See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/350249527

Measuring destination-based segregation through mobility patterns:

Application of transport card data

Article  in  Journal of Transport Geography · April 2021

DOI: 10.1016/j.jtrangeo.2021.103025

CITATIONS READS

0 64

3 authors, including:

Sorath Abbasi
Hanyang University
2 PUBLICATIONS   2 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Measuring Sustainable Accessibility of Seoul View project

Carsharing In Seoul View project

All content following this page was uploaded by Sorath Abbasi on 22 March 2021.

The user has requested enhancement of the downloaded file.

 Journal of Transport Geography 92 (2021) 103025

Contents lists available at ScienceDirect

Journal of Transport Geography

journal homepage: www.elsevier.com/locate/jtrangeo

Measuring destination-based segregation through mobility patterns:

Application of transport card data
Sorath Abbasi a, Joonho Ko a, *, Jaehong Min b
a
Department of Urban and Regional Development, Hanyang University, 222 Wangsimni-ro, Seongdong-gu, Seoul, 04763, Republic of Korea
b
Transportation Systems Research Team, Korea Railroad Research Institute, 176 Railway Museum Rd, Uiwang, Gyeonggi-do, 16105, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Keywords: This study explores the level of segregation experienced by seniors, children/youth, and passengers with dis­
Social interaction space abilities, compared to normal-fare passengers at their trip destination when using public transportation. One
Segregation week’s travel records of public transit passengers were extracted from Seoul’s transport card data to compute
Social groups
dissimilarity and exposure indices, theoretically equivalent to those developed in segregation research, to cap­
Transport card data
ture destination-based segregation through mobility patterns. Additionally, a multigroup entropy index was
computed to measure diversity by assessing the social mixture of all passenger flows in a spatial unit. The results
revealed that segregation levels experienced by passengers based on their social groups are notably different
depending on the time of day and the day of the week. The computed exposure measure illustrates that the
potential interaction between the selected social groups and normal-fare passengers is relatively higher during
peak hours on weekdays. The results also show that subway stations provide more opportunities for interaction
among different social groups. These findings can contribute to a better understanding of social segregation
through mobility patterns as well as the effective quantification of the public transport network performance in
terms of providing an interaction opportunity for the groups.

1. Introduction and Shaw, 2011). In this case, mobility is considered to be the result of
location choices over time (Shen, 2019) as people’s residences are
The increased mobility of people with the emergence of enclave subject to the accessibility to the urban resources required in their daily
urbanism has reshaped the social systems in urban areas (Urry, 2007). routine. Consequently, the number of people who travel, commute, and
Consequently, the interaction between people occurs when urban spaces move for all sorts of activities have increased markedly. However, the
are configured for different activities that have a spatiotemporal nature. problem arises when mobility does not increase in the same way for all,
This added configuration suggests that new insights can be gained when contributing to mobility-related social exclusion for some social groups
commonly used spatial concepts take into account human mobility and (Fu and Gu, 2018; Kandt and Leak, 2019; Lucas, 2012; Lucas et al.,
incorporate various facets of time as indispensable elements (Le Roux 2016). This is possible due to enclave urbanism that benefits selective
et al., 2017). This notion addresses a wide range of social issues such as sites (typically those where elites live, work, and consume), resulting in
social segregation, environmental justice, and access to social services. the growing socio-spatial fragmentation of cities (Deville et al., 2014;
As a result, arguments have been provided to expand the focus of Sevtsuk and Ratti, 2010). Social inclusion or exclusion, therefore, should
segregation studies to a greater consideration of “the dynamic flows of be understood through the lens of mobility patterns to quantify social,
everyday life” both within and outside the fields of residential interac­ spatial, and temporal dimensions of segregation, thereby providing a
tion and daily experience (Atkinson and Flint, 2004; Ellis et al., 2004). multilayered perspective of how people are dynamically interconnected
For instance, although social segregation is conceived as the extent to over space and time.
which members of one social group live apart from others, people’s Place-based segregation measures, in this regard, have been mainly
segregation experiences are also affected by how much time they spend applied to investigate the cumulative potential of spatial interactions
in areas outside of their residential areas and how much social interac­ between any pair of questioned groups (Morency et al., 2011;
tion they have with other social groups there (Ellis et al., 2004; Wong Schönfelder and Axhausen, 2003). Considering people’s daily mobility

* Corresponding author.
E-mail addresses: [email protected] (S. Abbasi), [email protected] (J. Ko), [email protected] (J. Min).

https://doi.org/10.1016/j.jtrangeo.2021.103025
Received 11 November 2020; Received in revised form 24 February 2021; Accepted 12 March 2021
Available online 21 March 2021
0966-6923/© 2021 Elsevier Ltd. All rights reserved.
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

and investigating other spaces (e.g., school, work, or leisure spaces) in potential social segregation experienced by social groups that are rarely
segregation studies have provided a more accurate assessment of peo­ explored such as people with disabilities and children through their use
ple’s experiences of segregation (Wong and Shaw, 2011). For instance, of public transport. The inclusion of these diverse groups in this explo­
studies have shown that segregation also occurs in workplaces although ration can additionally provide a comprehensive picture of to what
people from different backgrounds are more integrated (Åslund and extent public transport offers an interaction opportunity for these groups
Skans, 2010). There are also claims that individuals often self-segregate to influence the social mix of passengers. Finally, the results of this study
in their leisure activities (Schnell and Yoav, 2001; White, 1983). will provide direct evidence on the relationship between the mixing of
Research on segregation has thus moved beyond the static residential social groups and built environment characteristics of their destination
areas to understand the segregation of workplaces and other out-of- locations through the “entropy index”. This evidence offers a practical
home non-employment places (Kwan, 2013; Shen, 2019). These approach to understand social groups’ mobility patterns influenced by
works, although equipped with various approaches to simulating ac­ the city’s physical characteristics, generating policy implications for
tivity spaces and their spatial interrelationships, measured ‘place-based planners to consider more carefully which functions can be placed and
segregation’ from static and aggregated perspectives, simplifying the where to increase the use of urban spaces by people from different social
dynamic nature and spatial configuration of mobility patterns. groups.
Considerable attention has also been devoted to the need to include The remainder of this paper is organized as follows. Section 2 rep­
the entire scope of activities undertaken by individuals to gain a fuller resents the literature review, mainly focusing on the concept and mea­
understanding of activity space segregation (Kwan, 2013; Schnell and sures of place-based segregation with its social and spatial implications
Yoav, 2001; Wang et al., 2012; Wong and Shaw, 2011). These studies concerning different population groups. Section 3 describes the trans­
used the concept of activity space and mobility patterns to reflect the port card data used, the study area, and the proposed methodology. The
levels of segregation experienced by social groups across these spaces. results are presented in Section 4 and discussed in Section 5. Finally, in
The underlying assumption was that the disadvantaged groups tend to Section 6, we draw the conclusion, discuss the usefulness and limitations
be deprived of social support, life opportunities, and essential services, of our research, and identify the directions for future studies.
etc. by the lack of access to resources (Wang and Li, 2016). Emerging
technologies that can trace people’s movements with reasonable accu­ 2. Literature review
racy have helped effectively capture these multi-dimensions in a novel
manner (Deville et al., 2014; Sevtsuk and Ratti, 2010). 2.1. Segregation and mobility patterns
Against this background, this study uses Seoul’s public transport card
data to identify passengers’ aggregated mobility patterns and estimate Multiple studies have examined segregation processes through the
the level of segregation at their trip destinations over time. Specifically, lens of mobility patterns, most commonly by race, ethnicity, income,
we introduce a destination-based segregation measure in which expo­ age, and gender (Lucas et al., 2018, 2016; Schwanen et al., 2015). This
sure between social groups is conceptualized as the social interaction lens can be of particular importance when it is evident that the different
potential at their daily destinations. The framework also introduces sectors of the population exhibit fundamentally different mobility pat­
sophisticated place-based measurements that are more sensitive to the terns, as is often the case with underprivileged groups (Lucas et al.,
configuration of the flow-based interaction potentials than are conven­ 2016). These different mobility patterns also point towards the concern
tional indices. This study seeks to address the following three research that such groups may have been facing increasing barriers and in­
questions: 1) do normal-fare passengers have different mobility patterns equalities in performing their day-to-day activities (Morency et al.,
as compared to other social groups? 2) do social groups experience 2011; Schönfelder and Axhausen, 2003). To measure social segregation
segregation over their destinations? 3) at what extent does public through mobility patterns, the index-based approach has been most
transport offer an interaction opportunity for diverse social groups? widely used across different ethnic groups, years, and cities (Johnston
Seoul’s public transport card data combined with other built environ­ et al., 2004; Peach, 2009; Simpson, 2007). The exposure/isolation
ment data are used in this study to answer these research questions. As a index, which assesses the social homogeneity in a residential space, has
methodology, two segregation indices (dissimilarity and isolation) were therefore become a popular measure of social segregation since its cre­
calculated on an hourly basis over a week to examine the segregation ation (Lieberson, 1981). This approach was taken from a systematic
experienced by seniors, children/youth, and passengers with disabil­ study conducted by Massey and Denton (1988) who identified spatial
ities, compared to normal-fare passengers. Further, entropy indices were dimensions that reflect the diverse aspects of residential segregation,
computed, and their temporal and spatial dynamics were investigated. which were categorized into five key types: evenness, exposure, con­
In this investigation, sociodemographic and built environment charac­ centration, centralization, and clustering. Among these, evenness
teristics that are related to the social mixture of passengers at their (Duncan and Duncan, 1955) and exposure (Lieberson, 1981) are the
destination were also identified. It is expected that examining the spatial most widely used indices for the analysis of activity places (e.g., places
and temporal dimension of social interaction potential via people’s of work and leisure) and for analyzing spatial segregation from a tem­
mobility patterns with the added information of built environment and poral perspective (Massey and Fischer, 1999; Reardon and O’Sullivan,
sociodemographic characteristics of their destinations will greatly 2004; Silm and Ahas, 2014b). Schnell and Yoav (2001) and Wong and
enrich our theoretical understanding of social segregation and provide Shaw (2011) extended the exposure index from the residential space to
valuable knowledge for developing effective social integration policies. social networks and activity space, by measuring racial segregation as
The contribution of this study is fourfold. First, this study adds to a the share of people in an individual’s daily life environment who are of
growing body of literature dealing with the application of location- the same race as the individual. The underlying assumption is that the
based big data in urban contexts, identifying the mobility patterns of chance of cross-group interaction depends on the presence of different
different social groups through transport card data. The digital foot­ others in one’s activity space. The higher the proportion of the different
prints of trips offered by transit card data can be used to study mobility others, the more likely one will interact with people outside of his or her
patterns of different groups and uncover how urban dynamics are group. In other words, he or she is then less socially isolated. The en­
structured by these groups. Second, our knowledge of the spatiotem­ tropy index provides another approach, which has been used to study
poral characteristics of different social groups’ mobility patterns in an segregation in spatial detail (Johnston et al., 2004). This index addresses
urban context is still limited. This study helps broaden this knowledge the diversity of segregation rather than the segregation of one group,
by identifying the mobility patterns of typical subgroups in the actual examining multivariate relationships instead of binary relationships to
urban population and revealing their differences in shaping urban dy­ depict the spatial patterns of concentration (Brown and Chung, 2006;
namics. Third, the findings of this study further offer insights into the Marcińczak et al., 2011; Peach, 2009).

2
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Consideration of spatial and temporal dimensions of segregation is and accumulate more minutes of social integration in the community.
another emerging topic in the empirical research of social inequality, Thus, investments in improving public transport infrastructure and
which refers to the segregation between the disadvantaged groups (e.g., services may also represent an important opportunity to increase ado­
the low-income group, ethnic minorities) and the majority group in lescents’ social integration and activity. The study is of particular
space and time (Kempen and Eva, 1994; Massey, 2018). For instance, importance as it focuses on adolescents, a population that is often
Wong and Shaw (2011) combined activity-space exposure measures to overlooked and understudied. From a life-course perspective, adoles­
reveal significant segregation in daily activity between different ethnic cents are significantly increasing their territorial range and their needs
groups in Southeast Florida. Wang and Li (2016) compared the activity and capacity for regional travel but are too young to be licensed to drive,
space and social environment between public and private housing res­ making them an ideal target group for interventions seeking to increase
idents in Hong Kong. Their findings revealed that the activity space of healthy and sustainable travel modes such as public transport.
the two groups was comparable in terms of spatial extent and diversity, Similarly, researchers have repeatedly observed distinctive ten­
yet significantly different in terms of contact with similar socioeconomic dencies in the mobility practices of older populations. Senior residents
groups. Similar findings have been reported in Beijing by Zhang et al. travel shorter distances, visit fewer destinations, with grocery stores,
(2019), which measured both potential and actual activity spaces of other retailers, and nice restaurants being the main ones, and make
residents of different types of housing using a 7-day GPS dataset. Using fewer trips by individualized modes, including cars and bicycles
mobile phone data, Silm and Ahas (2014a) found that both the majority (Haustein, 2012; Schwanen et al., 2015; Siren and Haustein, 2013).
(i.e., Estonian) and minority groups (i.e., Russian-speaking population) Different travel patterns by the elderly population can regularly be
living in Tallin, tend to experience segregation in temporal shifts at attributed to a higher incidence of mobility-limiting conditions and a
daily, weekly, and seasonal scales. Their research identified that lower incidence of social cohesion (Kandt and Leak, 2019). Empirical
workday levels of segregation are far lower than evening and weekend evidence also suggests that travel patterns of the elderly population are
levels. Another study by Lee and Kwan (2011) developed four visual changing. Van den Berg et al. (2011) found in the Netherlands that to­
methods—3D space-time paths, time windows, 3D activity density sur­ day’s older populations are as mobile as younger ones concerning non-
faces, and ring-based visualization of social networks—to identify socio- work trips. Average travel distances did not decrease with age; instead,
spatial segregation among South Koreans living in Columbus, Ohio, over they were often affected by other factors, such as retirement, employ­
space and time. Finally, Farber et al. (2015) proposed a reproducible ment, residential location, gender, and household structure. For
exposure measure based on the potential opportunities of social contacts example, Siren and Haustein (2016) analyzed how retirement affects
for members of different social groups by taking into account the elderlies’ travel and the related future travel demand by interviewing
intersection of their spatiotemporal activity patterns. These studies used 864 people in 2009 and 2012. Their findings suggested that retirement is
the concept of space where daily activities take place and mobility a transition point associated with decreasing transportation demand.
patterns to reflect the levels of segregation experienced by social groups However, the emergence of leisure and consumption as major cultural
across time. and social frameworks of the third age is likely to alter this transition. In
addition, the elderlies are a heterogeneous group with varying mobility
2.2. Understudied social groups in transport systems preferences and resources, as reflected in their mobility choices (Siren
and Haustein, 2013). These findings suggest that as older population
Determining the process of segregation by examining the patterns of cohorts become more socially and culturally diverse, maintain profes­
public transport use is an effective tool (Kenyon et al., 2002). This tool, sional engagements, and access new leisure opportunities, the ways
in particular, assesses to what extent transport systems provide an op­ senior residents will travel in the future remain uncertain (Siren and
portunity for different populations to interact and perform their day-to- Haustein, 2016). Therefore, there is a constant need to study the
day activities seamlessly (Dargay, 2007) and identifies a viable or sus­ evolving mobility patterns and practices of the older population and
tainable solution to mobility-related aspects of social exclusion (Jar­ how they are realized through the available means of transit services
amillo et al., 2012; Kenyon et al., 2002). The inclusion of diverse social (Luiu et al., 2017).
groups such as seniors, children, and people with disabilities in this tool
can further deepen our understanding of social segregation. Some pre­ 2.3. Relationship between social interaction and built environment
vious studies have already attempted to enhance this understanding and
have successfully identified that social exclusion and mobility are inti­ In addition to delineating segregation through trip destination, the
mately linked (Xia et al., 2016) and that deficient transport systems present study takes a further step to examine the relationship between
contribute to social exclusion by restricting socially disadvantage destination’s built environment and sociodemographic characteristics
groups’ access to work, education, and leisure (Jaramillo et al., 2012; and the social mixture of passenger flows arriving in the area. There is no
Shirmohammadli et al., 2016; Xia et al., 2016). For example, Henly and consensus on exactly what these characteristics are or whether these
Brucker (2019) demonstrated inequalities in community participation characteristics affect the choices that people make to choose the places
for working-age Americans with disabilities through travel patterns. to perform their daily activities. The major reason for this is the complex
They utilized the 2017 National Household Travel Survey to examine mechanisms of people’s travel behavior (Jones and Pebley, 2014). Some
travel patterns for persons with disabilities. Their analysis found that of the complexity might be associated with the bidirectional causal
having a disability is associated with lower odds of taking a trip for relationship; for example, preferences for out-of-home activities might
shopping, for social or recreational purposes, for running errands, or for incentivize people to choose areas in or near activity centers, and
going to work. They further suggested that travel patterns for people naturally, such spatial proximity can encourage out-of-home activities.
with disabilities through public transportation should be explored in Although the relationships are complex, studies generally suggested that
order to provide a comprehensive picture of what barriers prevent the built environment and sociodemographic characteristics of neigh­
people from leaving their homes for community participation, inde­ borhoods are associated with social segregation or a mixture of diverse
pendent living, and economic involvement. social groups. Jones and Pebley (2014), for example, identified some
A study conducted in Melbourne, Australia seems meaningful in that socioeconomic characteristics, such as income and race/ethnicity that
it explored the association between public transport and the activity can affect social segregation. Their findings suggest that individuals
space of adolescents (younger than 18 years) (Zulkefli et al., 2020). with higher socioeconomic status have more financial and trans­
Based on Victorian Integrated Survey of Travel and Activity data, it portation resources as well as greater travel needs, and thus have larger
provided evidence that adolescents with higher public transport acces­ activity spaces. Similarly, socioeconomically privileged residents have
sibility around homes and schools are more likely to use public transport smaller activity spaces in both Hong Kong (Wang and Li, 2016) and

3
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Beijing, China (Wang et al., 2012). A possible reason is that privileged transit. At the time of this study, only 40% of regular city buses were low
residents retreat to avoid social interaction with other people (Wang floor (have no steps, which makes it easier for the elderly and disabled to
et al., 2012). Socioeconomic characteristics in a neighborhood, such as get on and off), and the city plans to increase the proportion up to 85%
the rate of the educated and employed population, reflect social ties and by 2022 with a goal to provide more comfortable services for the
social networks (Hu and Downs, 2019; Manca et al., 2019), which in disabled and elderly population. Moreover, among 728 subway stations,
return motivate social activities that are associated with activity spaces 40 stations in Seoul are still inaccessible to vulnerable transportation
(Miller, 2004; Silm and Ahas, 2014a). Spatial characteristics of neigh­ users as they are not equipped with ramps and elevators (TOPIS, 2021).
borhoods play important roles as well, and again empirical results are Fig. 1 shows the temporal distribution of transit trips by each
mixed. Among the common spatial variables, the density of population selected social group in the city. The data were further segmented into
and/or employment (Shen, 2019; Tan et al., 2017), road density (Tana two subsets of weekdays and the weekend for capturing hourly cross-
Kwan and Chai, 2016), and access to transit have positive impacts on sectional pictures of individuals’ final destinations. It can be observed
social interaction. This suggests that the relationship between spatial that during weekdays, normal-fare passengers tend to travel with two
characteristics of the neighborhood and social interaction is an impor­ load peaks: first in the morning (from 8 am to 9 am) with 11% of total
tant aspect and that should be explored for a better understanding of trips made, and the second in the evening (from 6 pm to 7 pm) ac­
social segregation, calling for further investigations. counting for almost 12% of total trips. Given that this is the general
This literature review suggests that an accumulating body of travel trend in Seoul, it is expected that travel by other social groups
research has employed mobility patterns to shed light on segregation would follow the same trend. However, notable differences are found in
experienced by different groups in their daily travel and other activities. the travel patterns of each social group. Seniors’ travel frequency does
These studies validated the capability of illustrating the dynamic nature not show a clear peak in the morning and children/youth’s morning and
and sophisticated distributions of segregation through fruitful repre­ evening peak times are slightly different from normal-fare passengers’
sentations. However, most of these studies have largely relied on sta­ peak hours. Only passengers with disabilities show a relatively similar
tistical regression that modeled activity space as a function of travel pattern to the normal-fare passengers.
socioeconomic status and other sociodemographic characteristics. In the case of the weekend, the travel frequencies of all the selected
Considering the complexity and multidimensional nature of the concept social groups show similar patterns with an occupational peak from 8
of social segregation, further investigations are required considering a am to 8 pm throughout the day. The maximum trips in terms of per­
broader range of social groups and their mobility patterns in space other centage were performed by children/youth at about 10% from 1 to 2 pm.
than residence and time. To address this research gap, this study com­ In sum, the travel patterns of social groups are significantly different,
bines the relative advantages of both spatiotemporal dimensions of compared to normal-fare passengers, particularly during weekdays.
place-based segregation and the application of location-based big data These temporal differences in the mobility pattern of social groups led to
to enrich the literature. While census data and other similar sources tend an in-depth analysis that can reveal the extent of segregation they
to focus on a narrow domain, this study uses location-based transport experience at their destinations.
card data which includes the travel records of millions of passengers in a
metropolitan area. Furthermore, it should be noted that senior citizens, 3.2. Study area
young adolescents, and persons with disabilities are still less investi­
gated social groups that may have dis/similar destination choices, The study focuses on Seoul, the capital of South Korea and the largest
compared to the general population. city in the country, with an area of 605 km2, categorized into 25
municipal districts and 424 administrative sub-units called Dong. It has a
3. Data, study area and methodology total population of about 10 million, of which 23% are seniors (aged 65
and above), 11% are young adolescents (below age 18), and 4% are
3.1. Data passengers with disabilities (KOSIS, 2019). Seoul provides an interesting
study context, given the prevalence of traffic congestion and the rapid
This study uses one week’s travel record data of public transit (bus urban development in line with the country’s fast-aging population. The
and subway) passengers extracted from Seoul’s transport card data from number of people who acquired disabilities because of their old age has
April 1–7, 2018. The dataset contains information about boarding and risen as well (Koo et al., 2017), resulting in the spatial separation of
alighting times and locations of public transit passengers composed of social groups. Decades of research have shown that the spatial separa­
four groups, normal-fare passengers, seniors (older than 65 years), tion of different social groups is associated with spatial inequalities such
children/youth, and passengers with disabilities. The latter three groups as disparities in health (Williams and Collins, 2001), economic outcomes
pay a fully or partially discounted fare for their travel. Metro systems (Massey and Denton, 1988), educational achievement of youth (Card
provide free tickets for seniors and passengers with disabilities while and Rothstein, 2006), and spatial mismatch between the locations of
they should pay a full fare for buses that are operated by private com­ low-wage workers and employment opportunities (Kain, 1968). There­
panies. The transit system in Seoul is fully integrated, allowing free fore, the identification of destination locations of each social group,
transfers between different modes of public transit. The free transfer especially seniors, could contribute to additional knowledge of overall
requires the passengers to tag their cards whenever they board and mobility patterns and help relevant authorities to mitigate issues related
alight. If we assume that the final alighting location is the passenger’s to traffic and spatial inequalities in the near future.
activity space, interactions between public transit passengers can be
measured based on their final destinations. Note that the final destina­ 3.3. Methodology
tions (alighting stations/stops) are likely to correspond to the passen­
ger’s activity space due to the dense service coverage of public transit in 3.3.1. Conceptual framework
Seoul. For the study period, around 6 million trips were collected. An alternative framework, named destination-based segregation was
Among them, 97.10%, 1.94%, 0.23%, and 0.73% of trips were made by developed to estimate the potential segregation experienced by pas­
normal-fare passengers, seniors, children/youth, and passengers with sengers at their daily destinations from static and aggregated perspec­
disabilities, respectively. The reason behind the hefty difference in the tives. The proposed framework simplifies the spatial configuration of
trip distribution of seniors, children/youth, and passengers with dis­ mobility patterns and the dynamic nature of segregation along with the
abilities as compared to their population distributions is that these added information of a social group of passenger flows arriving in the
groups are generally less mobile. In addition, there exist substantial Dong. The notion of destination-based segregation is drawn from the
physical barriers for “the vulnerable transportation users” to ride public concept of social interaction potential, originating from joint

4
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Fig. 1. Temporal distribution of aggregated trips by individuals.

accessibility measurement (Neutens et al., 2008, 2007), which is to work together. The extent can also have relations across social groups,
measure the intensity of social segregation across urban spaces based on opportunities for communication, and social interaction for their com­
the space-time framework (Farber et al., 2012). However, due to the mon good (Liu et al., 2015). However, the extent of segregation does not
structural particularities of the data, the theoretical development fo­ represent the overall situation of selected social groups in the study are
cuses on a potential social interaction within a single static space because it only represents the segregation from part of the social groups
(destination) over time of the day, but the framework can be extended to who use public transit.
multiple spaces and for any part of the day with the appropriate modi­ The spatial unit in this study was the destination Dong visited by
fications. Some other researchers have also emphasized the evolvement passengers to perform their activities. However, passenger’s duration of
of spatial dimensions of activity space (Kwan, 2000; Yu and Shaw, 2004) stay in the Dong, their afterword activities, and the presence of other
due to the growing availability of new data sources and the advance­ people in the Dong were not taken into account due to data structure and
ment in computational powers. limitations. Concerning the temporal dimension, trips were aggregated
Another conceptual adjustment was drawn towards the difference on an hourly basis. In the approach, trips occurring outside 5 am to
between the individual or grouped-based measure of segregation. This midnight were eliminated from the dataset due to the meager number of
study used aggregated trips based on space and time. This approach is trips generated.
easy to understand and can be extended to other measures of place-
based segregation when dealing with the data that consists of a huge 3.3.2. Measures of segregation
number of individuals. Once trips are aggregated by social groups over To evaluate the destination-based segregation, the classic measures
space and time, the level of segregation can be calculated, resulting in a of segregation—dissimilarity index (ID), index of exposure (IX), and
single indicator for the entire passengers that can be visualized for entropy index (IE)—were selected in light of their mathematical prop­
exploring the variations among the groups collectively. Such an example erties and their complementarity to measure different dimensions of
can be observed in Wong and Shaw (2011). They aggregated activity segregation. The measures of ID and IX quantify the degree of dissimi­
spaces for each group in each residential neighborhood, creating a larity or exposure between a selected social group and the normal-fare
group-location-based exposure index, which is arguably more applicable passengers in a pairwise manner over the entire study area. Mean­
to large samples. while, IE provides the measure of diversity that assesses the degree of
As a result, total trips were aggregated based on destination location social mix within the spatial unit. The varying values of these indices
(where they occur), time (when they occur), and social groups (by whom over time are examined for both weekdays and the weekend. The defi­
they occur). This three-level aggregation can reflect numerous di­ nition of the three measures are as follows.
mensions of segregation. For example, the segregation level based on
space can reveal spatial characteristics of neighborhoods that play 3.3.2.1. Index of dissimilarity (ID). Evenness has been measured most
important roles in shaping the social structure of urban areas in the often by ID proposed by Duncan and Duncan (1955) as a measure of
daytime. The time dimension of segregation can demonstrate the pairwise segregation (e.g., Black versus White). ID measures the extent
temporally dynamic nature of segregation over a certain period of time to which two social groups are distributed differently. It ranges from
(e.g., day or week). Similarly, the extent of segregation experienced by 0 (complete integration) to 1 (complete segregation). The value of ID
social groups reveals some important societal outcomes of regions like reaches a maximum when each spatial unit contains only one group and
social cohesion or the degree to which different members of the society it is minimum when the proportion of each group in each spatial unit is

5
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

the same as the proportion in the population as a whole. For the study experienced by inflows of passengers arriving in Dong based on their
area consisting of n Dongs, ID is calculated as: social groups before and after peak hours can be captured.
1∑n ⃒⃒xi yi ⃒⃒
ID = ⃒ − ⃒ 4. Results
2 i=1 X Y

where xi is the total number of passengers of a subgroup (e.g., seniors, 4.1. Social segregation through an index of dissimilarity (ID)
children/youth, and passengers with disabilities) in Dong i, and X is the
total number of passengers of a subgroup. In addition, yi and Y indicates Table 1 and Fig. 2 display the expected contributions of each type of
the number of passengers of a referenced group (normal-fare passenger) between-group interaction potential over their destinations. As seen in
in Dong i and the total number of passengers of the referenced group, Fig. 2 (a) and (b), segregation indices vary significantly by time of the
respectively. day and the day of the week. The mean ID values for seniors, children/
youth, and passengers with disabilities during weekdays are 0.258,
3.3.2.2. Index of exposure (IX). Another important element of segre­ 0.410, and 0.249, respectively, suggesting that children/youth are the
gation is exposure, often measured by IX proposed by Lieberson (1981) most segregated social group compared to normal-fare passengers
to assess the social homogeneity over an area. IX measures the contact (Table 1). The segregation experienced by children/youth is more
probability of one social group to another within the same arriving Dong. visible during the time period from 5 am to 8 am. From 1 pm, all the
The value of this index also ranges between 0 (lower exposure) and 1 social groups seem to experience the same level of segregation until
(higher exposure). Therefore, conceptually, this measure tends to be nighttime, when the extent of segregation begins to increase. Overall,
opposite to ID. The degree of exposure of group x to group y over an area the city is less segregated during the time period between 1 pm and 9 pm
composed of n sub-areas (in our case 424 Dongs) is computed as follows: and most segregated during the nighttime from 10 pm to midnight
(average ID for all groups during the nighttime = 0.38). This is expected
∑n [( ) ( ) ]
IX =
xi yi because people in social groups other than normal-fare passengers tend
i=1
X ti to be less mobile during the nighttime.
The weekend graph shows a slightly different ID pattern from that of
where xi, yi, and X are the same as mentioned before, and ti refers to the weekdays. The mean ID values during the weekend in Table 1 suggest
total number of passengers in Dong i. that children/youth and passengers with disabilities are the more
segregated social groups than seniors (ID = 0.48, 0.47, 0.30 for the three
3.3.2.3. Entropy score (IE). Both ID and IX are uni-group indices that groups, respectively). The detailed variation in the extent of segregation
can only measure the segregation of two groups compared with each level can be seen in Fig. 2, where children/youth are highly segregated
other. Meanwhile, IE (a multigroup index) is a measure of “diversity,” during the morning (mean ID = 0.66 between 5 am and 8 am) and
the extent to which groups are evenly distributed across the study area nighttime (mean ID = 0.63 between 9 pm and midnight). However,
(Massey and Denton, 1988). The value of IE ranges between 0 (full between 8 am and 9 pm, passengers with disabilities experience the
homogeneity) and 1 (full heterogeneity). IE for Dong i is calculated as: highest level of segregation (mean ID = 0.45). Overall, children/youth
experience the highest level of segregation during weekdays and early in
∑
k
( )
IE = − Pij ln Pij the morning, and late at night during the weekend. For seniors, the
j=1 pattern and size of the ID values for weekdays and the weekend appear
to be similar. Only the group comprising passengers with disabilities
where k is the number of selected social groups (i.e., senior, children/ shows statistically significantly different ID values for weekdays and the
youth, and passengers with disabilities) and Pij is the ratio of the number weekend (p = 0.000 in Table 1), suggesting that individuals with dis­
of passengers for a social group j to the total number of passengers in abilities are relatively more segregated during the weekend compared to
Dong i. weekdays.

3.3.3. Statistical tests and correlation analysis 4.2. Social interaction through an index of exposure (IX)
To highlight the different levels of ID and IX experienced by the
social groups depending on the day of the week (i.e., weekdays vs. Space-time segregation among social groups was further investigated
weekends), the statistical tests of two-sample t-tests were carried out. based on IX. Fig. 2 (c) and (d) depict the difference in social groups’
Further, the coefficients of correlation between IE scores and the built exposure patterns for weekdays and the weekend. The mean values of IX
environment and sociodemographic attributes of each Dong were for all the selected groups indicate that the probability that seniors,
calculated to identify factors that derive changes in IEs. To understand children/youth, and passengers with disabilities will encounter normal-
spatiotemporally changing IEs, a quantitative analysis was conducted by fare passengers in the same space is almost the same. Undoubtedly,
comparing IEs from different periods and for Dongs in peripheral and mean IXs for all the groups on both weekdays and the weekend are
central areas. By doing so, gaps between the segregation levels around 0.05 (Table 1). Interestingly, two exceptionally high exposure

Table 1
The values of ID and IX during weekdays and the weekend.
Index of dissimilarity (ID) Index of exposure (IX)

Senior Children/youth Passengers with disabilities Senior Children/youth Passengers with disabilities

Weekdays
Mean 0.258 0.410 0.249 0.052 0.051 0.054
Standard deviation 0.067 0.167 0.040 0.032 0.027 0.034
Weekend
Mean 0.298 0.477 0.473 0.050 0.049 0.049
Standard deviation 0.139 0.193 0.093 0.022 0.022 0.021
Statistical difference between weekdays and the weekend
p-value 0.256 0.244 0.000 0.839 0.859 0.601

Note: These statistics were obtained based on hourly measures over 20 h.

6
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Fig. 2. The hourly estimates of ID and IX during weekdays and the weekend.

peaks were found during the weekdays. The two peaks seem to perfectly morning time start to experience a high heterogenization of the social
correspond to the morning and evening commute times (8 am to 9 am profile during the evenings and nights of weekdays. Meanwhile, week­
and 6 pm to 7 pm), suggesting the high degree of destination choice ends are not subject to a significant spatiotemporally changing social
overlapping during the commute times. However, overall exposure mix of passengers arriving in the Dong (Fig. 3.). The mean IE values of
starts to decline after 10 pm; mean IXs for all groups are around 0.02 Dongs in central areas during morning and afternoon times are almost
between 10 pm and midnight. During the weekend, the computed IXs similar. During the evening and nighttime, the mean IE values of Dongs
reveal a totally different pattern, compared to weekdays (Fig. 2 (d)). in business areas are slightly higher than those of Dongs in peripheral
There were no high exposure peaks, and the IX values remain moderate areas. Central Dongs with high values of IE during evenings and the night
and rather stable across the day. Additionally, little difference is found on weekdays revealed a relatively lower value of IE scores on the
among the three social groups. weekend during the same time periods. This confirms that overall,
In summary, the overall levels of exposure are not differentiated during weekdays, some parts of the city are more heterogeneous,
among social groups and between weekdays and the weekend. However, particularly central business areas. The ability to map changes in en­
the exposure level is temporally dynamic as it significantly changes over tropy scores over the time and day of the week demonstrates that the
time, especially during weekdays. This suggests that social segregation is scale of day-to-day activities impacts the social structure of passenger
not static but that it has a temporal dimension. In addition to the tem­ flows in urban areas to a greater extent.
poral dimension, examining the spatial dynamics of social segregation To identify factors that derive changes in IEs, the coefficients of
would be interesting. The next section investigates how social segrega­ correlation between IE and the built environment and sociodemographic
tion is spatially differentiated based on the multi-group exposure mea­ attributes of each Dong were calculated, as shown in Table 2. The results
sure, IE. show that IE is positively correlated with the population-related vari­
ables: total population, the population of 60-to-70-year-olds, the popu­
lation of children/youth, and residential area size. The number of
4.3. Level of social mix/interaction through entropy score (IE)
subway stations were also found to have a significantly positive rela­
tionship with IE, implying that subway station areas provide opportu­
For spatiotemporal analyses, four time periods—(1) morning hours
nities for more diverse activities. Similarly, recreation and leisure areas
from 7 am to 11 am, (2) midday from 11 am to 5 pm, (3) evening from 5
can also be considered as areas where a greater mix of social groups is
pm to 7 pm, and (4) night from 8 pm to 10 pm—were selected to map
observed. Conversely, IE is negatively correlated with the number of
entropy scores over the day. The spatial representation of IE scores on a
jobs and commercial areas. This is rather counter-intuitive since the
Dong basis, shown in Fig. 3, successfully provided the dynamics of
literature has shown the opposite trend that workplaces and commercial
interaction among the social groups throughout the day. The color in­
areas are more diverse and less segregated than residential areas (Åslund
tensity in the map indicates the level of a mix of all social groups over the
and Skans, 2010; Silm and Ahas, 2014a). However, this finding may be
space. It is important to note that the dark areas in the map (high level of
reliable since the considered social groups, except for normal-fare pas­
IE) shift from peripheral areas (mostly residential areas) to central areas
sengers, in this study belong to non-working and less active groups.
(mostly business and commercial areas) over time, highlighting the
spatially changing social mix over time. To understand this spatiotem­
5. Discussion
porally changing social interaction trend, a quantitative analysis was
carried out by comparing IEs for Dongs in peripheral and central areas.
Segregation is a multidimensional phenomenon where social,
(For this analysis, 69 and 83 Dongs were selected for peripheral and
spatial, and temporal dimensions are interrelated. Given the changing
central areas, respectively.)
form of segregation and societal developments at large, the exploration
Fig. 4. confirms that the mean IE of most peripheral Dongs during
of the spatial mobility of individuals is becoming increasingly important
weekdays is significantly higher in the morning but drops subsequently.
in gaining a more comprehensive understanding of socio-spatial in­
The mean IE value of these Dongs is significantly lower during the
equalities (Kwan, 2013; Urry, 2007). This implies that researchers need
nighttime, implying that these are the most homogeneous urban areas at
to move beyond reliance on purely residential-based analyses and a
night. Conversely, socially homogeneous central Dongs during the

7
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Fig. 3. Comparison of average entropy scores by time and location.

Note: Bars around means indicate 95% confidence intervals.

8
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Fig. 4. The IE during weekdays and the weekend.

further step towards the new perspectives of place-based segregation spatial limitations that are defined as the final destinations visited by
that takes spaces other than residential, time, and the dynamic flows of transit passengers. In spite of this, the selected indices successfully
everyday life into account. Mobility in this case plays an essential role in captured the part of activity space of passengers, depicting the shifts in
people’s spatiotemporal experiences (Kandt and Leak, 2019; Lucas et al., the social structure of passenger flows in Seoul throughout the day and
2016), and these complex experiences cannot be fully understood by just the day of the week. Thus, it seems that the results can be utilized by
looking at where people live. The places people can reach and at what public transport agencies and disability or senior citizen advocates to
time they can reach them are important determinants of their exposures highlight and quantify the performance of the public transport network
to various social groups and physical environments of urban areas in terms of providing an interaction opportunity for the groups.
(Kwan, 2009). Several findings in this study merit attention for providing a more
This study aimed to contribute to place-based segregation research contemporary and potentially more accurate estimation of destination-
by expanding the analytical focus of static residential places to other based segregation. First, the level of segregation varies depending on
relevant places where people may interact or experience segregation the time of the day, the day of the week, and the type of social group. For
over time. For this, the flows of passengers were aggregated at their example, social groups are much more evenly distributed across the city
daily destinations based on their social group over time and then the in the morning and evening peak hours during weekdays. The potential
segregation and exposure levels of those social groups were computed. interaction between selected social groups and normal-fare passengers is
This approach demonstrated the potential of transport card data in also very high during those peak times. Children/youth, on the contrary,
revealing hidden spatiotemporal differences in mobility patterns of experienced a higher level of segregation regardless of the time of the
diverse social groups, advancing segregation research. Although weekday. The results during the weekend, however, indicate a different
aggregated trips were used instead of individual-level data to evaluate picture with no specific peak times, with the overall potential interac­
the exposure levels in the Dongs where passengers alight, the proposed tion remaining relatively low. The exposure to normal-fare passengers
approach represented an improvement over the traditional methods for also remained relatively low during the weekend, compared to week­
evaluating group-based segregation levels. The empirical results from days. When comparing the level of segregation between weekdays and
this study may not fully reflect the actual situation of segregation the weekend, all the social groups experience a similar level of segre­
experienced by the social groups because of the data particularities, gation except passengers with disabilities who are more segregated
uneven population and trip distribution of selected social groups, and during the weekend. Policymakers should specifically focus on

9
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Table 2 sociodemographic characteristics of their destinations will greatly

Correlation table between entropy index and selected area attributes. enrich our theoretical understanding of social segregation, providing
Sociodemographic and built environment Coefficients of t-value (p- valuable knowledge for developing effective social integration policies.
attributes of destination correlation value) The produced outcomes will help public transport and local government
Total population 0.168*** 3.516 stakeholders to implement appropriate interventions in areas with a
(0.000) high concentration of specific social groups around the clock and to
Population of 60-to-70-year-olds 0.237*** 5.024 reduce social inequalities more effectively within the metropolitan area.
(0.000) From a theoretical point of view, this analysis allowed the repro­
Population of children/youth 0.102* 2.117
(0.034)
duction of our understanding related to two major arguments in the
Number of students 0.059 1.221 activity space segregation field. The first argument supported the
(0.222) consensus that segregation methodologies and conceptualization ap­
Number of jobs − 0.238*** − 5.031 proaches are continuously evolving, particularly with the consideration
(0.000)
of the complex and multidimensional nature of the concept. The pro­
Industrial area − 0.026 − 0.536
(0.591) posed framework incorporated both temporal and spatial dimensions
Recreation and leisure area 0.087* 1.796 into the segregation measures based on aggregated mobility patterns. In
(0.023) terms of aggregation, although this study aggregated travel data at
Commercial area − 0.063† − 1.297 administration area (Dong) and hourly bases, any reasonable spatial
(0.084)
Residential area 0.167*** 3.477
(such as neighborhood, census-based, or non-consensus-based) or tem­
(0.000) poral units can be applied for future research depending on study pur­
Number of subway stations 0.104*** 2.164 poses. The second argument signified the viability of recent location-
(0.000) based big data technologies and their application in the field of activ­
Number of bus stops − 0.119* − 2.465
ity space segregation to identify the mobility patterns of different social
(0.014)
Number of apartments 0.121* 0.012 groups who play a major role in changing the social structure of urban
(0.012) areas. The integrated approach was applied jointly and vindicated both
arguments by utilizing such big data to assess the spatiotemporal aspects
*** < 0.001, **p < 0.01, *p < 0.05, † p < 0.1; Values in parentheses indicate p-
values.
of segregation along with an individual’s social background informa­
tion. A combination of pairwise and multi-group indices were derived
through mobility patterns of different social groups.
identifying the reasons for this segregation during the weekend based on
One key limitation of this study is that it is only confined to public
this finding.
transit users’ one-week travel data with uneven trip distribution patterns
The second interesting outcome of this study was found by IE, which
based on their social group. Future studies may consider yearly, sea­
represented the shifting concentration of a combination of social groups
sonal, and monthly variations by applying temporally extensive datasets
from peripheral areas to central areas over time. This result highlighted
with appropriate sample sizes of the representative social groups. The
the changing social structure of passenger flows arriving in a Dong where
sources of such datasets can include a mobile phone, social media, and
they may have a similar social structure during morning hours, but it can
credit card as well as a transportation card. An extensive study will
evolve in a very different way during the afternoon and the evening.
ensure the complete representation of the activity space of numerous
However, this dynamic can only be observed during weekdays. The
social groups in a city. Another important theme for future studies is to
weekend is less susceptible to the spatiotemporally changing social
consider using more sophisticated statistical methods to explain the
structure of the city, reflecting that during weekdays, the city is most
relationship between social mix and destination locations’ built envi­
heterogeneous with intense activity and mobile social groups, compared
ronment and sociodemographic characteristics. This effort will certainly
to the weekend.
bring opportunities to further explore diverse perspectives of social
The final meaningful outcome of the present study was drawn
segregation and changes in cities.
through the correlation between IE scores and sociodemographic and
the built environment attributes of each Dong. The correlation co­
CRediT authorship contribution statement
efficients confirmed that subway stations and recreational areas provide
more opportunities for the interaction of different social groups. This
Sorath Abbasi: Data curation, Formal analysis, Writing - original
finding offers a practical approach to understanding social groups’ ac­
draft. Joonho Ko: Conceptualization, Supervision, Writing - review &
tivity patterns influenced by the city’s physical characteristics. Such an
editing. Jaehong Min: Conceptualization, Data curation, Resources,
understanding generates direct policy implications for planners to
Validation, Writing - review & editing.
consider more carefully which functions can be placed and where to
increase the use of urban spaces by people from different social groups.
References
Instead of relying fully on market forces, policymakers should consider
contemporary social infrastructure investment and urban regeneration Åslund, O., Skans, O.N., 2010. Will I see you at work? Ethnic workplace segregation in
practices to promote social diversity and rejuvenate the city. Sweden, 1985-2002. Ind. Labor Relat. Rev. 63, 471–493. https://doi.org/10.1177/
001979391006300306.
Atkinson, R., Flint, J., 2004. Fortress UK? Gated communities, the spatial revolt of the
6. Conclusion elites and time-space trajectories of segregation. Hous. Stud. 19, 875–892. https://
doi.org/10.1080/0267303042000293982.
This study examined the level of segregation experienced by seniors, Brown, L.A., Chung, S.-Y., 2006. Spatial segregation, segregation indices and the
geographical perspective. Popul. Space Place 12, 125–143. https://doi.org/10.1002/
children/youth, and passengers with disabilities, compared to the psp.403.
normal-fare passengers at their trip destination when using public Card, D., Rothstein, J., 2006. Racial segregation and the black-white test score gap. Nat.
transportation. One week’s travel records of public transit passengers Bureau Econ. Res. https://doi.org/10.3386/w12078.
Dargay, J., 2007. The effect of prices and income on car travel in the UK. Transp. Res.
with imputed social backgrounds were extracted from Seoul’s transport Part A Policy Pract. 41, 949–960. https://doi.org/10.1016/j.tra.2007.05.005.
card data to compute dissimilarity and exposure indices, to capture Deville, P., Linard, C., Martin, S., Gilbert, M., Stevens, F.R., Gaughan, A.E., Blondel, V.D.,
destination-based segregation. It was believed that examining the tem­ Tatem, A.J., 2014. Dynamic population mapping using mobile phone data. Proc.
Natl. Acad. Sci. U. S. A. 111, 15888–15893. https://doi.org/10.1073/
poral and spatial aspects of segregation via passenger’s aggregated
pnas.1408439111.
mobility patterns with the added information of built environment and

10
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

Duncan, O.D., Duncan, B., 1955. A methodological analysis of segregation indexes. Am. Marcińczak, S., Musterd, S., Stępniak, M., 2011. Where the grass is greener: social
Sociol. Rev. 20, 210. https://doi.org/10.2307/2088328. segregation in three major polish cities at the beginning of the 21st century. Eur.
Ellis, M., Wright, R., Parks, V., 2004. Work together, live apart? Geographies of racial Urban Reg. Stud. 19, 383–403. https://doi.org/10.1177/0969776411428496.
and ethnic segregation at home and at work. Ann. Assoc. Am. Geogr. 94, 620–637. Massey, D.S., 2018. Residential Segregation, in: A Companion to Racial and Ethnic
https://doi.org/10.1111/j.1467-8306.2004.00417.x. Studies. Blackwell Publishers Ltd, London, pp. 348–354. https://doi.org/10.1111/
Farber, S., Páez, A., Morency, C., 2012. Activity spaces and the measurement of b.9780631206163.2002.00030.x.
clustering and exposure: a case study of linguistic groups in Montreal. Environ. Plan. Massey, D.S., Denton, N.A., 1988. The dimensions of residential segregation. Soc. Forces
A 44, 315–332. https://doi.org/10.1068/a44203. 67, 281. https://doi.org/10.2307/2579183.
Farber, S., O’Kelly, M., Miller, H.J., Neutens, T., 2015. Measuring segregation using Massey, D.S., Fischer, M.J., 1999. Does rising income bring integration? New results for
patterns of daily travel behavior: a social interaction based model of exposure. blacks, Hispanics, and Asians in 1990. Soc. Sci. Res. 28, 316–326. https://doi.org/
J. Transp. Geogr. 49, 26–38. https://doi.org/10.1016/j.jtrangeo.2015.10.009. 10.1006/ssre.1999.0660.
Fu, X., Gu, Y., 2018. Impact of a new metro line: analysis of metro passenger flow and Miller, H.J., 2004. Activities in space and time. Transp. Geogr. Spat. Syst. 5, 647–660.
travel time based on smart card data. J. Adv. Transp. 2018 https://doi.org/10.1155/ https://doi.org/10.1108/9781615832538-036.
2018/9247102. Morency, C., Paez, A., Roorda, M.J., Mercado, R., Farber, S., 2011. Distance traveled in
Haustein, S., 2012. Mobility behavior of the elderly: an attitude-based segmentation three Canadian cities: spatial analysis from the perspective of vulnerable population
approach for a heterogeneous target group. Transportation (Amst). 39, 1079–1103. segments. J. Transp. Geogr. 19, 39–50. https://doi.org/10.1016/j.
https://doi.org/10.1007/s11116-011-9380-7. jtrangeo.2009.09.013.
Henly, M., Brucker, D.L., 2019. Transportation patterns demonstrate inequalities in Neutens, T., Witlox, F., Van De Weghe, N., De Maeyer, P., 2007. Human interaction
community participation for working-age Americans with disabilities. Transp. Res. spaces under uncertainty. Transp. Res. Rec. 28–35 https://doi.org/10.3141/2021-
Part A Policy Pract. 130, 93–106. https://doi.org/10.1016/j.tra.2019.09.042. 04.
Hu, Y., Downs, J., 2019. Measuring and visualizing place-based space-time job Neutens, T., Schwanen, T., Witlox, F., De Maeyer, P., 2008. My space or your space?
accessibility. J. Transp. Geogr. 74, 278–288. https://doi.org/10.1016/j. Towards a measure of joint accessibility. Comput. Environ. Urban. Syst. 32,
jtrangeo.2018.12.002. 331–342. https://doi.org/10.1016/j.compenvurbsys.2008.06.001.
Jaramillo, C., Lizárraga, C., Grindlay, A.L., 2012. Spatial disparity in transport social Peach, C., 2009. Slippery segregation: discovering or manufacturing ghettos? J. Ethn.
needs and public transport provision in Santiago de Cali (Colombia). J. Transp. Migr. Stud. 35, 1381–1395. https://doi.org/10.1080/13691830903125885.
Geogr. 24, 340–357. https://doi.org/10.1016/j.jtrangeo.2012.04.014. Reardon, S.F., O’Sullivan, D., 2004. Measures of spatial segregation. Sociol. Methodol.
Johnston, R., Poulsen, M., Forrest, J., 2004. The comparative study of ethnic residential 34, 121–162. https://doi.org/10.1111/j.0081-1750.2004.00150.x.
segregation in the USA, 1980–2000. Tijdschr. Econ. Soc. Geogr. 95, 550–569. Schnell, I., Yoav, B., 2001. The sociospatial isolation of agents in everyday life spaces as
https://doi.org/10.1111/j.0040-747x.2004.00339.x. an aspect of segregation. Ann. Assoc. Am. Geogr. 91, 622–636. https://doi.org/
Jones, M., Pebley, A.R., 2014. Redefining neighborhoods using common destinations: 10.1111/0004-5608.00262.
social characteristics of activity spaces and home census tracts compared. Schönfelder, S., Axhausen, K.W., 2003. Activity spaces: measures of social exclusion?
Demography 51, 727–752. https://doi.org/10.1007/s13524-014-0283-z. Transp. Policy 10, 273–286. https://doi.org/10.1016/j.tranpol.2003.07.002.
Kain, J.F., 1968. Housing segregation, negro employment, and metropolitan Schwanen, T., Lucas, K., Akyelken, N., Cisternas Solsona, D., Carrasco, J.A., Neutens, T.,
decentralization. Q. J. Econ. 82, 175. https://doi.org/10.2307/1885893. 2015. Rethinking the links between social exclusion and transport disadvantage
Kandt, J., Leak, A., 2019. Examining inclusive mobility through smartcard data: what through the lens of social capital. Transp. Res. Part A Policy Pract. 74, 123–135.
shall we make of senior citizens’ declining bus patronage in the west midlands? https://doi.org/10.1016/j.tra.2015.02.012.
J. Transp. Geogr. 79, 102474. https://doi.org/10.1016/j.jtrangeo.2019.102474. Sevtsuk, A., Ratti, C., 2010. Does urban mobility have a daily routine? Learning from the
Kempen, V., Eva, T., 1994. The dual city and the poor: social polarisation, social aggregate data of mobile networks. J. Urban Technol. 17, 41–60. https://doi.org/
segregation and life chances. Urban Stud. 31, 995–1015. https://doi.org/10.1080/ 10.1080/10630731003597322.
00420989420080911. Shen, Y., 2019. Segregation through space: a scope of the flow-based spatial interaction
Kenyon, S., Lyons, G., Rafferty, J., 2002. Transport and social exclusion: investigating the model. J. Transp. Geogr. 76, 10–23. https://doi.org/10.1016/j.
possibility of promoting inclusion through virtual mobility. J. Transp. Geogr. 10, jtrangeo.2019.02.007.
207–219. https://doi.org/10.1016/S0966-6923(02)00012-1. Shirmohammadli, A., Louen, C., Vallée, D., 2016. Exploring mobility equity in a society
Koo, K.M., Park, C.H., Kim, C.J., 2017. Development of strategies for changing in undergoing changes in travel behavior: a case study of Aachen, Germany. Transp.
physical activity behaviors on older adults with disabilities. J. Exerc. Rehabil. 13, Policy 46, 32–39. https://doi.org/10.1016/j.tranpol.2015.11.006.
676–683. https://doi.org/10.12965/jer.1735144.572. Silm, S., Ahas, R., 2014a. The temporal variation of ethnic segregation in a city: evidence
KOSIS (Korean Statistical Information Service), 2019. Accessed September 22, 2019. from a mobile phone use dataset. Soc. Sci. Res. 47, 30–43. https://doi.org/10.1016/
http://www.kosis.kr. j.ssresearch.2014.03.011.
Kwan, M., 2000. Interactive geovisualization of activity-travel patterns using three- Silm, S., Ahas, R., 2014b. Ethnic differences in activity spaces: a study of out-of-home
dimensional geographical information systems: a methodological exploration with a nonemployment activities with mobile phone data. Ann. Assoc. Am. Geogr. 104,
large data set. Transp. Res. Part C Emerg. Technol. 8, 185–203. https://doi.org/ 542–559. https://doi.org/10.1080/00045608.2014.892362.
10.1016/s0968-090x(00)00017-6. Simpson, L., 2007. Ghettos of the mind: the empirical behaviour of indices of segregation
Kwan, M., 2009. From place-based to people-based exposure measures. Soc. Sci. Med. 69, and diversity. J. R. Stat. Soc. Ser. A (Statistics Soc.) 170, 405–424. https://doi.org/
1311–1313. https://doi.org/10.1016/j.socscimed.2009.07.013. 10.1111/j.1467-985x.2007.00465.x.
Kwan, M., 2013. Beyond space (as we knew it): toward temporally integrated Siren, A., Haustein, S., 2013. Baby boomers’ mobility patterns and preferences: what are
geographies of segregation, health, and accessibility. Ann. Assoc. Am. Geogr. 103, the implications for future transport? Transp. Policy 29, 136–144. https://doi.org/
1078–1086. https://doi.org/10.1080/00045608.2013.792177. 10.1016/j.tranpol.2013.05.001.
Le Roux, G., Vallée, J., Commenges, H., 2017. Social segregation around the clock in the Siren, A., Haustein, S., 2016. How do baby boomers’ mobility patterns change with
Paris region (France). J. Transp. Geogr. 59, 134–145. https://doi.org/10.1016/j. retirement? Ageing Soc. 36, 988–1007. https://doi.org/10.1017/
jtrangeo.2017.02.003. S0144686X15000100.
Lee, J.Y., Kwan, M.P., 2011. Visualisation of socio-spatial isolation based on human Tan, Y., Kwan, M.P., Chai, Y., 2017. Examining the impacts of ethnicity on space-time
activity patterns and social networks in space-time. Tijdschr. Econ. Soc. Geogr. 102, behavior: evidence from the City of Xining, China. Cities 64, 26–36. https://doi.org/
468–485. https://doi.org/10.1111/j.1467-9663.2010.00649.x. 10.1016/j.cities.2017.02.003.
Lieberson, S.E., 1981. An asymmetrical approach to segregation. Ethn. Segreg. Cities 81, Tana Kwan, M.P., Chai, Y., 2016. Urban form, car ownership and activity space in inner
61–82. suburbs: a comparison between Beijing (China) and Chicago (United States). Urban
Liu, Y., Liu, X., Gao, S., Gong, L., Kang, C., Zhi, Y., Chi, G., Shi, L., 2015. Social sensing: a Stud. 53, 1784–1802. https://doi.org/10.1177/0042098015581123.
new approach to understanding our socioeconomic environments. Ann. Assoc. Am. TOPIS (Seoul Transport Opertaion & Information Service), 2021. Accessed February 18,
Geogr. 105, 512–530. https://doi.org/10.1080/00045608.2015.1018773. 2021. https://topis.seoul.go.kr.
Lucas, K., 2012. Transport and social exclusion: where are we now? Transp. Policy 20, Urry, J., 2007. Mobilities. Polity Press, London.
105–113. https://doi.org/10.1016/j.tranpol.2012.01.013. Van den Berg, P., Arentze, T., Timmermans, H., 2011. Estimating social travel demand of
Lucas, K., Bates, J., Moore, J., Carrasco, J.A., 2016. Modelling the relationship between senior citizens in the Netherlands. J. Transp. Geogr. 19, 323–331. https://doi.org/
travel behaviours and social disadvantage. Transp. Res. Part A Policy Pract. 85, 10.1016/j.jtrangeo.2010.03.018.
157–173. https://doi.org/10.1016/j.tra.2016.01.008. Wang, D., Li, F., 2016. Daily activity space and exposure: a comparative study of Hong
Lucas, K., Phillips, I., Mulley, C., Ma, L., 2018. Is transport poverty socially or Kong’s public and private housing residents’ segregation in daily life. Cities 59,
environmentally driven? Comparing the travel behaviours of two low-income 148–155. https://doi.org/10.1016/j.cities.2015.09.010.
populations living in central and peripheral locations in the same city. Transp. Res. Wang, D., Li, F., Chai, Y., 2012. Activity spaces and sociospatial segregation in Beijing.
Part A Policy Pract. 116, 622–634. https://doi.org/10.1016/j.tra.2018.07.007. Urban Geogr. 33, 256–277. https://doi.org/10.2747/0272-3638.33.2.256.
Luiu, C., Tight, M., Burrow, M., 2017. The unmet travel needs of the older population: a White, M.J., 1983. The measurement of spatial segregation. Am. J. Sociol. 88,
review of the literature. Transp. Rev. 37, 488–506. https://doi.org/10.1080/ 1008–1018. https://doi.org/10.1086/227768.
01441647.2016.1252447. Williams, D.R., Collins, C., 2001. Racial residential segregation: a fundamental cause of
Manca, F., Sivakumar, A., Polak, J.W., 2019. The effect of social influence and social racial disparities in health. Public Health Rep. 116, 404–416. https://doi.org/
interactions on the adoption of a new technology: the use of bike sharing in a student 10.1016/s0033-3549(04)50068-7.
population. Transp. Res. Part C Emerg. Technol. 105, 611–625. https://doi.org/ Wong, D., Shaw, S.-L., 2011. Measuring segregation: an activity space approach.
10.1016/j.trc.2019.02.010. J. Geogr. Syst. 13, 127–145. https://doi.org/10.1007/s10109-010-0112-x.
Xia, J., Nesbitt, J., Daley, R., Najnin, A., Litman, T., Tiwari, S.P., 2016. A multi-
dimensional view of transport-related social exclusion: a comparative study of

11
S. Abbasi et al. Journal of Transport Geography 92 (2021) 103025

greater Perth and Sydney. Transp. Res. Part A Policy Pract. 94, 205–221. https://doi. Zulkefli, S.H.B., Barr, A., Singh, A., Carver, A., Mavoa, S., Scheurer, J., Badland, H.,
org/10.1016/j.tra.2016.09.009. Bentley, R., 2020. Associations between public transport accessibility around homes
Yu, H., Shaw, S., 2004. Representing and visualizing travel diary data: a spatio-temporal and schools and walking and cycling among adolescents. Children 7, 30. https://doi.
GIS approach. 2004. Esri Int. User Conf. 1–13. org/10.3390/children7040030.
Zhang, X., Wang, J., Kwan, M.-P., Chai, Y., 2019. Reside nearby, behave apart? Activity-
space-based segregation among residents of various types of housing in Beijing,
China. Cities 88, 166–180. https://doi.org/10.1016/j.cities.2018.10.009.

12

View publication stats