Geographical Analysis (2019) 0, 1–22
doi: 10.1111/gean.12224
© 2019 The Ohio State University
1
Special Issue
Neighborhood Dynamics with Unharmonized
Longitudinal Data
Fabio Dias
1
, Daniel Silver
2
1
Department of Mechanical & Industrial Engineering, University of Toronto, Toronto, ON Canada,
2
Department of Sociology, University of Toronto, Toronto, ON Canada
This article proposes a novel method for data-driven identifcation of spatiotemporal
homogeneous regions and their dynamics, enabling the exploration of their composition
and extents. Using a simple network representation, the method enables temporal
regionalization without the need for geographical harmonization. To allow for a transparent
corroboration of our method, we use it as a basis for an interactive and intuitive interface
for the progressive exploration of the results. The interface guides the user through the
original data, enabling both experts and nonexperts to characterize broad patterns of
stability and change and identify detailed local processes. The proposed methodology is
suitable for any region-based data, and we validate our method with illustrative scenarios
from Chicago and Toronto, with results that match the established literature. The system is
publicly available, with demographic data for over forty regions in the USA and Canada
between 1970 and 2010.
Introduction
Neighborhoods have increasingly become a central concept in social research and targets for
social policy (Sampson 2012; Looker 2015; Stone et al. 2015; Galster 2019). To be sure, a focus
on neighborhoods extends to the formative period of the modern social sciences (Abbott 1997).
Recent interest has at least partly been rekindled through newly available longitudinal demo-
graphic data sets (Logan, Xu, and Stults 2014; Manson et al. 2017), convenient computational
tools (Rey et al. 2018), and new sources of data (Poorthuis 2018).
Yet new challenges have also emerged, especially at the convergence of research on neigh-
borhood effects and neighborhood dynamics. Neighborhood effects research assumes knowledge
about the nature and scope of “the neighbourhood” that presumably shapes individual outcomes
(Kwan 2018; Shelton Poorthuis 2019). Concurrently, researchers note that neighborhoods are
not necessarily fixed containers in which other processes occur, but themselves dynamically
evolve (Delmelle 2017; Reades, Souza, and Hubbard 2019; Li Xie 2018). The result is to open
Correspondence: Fabio Dias, Department of Mechanical & Industrial Engineering, University of Toronto,
Toronto, Canada.
e-mail: fabio.dias@gmail.com
Submitted: June 15, 2019. Revised version accepted: October 21, 2019.