Residential Relocation and Changes in Household Vehicle

Transportation Research Part A 164 (2022) 134–155
Contents lists available at ScienceDirect
Transportation Research Part A

journal homepage: www.elsevier.com/locate/tra
Residential relocation and changes in household vehicle

ownership and travel behavior: Exploring the context of
Hubli-Dharwad twin-cities in India from a planning viewpoint
Chetan Doddamani , M. Manoj *
Department of Civil Engineering, Indian Institute of Technology, Delhi, New Delhi 110016, India
A R T I C L E I N F O A B S T R A C T
Keywords: This article simultaneously analyzes residential relocation, vehicle ownership changes, and travel
Built environment demand changes while accommodating potential self-selection effects to unravel the causal im
Vehicle ownership pacts of built environment features on vehicle ownership and travel behavior. Earlier studies on
Commute travel, Residential relocation
residential mobility have not considered the causal link from objective built environment features
Residential self-selection
India
to subjective built environment, and rarely control for attitudes and objective built environment
features in the empirical analysis. Besides, the literature has not discussed how households
perceive neighborhood in terms of children’s/women’s travel and how that changes by residential
relocation and impact mobility decisions. The paper compares built environment impacts on
mobility decisions in two adjacent areas in India, Hubli-Dharwad twin cities, that vary by socio-
economic and land use contexts. The study notes that Dharwad residents acquired more personal
vehicles than Hubli residents, while the proportion of households who relinquished their vehicles
is high in Hubli. Post movement, Dharwad residents opted for areas with higher diversity and
density than their counterparts in Hubli. In Hubli, households find that their current location does
not help their children use active modes to schools, and women find jobs closer to home. The
statistical analysis revealed that Hubli residents prefer locations with improved travel conve
niences for women and children. With improvements in the children’s travel friendliness,
households of both cities decreased vehicle ownership, whereas household vehicle ownership of
both cities increased with improvements in women’s travel friendliness. In Dharwad, moving to
residential areas characterized by high land-use diversity reduces commute travel, while an in
verse association is noted in Hubli. The statistical analysis did not confirm self-selection effects
due to travel attitudes. In Dharwad, the latent factor on women’s travel friendliness featured in
self-selection and shows an independent effect on vehicle ownership. In Hubli, the factor did not
have independent effects on vehicle ownership after accommodating travel attitudes and socio-
demographics.
* Corresponding author.
E-mail addresses: [email protected] (C. Doddamani), [email protected] (M. Manoj).
https://doi.org/10.1016/j.tra.2022.08.008
Received 20 March 2020; Received in revised form 25 July 2022; Accepted 9 August 2022
Available online 17 August 2022
0965-8564/© 2022 Elsevier Ltd. All rights reserved.
C. Doddamani and M. Manoj Transportation Research Part A 164 (2022) 134–155
1. Introduction
Vehicle ownership in India has been growing1 (Ministry of Road Transport and Highways, 2018; Ministry of Statistical and Pro
grame Implementation, 2018) and has put immense pressure on transportation infrastructure and the environment (NITI Aayog and
The Boston Consulting Group, 2018). Although land-use planning in India has been criticized for its lack of control in outward sprawl
and contributions to longer trips and vehicle ownership (Pucher et al., 2005), built environment-related policies – densification, mixed
land-use, and compact urban areas – are being promoted to encourage low carbon mobility (National Urban Transport Policy, 2014;
Ministry of Housing and Urban Affairs, 2015, 2018). However, the causal impacts of built environment policies on vehicle ownership
and travel demand are seldom explored in Indian cities, and the viability of such policies is rarely discussed.
Globally, there is little consensus among researchers and practitioners on whether the built environment matters in mobility de
cisions (Wang and Lin, 2019). Past studies have revealed that residential self-selection2 influences the association between the built
environment and mobility decisions (Cao et al., 2007a, 2009). The observed correlation between the built environment and mobility
decisions can be due to (confounding) travel attitudes (Wang and Lin, 2019) when residential self-selection exists. If travel choices and
neighborhood preferences influence housing location choice, the observed association between the built environment and travel
decisions will be hindered by residential self-sorting (self-selection) issues (Cao et al., 2009). In such cases, travel behavior (e.g.,
bicycle mode use) is contingent not on the built environment (e.g., availability of bicycle lanes). Numerous studies (e.g., Cao et al.,
2007a) have investigated the causal mechanism underlying the built environment travel behaviour connections; however, most of
these studies employed cross-sectional data. Cross-sectional data helps reveal correlations between the built environment and travel
behavior (e.g., Cervero and Koppelman, 1997; Munshi, 2016; Manoj and Verma, 2016; Doddamani and Manoj, 2021a,b) and fails to
portray causality. Although several longitudinal studies have been undertaken to study the causal impacts of the built environment on
travel behavior, the link from objective built environment features to perceptions of the built environment has been seldom considered
This paper simultaneously analyzes residential relocation, vehicle ownership changes, and travel demand changes while accom
modating potential self-selection effects to unravel the causal impacts of built environment features on vehicle ownership and travel
behavior. Using quasi-longitudinal data, the analysis controls for the impacts of the changes in objective built environment features on
subjective built environment characteristics while assessing the causal impacts of built environment features on vehicle ownership and
travel demand. This approach accommodates the impacts of objective built environment changes on perceptions of residential
neighborhoods, and the ’true’ causal impacts of built environment features on vehicle ownership and travel demand are revealed.
Besides, the paper explores the role of perceived neighborhood social characteristics related to women’s travel friendliness and
children’s travel friendliness in changes in mobility decisions. It is known that children’s/women’s activity participation and travel
needs impact household vehicle ownership and travel (e.g., Rosenbloom, 1987; Verma et al., 2017; Sarangi and Manoj, 2020; Verma
et al., 2020), and built environment measures are correlated with household decisions about children’s/women’s travel (e.g., Bianco
and Lawson, 1997; Manoj and Verma, 2016). However how households perceive residential neighborhoods regarding children’s/
women’s travel and how the perceptions change in a residential relocation and self-selection context and subsequently impact
household mobility decisions are yet to be revealed. This paper reports the findings of the first research incorporating neighborhood
perceptions relating to children and women’s travel in a residential relocation-based travel demand analysis framework. The study also
investigates mobility decisions in two adjacent urban areas planned to be unified and compares and contrasts built environment effects
in the individual urban areas. It is expected that information on such differences and commonalities could inform land-use policy for
developing individual cities into an urban agglomeration.
The remainder of the paper is organized as follows. A brief overview of the related works is presented in the next section. Section
2.1 describes the scopes of the analytical frameworks that explored the impacts of built environment changes on vehicle ownership and
travel behavior. Section 2.2 provides a broad overview of the impacts of different built environment features on vehicle ownership and
travel behavior. A brief summary of studies conducted in the Indian setting is the focus of Section 2.3. The possible extensions to the
existing research are summarized in Section 2.4. Section 3 presents the hypothetical framework utilized to explore research questions,
and the empirical context is discussed in Section 4. A descriptive summary of the data is given in Section 5, and the statistical models
are presented in Section 6. The last section concludes the article.
2. Background
2.1. Simultaneous analysis of built environment changes, long- and short-term mobility decisions
Several studies have explored the impacts of the changes in built environment features (or causal effects) on medium- (car
ownership) and short-term (e.g., mode choice) mobility decisions. Differences exist, however, in the scopes of the analytical frame
works. Cao et al. (2009) investigated how changes in residential preferences influenced auto ownership and driving behavior. The
framework controlled for changes in spaciousness and accessibility to investigate the impacts of residential relocation on mobility
decisions. However, the study did not control for objective land-use measures such as density, diversity, etc., leading to the lack of
1
First position (among 192 countries) in terms of motorcycle and moped (scooter) penetration rate (128 per 1000 persons) 10th ranked for car
penetration rate (23 cars per 1000 persons).
2
According to Ettema and Nieuwenhuis (2017), “Self-selection implies that households locate in places that provide them with conducive
conditions for their preferred way of travelling.”.
135
isolation of confounding issues of such measures with the perceived built environment features. Aditjandra et al. (2012) studied car
ownership and travel behavior changes in a residential relocation analysis framework. The study controlled for travel attitudes in the
framework; however, it did not include objective built environment features such as distance, diversity, and density, making the
models fail to show the ’true effects’ of subjective and objective built environment features on mobility choices. Lin et al. (2018)
followed the framework proposed by Cao et al. (2009) in analyzing the impacts of changes in social networks and neighborhood social
features on vehicle ownership (decisions on car purchase/disposal) and travel behavior (trip frequency, travel time, etc.). However, the
study did not control for travel attitudes in the analytical framework, limiting the scope of detecting residential sorting issues.
Engebretsen et al. (2018) proposed a framework similar to Cao et al. (2009) to investigate the effects of residential location on vehicle
ownership and travel behavior. The study primarily investigated how urban structures impacted vehicle ownership and travel
behavior. Wang and Lin (2019) proposed a panel data-based framework to study the influences of built environment changes on
vehicle ownership and travel behavior. De Vos et al. (2021) identified the indirect effects of the built environment on travel behavior
by examining the mediating effects of built environment features through vehicle ownership and travel attitudes. The study did not
investigate residential sorting issues; however, it shed light on the direct and indirect effects of built environment features on travel
behavior. Scheiner and Holz-Rau (2013) explored the concepts of mobility biographies (e.g., residential relocation) to understand built
environment effects on travel behavior. The analytical approach indicated that changes in household structure influenced relocation to
suburbs, which in turn influenced vehicle ownership and mode choice adjustments. Clark et al. (2016b) utilized longitudinal data to
study how residential relocation impacts car ownership levels after controlling for the direction of movements. The study showed that
relocating to suburban areas increased personal vehicle dependency. Although the study contributed to understanding the causal
impacts of built environment features on mode choice decisions, it did not control for density measures and diversity measures leading
to the confounding effects of such measures in other built environment features. Compared with other analytical frameworks, this
study explicitly accommodated the impacts of the built environment, attitudes, vehicle ownership, and travel behavior at the previous
residential locations on the corresponding variables at the current locations. In addition, the study also controlled for cross-lagged
effects, where the travel attitudes at the previous residential location impact the built environment at the current location.
2.2. Residential relocation and changes in vehicle ownership and travel behavior
Residential relocation provides data at two instances, before and after the movement, and reveals the causal link between (and self-
selection issues) the changes in the built environment and travel decisions (Lin et al., 2018). The effect of built environment measures
on vehicle ownership, after controlling for self-selection, varies by study area and research design. For instance, a quasi longitudinal
analysis of vehicle ownership decisions in Northern California found that changes in the distance to the nearest food outlet maintained
positive impacts on car ownership changes after controlling for travel attitudes and residential preferences (e.g., Cao et al., 2007a).
However, Cao et al. (2007b) confirmed self-selection in cross-sectional analysis using the Northern California data. The quasi-
longitudinal analysis revealed only a weak correlation between built environment measures and vehicle ownership. Cao and Cao
(2014) conducted a quasi-longitudinal analysis to identify the effects of light rail transit on car ownership in the Minneapolis-St. Paul
metropolitan area, USA. The cross-sectional analysis revealed the impacts of travel attitudes than built environment features on vehicle
ownership, pointing to self-selection issues. However, none of the travel attitude measures were found significant in the quasi-
longitudinal analysis of car ownership. In addition, the study did not find an independent effect of the light rail on car ownership
levels. The spatial variations of built environment impacts on vehicle ownership are also noted in a longitudinal analysis. Cao et al.
(2019) noted that residents of Oslo decreased auto ownership than Stavanger households, perhaps due to the transit-amenable land use
of Oslo. Although travel attitudes are known to impact built environment features (self-selection), reverse causality does matter in the
built environment travel behavior nexus. Ignoring reverse causality might over/underestimate self-selection issues in cross-sectional
and quasi-longitudinal analyses.
Longitudinal/quasi longitudinal studies related to travel behavior have reported significant impacts of built environment features
on travel behavior. Vehicle miles traveled are observed to decrease with increased accessibility (Krizek, 2003; Handy et al., 2005; Cao
et al., 2007a; Prillwitz et al., 2007). Accessibility measures are also found to exert stronger effects than travel attitudes on vehicle miles
traveled (Handy et al., 2005; Aditjandra et al., 2012). However, neighborhood accessibility can also increase the number of tours and
the number of trips/tours, perhaps indicating the effects of travel cost reductions with improvements in neighborhood accessibility
(Handy et al., 2005). Travel demand is also impacted by relocation distance as well as relocation sites. In Copenhagen, trip frequency
and trip length were increased for those who moved away from the CBD of the city (Næss, 2005). Buchanan and Barnett (2006) reached
the same conclusion in New Zealand. Distances to work, shopping, and education were increased for those who relocated to rural areas
in New Zealand.
The (quasi) longitudinal studies have also shown that changes in social connections and social dimensions of neighborhoods impact
mobility decisions (e.g., Clark et al., 2016b). In the British context, Aditjandra et al. (2012) noted that moving to a socially active
neighborhood decreased the car use of residents. While relocating to areas having higher social deprivation increases car use, as Clark
et al. (2016a) concluded in a study conducted in the UK. The changes in social connections in the neighborhood can affect non-
motorized mode use. Lin et al. (2018) found that an increase in social contact after the residential relocation decreases the share of
non-motorized modes in daily travel. Recent studies have shown that neighborhood preferences related to amenities, exercise, etc., can
influence mobility decisions. In Oslo and Stavanger, the ‘children factor’ increased car ownership, and the positive correlation was
significant in cross-sectional and quasi-longitudinal analyses (Cao et al., 2019).
136
2.3. Built environment-travel behavior studies in India
Compared to the international context, relatively fewer studies have investigated built environment travel behavior connections in
Indian cities, and most studies utilize cross-sectional data. Indian studies have found that vehicle ownership varies by city and land-use
settings. Compared with households of other megacities (Delhi, Bangalore, etc.), Kolkata residents are less likely to own cars, as Bansal
et al. (2018) concluded. The streetcar network in Kolkata could be a disincentive to vehicle ownership, all else equal. Many Indian
studies have noted that population density decreases car ownership levels and increases motorcycle ownership (Goel and Mohan,
2020; Banerjee, 2012). Traffic congestion and parking limitations concur with densification and could act as a disincentive to car
ownership. In contrast, a motorcycle could be easily operated in congested environments and requires only a small area for parking.
However, commute distance, housing density, and transit accessibility can decrease motorcycle ownership (Shirgaokar, 2012).
Population density can promote walking and transit use in Indian cities (Manoj and Verma, 2015a; Munshi, 2016), perhaps indicating
the closeness of activities and transit favoring conditions in dense areas. The likelihood of using motorcycles also increases with
population density (Manoj and Verma, 2015a). Land use heterogeneity is another important built environment feature impacting
vehicle ownership and mode choice decisions in Indian cities. It can increase the likelihood of using public transit and walking
(Munshi, 2016) and reduce non-work trip distances (Sarkar and Mallikarjuna, 2016; Rahul and Verma, 2017). Land use heterogeneity
can also increase out-of-home maintenance activity durations (Manoj and Verma, 2015b), the likelihood of recreational and shopping
activity participation (Manoj and Verma, 2015a), and travel distances of non-workers (Manoj and Verma, 2017).
Studies utilizing longitudinal data to unravel built environment impacts on travel decisions are minimal in Indian cities. The studies
based on (quasi) longitudinal design data have shown that the availability of shopping opportunities in the proximity of residents
decreases car ownership (Srinivasan et al., 2007). Such studies have also concluded that perceived neighborhood dimensions such as
driving safety and walking safety (Gopisetty and Srinivasan, 2013) and social connectivity factors, such as mobile phones (Srinivasan
et al., 2007) also impacts vehicle ownership and travel behavior.
2.4. Gaps in knowledge and scope for further reserach
The literature review presented above suggests the following points for further research:
i. Studies rarely control for all objective (density, diversity and distance) built environment measures. The lack of control for such
measures could lead to omitted variable bias and confounding issues (e.g., accessibility factor could mask effects of land use
diversity) in the models.
ii. The perceptions of the built environment (or subjective built enviromet features) could be influenced by the objective built
environment at a residential location. Studies, so far, have not considered the causal link from objective built environment
featurs to subjective built environment, limiting the possibility of revealing the ‘true’ causal impacts of objective built envi
ronment on mobility decisions through perceived built environment features.
Fig. 1. Conceptual Framework.
137
iii. Most studies have focused on car ownership decisions. However, in several countries motorcycles are also part of household
mobility ‘bundles’. Hence, the focus on total motorized household vehicles could be another avenue for research.
iv. Several built environment policies aim to enhance the quality of life and safety of women and children (e.g.,Ministry of Urban
Development, 2015). However, the literature has not discussed how households perceive neighborhood in terms of children’s/
women’s travel and how that changes by residential relocation and impact mobility decisions.
v. Little knowledge exists about the effects of residential relocation on vehicle ownership and travel behavior in Indian cities.
This paper proposes an empirical frameowork and accommodates the above-identifed points for future research.
3. Conceptual framework
The conceptual framework proposed in the study is inspired by the earlier works of Handy et al. (2006), Cao et al. (2007b),
Aditjandra et al. (2016), and Lin et al. (2018). These studies are based on the travel demand analysis framework proposed by Ben-Akiva
et al. (1996). The main differences between the present study and earlier works are: First, the analysis framework investigates the
impacts of objective built environment variables on subjective built environment features. Such association has rarely been tested in
analyzing the impacts of residential relocation on vehicle ownership and travel behavior. Second, our work considers total household
vehicles (sum of cars and motorcycles) instead of car ownership levels in previous works. Several Indian households own motorcycles
besides cars, and the focus on total household vehicles for modeling captures the context of Indian cities. Third, the present study
includes subjective built environment features such as accessibility, perceived neighborhood social aspects such as safety, women
travel friendliness, children travel friendliness, attractiveness, and cleanliness. As observed in past studies (e.g., Lin et al., 2018),
changes in the residential environment change a household’s social environment, further triggering travel behavior changes. While
Fig. 2. Map of the study area.
138
past studies have also included latent neighborhood safety measures, the inclusions of latent perceived measures related to women and
children travel friendliness is an important contribution. Fourth, the present study includes commute time and distance as travel
behavior variables. From an activity-based approach to demand analysis (Golob, 2000), time allocation to activity and travel shapes
individuals’ travel mode and distance decisions. However, very few studies have incorporated the analysis of commute time and
distance changes in a residential relocation framework.. The conceptual framework proposed in the study is shown in Fig. 1.
In the framework, the built environment changes –subjective and objective land-use measures and social aspects, changes in total
household vehicles, and household heads’ travel behavior are considered endogenous variables. The main exogenous variables are
changes in the socio-demographic variables and current travel attitudes. The ’causal’ model explores the following associations.
Changes in the built environment are influenced by changes in socio-demographics and current travel attitudes. Total vehicle
ownership and travel behavior are influenced by changes in the socio-demographic variables and current travel attitudes (Handy et al.,
2006; Cao et al., 2007b). Built environment changes are thought to influence vehicle ownership, travel behavior changes, and the
reverse associations are assumed to be invalid from a behavioral viewpoint. A bidirectional association is assumed between changes in
total vehicle ownership and travel behavior and between travel behavior indicators: commute time and commute distance. The
objective built environment features are conceived to impact the subjective built environment, and the revese association is invalid.
The self-selection effect is controlled for by regressing changes in socio-demographics and travel attitudes on changes in both sub
jective and objective built environment measures. The conceptual framework is estimated using the structural equations modeling
(SEM) approach.
4. Empirical context
The research utilizes primary travel behavior data gathered from Hubli-Dharwad twin cities in South India. Fig. 2 shows the map of
the twin cities, highlighting land use mix variations, bus stop locations, and the BRTS corridor linking the cities. The variables used to
assess the land use composition measure is the percentages of residential, commercial, and other land uses. The two cities differ in their
socio-economic conditions, spatial developments and growth, and vehicle ownership (Census of India, 2011; Hubli-Dharwad Bus
Rapid Transit System Company ltd., 2018a; Dharwad District Statistics, Government of Karnataka, 2018). For instance, Dharwad’s
population is nearly-one-third of that of Hubli and has low vehicle ownership. Regarding spatial development, Dharwad city hosts
many institutional, public, and semi-public activities. Besides, the city also has dedicated areas for parks and open spaces. While low on
parks and open areas, Hubli city has a high intensity of industrial and commercial activities than Dharwad.
The study uses five sets of variables: socio-demographic characteristics, household vehicle ownership and travel details, objective
neighborhood features, built environment perceptions, and travel attitudes. The changes in vehicle ownership and travel behavior
were captured using a retrospective, quasi-longitudinal survey targeting the residents who relocated within and between Hubli and
Dharwad cities from 2008 to 2018. The survey was conducted by trained volunteers between September 2018 and January 2019,
following the face-to-face survey administration method. The survey recorded household demographics, travel behavior, vehicle
ownership levels, and perceptions of the built environment before and after the relocation. Travel attitudes pertain only to the current
location, as it is assumed that attitudes would not have changed over time (e.g., Cao et al., 2007b; Ettema and Nieuwenhuis, 2017).
The sample size was estimated based on the Cochran formula (Cochran, 2007) for proportions. In this study, the proportion of
residential movers in Dharwad (9 %) and Hubli (13.5 %) are considered to arrive at the minimum sample size. The final survey resulted
in responses from 392 households in Dharwad and 398 households in Hubli. The survey response rate is 54 % in Dharwad, whereas, in
Hubli, 46 % of the approached households provided complete information. The distributions of key sample statistics are similar to their
population trends. For Hubli, the sample averages of household size and the number of married individuals are 4.53 and 1.61,
respectively. The population values of the corresponding variables are 4.65 and 1.72, respectively. The sample averages of these
variables in Dharwad are 4.56 and 1.25, respectively. The corresponding population values are 4.44 and 1.32, respectively. The
differences between the sample and population values are insignificant at the 5 % level in both cities. The sample statistics are also
comparable with their distributions reported from other Indian cities. For instance, in a related study conducted in Chennai, Gopisetty
and Srinivasan (2013) reported an average household size of 4.49 persons, nearly equal to the values reported above for Hubli and
Dharwad.
5. Descriptive analysis
5.1. Changes in socio-demographic characteristics
Table 1 presents a descriptive summary of the changes in the socio-demographic and economic environments of the households.
The table entries show the difference between a variable’s values at the current residential location and the value of the corresponding
variable at the previous housing location (e.g., Cao et al., 2007b, Aditjandra et al., 2016). Thus, a negative entry (e.g., ‘-1′ ) in the table
represents ’loss’ (‘-1′ implies decrease by 1 unit) and a positive value indicates ’gain’ in a particular variable. Regarding vehicle
ownership (cars, motorcycles, and total vehicles), the trends suggest that Dharwad residents acquired more personal vehicles than
Hubli residents. In contrast, the proportion of households who relinquished their vehicles is high in Hubli. The differences might be
pointing to households’ aspirations of an emerging city (Dharwad) to acquire vehicles and the difficulties faced by households of a
developed city (Hubli) in managing vehicle operating costs. In both cities, nearly 58 % of families secured homes, i.e., moved from a
rented house at the previous location to their own home at the current location or from their own house in the last location to own a
home in the current location. The proportion of households moving to an independent house is relatively high (72 %) in Dharwad than
139
Table 1
Summary of changes in socio-demographic variables, land-use measures, and commute behavior.
Variables Levels Summary
Dharwad City Hubli City
Car ownership changes [%] − 1 5.40 11.80

0 58.90 62.3
≥1 35.70 26.40
Motorcycle ownership changes [%] − 1 6.90 8.80
0 50.50 58.5
≥1 42.70 32.61
Total vehicle ownership changes [%] − 1 8.22 11.62
0 33.70 45.00
≥1 58.20 43.51
Homeownership [%] Rented to Own or Own to Own [1] 57.40 59.30
Own to Rented or Rent to Rented [0] 42.60 40.70
Dwelling type [%] Flat to Independent or Independent to Independent 72.20 34.20
[1]
Flat to Flat or Independent to Flat [0] 27.80 65.80
Changes in the number of rooms [%] 3 to 2 BHK# 8.50 14.40
No change in BHK# 44.90 43.70
2 to 3 BHK# 43.40 36.90
2 to 4 BHK# 3.30 5.00
Changes in earning members [%] ≤− 1 10.20 8.90
0 70.70 77.10
1 19.10 14.10
Changes in household size [%] − 1 25.50 26.30
0 53.60 63.80
1 20.90 10.80
Changes in driving license [%] − 1 8.70 8.80
0 71.40 74.10
1 19.90 17.10
Household Earning and Expenditure (INR) [INR 1 = USD Household Income/month 6055 (23035) 5538 (20491)
0.013] Household Expenditure /month 551 (4861) 1363 (5340)
Household travel expenditure /month 53 (2275) 414 (2434)
Property tax 1108 (2471) 6235 (2783)
Changes in the number of children (age ≤ 16 years) [%] ≤− 1 21.90 16.40
0 54.10 68.10
1 24.10 15.60
Changes in the number of married individuals [%] − 1 18.60 15.30
0 67.10 77.90
1 14.30 6.80
Changes in distance from home to [meter] Bus stop − 62 (536) − 695 (1908)
Hospital − 1637 (3820) − 1515 (4947)
Parks − 1463 − 587 (2056)
(3305.34)
Education − 2395 (5214) − 147 (4767)
Place of worship − 176 (940) − 789 (2635)
City center − 326 (1762) 423 (1248)
5
Changes in land-use diversity 0.019 (0.23) − 0.026 (0.20)
Changes in density measures [prs/ha] Population density 0.30 (28.39) − 98.41
(283.76)
Density of literate individuals 2.01 (33.19) − 76.22
(220.02)
Public transport network density [m/m2] − 0.0043 (0.047) − 0.0370
(0.216)
Gender [%] Male [1] 65.30 71.35
Female [0] 34.70 28.64
Average of completed age (in years) 45.78 (8.97) 45.99 (8.83)
Employment [%] Employed in the Government sector 42.59 44.97
Employed in the Private sector 40.25 26.38
Retired person /homemakers 4.93 5.27
Running business 12.20 23.12
Average of changes in commute distance 12.85 (14.74) 8.13 (9.51)
Average of changes in commute time 13.67 (16.28) 10.92 (12.26)
Sample Size 392 398
#
BHK – Bedroom Hall Kitchen; Standard deviation in brackets.
r 1 c 1 o 1
| − |+| − |+| − |
5
Land use mix diversity = 1 - L 3 L 3 L 3 (ref: Bhat and Guo, 2007); Where r = residential, c = commercial, o = others, L = total
4
3
area.
140
Hubli, perhaps indicating the differences in land availability and housing costs between the two cities. The share of households that
acquired bigger homes (in terms of BHK) is slightly higher in Dharwad compared with Hubli (47 % vs 41 %), possibly showing the
effects of low housing cost in Dharwad.
Regarding the changes in household size and the number of earners post-relocation, the findings are interesting. It is observed that
about 21 % of households in Dharwad saw the addition of a family member after the relocation. The share of households that saw the
addition of an earner is close to 20 %. In Hubli, the rate of addition of an earner and a household member is lower than that for
Dharwad. Post-relocation, the increase in driving license holders in Dharwad city was marginally higher (20 %) than Hubli (17 %). The
household earnings and expenditure details are evaluated in terms of monthly income changes, monthly household expenditure,
monthly travel expenditure, and property (house) tax.
In general, both households earn an additional INR 6000 post-relocation. However, a Hubli resident experienced increased monthly
household expenditure, travel expenditure, and property tax after the relocation. An average Hubli resident experienced a property tax
increase nearly-six times more than that experienced by Dharwad residents. In terms of the number of children, about 24 % of Dharwad
residents witnessed the addition of at least one child. In contrast, the proportion of households that seen the addition of at least one
child in Hubli is comparatively lower than Dharwad. A similar trend is observed in the addition of married individuals. The average age
of household heads is found to be the same between the cities. Also, in both cities, household heads are predominantly males, which is a
Table 2
Summary of variations in perceptions about the built environment and current travel attitude.
Variables Dharwad Hubli
Current location Previous location Current location Previous location
Accessibility in the neighborhood

Easy access to groceries 3.20 3.06 3.32 3.26
Paratransit is not easily accessible 2.27 2.18 2.58 2.40
Easy access to main road 3.25 3.03 3.38 3.36
Easy access to bus stop 3.12 2.98 3.24 3.26
Other amenities are easily accessible 3.26 3.06 3.28 3.30
City center is easily accessible (CBD) 2.69 2.64 2.71 3.13
Parks and playground are not easily accessible 2.87 2.63 2.92 3.65
Safety in the neighborhood
Low crime rate 3.56 3.28 3.50 3.25
Safe from pollution and accidents 3.31 2.95 3.37 3.05
I feel safe walking in this neighborhood 3.57 3.30 3.57 3.33
I feel unsafe while driving in this neighborhood 1.77 1.88 1.87 2.1
Good street light 3.41 3.08 3.34 3.27
Safe for children 3.57 3.16 3.49 3.25
Safe for women 3.53 3.25 3.57 3.28
Safe from stray animals and strangers 3.33 3.04 3.30 3.16
Social factors in the neighborhood
Worship place availability 2.34 2.12 2.84 2.59
Similar status 3.30 2.84 3.25 2.34
Active welfare and interactive neighbors 3.10 3.01 3.17 3.20
Frequent social gatherings 3.01 2.95 3.20 3.07
Diverse age, gender and religion 3.20 3.10 3.45 3.09
Children have a lot of friends 3.25 3.09 3.22 3.19
Homemakers are able to make friends 3.29 3.11 3.30 3.23
Attractiveness and cleanliness
Good landscaping 3.51 2.63 3.25 2.72
Attractive streets 3.42 2.58 3.29 2.76
Good housing styles 3.52 2.54 3.43 2.84
Dust free roads 3.16 2.44 3.02 2.71
Choke free drains 3.37 2.80 3.25 2.89
Children travel needs in the neighborhood
Neighborhood helps my children to walk/cycle to school 2.30 2.27 2.32 2.49
Schools are nearby 2.72 2.62 2.83 2.82
I do not have to escort my children to school 2.99 2.90 2.95 2.89
Women travel needs in the neighborhood
Women can walk to work and other places 2.64 2.44 2.83 2.68
Women can find job closer to home 2.46 2.38 2.70 2.97
Women do not have to make long trips 3.03 2.84 2.97 2.89
Shopping is easier for women 3.18 3.06 3.16 3.18
Current travel attitude#
I like to drive car 4.21 3.99
I like to use car only for long distance travel 4.42 4.81
I like to travel a lot 3.72 3.82
I like to make short trips using two-wheeler 4.35 4.27
Vehicle ownership improves ones status in the society 3.20 2.59
Sample Size 392 398
Table contents after Doddamani and Manoj (2021a); Doddamani and Manoj (2021b).
141
typical Indian scenario. Finally, most household heads are currently employed in the government sector, followed by the private
sector. The above-described variations in the socio-demographic levels may have implications for mobility patterns in the cities.
5.2. Changes in built environment characteristics
Residents experience changes in built environment characteristics due to relocation, as evident from Table 1. The objective built
environment measures considered for the analysis include distances to various activities from a respondent’s home, land use mix
diversity of residential zones and various density measures. The land-use diversity measures for the study zones were estimated using
the expression given in Bhat and Guo (2007). Table 1 shows that a household in Hubli city moved closer to a bus stop than a resident in
Dharwad, where the proximity to bus stops does not change significantly. The reductions in the distance to hospitals post-residential
relocation in both cities are almost the same. In Hubli, post-relocation, an average household did not go far from playgrounds/parks. As
discussed before, Hubli city has fewer areas allocated for parks and open spaces than Dharwad. Thus, an average Hubli resident would
be valuing their presence more than a Dharwad resident does and decide to stay closer to the facilities. In Dharwad, an average resident
moved closer to the CBD area, as indicated by the negative value for the changes in the CBD distance. This finding may indicate
households’ preferences for living closer to various activities and opportunities in an emerging city context. Whereas in Hubli, an
average resident moved away from the CBD, perhaps indicating the influences of living cost and congested neighborhoods on location
choices in a developed city context. Post-relocation, Dharwad residents opted for areas with higher diversity than their counterparts in
Hubli. In addition, Dharwad residents also moved to locations having high densities in comparison with their previous locations. In
Hubli and Dharwad, households witnessed reductions in transit network density post-relocation. An average Hubli household wit
nessed a reduction in network density at a greater magnitude than an average Dharwad resident. These differences point to the
variations in living conditions and aspirations for better living situations that suit households’ economic situations.
5.3. Travel attitudes and changes in perceptions about the built environment
Table 2 presents the variations in households’ ratings of built environment features at the current and previous residential loca
tions. Household heads were asked to indicate their perceptions about the built environment (e.g., neighborhood accessibility) and
social aspects (e.g., attractiveness, and cleanliness women and children travel convenience) at their current and previous residential
locations on a Likert scale ranging from 1 (not at all true) to 4 (entirely true). Travel attitude is measured at the current location and on
a five-point Likert-type scale (1 = strongly disagree to 5 = strongly agree). Aditjandra et al. (2012), Handy et al. (2006), Cao et al.
(2007b), and Handy et al. (2005) also measured travel attitude at the current residential location owing to the reason that retrospective
measurements of attitudes lead to unreliable data. In addition, due to non-response issues, residential preferences (across questions for
neighborhood perceptions) were not captured in the survey; this is an limitation of the study.
Regarding accessibility-related measures, the trends in responses suggest that an average Dharwad respondent moved to places that
have improved accessibility to various facilities. On the other hand, Hubli residents observed slight reductions in accessibility to
various amenities after residential relocation. The trends in responses towards various safety aspects of neighborhoods are consistent
between the two cities: the households witnessed improvements in neighborhood safety after the relocation. In general, Hubli and
Dharwad cities’ households saw improvements across several social aspects of the neighborhood considered in the study. As per Lin
et al. (2018), changes in social factors influence residential location choices and mobility decisions. A significant change is seen
regarding the neighborhood’s attractiveness and cleanliness. In general, residents of the cities opine that they have an attractive and
clean neighborhood at the current location compared to their previous residential areas. In Hubli, households find that their current
location does not help their children use active modes to schools compared to their previous location. As revealed by the ratings,
residents of both cities witnessed improvements for all other measurement items related to children’s travel convenience. Both city
dwellers find that they do not have to escort children at the current location compared to the previous locality. Dharwad residents
reported improvements in new locations regarding women’s travel convenience than their previous residential areas. In terms of
absolute magnitudes, greater improvements can be seen with the ease of conducting shopping and the possibility of making short trips.
Both city dwellers are almost equal in their rating on shopping convenience at the current residential location. In Hubli, the residents
find that women can easily find a job in their previous residential location than the current one. For the remaining items, Hubli
residents find improvements at the current location. The travel attitudes3 at the current locations show that Dharwad residents have
more propensity to drive cars. The residents also opine that car ownership improves their status in society. Hubli residents attach
higher importance to long-distance travel by car when compared to Dharwad households.
6. Model results
The conceptual model presented in Fig. 1 was implemented separately for the two cities using SEM, and the estimation results are
summarized in Tables 3 to 5. The model estimation started with a specification with only correlations between variables. Subsequently,
links/paths between variables were added where the correlations were found significant. The iterative approach revealed that vehicle
ownership decisions influence travel behavior, and the reverse association is invalid. The finding shows that vehicle ownership is a
3
The survey also included measurement items related to public transit, NMT modes, and the general environment. However, the factor related to
private vehicle was the only significant latent construct appeared in the multivariate model presented in the next section.
142
Table 3
Direct Effects of socio-demographic characteristics on subjective and objective built environment features.
Dharwad (N ¼ 392)
Change in Subjective built environment characteristics Objective built

explanatory environment
variables characteristics
Accessibility Safety Social Attractiveness and Children Women Land use Distance to
factors cleanliness travel travel mix CBD
Gender (Male) − 0.057 − 0.088 − 0.015 − 0.042 (-0.044) − 0.016 − 0.067 0.111* − 0.036*
(-0.55) (-0.086) * (-0.016) (-0.017) (-0.069)
Housing tenure − 0.068 − 0.034 0.028 − 0.020 (-0.022) − 0.041 0.057 − 0.125* 0.053
(-0.066) (-0.033) (0.030) * (-0.043) (0.059)
Dwelling type − 0.076 0.086 − 0.069 − 0.074 (-0.076) − 0.014 0.053 − 0.044 − 0.022
(-0.077) (0.089) * (-0.068) (-0.016) (0.055)
No. of earners in 0.053 (0.055) 0.095 − 0.014 − 0.032 (-0.033) 0.018* 0.055 0.098* − 0.080*
household (0.098) * (-0.016) (0.021*) (0.056) *
Households income − 0.038 − 0.032 0.089* − 0.036 (-0.38) 0.022 0.044 0.075* 0.053
(-0.039) (-0.034) (0.091) (0.024) (0.046)
Size of house 0.104* − 0.038 0.088 0.113 (0.116) * 0.018 0.072 0.015 − 0.009
(0.111*) (-0.038) (0.090) (0.021) (0.074) *
Household members 0.045 (0.048) 0.022 0.026 − 0.136 (-0.137) * − 0.023 − 0.022 0.059* − 0.015
* (0.025) (0.027) (-0.024) (-0.024)
No. of children in 0.061 (0.063) 0.034 0.018* 0.075 (0.076) * 0.038 0.015 0.064* 0.033
home * (0.036) (0.021) (0.041) (0.018)
No. of married 0.043 (0.045) − 0.022 0.042 − 0.012 (-0.014) 0.162 0.016 − 0.164* 0.102*
individuals (-0.026) (0.044) (0.165) * (0.018)
No. of driving licenses − 0.102 − 0.032 − 0.026 0.018 (0.021) − 0.125 − 0.078 0.019 − 0.028
(-0.105) * (-0.033) * (-0.025) (-0.126) * (-0.079)
Property tax − 0.065 0.042 0.049 0.082 (0.084) * 0.052 0.052 − 0.072* 0.044
(-0.066) (0.044) (0.051) (0.055) * (0.055)
Pro-private vehicle − 0.061 0.088
Land use mix 0.074 (0.076) 0.412 − 0.091 − 0.045 (-0.048) 0.081 0.044
(0.421) * (-0.092) (0.082) * (0.052)
Distance to CBD − 0.170 0.062 − 0.106 − 0.084 (-0.086) 0.178 − 0.084
(-0.168) * (0.063) (-0.107) * (0.180) * (-0.086)
Hubli (N ¼ 399)
Gender (Male) 0.012 (0.14) 0.015 0.036 − 0.036 (-0.038) − 0.033 − 0.070* 0.089* − 0.084*
(0.016) (0.037) (-0.036) (-0.072*)
Housing tenure 0.015 (0.016) 0.045 0.084 0.081* (0.082) − 0.011 0.014 − 0.026 0.048
(0.044) (0.086) * (-0.012) (0.016)
Dwelling type − 0.122 − 0.010 − 0.035 0.008 (0.009) − 0.065* − 0.060* − 0.046 0.136*
(-0.123) * (-0.011) (-0.036) (-0.068) (-0.062)
No. of earners in − 0.197 0.112* 0.044 − 0.099* (-0.097) − 0.036 − 0.161* 0.029 0.028
household (-0.198) * (0.115) (0.046) (-0.038) (-0.165*)
Household income − 0.098 0.046 0.036 0.031 (0.032) 0.163* 0.026 0.089* 0.018
(-0.099) * (0.048) (0.038) (0.165*) (0.028)
Travel expenditure 0.083 (0.86) * 0.064* 0.047 0.086* (0.088*) − 0.072* − 0.012 − 0.128* − 0.035*
(0.066) (0.049) (-0.074*) (-0.015)
Size of house − 0.025 − 0.097 0.016 0.098*(0.099*) 0.097* 0.137* 0.006* 0.123
(-0.026) (-0.098) (0.018) (0.099*) (0.138*)
Household members − 0.025 − 0.028* 0.014 − 0.109*(-0.111*) − 0.009 − 0.036 0.049 − 0.008
(-0.028) (-0.031) (0.016) (-0.008) (-0.034)
No. of children in − 0.126 − 0.104* − 0.108* − 0.149 (-0.151) − 0.024 0.084* 0.044 − 0.066*
home (-0.125) * (-0.106) (-0.109*) (-0.026) (0.086)
No. of married 0.154 (0.155) 0.016* − 0.036 − 0.020 (-0.022) 0.146* 0.065* − 0.028 − 0.008
individuals * (0.018) (-0.038) (0.148*) (0.068)
No. of driving license − 0.009 − 0.009 − 0.034 0.048 (0.049) 0.035 0.119* − 0.068* 0.069*
(-0.008) (-0.010) (-0.036) (0.036) (123*)
Property tax 0.147 (0.148) − 0.048 0.013 − 0.109* (-0.111*) − 0.117* − 0.156 − 0.035* − 0.041
* (-0.052) (0.015) (-0.118*) (-0.158)
Pro-private vehicle 0.030 − 0.092
Land use mix 0.012 (0.014) − 0.073 0.119 − 0.037(-0.038) − 0.096 − 0.099 – –
(-0075) (0.121) * (-0.099) * (-0.012)
Distance to CBD − 0.241 − 0.090 − 0.025 0.067 (0.068) * − 0.058 − 0.170 – –
(-0.248) * (-0.091) (-0.026) (-0.059) (-0.172) *
**
Significant at p ≤ 0.05; Total effects in parenthesis.
143
Table 4
Dharwad city: Direct and total effects of exogenous variables on endogenous variables.
Change in explanatory variables Changes in
Total vehicle ownership Commute distance Commute time
Gender 0.028 (0.034) * − 0.048 (-0.079) * − 0.018 (-0.014)

Housing tenure 0.108 (0.205) * 0.126 (0.214) * 0.132 (0.206) *
Dwelling type 0.085 (0.110) * 0.085 (0.092) * 0.058 (0.066) *
No. of earners in household 0.208 (0.326) * 0.007 (0.009) − 0.008 (-0.006)
Households’ income 0.122 (0.325) * 0.226 (0.228) * 0.168 (0.212) *
Household travel expenditure 0.028 (0.054) 0.096 (0.072) * 0.098 (0.079) *
Size of house 0.065 (0.126) * 0.036 (0.056) * 0.054 (0.062)
Household members 0.085 (0.118) * 0.062 (0.078) * 0.062 (0.068) *
No. of children in home 0.058 (0.046) * − 0.022 (-0.018) − 0.022 (-0.014)
No. of married individuals − 0.065 (-0.058) * − 0.016 (-0.021) − 0.048 (-0.052)
No. of driving license 0.206 (0.208) * − 0.024 (-0.038) − 0.014 (-0.032)
Property tax − 0.061 (-0.048) − 0.038 (-0.044) * − 0.026 (-0.014)
Pro private vehicle 0.042 (0.043) 0.072 (-0.054) * 0.034 (-0.032)
Accessibility − 0.051 (-0.051) 0.024 (0.038) − 0.040 (-0.044)
Safety − 0.075 (-0.089) * 0.034 (0.056) 0.008 (0.022)
Social − 0.037 (-0.042) − 0.018 (-0.024) − 0.036 (-0.024)
Attractiveness and cleanliness 0.162 (0.215) 0.008 (0.011) 0.052 (0.056)
Children travel convenience − 0.056 (-0.058) * 0.076 (0.044) * 0.092 (0.064) *
Women travel convenience 0.116 (0.287) * − 0.068 (-0.014) − 0.068 (-0.088) *
Land use mix 0.014 (0.076) − 0.028 (-0.026) − 0.009 (-0.012)
Distance to CBD − 0.012 (-0.015) 0.084 (0.016) * 0.068 (0.076)
Change in total vehicle ownership − 0.028 (-0.016) − 0.014 (-0.026)
*
Table 5
Hubli city: Direct and total effects of exogenous variables on endogenous variables.
Change in explanatory variables Changes in
Total vehicle ownership Commute distance Commute time
Gender − 0.068 (-0.064) * 0.022 (0.014) 0.026 (0.034)

Housing tenure − 0.019 (-0.023) 0.038 (0.021) 0.031 (0.043)
Dwelling type 0.114 (0.326) * 0.055 (0.011) 0.018 (0.016)
No. of earners in household 0.158 (0.071) * 0.059 (0.061) * 0.033 (0.052)
Households’ income 0.056 (0.052) * 0.059 (0.028) 0.019 (0.029) *
Household travel expenditure 0.154 (0.184) * 0.139 (0.126) 0.184 (0.282) *
Size of house 0.019 (0.027) 0.088 (0.094) 0.050 (0.067) *
Household members 0.047 (0.039) 0.071 (0.034) 0.049 (0.015)
No. of children in home 0.054 (0.034) * − 0.146 (-0.134) * − 0.144 (-0.417)
No. of married individuals 0.029 (0.037) − 0.065 (-0.059) * − 0.075 (-0.071) *
No. of driving license − 0.014 (0.004) * 0.013 (0.032) 0.005 (0.003)
Property tax − 0.162 (0.067) * 0.015 (0.013) 0.064 (0.083)
Pro-private vehicle 0.071 (0.029) * − 0.023 (-0.051) 0.025 (0.624)
Accessibility − 0.153 (-0.08) * 0.023 (0.044) − 0.006 (-0.005)
Safety 0.009 (0.024) 0.076 (0.064) * 0.022 (0.044)
Social 0.108 (0.092) * − 0.027 (-0.039) − 0.077 (-0.073)
Attractiveness and cleanliness 0.083 (0.082) * − 0.173 (-0.157) * − 0.042 (-0.063) *
Children travel convenience − 0.089 (-0.079) * 0.005 (0.003) − 0.071 (-0.066)
Women travel convenience 0.017 (0.026) − 0.031 (-0.028) − 0.018 (-0.053)
Land use mix 0.064 (0.036) 0.014 (0.012) 0.044 (0.041) *
Distance to CBD 0.064 (0.041) * 0.034 (0.043) 0.066 (0.061)
Change in total vehicle ownership − 0.088 (-0.265) − 0.065 (-0.039)
*
medium-term decision that further affects short-term choices such as daily travel behavior (Van-Acker and Witlox, 2010). In an earlier
study, Cao et al. (2007b) found the causal link between travel behavior and vehicle ownership insignificant in the US data.
Furthermore, the bidirectional links between commute time and commute distance changes did not yield statistically significant es
timates. Hence, the study could not reveal the association between time-use decisions and travel demand (Golob, 2000) and estimated
correlation coefficients between these variables.4 The model-fit indices are satisfactory. The ratio between the sample size and the
number of independent variables is>15 for both models and supports the application of maximum likelihood estimation (Stevens,
4
For Dharwad, the coefficient is 0.924 (p=0.00) and for Hubli it is 0.846 (p=0.00).
144
1996).
Besides, the samples meet minimum sample size criteria (minimum of 200 is acceptable, e.g., Schumacker and Lomax, 2004; Kline,
2015). The model fit indices for Dharwad are as follows: χ2 = 2327.724, d.o.f. = 1221; CFI = 0.930; TLI = 0.919; χ2/df = 2327.72/
1221 = 1.90; IFI = 0.931; RMSEA = 0.048, SRMR = 0.072. For Hubli city, the fit indices are χ2 = 2385.673, d.o.f. = 1205; CFI = 0.938;
TLI = 0.927; χ2/df = 2429.647/1233 = 1.97; IFI = 0.938; RMSEA = 0.050, SRMR = 0.074. The models with a path connecting from
objective built environment measures to subjective measurement items improve the model without those links (Δχ2 = 36 and 43.97,
for Dharwad (P < 0.005) and Hubli (P < 0.005), respectively). Coefficients significant at p ≤ 0.10 are kept in the model. Except for the
measurement items for current travel attitudes, all other variables represent changes to their values: i.e., the difference between a
variable’s level at the current and the previous home locations. For latent items (travel attitudes and built environment perceptions),
principal component analysis (PCA) and confirmatory factor analysis (CFA) were conducted to reduce dimensions and identify latent
factors [PCA outputs are in APPENDIX]. The changes in perceptions about the built environment were calculated by subtracting the
factor scores at the previous residential locations from the scores at the current locations (e.g., Cao et al., 2007; and Aditjandra et al.,
2012).
6.1. Effect of changes in socio-demographic variables on changes in built environment features
Table 3 presents the association between socio-demographic changes and changes in built environment measures for the two cities.
Initially, the models had also controlled for socio-economic attributes at the current location. However, the coefficients associated with
the current location’s variable levels were dropped due to low significance levels (p > 0.10). In Dharwad, males were less likely to opt
for neighborhoods with improved accessibility and safety, whereas in Hubli, the inverse relationship was seen. In Hubli and Dharwad,
males preferred locations with higher land-use mix diversity and those located away from the CBD. In both cities, households that
moved to their own homes are more likely to opt for a neighborhood with improved social features and areas away from the CBD, as
evident from the positive coefficients. In Hubli city, homeowners preferred locations with higher accessibility, safety, attractiveness,
and a women-friendly travel environment. These findings may be pointing to the differences in economic conditions between Dharwad
and Hubli and the housing choices of the individuals of a developed city (Hubli) versus a sub-urban setup (Dharwad). In Dharwad,
residents who moved to independent homes opted to stay close to the CBD, whereas, in Hubli, residents stayed far away from the city,
post-relocation. Hubli is more urbanized, and the cost of acquiring an independent home close to the CBD would be much higher than
Dharwad. With the increase in the number of earners, both cities’ residents opted for locations having improved land-use diversity, as
implied by the positive parameter estimates. In Dharwad, an increase in the number of earners also led to the choice of locations having
increased accessibility and being close to CBDs. In Hubli, an increase in household income also demanded safer neighborhoods,
possibly indicating the implications of safety and security issues in urban environments.
The effects of an increase in the house size on locations having higher land-use diversity, better social environment, etc., are
consistent between both cities. An increase in household members in both cities motivated households to opt for locations with higher
land-use diversity, those near the CBD, and those with a good social environment. In Dharwad, an increase in the number of household
members was also associated with the choice of locations having higher accessibility and safety, possibly indicating households’ re
quirements to meet the daily travel and other needs of household members. An increase in the number of married individuals in
households required locations having a good social environment in Dharwad, whereas an inverse association was observed in Hubli.
An increase in the number of children at households motivated Hubli residents to locate closer to the CBD and Dharwad residents to
move away from the CBD.
An increase in driving licenses motivated households to choose neighborhoods with improved land-use diversity and those close to
the city’s CBD in Dharwad. In Hubli, an increase in household driving licenses led to the choice of locations having reduced land-use
diversity and those far away from the CBD. These differences may be pointing to the variations in the cost of vehicle ownership
associated with the cities. In Hubli, having more vehicles may be making it difficult for a household to find parking places in the CBD.
Whereas in Dharwad, due to limited opportunities, attractive driving environment and maintenance facilities would be available only
in the CBD areas. The effects of ’pro-private vehicle’ attitude on built environment measures vary between the cities. The attitudinal
factor does not significantly influence subjective built environment features in both cities (p > 0.10). In Dharwad, households with
’pro-private vehicle’ attitudes are more likely to stay far from the CBD and do not opt to stay in highly diverse neighborhoods. In Hubli,
individuals of ’pro-private vehicle’ attitude remains close to the CBD and in diverse neighborhoods, probably indicating their pref
erences for traveling at a lesser cost.
Certain variations in the impacts of socio-demographic changes on perceived built environment features related to women and
children’s travel convenience are also noted. In Dharwad, with an increase in the number of children, households opted for neigh
borhoods with improved conveniences for women and children travel conditions. In contrast, an inverse effect is seen in Hubli. Due to
the semi-developed setting, not all areas in Dharwad have improved opportunities for women and children. Therefore, households
might be selecting locations that help children and women meet their needs with an increase in children. In Hubli, an increase in
opportunities would also be related to higher housing costs; hence the negative correlation is plausible. With increased driving
licenses, Hubli residents prefer locations with improved travel conveniences for women and children. An inverse effect of driving
licenses is seen in Dharwad. In Dharwad, driving license ownership could be leading to improved access to locations and travel
convenience. Therefore, households would not be considering the travel convenience of women and children when they hold more
driving licenses. In Hubli, with an increased driving license, women would also be driving to work, household activities, and children’s
needs. Therefore, a positive association is plausible. The impacts of improvements in property tax are also different between the two
cities. In Hubli, greater property tax leads to a lower likelihood of choosing neighborhoods with improved conveniences for women
145
and children. In a developed city like Hubli, the property tax may be even higher if the neighborhood has all opportunities. Hence,
households could be striking a balance between tax and opportunities for women and children. In Dharwad, on the other hand, lo
cations with higher property taxes may also have increased amenities and convenience for children and women. With the increase in
household income, households of both cities preferred locations with improved social conditions, friendliness for women and chil
dren’s travel convenience, and increased land-use diversity.
The objective built environment features influence subjective built environment items. In both cities, households that relocated to
areas characterized by higher land use mix or close to the CBD see improvements in accessibility. However, improvements in land use
mix diversity are also negatively correlated with neighbourhood attractiveness and cleanliness. In Dharwad, families situated far away
from the CBD observed improvements in neighbourhood safety; while an opposite trend was seen Hubli. The effects of objective
features – land use mix and distance to CBD locations – on subjective built environment factors related to children and women travel
friendliess are different between Hubli and Dharwad cities. In Hubli, households relocated to areas having improved land use mix
diversity have witnessed improvements in women and children travel convenience. However, in Dharwad an improvement in land use
mix diversity reduced women and children travel convenience.
6.2. Effect of changes in socio-demographic variables and built environment features on vehicle ownership changes
Tables 4 and 5 present the standardized direct and total effects of changes in socio-demographics, built environment factors, and
current travel attitudes on vehicle ownership changes in Dharwad and Hubli, respectively. As evident from the tables, male household
heads are more likely to add vehicles to the household fleet in Dharwad, whereas an inverse relationship is observed in Hubli. The
differences in decisions could be related to the variations in the cities’ spatial context and economic conditions. Hubli is more diverse
and developed than Dharwad, and Hubli’s male heads might not find an additional vehicle necessary to the household mobility post-
relocation. In Beijing, Lin et al. (2018) found that male household heads were more likely to acquire cars post-relocation and Cao and
Cao (2014) observed a similar correlation in the USA. In Dharwad, households that purchase homes are more likely to add vehicles to
their fleet. In contrast, a negative correlation is observed between moving to their own home and Hubli households’ vehicle ownership
levels. Buying a home could be costly in Hubli due to its urbanized setup, and families would be compensating vehicles for a home in
the city.
In both cities, moving to independent houses increases the likelihood of having a large vehicle fleet. The impact of change in
dwelling type is more visible in Hubli than in Dharwad. The influences of variables – the number of earners and household members,
household income and travel expenditure, and increased house spaciousness – are comparable. Lin et al. (2018), Clark et al. (2016a),
Cao et al. (2007a, 2007b), etc., have observed similar variable associations. The effect of an increase in the variables’ levels on vehicle
ownership is more important for Dharwad than Hubli, perhaps indicating the variations in households’ economic status of the two
cities. The impacts of the addition of married individuals on household vehicle ownership are different between cities. In Dharwad, the
addition of married individuals leads to a reduction in vehicle ownership, probably an indication of economic constraints in bigger
families. In Hubli, however, the addition of married members leads to more vehicles, perhaps showing the decisions to improve
household mobility. Concerning the influence of an increase in the number of children in a family, the twin cities share the same
picture: households increase the fleet size for every addition of a child (see Oakil et al. (2014) for similar results), and the variable
influences are approximately same for both cities. The acquisition of a driving license would contribute to more household vehicles in
Dharwad than in Hubli, as evident from the tables. Due to its lower spatial development, keeping a vehicle might be more convenient in
Dharwad when compared with Hubli. Finally, individuals possessing ’pro-private vehicle’ attitudes are more likely to increase vehicles
(e.g., Zhao and Zhang, 2018). The variable impact is almost double (direct effect) in Hubli in comparison with that of Dharwad.
The effects of built environment features on household vehicle ownership decisions are also presented in the tables. In both cities,
households attaching higher scores on neighborhood accessibility are observed to have fewer vehicles (see Cao et al. 2007b for similar
association). The impact of improvements in accessibility is more visible in Hubli than in Dharwad. In Dharwad city, an improvement
in neighborhood safety and social features contributes to fewer household vehicles. Whereas in Hubli, improvements in such features
contribute to larger fleets. Interestingly, households opt for fewer vehicles if there is an improvement in the factor assessing the
neighborhood’s contributions to children’s needs. The factor has a strong impact in Hubli than in Dharwad. In Hubli city, the cost of
owning a vehicle could be higher, and households may be shelving vehicles if they find improvements in neighborhood convenience to
children’s travel needs. The impacts of the latent factor associated with women’s travel convenience are positive in both cities. The
trends suggest that if the neighborhoods are conducive to women’s travel needs, residents of both cities add vehicles to the fleet. An
average Dharward resident is more likely to increase total household vehicles for an increase in the factor. This effect might perhaps be
an aspiration of households of a semi-developed setup or household decisions to encourage independent mobility among women by
adding personal vehicles. Vehicle ownership decisions are also related to neighborhood attractiveness and cleanliness. Households
keep more vehicles if they perceive that the neighborhood is attractive and clean. With an increase in land-use diversity of residential
locations in both cities, households tend to increase vehicle fleet. The impact is more visible in Hubli than in Dharwad. With the
increase in distance to the CBD, Hubli residents tend to acquire more vehicles than Dharwad households.
6.3. Changes in commute travel behavior
In Dharwad city, males are more likely to reduce commute distance and commute time after residential relocation, whereas, in
Hubli, males are found to have longer commute post-relocation. The differences may be associated with the cost of buying a home close
to the workplace in the cities. In general, the effects of changes in household contexts – tenure, type of dwelling, income, the number of
146
earners, household size, and travel expenditure – are comparable between two cities. In both cities, an increase in the number of
children decreases the household head’s commute length, and the effect is more prominent in Hubli. This difference might be reflecting
greater work constraints of household heads in Hubli and the decisions to cut down commute to spend more time with children and
family, all else being equal. In addition, an increase in the number of married individuals is likely to bring down the commute travel of
household heads in both cities, perhaps indicating an individual’s decision to spend more time with family members or allocation of
household non-work travel among household members.
An increase in the number of driving licenses increases the commute travel of Hubli residents. On the other hand, a Dharwad
resident is found to shorten the commute with every addition of a household driving license, probably indicating task sharing among
household members. In Dharwad, individuals of pro-private vehicle attitudes are observed to have increased commute distance and
commute time, as evident from the direct effects. However, the indirect effects of the variable are negative. In Hubli, however, the
effects are not in agreement with that for Dharwad. Individuals of pro-private vehicle attitudes are observed to travel short distances as
implied by the variable’s direct and total effects on commute distance. The time allocated for commuting by individuals possessing pro-
private vehicle attitudes, on the other hand, is increased after the relocation.
Improvements in accessibility of neighborhoods reduce the commute travel time and increase commute distance for individuals of
both cities. Since many activities remain closer in locations having improved accessibility, a reduction in travel time is expected. Due to
improved accessibility, several non-work activities could be linked to work travel, which would increase the total commute distance.
An improvement in the perceptions of the women’s travel convenience leads to reductions in the commute travel distance of in
dividuals of both cities. In such neighborhoods, women can meet their travel independently and would not require the household
head’s support to implement the same, all else being equal. Hence, reductions in the travel distances of the household head. However,
improvements in neighborhood conveniences for children travel lead to longer commute distances in both cities, possibly due to
assisting children in their activities. As intuitions suggest, an increase in the CBD distance increases commute distance and commute
time of individuals of both cities. The impact of an increase in the distance to the CBD on commute distance is greater in Dharwad when
compared with Hubli. However, the effects on commute travel time are the same. This correlation may be a reflection of the congestion
in Hubli. In Dharwad, moving to residential areas characterized by high land-use diversity reduces commute travel. On the other hand,
an increase in land use diversity increases Hubli residents’ commute travel features. In an urbanized setup like Hubli, diverse
neighborhoods may be characterized by congestion, and individual route options in such a setup would lead to longer commutes, all
else being equal. In both cities, improvements in vehicle ownership would lead to shortened commute, possibly an indication of
household travel need allocation across multiple vehicles. A more significant reduction in commute travel is observed in Hubli when
households own multiple vehicles.
6.4. Discussion on residential self-selection
The models did not provide enough evidence for residential self-selection due to the influence of travel attitude, ’Pro-private
vehicle,’ (at 95 % confidence interval) on both vehicle ownership and travel behavior. However, as noted before, the variable asso
ciations are different between the cities and are interesting. Travel attitude could not explain variations in the perceptions about the
built environment at the 95 % level. Travel attitude remained significant at the 90 % level when predicting the changes in objective
built environment measures – land-use diversity index and the CBD distance – in both cities. In Dharwad, the causal links – travel
attitude → built environment measures, travel attitude → vehicle ownership, and built environment measures → vehicle ownership –
are significant only at the 90 % level. In Hubli, the causal links between travel attitude → vehicle ownership and built environment
measures → vehicle ownership are significant at the 95 % level. The link, travel attitude → built environment measures stayed at the
90 % level. At the 95 % confidence level, the associations show that households of pro-vehicle attitude increase vehicle ownership, and
households staying away from the CBD opt for a larger vehicle fleet, clearly showing the built environment’s influence on vehicle
ownership. Omitting the impact of attitude on location preferences, the study shows that distance to the CBD shapes vehicle ownership
decisions in Hubli – an increase in the distance to home from the CBD increases vehicle fleet size. Further, households of pro-vehicle
attitudes also increase fleet size. At the 95 % confidence level, the distance to the CBD influenced the commute distance of Dharwad
residents. As intuition suggests, locating away from the CBD increases commute distance. For a Hubli resident, the increase in land use
mix at residential locations leads to increased commute travel time (95 % confidence level). The self-selection effect is visible in
Dharwad regarding the effect of gender on commute distance (at 95 % level). Male household heads opt for locations close to the CBD
and reduce commute distance. Distance to the CBD, after controlling for the two associations, appears to lengthen commute travel. In
Hubli, distance to the CBD significantly affects vehicle ownership (95 % level) after controlling for the sorting effect of gender. Male
heads are interested in staying close to the CBD and are also more likely to opt for a smaller fleet. Distance to the CBD, however,
independently increases vehicle ownership level. The two perceived neighborhood features – children’s travel convenience and
women’s travel convenience – are also related to residential self-selection. In Dharwad, with an increase in the number of earners,
households chose locations with improved neighborhood conditions for women and children (at the 5 % level). An increase in the
number of earners leads to a larger vehicle fleet (at the 5 % level). At the 5 % significance level, with an increase in neighborhood
conditions in favor of women’s travel, households opt for a higher number of vehicles. An increase in the neighborhood conditions in
favor of children contributes to reducing total household vehicles (at the 5 % level). These effects show self-selection and independent
effects of the perceived neighborhood measures on household vehicle ownership. In Hubli city, at 95 % level, the link from the number
of earning members to the perceived factor women travel convenience is significant. However, the factor’s independent effect is not
significant at the 5 % level. After accommodating self-selection effects, this effect implies that the perceived neighborhood measure
does not have any independent impact on total household vehicles.
147
6.5. Policy implications
The study has important implications for planning and policy, especially towards the integrated development of Dharwad and
Hubli into Hubli-Dharwad twin cities. The study shows that promoting built environment measures such as mixed land use and density
measures may not effectively control vehicle ownership and travel demand for the integrated cities as the impacts vary between cities..
In Hubli, coefficients attached to the CBD distance and land use diversity index are positive and significant at the 5 % level. From a
broad policy viewpoint, job-hosing balance (for commute distance reduction) and the current TOD policy developments (for mixed-
use) (Hubli-Dharwad Bus Rapid Transit System, 2018a, 2018b) would be effective in the context of Hubli city. However, promoting
any such measures might not be effective for Dharwad as the coefficients attached to the CBD distance, and land use diversity index are
nearly zero. Thus, the promotion of land use measures such as TOD and mixed residential developments (Hubli-Dharwad Bus Rapid
Transit System, 2018a, 2018b) might not yield the desired results of reductions in vehicle ownership, at least for Dharwad city. This
finding clearly shows that the planning bodies may devise city-specific land-use policies for sustainable management of cities. The
study also indicates that perceived neighborhood features influence households’ vehicle ownership decisions, and the impacts vary
between cities. Improving accessibility to various amenities might bring down Hubli residents’ vehicle ownership, whereas it might
not be an effective strategy for Dharwad. However, in Dharwad, improvements in neighborhood safety would lead to reductions in
household vehicle fleet size. Thus, in Dharwad improving neighborhood safety through infrastructure interventions and law
enforcement may effectively curb vehicle ownership.
According to our models, the latent built environment factor measuring neighborhoods’ support to women’s travel maintains a
positive correlation with vehicle ownership levels in both cities. As more women are entering the workforce, the travel needs of women
might increase. If neighborhood interventions support women’s travel, there would be an increase in the demand for private vehicles,
as found in this study. This finding clearly shows that neighborhood improvements could also consider measures to ensure safe and
efficient travel for women. The provision of safe transit, safe NMT infrastructure, etc., could reduce personal vehicle dependency. The
variable’s impact was strong in Dharwad, presumably displaying the aspirations of the populace of an emerging urban area. Whereas
Hubli has a developed setup and has more options for safe and efficient travel for women. This finding also shows the need to devise
city-specific policies to address vehicle ownership. In both cities, the latent built environment constructs concerning neighborhoods’
help to children travel are negatively correlated with vehicle ownership, suggesting that if the children can travel safely and have
various opportunities, households decrease vehicle ownership. This is encouraging from a societal standpoint. However, the latent
factor also leads to an increase in travel in terms of distance and time, which might be considered for managing travel demand in the
cities. The national transit-oriented development (TOD) policy (Ministry of Urban Development, 2015) envisions improving women’s
and children’s safe travel. According to our study, improvements in the neighborhood conditions in favor of children’s travel needs
could reduce vehicle ownership; however, with the secondary effect of an increase in travel distance. However, the neighborhood
developments favoring women’s travel might lead to increased vehicle ownership, as found from the analysis. These findings show the
need to incorporate additional measures, besides neighborhood developments, to manage vehicle ownership and travel demand.
7. Summary and conclusions
This paper investigates the impacts of built environment measures on vehicle ownership and travel behavior in an Indian setting.
This paper is the first application of quasi-longitudinal data with travel attitudes and neighborhood perceptions to unravel the rela
tionship between the built environment and mobility decisions by considering the impacts of objective built environment features on
subjective built environment items. The study also compares built environment impacts of mobility decisions between two adjacent
urban areas, Hubli and Dharwad, in India. In addition, the paper also investigated how neighborhood perceptions regarding children
and women travel conveniences impact mobility decisions and are linked with residential self-selection.
The study finds that Dharwad residents acquired more personal vehicles than Hubli residents after the movement, while the
proportion of households who relinquished their vehicles is high in Hubli. Hubli city is more developed than Dharwad, and the dif
ferences could be revealing the aspirations of households of emerging and developed urban contexts. The urban development patterns
are also reflected in households’ relocation in two cities. A household in Hubli city moved closer to a bus stop than a resident in
Dharwad, where the proximity to bus stops does not change significantly. This finding might perhaps be suggesting the effect of
congestion and increased travel cost in Hubli city. In Hubli, post-relocation, an average household did not go far from playgrounds/
parks, and in Dharwad, an average resident moved closer to the CBD area. These findings are also aligned with the spatial development
variations between the two cities. Since Hubli has fewer parks and playgrounds, relocation close to the recreational amenities is
intuitive. Since Dharwad is emerging, relocation close to the CBD and areas featured by increased land-use heterogeneity is also
aligned with general expectations. The residential location choices also varied by the cities. In Dharwad, an increase in driving licenses
motivated households to choose neighborhoods with improved land-use diversity and those close to the city’s CBD. In Hubli, an in
crease in household driving licenses led to the choice of locations having reduced land-use diversity and those far away from the CBD.
Perhaps, these differences may be pointing to the variations in the cost of vehicle ownership associated with the cities.
The comparison of built environment effects on vehicle ownership and travel behavior reveals interesting findings connected with
the context of the cities. As per our analysis, improvements in neighborhood accessibility could reduce household vehicle ownership.
However, land-use diversity is positively correlated with vehicle ownership in Hubli. In addition, Hubli residents are more likely to
increase vehicle ownership with an increase in the distance to the CBD. In Dharwad, both land-use diversity and distance to the CBD
have negligible effects on vehicle ownership. Vehicle ownership may be related to aspirations than a response to land use hetero
geneity in Dharwad. In Dharwad, moving to residential areas characterized by high land-use diversity reduces commute travel. On the
148
other hand, an increase in land use diversity increases Hubli residents’ commute travel features. Being an urbanized context, Hubli city
may have higher congestion, and individuals’ route decisions could be leading to longer trip lengths. The statistical models also
revealed that objective built environment features affects households perceptions about the subjective built environment, and there
are variations in the influences. In Hubli, households relocated to areas having improved land use mix diversity have witnessed im
provements in women and children travel convenience. However, in Dharwad an improvement in land use mix diversity reduced
women and children travel convenience. These differences may be associated with the variations in the socio-economic and land use
settings of the cities.
The study also included two perceived neighborhood factors about children and women travel friendliness of neighborhoods. In
terms of the rating attached to the measurement items of the factors, the trends suggest that in Hubli, the current location does not help
children use active modes to schools compared to their previous location. Regarding the items related to women’s travel, greater
improvements can be seen with the ease of conducting shopping and the possibility of making short trips in Dharwad. The impacts of
the latent factor associated with women’s travel convenience on vehicle ownership are positive in both cities. The trends suggest that if
the neighborhoods are conducive to women’s travel needs, residents of both cities add vehicles to the fleet. An average Dharwad
resident is more likely to increase total household vehicles for an increase in the factor, perhaps suggesting the aspiration of emerging
city dwellers. These findings show that neighborhood interventions to enhance the travel conveniences of women could lead to
increased vehicle ownership. The models also reveal that an improvement in the perceptions of the women’s travel convenience leads
to reductions in the commute travel distance of individuals of both cities. In such neighborhoods, women can meet their travel
independently and would not require the household head’s support to implement the same. However, improvements in neighborhood
conveniences for children travel lead to longer commute distances in both cities, possibly due to assisting children in their activities.
Overall, the impacts of the latent neighborhood measures reveal that vehicle ownership and travel behavior would be increased with
interventions towards improving women’s travel environment.
The study did not reveal self-selection due to travel attitudes. The self-selection effects are visible concerning a few socio-
demographic variables, objective built environment factors, and the latent factors associated with women and children’s travel
conveniences. In Dharwad city, with an increase in the number of earners, households opt for neighborhoods characterized by
improved travel convenience for women and children. The neighborhood factor also significantly impacts vehicle ownership. Whereas
in Hubli, after accommodating self-selection, the factors did not affect vehicle ownership independently.
Overall, the study finds that the built environment impacts vehicle ownership and travel decisions differ in Hubli and Dharwad. The
study also highlights the need to compare built environment travel behavior connections across different areas to promote land-use
measures to alter travel choices. The study, however, has a few limitations. Due to data collection challenges, the study could not
control for residential preferences, and future Research might include such features to isolate self-selection effects. Future Research
could also be benefitted by incorporating changes in attitudes as residential relocation can influence individuals’ attitudes towards
travel. The present study explored total vehicle ownership. However, the factors influencing household car and two-wheeler
ownership could be different. The literature would be benefitted by undertaking a study to examine car and two-wheeler owner
ship decisions separately while accommodating possible interaction between the decisions related to the levels of these vehicles.
CRediT authorship contribution statement
Chetan Doddamani: Conceptualization, Methodology, Data curation, Software, Writing – original draft, Visualization, Investi
gation. M. Manoj: Supervision, Conceptualization, Methodology, Writing – review & editing.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.
Data availability
Data is part of an ongoing ph.d research
Acknowledgments
The authors appreciate the comments and suggestions of four anonymous reviewers on an earlier version of the manuscript. The
authors gratefully acknowledge the support of the office of Hubli-Dharwad BRTS and Directorate of Urban Land Transport Department
(DULT), Bangalore, Karnataka, India, for providing the land use data of Hubli and Dharwad. The authors are responsible for analyzing
and interpreting the data, and they take responsibility for any omissions and errors. The views presented in the paper are of the authors
and do not represent that of the offices of Hubli-Dharwad BRTS and Directorate of Urban Land Transport Department (DULT),
Bangalore.
Author Contributions
The first author collected the primary data, performed the statistical analysis, and drafted the paper. The second author contributed
149
to survey design, model refinement, and manuscript preparation and revisions.
Appendix
Outputs of confirmatory factor analysis – Built environment perceptions
Table contents are after Doddamani and Manoj (2021a); Doddamani and Manoj (2021b).
Dharwad city – Current location
Factor name Items Standardized Squared multiple t-value Covariance Reliability coefficient
loadings correlation
Attractive & Clean Attractive streets 0.79 0.63 Attractive & clean ~~ Safe in 0.821
Good housing style 0.73 0.54 14.1 neighborhood = 0.488
Choke free drains 0.63 0.40 12.11
Dust free roads 0.61 0.38 11.65
Good landscape 0.7 0.50 13.56
Safe Neighborhood I feel safe walking in this 0.63 0.40 Safe neigborhood ~~ 0.772
neighborhood Socially active neighborhood
Low safe from pollution and 0.46 0.21 7.502 = 0.423
accidents
Safe for children 0.6 0.37 9.11
Safe for women 0.68 0.47 9.72
Low crime rate 0.65 0.43 9.35
Safe from stray animals 0.52 0.28 8.35
Good streetlight 0.4 0.27 6.71
Socially active Frequent social gathering 0.77 0.60 Socially active neighborhood 0.793
Homemakers are able to make 0.59 0.35 10.2 ~~ Accessibility = 0.362
friends
Diverse in terms of age, 0.6 0.37 10.42
gender, and religion
My children have lot of 0.48 0.23 8.26
friends
Active welfare association 0.7 0.50 11.57
and interactive neighbors
Accessibility Easy access to main road 0.63 0.40 Accessibility neighborhood 0.801
Easy access to bus stop 0.69 0.48 12.71 ~~ Women-friendly
Other amenities are available 0.73 0.55 10.35 neighborhood = 0.480
nearby
Easy access to groceries 0.68 0.47 10
City center is easily available 0.52 0.27 8.35
Women-friendly Women can find job close to 0.56 0.32 Women-friendly 0.626
home neighborhood ~~ Child-
Shopping is easy for women 0.62 0.40 7.47 friendly neighborhood =
Women can walk to work and 0.58 0.35 7.38 0.390
other place
Child-friendly Schools are nearby 0.8 0.65 0.602
This neighborhood helps my 0.51 0.26 4.82
children to walk/cycle to
school
Parks and playgrounds are 0.38 0.25 2.72
not easily accessible
Note: Model fit indices χ 2 = 660.41, d.o.f. = 334; CFI = 0.899; TLI = 0.883; RMSEA = 0.050, SRMR = 0.064. GFI = 0.898. Average
variance explained = 61.91 %.
Dharwad city-Previous location
(continued on next page)
150
(continued )

Low safe from pollution 0.38 0.25 7.64 = 0.423
and accidents
Homemakers are able to 0.74 0.56 11.96 ~~ Accessibility = 0.362
make friends
friends
Active welfare 0.64 0.41 10.66
association and
interactive neighbors
Other amenities are 0.78 0.61 14.52 neighborhood = 0.480
available nearby
Easy access to groceries 0.67 0.45 12.54
City center is easily 0.52 0.27 9.69
available
Women-friendly Women can find job close 0.57 0.33 Women-friendly 0.59
to home neighborhood ~~ Child-
Shopping is easy for 0.61 0.37 7.35 friendly neighborhood =
women 0.390
Women can walk to work 0.55 0.30 7.05
and other place
This neighborhood helps 0.61 0.37 6.87
my children to walk/
cycle to school
Parks and playgrounds 0.36 0.23 3.25
are not easily accessible
Note: Model fit indices χ 2 = 718.064, d.o.f. = 334; CFI = 0.909; TLI = 0.898; RMSEA = 0.046, SRMR = 0.057. GFI = 0.896, Average
Hubli city-Current location
Factor name Items Standardized Squared t-value Covariance Reliability coefficient

loadings multiple correlation
accidents
friends
151
(continued )
loadings multiple correlation

friends
nearby
other place
This neighborhood helps my 0.86 0.73 3.74
school
Hubli city-Previous location

loadings multiple
correlation
accidents
friends
friends
nearby
other place
0.40 0.26 9.48
152
(continued )
loadings multiple
correlation
This neighborhood helps my

school
Note: Model fit indices χ 2 = 751.424, d.o.f. = 334; CFI = 0.896; TLI = 0.879; RMSEA = 0.051, SRMR = 0.054. GFI = 0.889. Average
Dharwad city- travel attitude
Factor name Items Standardized Squared multiple t- Covariance Reliability

loadings correlation value coefficient
Pro public I like travelling by PT 0.682 0.465 Pro public transport~~ 0.79
transport PT is reliable 0.581 0.338 9.17 Pro NMT = 0.348
PT is safe 0.551 0.304 8.58
PT is cheap 0.649 0.421 6.67
Pro NMT I like walking 0.419 0.276 Pro NMT ~~ Pro private 0.72
Walking and cycling improves 0.896 0.803 6.15 vehicle = 0.262
health
Walking and cycling do not 0.760 0.578 3.78
harm environment
Pro private I like to travel a lot 0.522 0.272 0.75
vehicle I like to drive car 0.475 0.226 3.69
I like to make short trips using − 0.346 0.220 3.55
TW
Vehicle ownership improves 0.329 0.208 3.41
ones status in the society
Hubli city-travel attitude
Factor name Items Standardized Squared multiple t- Covariance Reliability

loadings correlation value coefficient
Pro public I like travelling by PT 0.735 0.540 Pro public transport~~ 0.79
transport PT is reliable 0.633 0.401 9.08 Pro NMT = 0.309
PT is safe 0.621 0.386 9.16
PT is cheap 0.447 0.200 6.81
Pro NMT I like walking 0.391 0.253 Pro NMT ~~ Pro private 0.71
Walking and cycling improves 0.750 0.563 vehicle = 0.239
health
Walking and cycling do not 0.457 0.209 3.45
harm environment
Pro private I like to travel a lot 0.615 0.378 0.73
vehicle I like to drive car 0.389 0.251 4.86
I like to make short trips using 0.445 0.298 3.25
TW
Vehicle ownership improves 0.361 0.230 2.98
ones status in the society
Appendix A. Supplementary material
Supplementary data to this article can be found online at https://doi.org/10.1016/j.tra.2022.08.008.
153
References
Aditjandra, P.T., Cao, X.J., Mulley, C., 2012. Understanding neighbourhood design impact on travel behaviour: an application of structural equations model to a
British metropolitan data. Transp. Res. Part A: Policy Pract. 46 (1), 22–32.
Aditjandra, P.T., Cao, X., Mulley, C., 2016. Exploring changes in public transport use and walking following residential relocation: a British case study. J. Transport
Land Use 9 (3), 77–95.
Banerjee, I., 2012. Automobility in India: A Study of Car Acquisition and Ownership Trends in the City of Surat (Doctoral dissertation, UC Berkeley).
Bansal, P., Kockelman, K.M., Schievelbein, W., Schauer-West, S., 2018. Indian vehicle ownership and travel behavior: a case study of Bengaluru, Delhi and Kolkata.
Res. Transp. Econ. 71, 2–8.
Ben-Akiva, M., de Palma, A., Kaysi, I., 1996. The impact of predictive information on guidance efficiency: an analytical approach. In Advanced methods in
transportation analysis. Springer, Berlin, Heidelberg, pp. 413–432.
Bhat, C.R., Guo, J.Y., 2007. A comprehensive analysis of built environment characteristics on household residential choice and auto ownership levels. Transp. Res.
Part B: Methodol. 41 (5), 506–526.
Bianco, M., Lawson, C., 1997. Trip Chaining, Childcare, and Personal Safety. In Women’s Travel Issues: Proceedings from the Second National Conference (Vol. 1, No.
9, pp. 124-142).
Buchanan, N., Barnett, R., 2006. Peripheral residential relocation and travel pattern change. Urban Policy Res. 24 (2), 217–236.
Cao, J., Cao, X., 2014. The impacts of LRT, neighbourhood characteristics, and self-selection on auto ownership: evidence from Minneapolis-St. Paul. Urban Stud. 51
(10), 2068–2087.
Cao, X., Mokhtarian, P.L., Handy, S.L., 2007a. Cross-sectional and quasi-panel explorations of the connection between the built environment and auto ownership.
Environ. Plan. A 39 (4), 830–847.
Cao, X., Mokhtarian, P.L., Handy, S.L., 2007b. Do changes in neighbourhood characteristics lead to changes in travel behaviour? A structural equation modelling
approach. Transportation 34 (5), 535–556.
Cao, X., Mokhtarian, P.L., Handy, S.L., 2009. Examining the impacts of residential self-selection on travel behaviour: a focus on empirical findings. Transp. Rev. 29
(3), 359–395.
Cao, X.J., Næss, P., Wolday, F., 2019. Examining the effects of the built environment on auto ownership in two Norwegian urban regions. Transp. Res. Part D: Transp.
Environ. 67, 464–474.
Cervero, R., Kockelman, K., 1997. Travel demand and the 3Ds: Density, diversity, and design. Transp. Res. Part D: Transp. Environ. 2 (3), 199–219.
Clark, B., Chatterjee, K., Melia, S., 2016a. Changes to commute mode: the role of life events, spatial context and environmental attitude. Transp. Res. Part A: Policy
Pract. 89, 89–105.
Clark, B., Chatterjee, K., Melia, S., 2016b. Changes in level of household car ownership: the role of life events and spatial context. Transportation 43 (4), 565–599.
Cochran, W.G., 2007. Sampling techniques. John Wiley & Sons.
De Vos, J., Cheng, L., Kamruzzaman, M., Witlox, F., 2021. The indirect effect of the built environment on travel mode choice: A focus on recent movers. J. Transp.
Geogr. 91, 102983.
Dharwad District Statistics Government of Karnataka, 2018. Retrieved from: http://www.dharwad.nic.in/en/documents/DistrictStatistics2017-18.pdf. Accessed on
05-10-2019 through, http://www.dharwad.nic.in/en/index.aspx.
Doddamani, C., Manoj, M., 2021a. Effects of the built environment and socio-demographics on the car and two-wheeler ownership levels: a case study of Dharwad
City. Transp. Developing Econ. 7 (2), 1–13.
Doddamani, C., Manoj, M., 2021b. Analysis of the influences of built environment measures on household car and motorcycle ownership decisions in Hubli-Dharwad
cities. Transportation 1–39.
Engebretsen, Ø., Næss, P., Strand, A., 2018. Residential location, workplace location and car driving in four Norwegian cities. Eur. Plan. Stud. 26 (10), 2036–2057.
Ettema, D., Nieuwenhuis, R., 2017. Residential self-selection and travel behaviour: what are the effects of attitudes, reasons for location choice and the built
environment? J. Transp. Geogr. 59, 146–155.
Goel, R., Mohan, D., 2020. Investigating the association between population density and travel patterns in Indian cities—An analysis of 2011 census data. Cities 100,
102656.
Golob, T.F., 2000. A simultaneous model of household activity participation and trip chain generation. Transp. Res. Part B: Methodol. 34 (5), 355–376.
Gopisetty, V., Srinivasan, K.K., 2013. Joint models for analysis of household trip frequency and vehicle ownership in Chennai city. Int. J. Adv. Eng. Sci. Appl. Math. 5
(2–3), 129–144.
Government of India census data. (2011). Retrieved from http://censusindia.gov.in Accessed on 06-10-2019.
Handy, S., Cao, X., Mokhtarian, P., 2005. Correlation or causality between the built environment and travel behavior? Evidence from Northern California. Transp. Res.
Part D: Transp. Environ. 10 (6), 427–444.
Handy, S., Cao, X., Mokhtarian, P.L., 2006. Self-selection in the relationship between the built environment and walking: empirical evidence from Northern California.
J. Am. Plan. Assoc. 72 (1), 55–74.
Hubli-Dharwad Bus Rapid Transit System Company Ltd. (HDBRTS), 2018a. Project report on: Hubli-Dharwad 2030, City Development framework.
Hubli-Dharwad Bus Rapid Transit System Company Ltd. (HDBRTS), 2018b. Project report on: Integrated Land Use Management and Transportation.
Kline, R.B., 2015. Principles and practice of structural equation modeling. Guilford publications.
Krizek, K.J., 2003. Residential relocation and changes in urban travel: Does neighborhood-scale urban form matter? J. Am. Plan. Assoc. 69 (3), 265–281.
Lin, T., Wang, D., Zhou, M., 2018. Residential relocation and changes in travel behavior: what is the role of social context change? Transp. Res. Part A: Policy Pract.
111, 360–374.
Manoj, M., Verma, A., 2015a. Activity–travel behaviour of non-workers from Bangalore City in India. Transp. Res. Part A: Policy Pract. 78, 400–424.
Manoj, M., Verma, A., 2015b. Activity-travel behaviour of non-workers belonging to different income group households in Bangalore, India. J. Transp. Geogr. 49,
99–109.
Manoj, M., Verma, A., 2016. Effect of built environment measures on trip distance and mode choice decision of non-workers from a city of a developing country, India.
Transp. Res. Part D: Policy Pract. 46, 351–364.
Manoj, M., Verma, A., 2017. A structural equation model based analysis of non-workers’ activity-travel behaviour from a city of a developing country. Transportation
44 (2), 241–269.
Ministry of Housing and Urban Affairs (MoHUA), 2015. Annual report -2015-16. Retrieved from: http://mohua.gov.in/cms/annual-reports.php.
Ministry of Housing and Urban Affairs (MoHUA, 2018). Report on National Mission on Sustainable Habitat (NMSH). Retrieved from http://mohua.gov.in/upload/
uploadfiles/files/NMSH_parameters_v4_1.pdf.
Ministry of Road Transport and Highways, (MORTH), Annual Report, 2018. Retrieved from: https://www.morth.nic.in/annual-reports.
Government of India, Ministry of Statistical and Programe Implementation, (MoSpI), (2018). http://mospi.nic.in/statistical-year-book-india/2018/189.
Ministry of Urban Development (MoUD), 2015. Report on National transit-oriented development (TOD) policy. New Delhi: Government of India. Retrieved from
http://mohua.gov.in/cms/annual-reports.php.
Munshi, T., 2016. Built environment and mode choice relationship for commute travel in the city of Rajkot, India. Transp. Res. Part D: Transp. Environ. 44, 239–253.
Næss, P., 2005. Residential location affects travel behavior—but how and why? The case of Copenhagen metropolitan area. Prog. Plan. 2 (63), 167–257.
National Urban Transport Policy, 2014. Ministry of Urban Development, Government of India (GOI). Available at http://www.urbantransport.kar.gov.in/National%
20Urban%20TransportPolicy.pdf.
NITI Aayog and The Boston Consulting Group, Report on: Transforming India’s Mobility: A perspective, 2018. Retrieved from https://niti.gov.in/writereaddata/files/
document_publication/BCG.pdf.
154
Oakil, A.T.M., Ettema, D., Arentze, T., Timmermans, H., 2014. Changing household car ownership level and life cycle events: an action in anticipation or an action on
occurrence. Transportation 41 (4), 889–904.
Prillwitz, J., Harms, S., Lanzendorf, M., 2007. Interactions between residential relocations, life course events, and daily commute distances. Transp. Res. Rec. 2021
(1), 64–69.
Pucher, J., Korattyswaropam, N., Mittal, N., Ittyerah, N., 2005. Urban transport crisis in India. Transp. Policy 12 (3), 185–198.
Rahul, T.M., Verma, A., 2017. The influence of stratification by motor-vehicle ownership on the impact of built environment factors in Indian cities. J. Transp. Geogr.
58, 40–51.
Rosenbloom, S., 1987. The impact of growing children on their parents’ travel behavior: a comparative analysis. Transp. Res. Rec. 1135, 17–25.
Sarangi, P., Manoj, M., 2020. Escorting and mode choice decisions of members of an urban university in New Delhi, India. Case Stud. Transpo. Policy 8 (4),
1440–1450.
Sarkar, P.P., Mallikarjuna, C., 2016. Effect of Perception and Attitudinal Variables on Mode Choice Behavior: A Case Study of Indian City. Travel Behaviour and
Society, Agartala, p. 10.
Scheiner, J., Holz-Rau, C., 2013. Changes in travel mode use after residential relocation: a contribution to mobility biographies. Transportation 40 (2), 431–458.
Schumacker, R.E., Lomax, R.G., 2004. A beginner’s guide to structural equation modeling. Psychology press.
Shirgaokar, M., 2012. The rapid rise of middle-class vehicle ownership in Mumbai.
Srinivasan, K.K., Bhargavi, P., 2007. Longer-term changes in mode choice decisions in Chennai: a comparison between cross-sectional and dynamic models.
Transportation 34 (3), 355–374.
Stevens, J., 1996. Categorical data: The log linear model. Applied multivariate statistics for the social sciences, 3rd ed. Lawrence Erlbaum Associates, Mahwah, NJ,
pp. 518–557.
Van Acker, V., Witlox, F., 2010. Car ownership as a mediating variable in car travel behaviour research using a structural equation modelling approach to identify its
dual relationship. J. Transp. Geogr. 18 (1), 65–74.
Verma, M., Manoj, M., Rodeja, N., Verma, A., 2017. Service gap analysis of public buses in Bangalore with respect to women safety. Transp. Res. Procedia 25,
4322–4329.
Verma, M., Rodeja, N., Manoj, M., Verma, A., 2020. Young Women’s Perception of Safety in Public Buses: A Study of Two Indian Cities (Ahmedabad and Bangalore).
Transp. Res. Procedia 48, 3254–3263.
Wang, D., Lin, T., 2019. Built environment, travel behavior, and residential self-selection: A study based on panel data from Beijing. China. Transp. 46 (1), 51–74.
Zhao, P., Zhang, Y., 2018. Travel behaviour and life course: Examining changes in car use after residential relocation in Beijing. J. Transp. Geogr. 73, 41–53.
155

Residential Relocation and Changes in Household Vehicle

Uploaded by

Copyright:

Available Formats

Residential Relocation and Changes in Household Vehicle

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Residential Relocation and Changes in Household Vehicle

Uploaded by

Copyright:

Available Formats

Transportation Research Part A 164 (2022) 134–155

Contents lists available at ScienceDirect

Transportation Research Part A

Residential relocation and changes in household vehicle

2.3. Built environment-travel behavior studies in India

2.4. Gaps in knowledge and scope for further reserach

Fig. 1. Conceptual Framework.

Fig. 2. Map of the study area.

5.1. Changes in socio-demographic characteristics

Dharwad City Hubli City

Car ownership changes [%] − 1 5.40 11.80

Current location Previous location Current location Previous location

Accessibility in the neighborhood

5.2. Changes in built environment characteristics

Change in Subjective built environment characteristics Objective built

Total vehicle ownership Commute distance Commute time

Gender 0.028 (0.034) * − 0.048 (-0.079) * − 0.018 (-0.014)

Total vehicle ownership Commute distance Commute time

Gender − 0.068 (-0.064) * 0.022 (0.014) 0.026 (0.034)

6.1. Effect of changes in socio-demographic variables on changes in built environment features

6.3. Changes in commute travel behavior

6.4. Discussion on residential self-selection

6.5. Policy implications

7. Summary and conclusions

CRediT authorship contribution statement

Declaration of Competing Interest

Data is part of an ongoing ph.d research

to survey design, model refinement, and manuscript preparation and revisions.

Outputs of confirmatory factor analysis – Built environment perceptions

Dharwad city – Current location

Dharwad city-Previous location

Good landscape 0.67 0.46 12.71

Hubli city-Current location

Factor name Items Standardized Squared t-value Covariance Reliability coefficient

My children have lot of 0.64 0.41 10.41

Hubli city-Previous location

Factor name Items Standardized Squared t-value Covariance Reliability coefficient

This neighborhood helps my

Dharwad city- travel attitude

Factor name Items Standardized Squared multiple t- Covariance Reliability

Hubli city-travel attitude

Factor name Items Standardized Squared multiple t- Covariance Reliability

Appendix A. Supplementary material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.tra.2022.08.008.

You might also like