographers",
issn = "2469-4452",
publisher = "Taylor & Francis Group",

}

RIS - suitable for import to EndNote

TY - JOUR

T1 - Spatially-Correlated Multilevel Models: A General Specification with Spatially-Local Regularization

AU - Wolf, Levi

AU - Anselin, Luc

AU - Arribas-Bel, Daniel

AU - Mobley, Lee

PY - 2018/5/24

Y1 - 2018/5/24

N2 - Multilevel (or variance components) models have been applied in many areas of regional science, epidemiology and polimetrics. They are most often used to model treatment nonstationarity in policy regimes, a form of spatial process heterogeneity. Multilevel models with spatially-correlated components are increasingly used to model the presence of both spatial heterogeneity and spatial dependence. Previous treatments typically focus on studying a single model specification in detail, deriving its estimators, and demonstrating its properties in a case study. Instead, a general approach is possible. In this paper, a generic Gibbs sampler for spatially-correlated multilevel models is developed and its empirical properties examined in model of cancer screening. Critically, there is tension between spatial dependence and spatial heterogeneity terms in the model. This results in apparent ``growth'' in some multilevel models where the dependence terms overpower the effects of estimate regularization. This results in locally-smoothed estimates, which, depending on the strength of spatial dependence, may be both larger and more precise than their corresponding single-level standard linear models. Thus, shrinkage always applies, but not always from the standard single-level linear model. Due to the complexity of these tradeoffs between model fit, dependence, and heterogeneity terms, this formal attention is critical to building systematic and transferable understanding.

AB - Multilevel (or variance components) models have been applied in many areas of regional science, epidemiology and polimetrics. They are most often used to model treatment nonstationarity in policy regimes, a form of spatial process heterogeneity. Multilevel models with spatially-correlated components are increasingly used to model the presence of both spatial heterogeneity and spatial dependence. Previous treatments typically focus on studying a single model specification in detail, deriving its estimators, and demonstrating its properties in a case study. Instead, a general approach is possible. In this paper, a generic Gibbs sampler for spatially-correlated multilevel models is developed and its empirical properties examined in model of cancer screening. Critically, there is tension between spatial dependence and spatial heterogeneity terms in the model. This results in apparent ``growth'' in some multilevel models where the dependence terms overpower the effects of estimate regularization. This results in locally-smoothed estimates, which, depending on the strength of spatial dependence, may be both larger and more precise than their corresponding single-level standard linear models. Thus, shrinkage always applies, but not always from the standard single-level linear model. Due to the complexity of these tradeoffs between model fit, dependence, and heterogeneity terms, this formal attention is critical to building systematic and transferable understanding.

UR - https://github.com/pysal/spvcm

U2 - 10.17605/OSF.IO/KS6T3

DO - 10.17605/OSF.IO/KS6T3

M3 - Article

JO - Annals of the American Association of Geographers

JF - Annals of the American Association of Geographers

SN - 2469-4452

ER -