Content uploaded by Srinivas Peeta
Author content
All content in this area was uploaded by Srinivas Peeta on Dec 31, 2019
Content may be subject to copyright.
1
Impacts of Personalized Accessibility Information on Residential Location Choice and
Travel Behavior
Yuntao Guo a, Srinivas Peeta b*
a Lyles School of Civil Engineering/NEXTRANS Center, Purdue University, 550 Stadium Mall
Drive, West Lafayette, IN 47907-2051
b School of Civil and Environmental Engineering, and H. Milton Stewart School of Industrial and
Systems Engineering, Georgia Institute of Technology, 790 Atlantic Drive, Atlanta, GA 30332-
0355 Tel: +1-404-894-2243 Email: peeta@gatech.edu
ABSTRACT
This paper investigates the impacts of personalized accessibility information provided to possible
relocators on their residential location decision-making process and travel behavior after relocation.
An experiment is designed and implemented using a sample of people relocating to Tippecanoe
County in Indiana, United States. The participants are randomly allocated to either a control or
treatment group. The treatment group participants are provided personalized neighborhood
accessibility information before relocation that characterizes the ease of access of each
neighborhood for six trip purposes (work, healthcare, social or recreational, restaurants, education,
and retail or grocery shopping) using walk, bike, transit, or car mode. The control group
participants are not given this information. Surveys are designed to capture participants’ self-
reported residential location decision-making process and travel behavior before and after
relocation. Simultaneous equation models are estimated to capture the potential interrelationship
between the accessibility of participants’ residential neighborhood and their self-reported weekly
driving time after relocation, and the factors that affect them. Descriptive statistics comparing
behavior before and after relocation, and model estimation results, show that personalized
accessibility information can potentially make relocators more informed about travel-related
information, and assists them in selecting a residence that better addresses their travel needs based
on higher accessibility to potential destinations. Ultimately, this information makes it more likely
that they will travel less using car (about 10% less weekly driving time, on average) and use walk
or transit mode more (about 10% and 5% more frequently on average, respectively) to their
destinations.
Keywords: Personalized accessibility information; residential location choice; travel behavior;
sustainability; mode choice
2
1. Introduction
Travel-related decisions are often made across a broad spectrum from the short-term to the
long-term. Over the short term, people make decisions on parking options (e.g., free street parking
or reserved parking structure) and non-work destinations (e.g., grocery shopping and restaurants).
Choice of travel mode may be a day-to-day decision, while route choice can be spontaneous. In
the long-term, individuals decide on their residential location, vehicle ownership, whether to make
life-style changes, and employment. Despite the broad range of time frames, current information
intervention strategies, which use transportation information to influence travel decisions, focus
on the short-term end of the spectrum such as using real-time travel information to influence day-
to-day mode and route choices (Emmerink et al. 1995; Peeta and Mahmassani 1995; Lam and
Chan 2001; Paz and Peeta 2009; Ben-Elia and Shiftan 2010).
Due to the challenges associated with ever-increasing road congestion, automobile
dependency, urban sprawl, and pollution, there is a critical need to develop effective strategies that
can foster more sustainable travel behavior by reducing the usage of private vehicles (hereafter
referred to as “car mode”) and increasing the usage of sustainable travel modes, such as walking,
riding a bicycle, and using public transit (hereafter referred to as “walk, bike, and transit modes”).
Information intervention strategies have been identified by researchers and policy-makers as an
effective solution to address these challenges by providing information to travelers and making
them more informed on short- and long-term travel decisions (Kenyon and Lyons 2003; Rodriguez
and Rogers 2014). An information intervention strategy designed to improve the sustainability of
travel behavior would ideally work along the full time-scale range, particularly since longer-term
choices (e.g., transit accessibility) frequently constrain the shorter-term options (e.g., transit usage).
Yet not enough is known about designing such information, the impact of such information on
decisions at varying time scales, or the impact of longer-term decisions on those made over the
shorter term. Strategies based on up-to-the-minute transportation information, while desirable for
multiple reasons, have demonstrated limited ability to significantly alter travel behavior, especially
for habitual travel behavior (e.g., mode choice), towards a more sustainable direction (Guo 2011;
Zhou 2012; Andersson et al. 2018).
The objective of this study is to design personalized neighborhood accessibility
information for intervention strategies to influence the residential location decision-making
process and foster formation of more sustainable travel behavior after relocation. Neighborhood
accessibility quantifies the ability to access different services and opportunities from a
neighborhood using the available transportation modes (Guo et al. 2016b). Six types of
neighborhood accessibility information for six types of potential trip purposes were considered in
this study, including work, healthcare, social or recreational, restaurants, education, and retail or
grocery shopping. An interactive online accessibility mapping application (IOAMA 2015) was
developed to deliver such information. An IOAMA can provide personalized neighborhood
accessibility information for four transportation modes (walk, bike, transit, and car) based on users’
work locations and travel needs. This information manifests itself in terms of the potential users’
ability to visualize five personalized accessibility levels for each mode, with accessibility ranging
from 1 (very low accessibility) to 100 (highest accessibility). Potential users can also find each
type of accessibility information of a neighborhood through IOAMA by clicking on each
neighborhood.
To evaluate the effectiveness of the designed information, an experiment was designed and
administered to a sample of participants selected from relocators (people who change their
residence from one city to another) to Tippecanoe County in Indiana, United States (U.S.) in 2014.
3
Relocators are used as the targeted population because they make more long- and short-term travel
decisions that can be observed compared to the general population and relocation can be an ideal
time to form new travel behavior. These participants were randomly allocated to a treatment group
or a control group. The treatment group participants were provided access to IOAMA to assist in
their residential location and travel-related decision-making process, while those in the control
group did not have access. Statistical analysis was performed to determine whether there are
statistically significant differences between the treatment and control group participants in terms
of their perceived importance of different factors affecting their residential location choice, the
chosen residential neighborhood’s accessibility for different trip purposes, and their travel
behavior, such as weekly “drive alone” trips made and mode share. Simultaneous equation models
are used to analyze the impacts of having personalized neighborhood accessibility information
along with other factors on participants’ residential location choice in terms of neighborhood
accessibility and car usage in terms of weekly driving time (minutes/week). These factors include
household sociodemographic and personal preferences, and other neighborhood characteristics
(such as school district) (Pinjari et al. 2011; Guo et al. 2018; Searcy et al. 2018).
The remainder of this paper is organized as follows. Section 2 reviews previous studies on
understanding the impacts of information on residential location choice and travel behavior.
Section 3 discusses the proposed experimental design and implementation, and the methodological
underpinnings of IOAMA. Section 4 discusses the statistical analysis and model estimation results
of the experiment and the impacts of the accessibility information on the relocators’ residential
location choice and travel behavior. Section 5 provides some concluding comments.
2. Literature Review
This section focuses on four main topics related to this study, including: (i) the importance
of residential location accessibility information in people’s residential location decision-making
process, (ii) methods used to quantify a residential location’s accessibility, (iii) the impacts of
residential relocation on travel behavior, and (iv) behavioral intervention strategies to influence
the residential location decision-making process and foster formation of more sustainable travel
behavior after relocation.
Traditionally, the residential location decision-making process has been studied under
rational choice theory in which individuals have complete information about the available choices
and have the ability to process such information (e.g., Muth 1969; Hechter and Kanazawa 1997).
Individuals choose their residence by comparing available options and performing trade-offs
among factors that contribute to their residential location decision-making process. Four main
categories of contributing factors that influence the residential location decision-making process
have been identified in previous studies, including a property’s physical characteristics (e.g.,
availability of garage and lot size), neighborhood environment (e.g., land use mix and crime rate),
transportation accessibility (e.g., work commute time, distance, and costs), and decision-maker
sociodemographic and preference (e.g., gender and race) (e.g., Bayoh et al. 2006; Prashker et al.
2008; Lee and Waddell 2010; Kortum et al. 2012). Some of these studies highlighted the
importance of transportation accessibility factors in the residential location decision-making
process (e.g., Anas 1985; Molin et al. 1999; Zondag and Pieters 2005; Lee et al. 2010). However,
several recent studies suggest that people may not have complete accessibility-related information
and/or the ability to process it, especially for relocators coming from other cities (Palm and Danis
2001; Schwanen and Mokhtarian 2004; Chorus et al. 2006; Simonsohn 2006; Rodriguez and
Rogers 2014). For example, it can be difficult to quantify level of access to different trip purposes
4
for each neighborhood despite knowing the locations of their potential destinations through
information sources, such as Google Maps and Waze. Hence, relocators tend to experience longer
average commute time and have higher automobile dependency after relocation compared to long-
time residents (Levinson 1998; Lau and Chiu 2013; Guo et al. 2018). In addition, some studies
have shown that some relocators choose to move a second time within the city partly due to longer
commute and non-commute travel time caused by the mismatch between their preferred and actual
accessibility of their initial residence (e.g., congestion, lack of transit access, etc.) (Salomon and
Mokhtarian 1997; Dökmeci and Berköz 2000; Cervero and Day 2008; Sultana and Weber 2008;
Rashidi et al. 2011), especially for those with strong habit of car use and those who consider
themselves as green transport advocates (Zahra et al. 2019). This can entail heavy social and
financial burden on relocators, especially for lower-income families (Simonsohn 2006; Sriraj et al.
2006). Some relocators may not be able to relocate and/or are reluctant to relocate due to reasons
such as high cost of relocation and strong social ties to their community. These relocators may
experience higher automobile dependency, lowered living standards, and decreased quality of life
due to relatively high commute times and low access to various types of services (Milakis et al.
2017; Behren et al. 2018). These studies illustrate the importance of accessibility information in
people’s residential location decision-making process.
A residential location’s accessibility is determined by two main components, including
opportunities and cost of travel (Pivo and Fisher 2011; Guo et al. 2017). Opportunities are defined
by the types and the amount of potential destinations that a residential location can access. The
cost of travel is often measured based on the distance or travel time from a residential location to
its potential destinations. A travel time-based method to quantify cost of travel can potentially
perform better than a travel distance-based one in a multimodal context for two main reasons. On
the one hand, a travel distance-based method cannot capture the amount of travel from a
neighborhood to a potential destination using transit mode. A door-to-door transit trip includes
walking from a neighborhood to the transit stop, time spent in transit, and walking from the transit
stop to the destination. For example, for identical travel distances from two neighborhoods to a
retail location, the travel time for a neighborhood with bad transit access (e.g., fewer bus stops or
routes) may be significantly higher than that for the one with good transit access. On the other
hand, previous studies suggest that a travel distance-based method cannot capture the impact of
congestion on accessibility (e.g., Ryan 1999; Guo et al. 2016b). The Hansen-gravity measure and
floating catchment methods have been widely used in the literature as robust approaches to
quantify a residential location’s accessibility. The Hansen-gravity measure (Hansen, 1959) was
introduced to measure job accessibility by applying a gravity-based model. Although the Hansen-
gravity accessibility measure is considered more conceptually complete than the floating
catchment methods, many studies found that it is not intuitive to interpret, especially for
practitioners and the general public (Joseph and Phillips 1984; Luo and Qi 2009; Guo et al. 2016b).
Floating catchment methods represent a special type of Hansen-gravity measure and are more
intuitive to interpret (Luo and Qi 2009; Guo et al. 2017).
Travelers’ residential location characteristics (e.g., neighborhood environment and
transportation accessibility), along with some other sociodemographic and behavioral
characteristics (e.g., income and subjective norms), play important roles in shaping their travel
behavior (Bhat and Guo 2007; Cao et al. 2010; Ewing and Cervero 2010; Majid et al. 2014;
Macfarlane et al. 2015). While travel decisions, in principle, can vary on a day-to-day basis (e.g.,
driving to work versus using transit), they are more often habitual and are rarely meaningfully
reconsidered, especially for daily home-work trips (e.g., most people know which mode they want
5
to use for work without reconsidering every time) (Matthies et al. 2002; Klöckner and Matthies
2004; Verplanken et al. 2008; Rasouli and Timmermans 2015; Li et al. 2019). However, these
habitual travel behaviors can change when triggered by different significant familial or
professional events, such as relocating to a new city. These events can result in changes to travel
needs, potential destinations, preferences, options, and abilities during the process (Handy et al.
2005; Prillwitz et al. 2007; Choocharukul et al. 2008; Müggenburg et al. 2015; Clark et al. 2016;
Klinger 2017; Haggar et al. 2019; Zarabi et al. 2019). Therefore, it is important for planners and
policymakers to leverage such windows of opportunity provided by these significant events to
promote more sustainable travel behavior.
In this context, several studies have recommended designing behavioral intervention
strategies that target relocators to influence their residential location decision-making process
before their relocation and/or promote new travel behavior that is more sustainable after their
relocation, especially for relocators with weaker travel habits (Müggenburg et al. 2015; Busch-
Geertsema and Lanzendorf 2017; Zarabi and Lord 2019). These behavioral intervention strategies
can be classified into two main types, including incentive-based (e.g., Bamberg 2006; Thøgersen
2012; Walker et al. 2015) and information-based (e.g., Fujii and Taniguchi 2005; Rodriguez and
Rogers 2014; Verplanken and Roy 2016). Incentive-based behavioral intervention strategies focus
on offering economic incentives (e.g., free bus pass) and/or disincentives (e.g., penalties in the
form of increased parking costs) to relocators after their relocation to promote the usage of
sustainable travel modes and/or discourage car mode usage (Arnott et al. 2014). These strategies
can promote short-term behavioral changes, but such changes may not be sustainable when
incentives and/or disincentives are removed, and the introduction of disincentives can potentially
lead to public pushback (Kearney and De Young 1996). In addition, the effectiveness of such
strategies can also be limited as shorter-term travel decisions are often constrained by longer-term
choices such as residential location characteristics, and most incentive-based strategies are
designed to influence travel behavior after relocation. For example, offering free bus services to
relocators in a neighborhood with low transit accessibility may not be effective to promote transit
usage among them as transit cannot support their travel needs. Information-based behavioral
intervention strategies aim to influence people’s residential location choices before relocation
and/or promote them to voluntarily change their travel behavior after relocation by providing
travel-related information and feedback (Richter et al. 2011; Sunio and Schmöcker 2017; Stark et
al. 2018). Such strategies focused on promoting behavioral changes after relocation have been
found to be effective in promoting sustainable travel behavior among some relocators. However,
most of these strategies implemented after relocation have similar limitations, as residential
location choice can constrain travel behavior after relocation. To the best of the authors’
knowledge, none of the existing studies have proposed a design of personalized residential location
accessibility information to influence relocators’ travel behavior after relocation, implemented and
evaluated their effectiveness in influencing the residential location decision-making process, or
studied their impacts on travel behavior after relocation.
To address these limitations, this study: (i) proposes a design of personalized accessibility
information for information-based behavioral intervention strategies based on relocators’ travel
needs and work/school locations to influence relocators’ residential location decision-making
process, and (ii) designs and performs experiments to evaluate the proposed information design’s
effectiveness in influencing the residential location decision-making process and travel behavior
after relocation among a group of relocators. The next section describes the experimental design
6
and implementation, the methods used to create the IOAMA, and the modeling methods used to
study the impacts of such information on residential location choice and travel behavior.
3. Methodology
3.1. Experimental Design
The experimental design contained two phases: Phase I (before participants made their residential
location choice) and Phase II (three months after participants relocated). In Phase I, a pilot test
with ten participants was conducted for improving user experience of IOAMA and maximizing
the study attractiveness to potential participants. A group of participants were recruited from
individuals relocating to Tippecanoe County (referred to as “relocators” hereafter). Tippecanoe
County is in the northwest quadrant of Indiana, with about 170,000 people in 2010 (U.S. Census
Bureau 2010). It consists of 13 townships and two cities (Lafayette and West Lafayette). Over 60%
of its population is in Lafayette (38.9%) and West Lafayette (24.2%).
Participants were recruited by contacting employers in the Tippecanoe County area to
distribute recruitment emails in Spring 2014 to their newly hired employees who would start work
in Fall 2014, which ensured an adequate sample of relocators and a higher chance that the
personalized accessibility information was given to participants before they made their residential
location choice. Participation in the study was voluntary and participants were able to quit at any
time. Participants were randomly assigned to a control or a treatment group. The treatment group
participants were given password-protected access to the IOAMA designed to assist their
residential location decision-making processes while control group participants did not receive this
information.
In Phase I, a pre-relocation survey (Pre-relocation survey for control group 2014; Pre-
relocation survey for treatment group 2014) was conducted in Spring 2014 for each group to obtain
information related to the participants’ self-reported current residence, residential location choice
preference and travel behavior before relocation, and other sociodemographic characteristics. The
IOAMA information was only available in the survey given to the treatment group participants
which was the only difference in the pre-relocation surveys received by both groups.
In Phase II, a post-relocation survey (Post-relocation survey for control group 2014; Post-
relocation survey for treatment group 2014), was conducted from August 2014 to October 2014 as
a follow-up survey to participants who had completed surveys in Phase I. This was done three
months after the participant’s confirmed relocation so that they were more likely to have
established a stable travel behavior. An assumption was made that most individual and household
sociodemographic characteristics (e.g., income, education, number of people in the household) of
a participant remained the same when completing pre-relocation survey and post-relocation survey.
Hence, some of the questions related to individual and household sociodemographic characteristics
were not included in the post-relocation survey. Another assumption made is that participants
would maintain such travel behavior for a long term after relocation if their residence remains the
same and no drastic life changing events happen (e.g., getting married or having children). The
survey included information related to the participants’ self-reported residential characteristics and
travel behavior after relocation, and the residential location choice. Only the treatment group
participants received questions related to the perceived usefulness of IOAMA, which was the only
difference in the post-relocation surveys received by both groups.
To quantify the impacts of having personalized accessibility information on the participants’
residential location choice and travel behavior, four sets of outcomes (two residence-related
7
outcomes and two travel-related outcomes) were selected to analyze the differences between the
control and treatment groups. The first set consisted of the participants’ perceived importance of
11 factors affecting their residential location decision-making process before and after relocation
on a scale of 1-5, where 1 indicates “not at all important” and 5 indicates “extremely important.”
The second set consisted of six types of neighborhood accessibility values (work, healthcare, social
or recreational, restaurant, education, and retail or grocery accessibility) using four modes (walk,
bike, transit, and car). The third and fourth sets were self-reported average driving time of “drive
alone” trips made after relocation and mode share for different purposes after relocation,
respectively.
3.2. Interactive Accessibility Information Design
The IOAMA provides users personalized accessibility information for each neighborhood
of interest based on their travel needs using each transportation mode. Each neighborhood
represented a census block group (The U.S. Census Bureau 2010), and Tippecanoe County had
102 neighborhoods. The weighted accessibility of neighborhood i using mode c for participant n
was calculated as:
!!!!!!!!!!!!!"#$%&'#("#$%
()'#*"#$%
*)'#+"#$%
+)'#,"#$%
,)'#-"#$%
-)'#."#$%
.!!!!!!!!!!!!!!!!!!!!!!!!!/01
and,
'#()'#*)'#+)'#,)'#-)'#.&0223!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/41
where
'#(
,
'#*
,
'#+
,
'#,
,
'#-
, and
'#.
are the weights assigned by participant n to work, healthcare,
social or recreational, restaurant, education, and retail or grocery shopping accessibility,
respectively, and
"#$%
(
,
"#$%
*
,
"#$%
+
,
"#$%
,
,
"#$%
-
, and
"#$%
.
are the corresponding accessibilities using
mode c for participant n. It is important to note that a neighborhood’s personalized accessibility
was very different among participants due to their diverse work locations and the weights assigned
based on their travel needs. Compared to providing six types of accessibility on six maps and
asking potential users to process such information, it is easier for potential users to understand
weighted neighborhood accessibility calculated based on their travel needs using one map.
To ensure that participants can easily process the provided personalized accessibility
information, floating catchment methods were used. It is important to note that the method used to
calculate work accessibility is different from other types of accessibilities because each participant
needs access to only one job location; in principle, they can access multiple locations for other
trips purposes within a threshold of travel time.
Work accessibility (
"#$%
(
) was determined using a floating catchment method (Luo 2004).
The work accessibility of neighborhood i is calculated as follows. Given a neighborhood j,
identified by participant n as his/her work location, search any neighborhood i within a threshold
value of travel time (
56%
) using mode c. Then,
"#$%
(&7
8
5$9%:56%
;
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/<1
where
5$9%
is the travel time between i and j using mode c with a threshold value of the travel time
(
56%
) under mode c, and
7
8
5$9%:56%
; is the travel time decay function.
5$9%
, collected using Google
Maps, is the travel time between the centroid point of neighborhood i to neighborhood j using
mode c. The travel time decay function captures the inverse relationship between travel time and
accessibility, and a kernel function is often used to reflect such relationship (Dai and Wang 2011;
Guo et al. 2017). The Epanechnikov function is selected (Dai and Wang 2011) to capture the travel
time decay. The distance decay function can be written as:
8
=
7
8
5$9%:56%
;
&<
>
?
0@
A
5$9%
56%
B
C
D
:E7!5$9% F56%!
7
8
5$9%:56%
;
&2!E7!5$9% G56%!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/>1
The threshold travel time implies that only destinations within a threshold value for travel
time are considered accessible from a neighborhood, and those outside are not. The threshold travel
time to work location for all four modes was set as 60 minutes. The main reason for this threshold
travel time is that, based on Google Maps, the longest driving time between two neighborhoods in
Tippecanoe County is about 57 minutes during morning peak hours and the longest bus travel time
between any two bus stops (without waiting time) is one hour and nine minutes. Based on this, we
assumed the threshold travel time to work location was 60 minutes. Then, the work accessibility
values calculated using Eq. (3) were normalized to the indexed accessibility score ranging from 0
to 100 for each mode. The normalization scales the values of accessibility with different orders or
magnitudes to a value between 0 and 100 (Luo and Qi 2009). These values were used to quantify
the work accessibility of a neighborhood.
A neighborhood’s non-work related accessibilities were calculated using a modified
floating catchment methods method proposed by Guo et al. (2017). Given a neighborhood i, search
the intended destinations k for participant n was searched within a threshold value of the travel
time (
56%
) under mode c. Then,
!!!!!!!!!!"#$%
H
＝
I
JK7/5$K%:56%1
KL/MNOPQMRP1!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/S1
where
5$K%
is the travel time between the centroid point of neighborhood i and intended healthcare
related destination k using mode c,
TL
U
V:W:X:Y:Z
[
!
and
JK
is the weight of destination k. In
this study, the weights of the intended destinations were assumed proportional to their physical
areas. For example, for the same amount of travel time, a larger healthcare facility (such as a
general hospital) was assumed to provide higher healthcare accessibility compared to a smaller
healthcare facility (such as a clinic).
7
/
5$K%:56%
1 represents the travel time decay function, and can
be written similar to Eq. (4),
!!!!!!!!!!!!!!
\
7
/
5$K%:56%
1
&<
>
?
0@
A
5$K%
56%
B
C
D
:E7!5$K%F56%!
7
/
5$K%:56%
1
&2!E7!5$K%G56%!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/]1
The threshold travel time for all four modes to all destinations other than the work location
was set as 30 minutes. There were three reasons to select 30 minutes as the threshold travel time
for non-work destinations. First, many studies (e.g., Krizek 2003; Guo et al. 2016c) have found
that people often travel a shorter distance or time for services and shopping compared to travel to
work. Second, the longest vehicular travel time between a neighborhood and a major non-work
location (such as a supermarket for retail or grocery shopping) was within 30 minutes in
Tippecanoe County during off-peak hours (assuming these types of travels were made mostly
during off-peak hours). Third, several recent studies in a related domain (e.g., Dai and Wang 2011)
identified 30 minutes as an appropriate threshold travel time for non-work destinations in the U.S.
Then, the accessibility values calculated using Eq. (5) were normalized to the indexed accessibility
score ranging from 0 to 100 for each combination of accessibility type and mode. These values
were used to quantify the healthcare, social or recreational, restaurant, education, and retail or
grocery shopping accessibilities of a neighborhood. The information for these destinations
(including their locations and sizes) was collected using Reference USA (Reference USA 2014).
9
When IOAMA was provided to the treatment group participants, a detailed description was
provided along with it related to how to use it and how such accessibility information was
calculated.
3.3. Simultaneous Equation Model Formulation
To understand the factors that affect participants’ chosen neighborhood accessibility and
weekly driving time, several dependent variables were considered. One possible modeling method
is to create six separate models with each model tasked to address one type of neighborhood
accessibility and its corresponding driving time for that purpose. However, by modeling them
separately, it is possible that the potential correlations among these accessibilities and travel time
may be ignored. These correlations can exist because: (i) participants make their residential
location choice based on the information related to all types of accessibility and made tradeoffs
among these accessibilities; and (ii) participants who prefer to drive for one type of trip purpose
over other transportation modes may be more likely to drive to other types of trip purposes (Steg
2005; Gatersleben 2007) which leads to their longer weekly driving time. Hence, neighborhood
average weighted accessibility (average of neighborhood weighted accessibility using each mode
of transportation) and weekly driving time (minutes/week) were used for model estimation to
capture the characteristics of participants who select neighborhoods with higher neighborhood
average weighted accessibility and shorter weekly driving time. It is possible that these two
dependent variables can be modeled separately. However, if separate ordinary least squares
regression models are used, the estimation results would not factor the potential correlation
between neighborhood accessibility and vehicle usage (e.g., Ewing and Cervero 2010; Cao et al.
2010); these two models are interrelated whereby the dependent variable (residential location’s
neighborhood average weighted accessibility) in one equation can be the independent variable in
the other. This limits the use of OLS regression, as a potential estimation problem exists due to the
violation of a key ordinary least squares assumption in that a correlation exists between regressors
and disturbances, and common unobserved factors may exist affecting both dependent variables
(Washington et al. 2010). Ignoring such endogeneity can lead to erroneous conclusions (Shankar
and Mannering 1998; Tielemans et al. 1998). To address this limitation of ordinary least squares
regression for estimating the two models separately, a simultaneous equation system was used:
!!!!!!!!!!!!!^_!/"1&`aba)ca!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/d1
!!!!!!!!!!!!^_!/e1&`fbf)gfhi!/"1)cf!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!/j1
where A is the average weighted accessibility of the neighborhood that an individual selected after
relocation, V is the weekly driving time (minutes/week), Z is the vector of exogenous variables
(other contributing factors related to participants’ sociodemographic characteristics) influencing A
and V, β are the vectors of the estimable parameters, λ is the estimable scalar, and ε is the
disturbance term. Given that the dependent variables are always positive, semi-logarithmic
transformations are used. Two types of estimation methods can be used to estimate the
simultaneous equation system, including single-equation methods (e.g., two-stage least squares)
and system estimation methods (e.g., three-stage least squares). Three-stage least squares was used
in this study as it produces more efficient parameter estimates (Washington et al. 2010).
10
4. Results
4.1. Descriptive Statistics
Only individuals who completed both the pre-relocation and the post-relocation surveys
were included in the analysis. A total of 282 completed responses were collected, including 147 in
the treatment group and 135 in the control group. The pre-relocation survey questions were
organized into three parts: (1) individual and household sociodemographic characteristics, (2)
travel behavior before relocation, (3) residence-related characteristics before relocation. Tables 1-
4 illustrate some of the descriptive statistics of these characteristics. Chi-square two sample tests
were used to evaluate if there are statistically significant differences between the control and
treatment groups in terms of these characteristics. The results suggest that the participants in the
control group and the participants in the treatment group are similar in these characteristics.
A few key observations can be made. First, most of the participants in both the control and
treatment groups are Caucasians between the ages of 25 and 54, had a more than high school
diploma or had more than two private vehicles in the household (Table 1). Second, both groups
had similar travel-related behavior before relocation, including using car mode (“drive alone” and
“drive with passenger(s)”) as their primary mode of transportation for both work and non-work
trips, and checking transportation-related information regularly (more than three times a week)
(Table 2). Third, most participants owned a single-family detached home and expected to purchase
a single-family detached home with a mortgage in Tippecanoe County at the time of the pre-
relocation survey (Tables 3 and 4). In terms of the housing market in Tippecanoe County in 2014,
monthly median list price throughout 2014 was between $129,000 (January 2014) and $139,000
(August 2014) with a unit price of $81-$83 per square foot (Zillow, 2019). Monthly median rent
list price throughout 2014 was between $350 (one bedroom of a house) and $1,000 (single family
house) (Zillow, 2019). As shown in Table 4, most participants expected the purchase price and
monthly rent to be around the monthly median list price and monthly median rent list price in
Tippecanoe County, respectively, in 2014.
The post-relocation surveys consisted of two parts: (1) self-reported residence type,
ownership, and residence’s neighborhood after relocation, and (2) importance of different factors
when choosing residence. Table 5 shows the self-reported residence type and ownership after
relocation.
11
Table 1. Sociodemographic characteristics of participants
Control
Group
(N =
135)
Treatment
Group
(N = 147)
kl
Gender
Male
50.4%
52.4%
0.11
Female
49.6%
47.6%
Race/Ethnicity
African American
14.8%
21.1%
6.68
Asian
23.7%
13.6%
Hispanic/Non-white
8.9%
6.8%
Hispanic/White
5.2%
4.1%
Caucasian
47.4%
54.4%
Other
0%
0%
Marital Status
Married
44.4%
47.8%
1.89
Single
45.2%
45.4%
Separated
3.7%
1.4%
Divorced
6.7%
5.4%
Education level
Some high school
5.2%
7.5%
1.49
High school diploma
13.3%
11.6%
Technical college degree
25.2%
27.9%
College degree
29.6%
30.6%
Post graduate degree
26.7%
22.4%
Annual household income
Under $14,999
5.9%
5.5%
1.04
$15,000 – $24,999
11.9%
13.6%
$25,000 – $34,999
15.6%
12.9%
$35,000 – $49,999
18.5%
17.0%
$50,000 – $74,999
16.3%
18.4%
$75,000 – $99,999
14.8%
13.6%
$100,000 or more
17.0%
19.0%
Age
Under 25
16.4%
15.6%
1.83
25 – 34
29.6%
36.7%
35 – 44
31.1%
25.9%
45 – 54
13.3%
12.9%
Over 54
9.6%
8.9%
Average number of people living in a household
1.9
2.1
N/A
Participants with children under 6
11.9%
15.0%
0.58
Participants with children between 6 and 17
14.8%
10.2%
1.38
Average number of licensed and operable motor
vehicles in a household
2.2
2.1
N/A
12
Table 2. Travel behavior before relocation
Control Group
(N = 135)
Treatment Group
(N = 147)
kl
Average number of single work trips per week (percentage of mode share)
Drive-alone
7.84 (74.6%)
7.52 (71.5%)
2.60
Drive with passenger(s)
0.44 (4.2%)
0.88 (8.4%)
Transit
1.70 (16.2%)
1.50 (14.2%)
Bike
0.37 (3.5%)
0.41 (3.9%)
Walk
0.15 (1.5%)
0.20 (2.0%)
Average number of single non-work trips per week (percentage of mode share)
Drive-alone
5.04 (33.0%)
6.20 (38.9%)
2.42
Drive with passenger(s)
4.77 (31.2%)
4.57 (28.7%)
Transit
1.35 (8.8%)
0.82 (5.1%)
Bike
1.41 (9.2%)
1.69 (10.6%)
Walk
2.71 (17.8%)
2.65 (16.7%)
Expected work-related parking behavior after relocation
Monthly parking pass
20.0%
25.2%
1.29
Paid daily parking
3.7%
2.7%
Free parking provided by employer
18.5%
17.7%
Free street parking
38.5%
37.4%
Not driving to work
19.3%
17.0%
Transit usage (percent)
Still using
29.6%
25.2%
0.71
Not using, but has experience
70.4%
74.8%
No experience
0.0%
0.0%
Frequency of accessing travel-related information per week
Never
12.6%
12.9%
0.36
Once or twice
19.3%
21.8%
3 – 5 times
30.4%
29.9%
Once a day
26.7%
24.5%
More than once a day
11.0%
10.9%
Most frequently used source of travel-related information
Radio
38.5%
32.8%
3.16
Television
23.7%
25.0%
Internet
22.2%
18.8%
Applications on cell phone
15.6%
23.4%
Others
0.0%
0.0%
13
Table 3. Residence characteristics and preference before relocation
Control
Group
(N = 135)
Treatment Group
(N = 147)
kl
Current residence unit type
Single-family detached home
48.9%
42.2%
2.90
Row house/townhouse
23.0%
32.0%
Apartment
28.1%
25.8%
Mobile home
0.0%
0.0%
Other
0.0%
0.0%
Ownership of current residence unit
Owning without mortgage
8.9%
10.2%
3.61
Owning with mortgage
56.3%
65.3%
Renting
34.8%
24.5%
Relocation purpose
Going to work
93.3%
94.5%
0.19
Attending school
6.7%
5.5%
Residence type of interest (multiple-
choice)
Single-family detached home
65.2%
63.3%
1.01
Row house/townhouse
33.3%
38.1%
Apartment
36.3%
31.3%
Mobile home
0.0%
0.0%
Other
0.0%
0.0%
Expected ownership after relocation
Owning without mortgage
15.6%
14.3%
0.92
Owning with mortgage
57.0%
53.1%
Renting
27.4%
32.6%
14
Table 4. Expected residential costs before relocation
Control
Group
Treatment
Group
kl
Expected total costs if decided to own a residence without mortgage after
relocation
Total number of participants who expected to
own a residence without mortgage
21
25
0.17
Under $150,000
38.1%
44.0%
$150,000 – $199,999
52.4%
48.0%
$200,000 – $299,999
9.5%
8.0%
$300,000 – $499,999
0%
0%
$500,000 or more
0%
0%
Median list price for buying all types of residence
in Tippecanoe County in 2014
$129,000 ~ $139,000
Expected monthly mortgage if decided to own a residence with mortgage after relocation
Total number of participants who expected to
own a residence with mortgage
75
78
0.40
Under $1,000
37.7%
42.3%
$1,000 – $1,499
61.0%
56.4%
$1,500 – $1,999
1.3%
1.3%
$2,000 or more
0.0%
0.0%
Expected rent if decided to rent a residence
Total number of participants who expected to rent
a residence
37
47
0.04
Under $500
62.2%
63.8%
$500 – $749
29.7%
27.7%
$750 – $999
8.1%
8.5%
$1,000 – $1,499
0.0%
0.0%
$1,500 or more
0.0%
0.0%
Median list price for renting all types of residence
in Tippecanoe County in 2014
$350 ~ $1,000
15
Table 5. Residence characteristics after relocation
Control Group
(N = 135)
Treatment Group (N
= 147)
Current residence type
Single-family detached home
40.0%
46.2%
Row house/townhouse
25.9%
32.0%
Apartment
34.1%
21.8%
Mobile home
0.0%
0.0%
Other
0.0%
0.0%
Ownership of current residence
Owning without mortgage
10.4%
14.9%
Owning with mortgage
54.0%
59.9%
Renting
35.6%
25.2%
Total costs of current residence if the ownership is owning without mortgage
Total number of participants who reported owning
a residence without mortgage
14
22
Under $150,000
14.3%
18.2%
$150,000 – $199,999
50.0%
45.5%
$200,000 – $299,999
35.7%
36.3%
$300,000 – $499,999
0.0%
0.0%
$500,000 or more
0.0%
0.0%
Monthly mortgage of current residence if the ownership is owning with mortgage
Total number of participants who reported owning
a residence with mortgage
73
88
Under $1,000
23.3%
36.4%
$1,000 – $1,499
60.3%
51.1%
$1,500 – $1,999
16.4%
12.5%
$2,000 or more
0.0%
0.0%
Rent of current residence if the ownership is renting
Total number of participants who reported renting a
residence with mortgage
48
37
Under $500
35.4%
29.7%
$500 – $749
29.2%
43.2%
$750 – $999
33.3%
27.1%
$1,000 – $1,499
2.1%
0.0%
$1,500 or more
0.0%
0.0%
Expected number of years of staying at the current residence
Less than 1 year
25.2%
17.7%
1 – 5 years
15.6%
10.9%
5 – 10 years
57.0%
68.0%
More than 10 years
2.2%
3.4%
Most of the treatment group participants (over 95%) reported that they relocated to a
residence consistent with their preference at the time of the pre-relocation survey in terms of the
residence type and ownership compared to control group participants (about 70%) (Table 3). More
than 10% of the control group participants reported that they choose to rent a residence instead of
buying one. In addition, the treatment group participants planned to stay longer in their current
property compared to the control group participants, suggesting greater satisfaction with their
residential location choice. The results show that participants who had accessibility information
16
were more likely to find residences that satisfied their needs, and they purchased the residence. By
contrast, some participants without accessibility information did not find an initial residential
location that satisfied their needs and were therefore more likely to rent a residence for a short
period with a higher likelihood of moving later to a residence meeting their needs within the region.
In addition, based on the aggregated self-reported residential locations in Tippecanoe County for
the control and treatment group participants, the average estimated distance for the treatment group
participants from their neighborhood to their work locations was about 25% shorter compared to
that for the control group participants.
4.2. Perceived Importance of Various Factors that Affected Residential Location Choice
Participants were requested to rate their perceived importance of various factors that
affected their residential location decision-making process on a scale of 1-5, where 1 indicates “not
important at all” and 5 indicates “extremely important.” Eleven factors in three categories were
considered: (1) physical characteristics of residence (cost, number of bedrooms/bathrooms, and
parking); (2) neighborhood environment (aesthetic value and safety); and (3) transportation
accessibility to education, work, park/recreational/public facilities, restaurants, retail/grocery, and
healthcare. Table 6 illustrates the average ratings of these factors.
Based on a series of Whitney U tests comparison of the factor means and Spearman’s rank
correlation coefficients for analyzing the statistical dependence for the within-group ranking (Guo
and Peeta 2015; Guo et al. 2016a), the results show that there is: (i) a high degree of similarity
among the participants of the control and treatment groups in terms of their perceived importance
of the factors in their residential location decision-making process before relocation, (ii) a high
degree of dissimilarity in the ratings on these factors of the treatment group participants before and
after relocation after receiving the personalized accessibility information, and (iii) a high degree
of similarity in the rating on these factors among the participants of the control group who did not
receive the information. These results were possibly because relocators with personalized
accessibility information were more informed on transportation accessibility and placed higher
importance on accessibility-related factors in their residential location decision-making process.
4.3. Neighborhood Accessibility to Different Trip Purposes
In the post-relocation surveys, participants were asked to identify the neighborhood where
their residence was located rather than their address for privacy reasons. Table 7 shows that the
averages of the neighborhood accessibility for the six trip purposes using the four modes for the
treatment group participants were higher than those of the control group, especially for
neighborhood accessibility using non-automobile modes. These results show that personalized
accessibility information can assist relocators in selecting neighborhoods with better access to their
potential destinations using different modes of transportation, especially for accessibility using
non-automobile modes, which normally is not easily accessible to relocators.
17
Table 6. Importance of different factors affecting participants’ residential location decision-making process
Before relocation
After relocation
Control
Group
Treatment
Group
p-value
Control
Group
p-value
Treatment
Group
p-value
Physical characteristics of the residence after relocation
Cost of renting or buying
3.90
3.95
0.72
3.96
0.68
3.79
0.42
Number of bedrooms/bathrooms
2.97
3.01
0.74
3.02
0.73
2.95
0.86
Parking availability
2.55
2.51
0.79
2.74
0.20
2.22
0.02*
Neighborhood environment
Safety of neighborhood
3.21
2.99
0.15
3.31
0.55
3.14
0.64
Aesthetic value
2.91
2.86
0.74
3.03
0.46
2.97
0.70
Transportation accessibility
Accessibility to work
3.03
2.99
0.79
3.06
0.86
2.88
0.31
Accessibility to restaurants
2.58
2.48
0.39
2.67
0.45
2.74
0.16
Accessibility to retail, grocery or other destinations
2.44
2.49
0.69
2.56
0.35
2.82
0.00*
Accessibility to parks, recreational, or public
facilities
2.39
2.37
0.91
2.45
0.66
2.85
0.00*
Accessibility to education
2.36
2.44
0.65
2.27
0.56
2.82
0.00*
Accessibility to healthcare
1.44
1.36
0.47
1.33
0.28
1.62
0.15
* denotes significance at a 95% level of confidence
18
4.4. Weekly “Drive Alone” Trips and Mode Share for Various Trip Purposes
Table 8 illustrates the aggregated travel behavior of participants in the control and
treatment groups. After relocation, for all trip purposes, the treatment group participants
experienced shorter average driving times using “drive alone” mode compared to the control group
participants, and these differences were statistically significant for work, social/recreational,
restaurants, and retail/grocery shopping trips. In addition, the shares of trips using non-automobile
modes (transit, bike, and walk) were higher for the treatment group participants compared to the
control group participants, especially for walk mode in social/recreational, restaurants, and
retail/grocery shopping trips.
Additional t-test analyses were performed to examine whether certain subgroups among
the treatment group participants experienced a larger impact from the personalized accessibility
information based on gender, age, household income, marital status, automobile ownership,
whether using transit before relocation, and the frequency of accessing transportation-related
information per week. The results show that married participants in the treatment group have a
statistically significant increase in “drive with passenger(s)” mode usage after relocation while no
statistically significant change is observed among unmarried participants. It is possible that these
married participants may find a residence that was more suitable for making joint travel decisions
after relocation compared to the one they had before relocation. The treatment group participants
who accessed transportation-related information more often (more than three times a week) chose
residences in neighborhoods with higher accessibility. This suggests that participants who
accessed transportation-related information more often may have used the IOAMA more
effectively or were more receptive to the information from IOAMA in their residential location
decision-making process.
4.5. Simultaneous Equation Estimation Results
Table 9 shows the simultaneous equation model estimation results. For comparison, the
two models were also run as separate ordinary least squares regression models. The comparison
results illustrated that the two separate ordinary least squares regression models showed noticeably
higher standard errors resulting in lower t-statistics compared to the simultaneous equation models.
Similar observations were also found in previous studies (e.g., Shankar and Mannering 1998).
As shown in Table 9, six variables were found to have a statistically significant correlation
(t ≥ 1.96) with the average weighted accessibility of the neighborhood than an individual selected
after relocation (hereafter labeled as the neighborhood average weighted accessibility), including
three variables related to individual and household sociodemographic characteristics, two variables
related to travel behavior before relocation, and one variable related to whether an individual was
in the treatment group or not.
Four variables were found to have a statistically significant correlation (t ≥ 1.96) with
weekly driving time (minutes/week) after relocation, including one variable related to individual
and household sociodemographic characteristics, one variable related to travel behavior before
relocation, one variable related to whether an individual was in the treatment group or not, and the
neighborhood average weighted accessibility.
19
Table 7. Average neighborhood accessibility for different trip purposes
Control Group
(N = 135)
Treatment Group
(N = 147)
p-value
Accessibility to work:
Car
72.75
89.63
0.67
Transit
62.83
84.52
0.03*
Bike
65.11
86.93
0.07*
Walk
61.34
77.84
0.05*
Accessibility to healthcare:
Car
50.24
57.21
0.62
Transit
52.42
55.72
0.80
Bike
56.48
58.67
0.72
Walk
55.90
59.72
0.52
Accessibility to social and recreational activities
Car
67.75
85.22
0.04*
Transit
61.04
86.27
0.00*
Bike
62.69
82.64
0.05*
Walk
63.10
87.62
0.03*
Average accessibility to restaurants
Car
70.25
82.56
0.40
Transit
69.02
84.55
0.32
Bike
65.42
86.21
0.08*
Walk
67.53
87.00
0.09*
Accessibility to educational activities
Car
72.42
74.62
0.75
Transit
70.20
73.45
0.80
Bike
71.25
75.69
0.69
Walk
72.21
76.01
0.65
Accessibility to retail/grocery activities
Car
64.38
88.34
0.04*
Transit
66.71
87.63
0.06*
Bike
65.17
89.21
0.02*
Walk
66.08
90.26
0.01*
Weighted accessibility
Car
67.74
80.60
0.00*
Transit
64.43
81.23
0.00*
Bike
65.42
84.54
0.00*
Walk
67.22
82.10
0.00*
20
Table 8. Comparison of travel-related outcomes after relocation
Control
Group
(N = 135)
Treatment
Group
(N = 147)
p-value
Work trips
Average driving time of using “drive alone” mode (minutes)
9.38
8.25
0.00*
Average weekly driving time using “drive alone” mode (minutes)
93.47
81.85
0.00*
Percentage of “drive with passenger(s)” mode share
7.41
11.60
0.23
Percentage of transit mode share
13.19
19.51
0.15
Percentage of bike mode share
3.26
3.68
0.84
Percentage of walk mode share
5.93
9.28
0.27
Healthcare-related trips
Average driving time of using “drive alone” mode (minutes)
11.33
9.44
0.60
Average weekly driving time using “drive alone” mode (minutes)
24.25
21.50
0.68
Percentage of “drive with passenger(s)” mode share
29.41
31.25
0.91
Percentage of transit mode share
5.88
0.00
0.32
Percentage of bike mode share
0.00
0.00
--
Percentage of walk mode share
0.00
6.25
0.32
Social/recreational trips
Average driving time of using “drive alone” mode (minutes)
8.21
7.66
0.08*
Average weekly driving time using “drive alone” mode (minutes)
32.65
27.60
0.04*
Percentage of “drive with passenger(s)” mode share
36.29
36.34
0.64
Percentage of transit mode share
7.87
4.76
0.13
Percentage of bike mode share
15.23
13.53
0.44
Percentage of walk mode share
19.04
28.82
0.07*
Restaurant-related trips
Average driving time of using “drive alone” mode (minutes)
8.65
7.71
0.00*
Average weekly driving time using “drive alone” mode (minutes)
36.15
30.32
0.00*
Percentage of “drive with passenger(s)” mode share
40.70
37.41
0.23
Percentage of transit mode share
4.91
6.47
0.70
Percentage of bike mode share
1.75
1.80
0.74
Percentage of walk mode share
7.02
22.30
0.08*
Education-related trips
Average driving time of using “drive alone” mode (minutes)
8.93
8.11
0.72
Average weekly driving time using “drive alone” mode (minutes)
52.29
45.47
0.84
Percentage of “drive with passenger(s)” mode share
32.69
28.68
0.92
Percentage of transit mode share
15.38
14.73
0.87
Percentage of bike mode share
5.77
3.88
0.74
Percentage of walk mode share
12.50
12.40
0.84
Retail/grocery shopping trips
Average driving time of using “drive alone” mode (minutes)
9.13
8.05
0.01*
Average weekly driving time using “drive alone” mode (minutes)
19.29
16.19
0.00*
Percentage of “drive with passenger(s)” mode share
39.89
36.84
0.77
Percentage of transit mode share
13.30
15.31
0.60
Percentage of bike mode share
0.00
0.96
0.16
Percentage of walk mode share
15.43
24.88
0.04*
21
Table 9. Simultaneous equation estimation results
Variables
Estimates
t-Statistics
Standard
error
estimates
Dependent variable: Neighborhood average weighted accessibility
Constant
3.17
10.11
0.31
Treatment group indicator: 1, if individual was in
treatment group; 0, otherwise
1.03
7.21
0.14
High income indicator: 1, if individual’s annual
household income is over $49,999; 0, otherwise
0.33
2.08
0.16
Married indicator: 1, if individual is married; 0,
otherwise
0.14
2.43
0.06
Average number of licensed and operable motor
vehicles in individual’s household
-0.46
-2.77
0.17
Automobile-habit indicator: 1, if at least 60% of
individual’s trips were made using “drive alone”
mode before relocation; 0, otherwise
-0.96
-7.30
0.13
Frequent transportation information access indicator: 1,
if individual’s frequency of accessing transportation-
related information per week is three times or more;
0, otherwise
1.01
3.65
0.28
Dependent variable: Weekly driving time after relocation (minutes/week)
Constant
4.13
14.19
0.29
Average weighted accessibility
-0.97
-7.47
0.13
Treatment group indicator: 1, if individual was in
treatment group; 0, otherwise
-0.83
-5.53
0.15
Average number of licensed and operable motor
vehicles in individual’s household
0.37
3.41
0.12
Automobile-habit indicator: 1, if at least 60% of
individual’s trips were made using “drive alone”
mode before relocation; 0, otherwise
-0.74
3.27
0.23
Number of observations
282
R-squared—Average weighted accessibility
0.41
R-squared—Automobile travel per week
0.47
3SLS system R-squared
0.46
The estimation results indicate that the treatment group participants who received
personalized accessibility information were more likely to choose a neighborhood with higher
average weighted accessibility and traveled less by private vehicle. This is consistent with the
results of the t-test comparison of the average neighborhood accessibility and travel-related
outcomes after relocation between the control and treatment group participants (Tables 6 and 7),
which suggest that the designed personalized accessibility information can assist participants in
selecting neighborhoods with a better average weighted accessibility and reduce their car usage.
22
The neighborhood average weighted accessibility was found to have a statistically
significant negative correlation with weekly driving time after relocation. Similar results were also
found in previous studies (e.g., Cao et al. 2010); that is, individuals who lived in neighborhoods
with higher accessibility traveled less using car mode compared to those who lived in
neighborhoods with lower accessibility. This outcome indicates that the proposed strategy can
foster sustainable long-term travel behavior, in terms of reducing car usage by assisting
participants’ with information that will assist their selection of a neighborhood that offers better
access to their potential destinations.
The average number of licensed and operable vehicles in a household was found to have a
statistically significant negative correlation with the neighborhood average weighted accessibility,
but was found to have a statistically significant positive correlation with weekly driving time after
relocation. A possible explanation for this is that households with more mobility resources may
value neighborhood accessibility less, but value other factors (such as costs of renting or buying)
more in their residential location decision-making process due to their high household mobility.
The residential property price was not found to have a statistically significant correlation
with a neighborhood’s average weighted accessibility. This may seem to contradict to several
previous studies (e.g., Guo et al. 2016b) in which a property’s neighborhood accessibility is often
found to be positively correlated with its property price. However, such correlation may not exist
between neighborhood average weighted accessibility and a property’s price, because of the
significant difference between neighborhood average weighted accessibility and neighborhood
accessibility. A property’s neighborhood average weighted accessibility depends not only on its
various types of non-work-related neighborhood accessibilities, but also on an individual’s work
location and travel needs. An individual’s work location determines the neighborhood work
accessibility, and his/her travel needs dictate how much he/she weights each type of accessibility.
This indicates that, for the same property, different people can have different assessments in terms
a property’s weighted accessibility. For example, among the participants, some individuals who
work in rural areas weighed work accessibility much more than other types of accessibility. They
may be more likely to select a property located near their work location, and the property’s price
may also be lower in a rural region. In this case, a residential property’s price may be negatively
correlated with the neighborhood average weighted accessibility among these participants.
Individuals with similar sociodemographic characteristics whose jobs are located downtown and
weighed non-work-related accessibilities much higher than work accessibility, may select a
property location that offers better access to non-work-related activities. Also, the property’s price
may be higher than the one located in a rural region. In this case, a residential property’s price is
positively correlated with the neighborhood average weighted accessibility. Hence, as people have
different work locations and diverse travel needs, it is reasonable that the estimation results show
that there is no statistically significant correlation between a residential property’s price and its
neighborhood average weighted accessibility. However, it does not imply that the residential
property price is not factored in the residential location decision-making process.
Two variables related to individual and household sociodemographic characteristics,
household income and marital status, were found to be statistically significantly correlated with
the neighborhood average weighted accessibility but not with weekly driving time after relocation.
If an individual’s annual household income is over $49,999, they are more likely to select a
neighborhood with higher average weighted accessibility. In 2014, the median annual household
income in Tippecanoe County was $44,474 (the U.S. Census Bureau 2015). An individual with a
higher annual household income may be less sensitive to costs of renting or buying, and other
23
factors such as accessibility may be more important in their residential location decision-making
process. Hence, they are more likely to relocate to neighborhoods with higher average weighted
accessibility. The results also show that if an individual is married, he or she is more likely to
select a neighborhood with higher average weighted accessibility after relocation because married
individuals are more likely to have diverse travel needs and travel behavior in their household
when making residential location choice, while unmarried or separated/divorced individuals may
only have to factor their own needs. Hence, married individuals are more likely to select a
neighborhood with high accessibility for different trip purposes using different modes of
transportation, which is consistent with the subgroup study results. If an individual made at least
60% of his/her trips using “drive alone” mode before relocation, he/she was more likely to choose
a neighborhood with lower average weighted accessibility and have a higher weekly driving time
after relocation. This is similar to findings in previous studies (e.g., Choocharukul et al. 2008), in
that individuals with a frequent car usage habit was less likely to relocate to a neighborhood with
convenient public transportation. This indicates that an individual’s travel behavior before
relocation has a strong impact on his/her residential location decision-making process and travel
behavior after relocation.
The results also illustrate that individuals who access transportation-related information
more frequently (three times or more per week) were more likely to select neighborhoods with
higher average weighted accessibility after relocation because such individuals often may be more
amenable to using accessibility-related information and value a higher level of accessibility more
when choosing their residential location, which also consistent with the insights in this paper.
4.6. Study Limitations
This study has several potential limitations. First, the voluntary nature of the online survey
without compensation, and the use of IOAMA which is an online application, may limit the types
of participants in the study in terms of their sociodemographic characteristics (e.g., they need to
have internet access) and may not be fully representative of relocators to Tippecanoe County.
Second, the education accessibility provided was measured based only on travel time to different
schools which may not reflect the actual access to educational services (e.g., school zone
boundaries and school quality). Third, the threshold travel time in this study was predetermined
based on the literature instead of personalized threshold travel time. A tradeoff was made to reduce
some interactive features and improve study participation based on pilot test results. Fourth, the
data analysis relies on self-reported survey data which has its limitations in terms of data accuracy
that can be difficult to verify. Fifth, the use of census blocks as neighborhood boundaries may not
be ideal because, in many cases, a census block does not represent a neighborhood (Coulton et al.
2001). Sixth, other types of models such as the class of random parameters models (Andrew et al.,
2013; Anastasopoulos and Mannering, 2016; Behnood and Mannering, 2017; Yamamoto et al.,
2018; Eker et al., 2019; Fountas et al., 2019) and latent class models (Fountas et al., 2018; Ding-
Mastera et al., 2019) can be considered to account for unobserved heterogeneity. Seventh, it is
assumed that a participant’s individual and household sociodemographic characteristics remained
the same when completing pre-relocation and post-relocation surveys. It is possible that some
participants may have experienced individual and household sociodemographic characteristics
changes considering the time span between pre-relocation and post-relocation surveys. Eighth, it
is assumed that participants in the treatment group would maintain their travel behavior as they
reported three months after relocation because most of them found a residence that addressed their
travel needs and intended to stay there for a long time. It is possible that some of the participants
24
may have experienced relapse and went back to their travel behavior before relocation. A longer-
term study can be conducted to address this limitation. Ninth, an assumption was made in the
model estimation that six types of accessibilities and driving time to different trip purposes can
potentially be correlated. Additional studies are needed to validate this assumption.
5. Concluding Remarks and Future Work
This study proposes a design of personalized accessibility information which factors an
individual’s work location, travel needs, and mode choice and evaluates its potential to foster
sustainable travel behavior by providing it to relocators’ to influence their long-term residential
location choices so that they can reduce car usage and increase the usage of sustainable travel
modes. Although other neighborhood-related characteristics (such as school district, crime rate,
etc.) also can influence the long-term residential location decision-making process, this study
focused on analyzing the impacts of personalized accessibility information on the residential
location decision-making process and travel behavior after relocation. A key motivation for this
focus is that advances in information and communication technologies can potentially be leveraged
to enable relocators to be more informed in their decision-making process through the provision
of intuitive visual information.
The descriptive statistics and model estimation results show that the designed personalized
accessibility information can influence relocators’ residential location decision-making process.
The treatment group participants who received personalized accessibility information were more
likely to place more importance on the accessibility-related factors of potential residence locations
in their residential location decision-making process and choose a residence in a neighborhood
that had better overall accessibility using different modes of transportation, and was more suitable
to their specific household travel needs. This can be attributed to the designed information itself
and the possibility that such information can inspire treatment group participants to check more
transportation-related information for the neighborhood which can further help them to make more
informed residential location decisions. The treatment group participants were also less likely to
consider changing their residence again after relocation, tended to drive alone less, drove with
other household members more, and walked, rode a bicycle, and used transit more if they
maintained their travel behavior after relocation compared to the control group participants as their
travel needs were better addressed in their residence after relocation.
By influencing the long-term residential location choice, the long-term travel behavior of
individuals also can be altered and a new travel behavior formed after relocation in terms of
reducing car usage and increasing their mode share of walk, bike and transit. These insights have
four important implications for planners and policy-makers in the context of designing information
for intervention strategies to improve the sustainability of travel behavior. First, a community-
based approach can be used to design information intervention strategies by leveraging community
characteristics. For example, community inputs can be collected through entities such as
community advisory committees, and community coalitions can be forged to assist in tailoring
interventions to specific target groups within the community. An effective information
intervention strategy in one community does not ensure its success in another without
understanding that community and its people. For example, information intervention strategies
that focus on providing up-to-the-minute transit information can be beneficial to increase transit
ridership in neighborhoods with relatively good transit accessibility and/or positive attitude
towards using transit for travel. However, such strategies may not be very effective in promoting
25
transit usage in neighborhoods with relatively low transit accessibility (e.g., using transit cannot
satisfy their travel needs) and/or for travelers with strong automobile dependency.
Second, strategies can be more effective in targeting people who are experiencing
significant familial or professional events and will make long-term decisions. These can include
but are not limited to relocators who will make their residential location decisions resulting in
travel behavioral changes, people who will change their jobs, younger generations who will decide
if they want to get a driver’s license, potential car owners who will decide if they want to buy a
car, or people who have weaker car use habit and want to develop a more sustainable travel
behavior. As long-term decisions will affect short-term travel decisions which can be habitual once
formed, it is important to target people while they are making long-term decisions and are more
likely to form a new set of travel-related habits once these decisions are made. Third, personalized
information delivery and visualization can potentially be more useful for people as an easier way
to process information, thus improving the information’s attractiveness and effectiveness
compared to strategies that provide generalized information based on a “one-size-fits-all” approach.
The results show that individuals who more frequently access transportation-related information
are more likely to be amenable to the influence of accessibility information for intervention
strategies. Information can be delivered through channels that people are more accustomed to, and
the application should be easy to access and use. In addition, information can be designed to be
more personalized not only for users themselves but also for their families to make joint short- and
long-term decisions that can better address travel and other related needs of all family members.
Fourth, the proposed information-based intervention strategies can be bundled with other long-
and short-term strategies (such as congestion pricing strategies, reward programs for using public
transit, and other incentive- and information-based strategies) to improve their ability to influence
individuals with strong automobile use habits.
Future study directions can include the following: (i) implement the proposed intervention
strategy in a larger metropolitan area with a larger sample size and develop separate econometric
models for the participants in the control and treatment groups to evaluate the proposed strategy’s
effectiveness; (ii) use the designed personalized accessibility information as a foundation to
develop a more comprehensive livability index from a transportation perspective with bundled
information related to accessibility and the neighborhood built environment (such as school district
quality); and (iii) include additional interactive features (e.g., adjustable threshold of travel time,
restaurant preference, etc.) to IOAMA and link it with real estate websites (e.g., Zillow, Trulia,
etc.) to provide easier access to both residential location choices and personalized accessibility
information which can better assist relocators to make more informed residential location choices.
Acknowledgments
This study was based on research supported by the NEXTRANS Center, the USDOT Region 5
University Transportation Center at Purdue University. The authors gratefully acknowledge the
individuals who participated in the experiment. The authors also thank Dr. Jonathan Levine of the
University of Michigan for his suggestions on the survey questionnaire design. Thanks are also
due to Shubham Agrawal for inputs on designing the IOAMA, Yue E for suggestions in data
analysis, and Jennifer Alexander for inputs on the survey distribution process. The authors are
solely responsible for the contents of this paper.
26
References
Anas, A. (1985). The combined equilibrium of travel networks and residential location
markets. Regional Science and Urban Economics, Vol. 15, No. 1, 1-21.
Anastasopoulos, P. C., Mannering, F. L., and Haddock, J. E. (2012). Random parameters
seemingly unrelated equations approach to the postrehabilitation performance of pavements.
Journal of Infrastructure Systems, Vol. 18, No. 3, 176-182.
Andersson, A., Hiselius, L. W., and Adell, E. (2018). Promoting sustainable travel behaviour
through the use of smartphone applications: A review and development of a conceptual
model. Travel Behaviour and Society, Vol. 11, 52-61.
Arnott, B., Rehackova, L., Errington, L., Sniehotta, F. F., Roberts, J., and Araujo-Soares, V. (2014).
Efficacy of behavioural interventions for transport behaviour change: systematic review, meta-
analysis and intervention coding. International Journal of Behavioral Nutrition and Physical
Activity, No. 11, No. 1, 133.
Bamberg, S. (2006). Is a residential relocation a good opportunity to change people’s travel
behavior? Results from a theory-driven intervention study. Environment and Behavior, Vol.
38, No. 6, 820-840.
Bayoh, I., Irwin, E. G., and Haab, T. (2006). Determinants of residential location choice: How
important are local public goods in attracting homeowners to central city locations? Journal of
Regional Science, Vol. 46, No. 1, 97-120.
Behnood, A., and Mannering, F. (2017). The effect of passengers on driver-injury severities in
single-vehicle crashes: A random parameters heterogeneity-in-means approach. Analytic
Methods in Accident Research, Vol. 14, 41-53.
von Behren, S., Puhe, M., and Chlond, B. (2018). Office relocation and changes in travel behavior:
Capturing the effects including the adaptation phase. Transportation Research Procedia, Vol.
32, 573-584.
Ben-Elia, E. and Shiftan, Y. (2010). Which road do I take? A learning-based model of route-choice
behavior with real-time information. Transportation Research Part A: Policy and
Practice, Vol. 44, No. 4, 249-264.
Bhat, C. R. and Guo, J. Y. (2007). A comprehensive analysis of built environment characteristics
on household residential choice and auto ownership levels. Transportation Research Part B:
Methodological, Vol. 41, No. 5, 506-526.
Busch-Geertsema, A., and Lanzendorf, M. (2017). From university to work life–Jumping behind
the wheel? Explaining mode change of students making the transition to professional life.
Transportation Research Part A: Policy and Practice, Vol. 106, 181-196.
Cao, X. J., Xu, Z., and Fan, Y. (2010). Exploring the connections among residential location, self-
selection, and driving: Propensity score matching with multiple treatments. Transportation
Research Part A: Policy and Practice, Vol. 44, No. 10, 797-805.
Cervero, R., and Day, J. (2008). Suburbanization and transit-oriented development in China.
Transport Policy, Vol. 15, No. 5, 315-323.
Choocharukul, K., Van, H. T., and Fujii, S. (2008). Psychological effects of travel behavior on
preference of residential location choice. Transportation Research Part A: Policy and Practice,
Vol. 42, No. 1, 116-124.
Chorus, C. G., Arentze, T. A., Molin, E. J., Timmermans, H. J., and Van Wee, B. (2006). The
value of travel information: Decision strategy-specific conceptualizations and numerical
examples. Transportation Research Part B: Methodological, Vol. 40, No. 6, 504-519.
27
Clark, B., Chatterjee, K., and Melia, S. (2016). Changes in level of household car ownership: the
role of life events and spatial context. Transportation, Vol. 43, No. 4, 565-599.
Collins, A. T., Rose, J. M., and Hensher, D. A. (2013). Specification issues in a generalised random
parameters attribute nonattendance model. Transportation Research Part B: Methodological,
Vol. 56, 234-253.
Coulton, C. J., Korbin, J., Chan, T., and Su, M. (2001). Mapping residents' perceptions of
neighborhood boundaries: a methodological note. American Journal of Community
Psychology, Vol. 29, No. 2, 371-383.
Dai, D. and Wang, F. (2011). Geographic disparities in accessibility to food stores in southwest
Mississippi. Environment and Planning B: Planning and Design, Vol. 38, No. 4, 659-677.
Ding-Mastera, J., Gao, S., Jenelius, E., Rahmani, M., and Ben-Akiva, M. (2019). A latent-class
adaptive routing choice model in stochastic time-dependent networks. Transportation
Research Part B: Methodological, Vol. 124, 1-17.
Dökmeci, V., and Berköz, L. (2000). Residential-location preferences according to demographic
characteristics in Istanbul. Landscape and Urban Planning, Vol. 48, No. 1-2, 45-55.
Eker, U., Ahmed, S. S., Fountas, G., & Anastasopoulos, P. C. (2019). An exploratory investigation
of public perceptions towards safety and security from the future use of flying cars in the
United States. Analytic Methods in Accident Research, 23, 100103.
Emmerink, R. H., Axhausen, K. W., Nijkamp, P., and Rietveld, P. (1995). Effects of information
in road transport networks with recurrent congestion. Transportation, Vol. 22, No. 1, 21-53.
Ewing, R., and Cervero, R. (2010). Travel and the built environment. Journal of the American
Planning Association, Vol. 76, No. 3, 265-294.
Fountas, G., Anastasopoulos, P. C., and Mannering, F. L. (2018). Analysis of vehicle accident-
injury severities: a comparison of segment-versus accident-based latent class ordered probit
models with class-probability functions. Analytic Methods in Accident Research, Vol. 18, 15-
32.
Fountas, G., Pantangi, S. S., Hulme, K. F., and Anastasopoulos, P. C. (2019). The effects of driver
fatigue, gender, and distracted driving on perceived and observed aggressive driving behavior:
A correlated grouped random parameters bivariate probit approach. Analytic Methods in
Accident Research, Vol. 22, 100091.
Fujii, S., and Taniguchi, A. (2005). Reducing family car-use by providing travel advice or
requesting behavioral plans: An experimental analysis of travel feedback programs.
Transportation Research Part D: Transport and Environment, Vol. 10, No. 5, 385-393.
Gatersleben, B. (2007). Affective and symbolic aspects of car use. In Threats from car traffic to
the quality of urban life: Problems, Causes and Solutions (pp. 219-233). Emerald Group
Publishing Limited.
Guo, Z. (2011). Mind the map! The impact of transit maps on path choice in public
transit. Transportation Research Part A: Policy and Practice, Vol. 45, No. 7, 625-639.
Guo, Y. and Peeta, S. (2015). Rail-truck multimodal freight collaboration: truck freight carrier
perspectives in United States. Journal of Transportation Engineering Vol. 141, No. 11, 1-11.
Guo, Y., Peeta, S., and Mannering, F. L. (2016). Rail-truck multimodal freight collaboration: A
statistical analysis of freight-shipper perspectives. Transportation Planning and Technology,
Vol. 39, No. 5, 2016a, 484-506.
Guo, Y., Peeta, S., and Somenahalli, S. (2015). The impact of walkable environment on single
family residential property values. Journal of Transport and Land Use, Vol. 10, No. 2, DOI:
http://dx.doi.org/10.5198/jtlu.2017.824.
28
Guo, Y., Agrawal, S., Peeta, S., and Somenahalli, S. (2016). Impacts of property accessibility and
neighborhood built environment on single-unit and multiunit residential property values.
Transportation Research Record: Journal of the Transportation Research Board, No. 2568,
2016b, 103-112.
Guo, Y., Wang, J., and Peeta, S. (2016). The impacts of household registration and head of
household travel behavior on travel mode choice in China. In compendium of Transportation
Research Board 95th Annual Meeting, Washington D.C., U.S., 2016c.
Guo, Y., Wang, J., Peeta, S., and Anastasopoulos, P. C. (2018). Impacts of internal migration,
household registration system, and family planning policy on travel mode choice in China.
Travel Behaviour and Society, Vol. 13, 128-143.
Haggar, P., Whitmarsh, L., and Skippon, S. M. (2019). Habit discontinuity and student travel mode
choice. Transportation Research Part F: Traffic Psychology and Behaviour, Vol. 64, 1-13.
Handy, S., Cao, X., and Mokhtarian, P. (2005). Correlation or causality between the built
environment and travel behavior? Evidence from Northern California. Transportation
Research Part D: Transport and Environment, Vol. 10, No. 6, 2005, 427-444.
Hansen, W. G. (1959). How accessibility shapes land use. Journal of the American Institute of
Planners, Vol. 25, No. 2, 73-76.
Hechter, M. and Kanazawa, S. (1997). Sociological rational choice theory. Annual Review of
Sociology, 191-214.
IOAMA. Accessed September, 2015 via: http://web.ics.purdue.edu/~guo187/mp/nextrans.php.
Joseph, A., and Phillips, R. Accessibility and Utilization: Geographical Perspectives on Health
Care Delivery. New York: Harper and Row, 1984.
Kearney, A., and De Young, R. (1996). Changing commuter travel behavior: Employer-initiated
strategies. Journal of Environmental Systems, Vol. 24, No. 4, 373-393
Kenyon, S. and Lyons, G. The value of integrated multimodal traveller information and its
potential contribution to modal change. Transportation Research, Part F: Traffic Psychology
and Behavior, Vol. 6, No. 1, 2003, 1-21.
Klinger, T. (2017). Moving from monomodality to multimodality? Changes in mode choice of
new residents. Transportation Research Part A: Policy and Practice, Vol. 104, 221-237.
Klöckner, C. A., and Matthies, E. (2004). How habits interfere with norm-directed behaviour: A
normative decision-making model for travel mode choice. Journal of Environmental
Psychology, Vol. 24, No. 3, 319-327.
Kortum, K., Paleti, R., Bhat, C., and Pendyala, R. (2012). Joint model of residential relocation
choice and underlying causal factors. Transportation Research Record: Journal of the
Transportation Research Board, No. 2303, 28-37.
Krizek, K. J. (2003). Neighborhood services, trip purpose, and tour-based travel. Transportation,
Vol. 30, No. 4, 387-410.
Lam, W. H., and Chan, K. S. (2001). A model for assessing the effects of dynamic travel time
information via variable message signs. Transportation, Vol. 28, No. 1, 79-99.
Lau, J. C. Y., and Chiu, C. C. (2013). Dual-track urbanization and co-location travel behavior of
migrant workers in new towns in Guangzhou, China. Cities, Vol. 30, 89-97.
Lee, B. H. and Waddell, P. (2010). Residential mobility and location choice: A nested logit model
with sampling of alternatives. Transportation, Vol. 37, No. 4, 587-601.
Lee, B. H., Waddell, P., Wang, L., and Pendyala, R. M. (2009). Operationalizing Time-Space
Prism Accessibility in a Building-Level Residential Choice Model. In Proceedings of the 88th
Annual Meeting of the TRB.
29
Levinson, D. M. (1998). Accessibility and the journey to work. Journal of Transport Geography,
Vol. 6, No. 1, 11-21.
Li, Y., Guo, Y., Lu, J., and Peeta, S. (2019). Impacts of congestion pricing and reward strategies
on automobile travelers’ morning commute mode shift decisions. Transportation Research
Part A: Policy and Practice, Vol. 125, 72-88.
Luo, W. (2004). Using a GIS-based floating catchment method to assess areas with shortage of
physicians. Health & Place, Vol. 10, No. 1, 1-11.
Luo, W. and Qi, Y. (2009). An enhanced two-step floating catchment area (E2SFCA) method for
measuring spatial accessibility to primary care physicians. Health & Place, Vol. 15, No. 4,
1100-1107.
Macfarlane, G. S., Garrow, L. A., and Mokhtarian, P. L. (2015). The influences of past and present
residential locations on vehicle ownership decisions. Transportation Research Part A: Policy
and Practice, Vol. 74, 186-200.
Majid, M. R., Nordin, A. N., and Medugu, I. N. (2014). Influence of housing development designs
on household designs on household vehicle miles traveled: A case of Iskandar Malaysia.
Transportation Research Part D: Transport and Environment, Vol. 33, 63-73.
Mannering, F. (2018). Temporal instability and the analysis of highway accident data. Analytic
Methods in Accident Research, Vol. 17, 1-13.
Matthies, E., Kuhn, S., and Klöckner, C.A. (2002). Travel mode choice of women: The result of
limitation, ecological norm, or weak habit? Environment and Behavior, Vol. 34, No. 2, 163-
177.
Milakis, D., Efthymiou, D., and Antoniou, C. (2017). Built environment, travel attitudes and travel
behaviour: quasi-longitudinal analysis of links in the case of Greeks relocating from US to
Greece. Sustainability, Vol. 9, No. 10, 1774.
Molin, E., Oppewal, H., and Timmermans, H. (1999). Group-based versus individual-based
conjoint preference models of residential preferences: A comparative test. Environment and
Planning A, Vol. 31, No. 11, 1935-1947.
Müggenburg, H., Busch-Geertsema, A., and Lanzendorf, M. (2015). Mobility biographies: A
review of achievements and challenges of the mobility biographies approach and a framework
for further research. Journal of Transport Geography, Vol. 46, 151-163.
Muth, R. F. Cities and Housing; The Spatial Pattern of Urban Residential Land Use. University
of Chicago Press, Chicago, IL, 1969.
Palm, R. and Danis, M. A. (2001). Residential mobility: The impacts of Web-based information
on the search process and spatial housing choice patterns. Urban Geography, Vol. 22, No. 7,
641-655.
Peeta, S. and Mahmassani, H. S. (1995). Multiple user classes real-time traffic assignment for
online operations: A rolling horizon solution framework. Transportation Research Part C:
Emerging Technologies, Vol. 3, No. 2, 83-98.
Prashker, J., Shiftan, Y., and Hershkovitch-Sarusi, P. (2008). Residential choice location, gender
and the commute trip to work in Tel Aviv. Journal of Transport Geography, Vol. 16, No. 5,
332-341.
Prillwitz, J., Harms, S., and Lanzendorf, M. (2007). Interactions between residential relocations,
life course events, and daily commute distances. Transportation Research Record, Vol. 2021,
No. 1, 64-69.
30
Paz, A. and Peeta, S. (2009). Information-based network control strategies consistent with
estimated driver behavior. Transportation Research, Part B: Methodological, Vol. 43, No. 1,
73-96.
Pinjari, A. R., Pendyala, R. M., Bhat, C. R., and Waddell, P. A. (2011). Modeling the choice
continuum: an integrated model of residential location, auto ownership, bicycle ownership, and
commute tour mode choice decisions. Transportation, Vol. 38, No. 6, 933-958.
Post-relocation survey for control group. 2014. Accessible via:
https://purdue.qualtrics.com/SE/?SID=SV_dbqFxfB2gwgQM0B.
Post-relocation survey for treatment group. 2014. Accessible via:
https://purdue.qualtrics.com/SE/?SID=SV_6xIknIURCaVyxa5.
Pre-relocation survey for control group. 2014. Accessible via:
https://purdue.qualtrics.com/SE/?SID=SV_8qAcmdU6L050LmB.
Pre-relocation survey for treatment group. 2014. Accessible via:
https://purdue.qualtrics.com/SE/?SID=SV_e9esAuD1Kis8wXH.
Rashidi, T. H., Mohammadian, A., and Koppelman, F. S. (2011). Modeling interdependencies
between vehicle transaction, residential relocation and job change. Transportation, Vol. 38,
No. 6, 909.
Reference USA. U.S. Business, 2014. Accessed November, 2014.
http://www.referenceusa.com/Home/Home.
Richter, J., Friman, M., and Gärling, T. (2011). Soft transport policy measures: Gaps in knowledge.
International Journal of Sustainable Transportation, Vol. 5, No. 4, 199-215.
Rodriguez, D. A. and Rogers, J. (2014). Can housing and accessibility information influence
residential location choice and travel behavior? An experimental study. Environment and
Planning B: Planning and Design, Vol. 41, No. 3, 534-550.
Salomon, I., and Mokhtarian, P. L. (1997). Coping with congestion: Understanding the gap
between policy assumptions and behavior. Transportation Research Part D: Transport and
Environment, Vol. 2, No. 2, 107-123.
Schwanen, T. and Mokhtarian, P. L. (2004). The extent and determinants of dissonance between
actual and preferred residential neighborhood type. Environment and Planning B: Planning
and Design, Vol. 31, No. 5, 759-784.
Searcy, S. E., Findley, D. J., Huegy, J. B., Ingram, M., Mei, B., Bhadury, J., and Wang, C. (2018).
Effect of residential proximity on university student trip frequency by mode. Travel Behaviour
and Society, Vol. 12, 115-121.
Shankar, V. and Mannering, F. L. (1998). Modeling the endogeneity of lane-mean speeds and lane-
speed deviations: A structural equations approach. Transportation Research Part A: Policy
and Practice, Vol. 32, No. 5, 311-322.
Simonsohn, U. (2006). New Yorkers commute more everywhere: Contrast effects in the
field. Review of Economics and Statistics, Vol. 88, No.1, 1-9.
Stark, J., Berger, W. J., and Hössinger, R. (2018). The effectiveness of an intervention to promote
active travel modes in early adolescence. Transportation Research Part F: Traffic Psychology
and Behaviour, Vol. 55, 389-402.
Steg, L. (2005). Car use: lust and must. Instrumental, symbolic and affective motives for car use.
Transportation Research Part A: Policy and Practice, 39(2-3), 147-162.
Sultana, S., and Weber, J. (2007). Journey-to-work patterns in the age of sprawl: Evidence from
two midsize southern metropolitan areas. The Professional Geographer, Vol. 59, No. 2, 193-
208.
31
Sunio, V., and Schmöcker, J. D. (2017). Can we promote sustainable travel behavior through
mobile apps? Evaluation and review of evidence. International Journal of Sustainable
Transportation, Vol. 11, No. 8, 553-566.
Sriraj, P. S., Minor, M., and Thakuriah, P. (2006). Impact of Information on Housing Relocation
Using Analytical Hierarchy Process and Interactive GIS. In Applications of Advanced
Technology in Transportation (pp. 816-821).
Thøgersen, J. (2012). The importance of timing for breaking commuters' car driving habits'. In
Helsinki: Helsinki Collegium for Advances Studies, 35-37.
Tielemans, E., Kupper, L. L., Kromhout, H., Heederik, D., and Houba, R. (1998). Individual-based
and group-based occupational exposure assessment: Some equations to evaluate different
strategies. Annals of Occupational Hygiene, Vol. 42, No. 2, 115-119.
The U.S. Census Bureau. 2010 Population Finder. Accessed September 2015 via:
http://www.census.gov/popfinder/index.php.
The U.S. Census Bureau. 2010-2014 American Community Survey 5-year Estimates. Accessed
September 2015 via: http://factfinder.census.gov/faces/nav/jsf/pages/community_facts.xhtml.
Verplanken, B., Walker, I., Davis, A., and Jurasek, M. (2008). Context change and travel mode
choice: Combining the habit discontinuity and self-activation hypotheses. Journal of
Environmental Psychology, Vol. 28, No. 2, 121-127.
Walker, I., Thomas, G. O., and Verplanken, B. (2015). Old habits die hard: Travel habit formation
and decay during an office relocation. Environment and Behavior, Vol. 47, No. 10, 1089-1106.
Washington, S. P., Karlaftis, M. G., and Mannering, F. L. (2010). Statistical and Econometric
Methods for Transportation Data Analysis. New York, New York: Chapman & Hall/CRC.
Yamamoto, T., Takamura, S., and Morikawa, T. (2018). Structured random walk parameter for
heterogeneity in trip distance on modeling pedestrian route choice behavior at downtown area.
Travel Behaviour and Society, Vol. 11, 93-100.
Zarabi, Z., Manaugh, K., and Lord, S. (2019). The impacts of residential relocation on commute
habits: A qualitative perspective on households’ mobility behaviors and strategies. Travel
Behaviour and Society, Vol. 16, 131-142.
Zillow. (2019). Tippecanoe County Home Prices and Values. Accessed September 2019 via:
https://www.zillow.com/tippecanoe-county-in/home-values/
Zhou, J. (2012). Sustainable commute in a car-dominant city: Factors affecting alternative mode
choices among university students. Transportation Research Part A: Policy and Practice,
Vol. 46, No. 7, 1013-1029.
Zondag, B. and Pieters, M. (2005). Influence of accessibility on residential location
choice. Transportation Research Record: Journal of the Transportation Research Board,
No. 1902, 63-70.
