Flexible Spatial Multilevel Modeling of Neighborhood Satisfaction in Beijing

Abstract

This article develops an innovative and flexible Bayesian spatial multilevel model to examine the sociospatial variations in perceived neighborhood satisfaction, using a large-scale household satisfaction survey in Beijing. In particular, we investigate the impact of a variety of housing tenure types on neighborhood satisfaction, controlling for household and individual sociodemographic attributes and geographical contextual effects. The proposed methodology offers a flexible framework for modeling spatially clustered survey data widely used in social science research by explicitly accounting for spatial dependence and heterogeneity effects. The results show that neighborhood satisfaction is influenced by individual, locational, and contextual factors. Homeowners, except those of resettlement housing, tend to be more satisfied with their neighborhood environment than renters. Moreover, the impacts of housing tenure types on satisfaction vary significantly in different neighborhood contexts and spatial locations.

本文发展一个创新且弹性的贝叶斯空间多层级模型, 并运用北京的大规模家户满意度调查, 检视邻里满意度感知的社会空间变异。我们控制家户与个人社会人口属性及地理脉络效应, 特别探讨多样的房屋租赁类型对于邻里满意度的影响。本文提出的方法论, 透过明确解释社会依赖与异质性的效应, 为广泛应用于社会科学研究的模式化在空间上群聚的调查数据, 提供了弹性架构。研究解果显示, 邻里满意度受到个人、区位及脉络因素所影响。房屋持有者, 除了再安置的住宅之外, 较租赁者倾向更满意其邻里环境。此外, 房屋租赁类型对于满意度的影响, 在不同的邻里脉络与空间区位中具有显着的差异。

Este artículo desarrolla un modelo bayesiano espacial de nivel múltiple, innovador y flexible, para examinar las variaciones socioespaciales en la satisfacción vecinal percibida, usando un estudio de Beijing sobre satisfacción familiar a gran escala. En particular, investigamos el impacto de una variedad de tipos de tenencia de la vivienda sobre la satisfacción vecinal, controlando los atributos sociodemográficos familiares e individuales y los efectos geográficos contextuales. La metodología propuesta ofrece un marco flexible para el modelaje de datos del estudio, concentrados ampliamente de manera espacial, que se usan en la investigación de las ciencias sociales, explícitamente tomando en cuenta la dependencia espacial y los efectos de la heterogeneidad. Los resultados indican que la satisfacción del vecindario es influida por factores individuales, locacionales y contextuales. Los propietarios, excepción hecha de los de viviendas de reasentamiento, tienden a mostrarse más satisfechos con su entorno vecinal que quienes son arrendatarios. Más aun, los impactos de los tipos de tenencia de la vivienda sobre la satisfacción varían de manera significativa en diferentes contextos vecinales y localizaciones espaciales.

Key Words:

• Beijing
• housing tenure
• multilevel modeling
• neighborhood satisfaction
• spatial statistics

关键词:

• 北京
• 房屋租赁
• 多层级模式化
• 邻里满意度
• 空间统计学。

Palabras clave:

• Beijing
• tenencia de la vivienda
• modelado a nivel múltiple
• satisfacción vecinal
• estadística espacial

Acknowledgments

The authors are grateful for the comments of the editor and the reviewers, which have improved the article. They also thank Professor Wenzhong Zhang from the Chinese Academy of Sciences for providing the valuable data used in this article.

Funding

This work was funded by the National Natural Science Foundation of China (Grant No. 41601148) and the Fundamental Research Funds for the Central Universities (Grant No. 2015NT17).

Notes

1 We acknowledge that there are other approaches to modeling spatial dependence, such as spatial econometrics, geostatistics, and other types of CAR models (e.g., Anselin 1988; Haining 2003; Banerjee, Carlin, and Gelfand 2004). We use an LCAR model because it has been shown to be more reliable than other CAR models (Lee 2011).

2 The order of the relative importance of neighborhood environment domains for each respondent is presented from 1 (least important) to 6 (most important). The weights assigned to each category are 5 percent (least important), 10 percent, 14 percent, 19 percent, 24 percent, and 28 percent (most important), respectively, following Zhang, Yin, and Zhang (2006). We also tried other weighting schemes but the modeling results remain similar.

3 For analyzing satisfaction levels of each individual dimension of neighborhood environment, it is arguable that an ordinal response model should be employed. As discussed earlier, though, the study is interested in the sociospatial variations of overall neighborhood satisfaction, which approximate well to a normal distribution. The development of a Bayesian spatial multilevel ordinal response model is, however, on our research agenda for appropriately analyzing individual domains of neighborhood satisfaction.

4 Residence length is based on two survey questions. The first is a binary question asking whether the respondent had lived in the current residence for more than ten years. If the answer is “no,” the respondent was further asked when he or she moved into the current residence. Therefore, residence length in our study is a right-censored variable. We extract two variables to capture the effect of residence length. The first is residence length (<10), which is a right-censored variable with a value of 10 indicating residence length above ten years. The second is a dummy variable, residence length (>10), in which one indicates residence length greater than ten years.

Additional information

Notes on contributors

Jing Ma

JING MA is an Assistant Professor in the School of Geography, Faculty of Geographical Science, Beijing Normal University, Beijing 100875, P.R. China. E-mail: [email protected]. Her main research interests include environmental justice, health inequality, spatial microsimulation, travel behavior, and transport carbon emission.

Yu Chen

YU CHEN is an Assistant Professor of Chinese Studies in the School of East Asian Studies, University of Sheffield, Sheffield S10 2TD, UK. E-mail: [email protected]. Her research interests include China's urbanization and urban development, migrant labor, and housing markets and policy.

Guanpeng Dong

GUANPENG DONG is an Assistant Professor of Geographic Data Science at Department of Geography and Planning, University of Liverpool, Liverpool L69 7ZQ, UK. E-mail: [email protected]. His core research interests include spatial and spatiotemporal statistics and multilevel modeling development and application, environmental evaluation, and urban economics.