Content uploaded by Konstadinos Goulias
Author content
All content in this area was uploaded by Konstadinos Goulias on Dec 06, 2016
Content may be subject to copyright.
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
1
Chapter 5
Travel Behavior Models
Konstadinos G. Goulias
goulias@geog.ucsb.edu
Department of Geography
University of California Santa Barbara
November 28, 2016
GEOTRANS 2016-11-01
Santa Barbara, CA USA
!
2!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
2
1. Definitions and Framework
Planning for transportation requires analysis, modeling, and simulation of the use and
performance of transportation systems. The ultimate objective is to design efficient
infrastructure and services to meet our needs for accessibility and mobility, but also to
implement policies improving sustainability. At the center of this design are analytical
methods requiring understanding of human behavior. Understanding human behavior
requires us to collect data, and analyze and develop synthetic models of human agency
in its most important dimensions and most elementary constituent parts. This includes
but is not limited to understanding individual development along life-cycle paths and
the complex interaction between an individual and the anthropogenic, natural, and
social environments. Travel behavior analysis aims to understand how traveler values,
norms, attitudes, and constraints lead to observed behavior. Traveler values and
attitudes refer to motivational, cognitive, situational, and dispositional factors that
determine human behavior. From early applications, studying travel behavior focused
primarily on the analysis and modeling of travel demand, with roots in theories and
analytical methods from a variety of scientific fields. These theories and methods
include, but are not limited to, the use of time and its allocation to activities and to
travel, methods for studying travel and time allocation in a variety of contexts and
stages in a person’s life, and the arrangement of artifacts and the use of space at
multiple levels of social organization such as the individual, the household, the
community, and other formal or informal groups. Travel behavior also includes the
movement of goods and the provision of services that have strong interfaces and
relationships with engagement in activities and the movement of persons. Travel
behavior analysis is heavily influenced by policy needs, pragmatic considerations in
data collection and analysis, and a multitude of both theoretical and computational
approaches.
The genesis of travel behavior analysis can be traced back to the 1950s, which saw
rapid motorization of major US cities, creation of a legislative framework to guide
transportation planning in cities with more than 50,000 residents, and the development
of a “rational” planning process as required for the receipt of federal funds (Weiner,
2012). Similar development also occurred in Europe with the conduct of traffic studies
resulting from technology transfer from the US (Jones, 2012). These very early studies
aimed at describing the impacts of motorization (e.g., increase in household car
ownership and use) and changes of land development (e.g., creation of large-scale
shopping centers) on trip making and, consequently, traffic flow on roadways and
ridership on buses. These impacts were quantified by estimating the demand for travel
and its “assignment” to a network of roadways and public transportation. As a result a
formal procedure and simulation of traffic in cities using mathematical models was
created, using techniques from considerations of population travel behavior and spatial
interaction. Recognizing the value of analyzing individual behavioral information to
support these computational techniques, data collection targeted individuals in cities.
Although the techniques used for urban simulation employed computational models
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
3
based on groups of individuals living and working in large geographical areas called
traffic analysis zones (TAZ), survey data were collected at fine detail and contained
information about behavior that is still the core information of contemporary travel
and activity surveys. Data collection in this planning effort was motivated by the need
to create a forecasting model to plan for locations where a new roadway (such as the
Interstate freeways of the 1950s and 1960s) should be constructed, and to learn how
this influenced the functioning of cities. Planning was also conceived as
comprehensive, and included land use and transportation as the main domains.
Moreover, data collection was viewed as the basic requirement to inventory all of the
city’s functions. The mode of the main survey was in-person with home interviews of
a large sample. In Chicago, 1 out of every 30 households in the study area was
recruited, yielding approximately 50,000 interviews and single-day travel diaries also
called the travel log. The survey unit was the household because this was considered
the fundamental decision-making unit for travel. Due to emphasizing the analysis of
traffic in the most congested conditions, travel during weekdays using a single
weekday travel log was used. The implied assumption is that weekdays have sufficient
similarity to each other to build a model of behavior on a typical weekday; weekend
traffic is not likely to exceed transportation system capacity and is, therefore, not
required in data collection. To capture travel demand variability, these early data-
collection designs also included questions about a few demographic variables such as
age and gender composition in the household, car ownership, employment and
industry type, driver’s license holding, and income. These were the fundamental
variables to explain travel behavior variability in the population of a study area. The
travel log contents were also relevant to an underlying behavioral model. The log
requested information about the locations of origins and destinations (address) for
each trip (one way movement from an origin to a destination), combining locations
and trip purpose (home, work, shop, school, social-recreational, personal business,
medical and dental, work related business, sightseeing, serve passenger, change means
of travel). The data recorded included social and demographic information about each
household and its members (age, gender, employment, income, driver’s license
holding, residence location), spatial determinants of travel (accessibility, land use
intensity, and roadway configuration), car ownership, and travel behavior descriptors
(number of trips, distance traveled, travel time, trip purpose, trip origins, trip
destinations, timing of trips, mode for each trip, and number of person with whom
each trip was made). A simulation model that was called the four step model emerged
as the preferred modeling approach, with its steps being trip generation, trip
distribution, modal split, and traffic assignment. The objective of this model was to
start with an area (a region today), its population, and the transportation networks, and
through a series of procedures show simulated traffic on the network and allow
comparison among different alternative designs of the network. The study area (e.g.,
the Chicago metropolitan area) was divided into a few hundred geographical zones
considered homogeneous in terms of their residents and land uses (e.g., downtown
central business districts vs residential suburbs). These were referred to as Traffic
Analysis Zones (TAZ). A sketch of a network was derived from “higher level”
roadways (e.g., freeways and major thoroughfares), and a graph representing this
network was created with nodes and links. Then, a shortest path algorithm was used to
4!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
4
identify the shortest paths from each origin to each destination within the area. This
was then used to compute all travel distances. The amount of travel “produced” by
each zone was modeled as a function of the resident characteristics and the land uses
in each zone. The relationships among resident characteristics, land uses, and daily
person trips were derived from a survey. Then, using zonal averages from the US
Census, travel predictions for each zone were obtained. The amount of travel
generated in each zone needed to be distributed among all the other zones of the study
area; this was done using one of two main competing behavioral theories created by
planning practitioners. The first was the gravity model, based on the following. Trips
between groups of people (the zones) occur proportional to the product of the
population sizes of the two groups and inversely proportional to a power function of
the distance separating them. This allowed computation of travel between two
geographical areas to be computed using the population of the two zones and the
distance. The power function, representing sensitivity of travelers to travel distance,
was computed using trip length observations. The second main behavioral theory was
the intervening opportunities model, which envisions a traveler searching from an
origin for an activity opportunity; as the traveler performs the search, the probability
of selecting a destination decreases, because the traveler has increasingly likely
already found a satisfactory destination. In this way, opportunities that are closer to an
origin have higher probability of selection, and destinations within zones of higher
density of opportunities also have higher probability of being selected. After one of
these trip distribution models was used and person trip flows among pairs of zones
were produced, the trips were divided into private car versus transit (public
transportation) trips using functions that show the percent of trips split as a function of
travel time and travel costs between specific origins and destinations. The final step
allocated vehicles onto the schematic network, based on travel time between origins
and destinations, assuming drivers would select the shortest travel time. The travel
time on a link of a network was also modified in increments to account for decreases
in average speeds when the amount of traffic increased on a link. Using this basic
model, a baseline was first created for the study area and verified using external data,
such as traffic counts on roadways; then scenarios of changes to the transportation
network were implemented in the model, and impacts on simulated traffic examined.
This process is still followed today, with the key difference that data and models used
in the simulation are more detailed, and many models are based on a deeper
understanding of human behavior.
There are many implied behavioral assumptions in the schematic model outlined
above, the most important being that of collective optimization and homogeneity. The
selection of a route from an origin to a destination via the shortest path implies that on
average drivers know all the paths, and they select the path of minimum travel time.
The adjustment of assigned vehicles to a link based on congestion also implies that
drivers know when the link will be congested, which leads them to select a different
path. The homogeneity assumption implies that the amount of travel on average is a
function of a few demographic characteristics of a person and a small number of land-
use types for each origin and destination. In terms of travel increase and traffic growth,
employment and income were considered as the important modifiers of behavior and
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
5
used to predict future car ownership and travel. Challenging these assumptions has
been the aim of many subsequent behavioral studies up to the present, significantly
improving our understanding. Increasing computational power has also enabled
models at a finer spatial resolution and a move towards individual dwelling units as
the spatial subdivision of models. But, most important, individual and group
behavioral models were developed, with a move away from deterministic models and
the development of probabilistic models which allowed the inclusion of uncertainty in
simulations. Data collection for travel behavior also evolved, taking advantage of new
technologies. However, like the earliest studies, the household is still considered to be
the fundamental decision-making unit, a household questionnaire and a person diary
are still the core elements of a travel-behavior survey, and a large sample is still
desirable.
Today, travel behavior modeling and simulation is asked to assess trends and policies
that include but are not limited to: (a) changing land use in order to increase density
and diversity of development, which in turn would curb greenhouse gas emissions; (b)
the impacts of changing demographics due to population aging, in- and out-migration,
changing fertility and mortality, changes in educational attainment, family formation
and dissolution, and in employment achievement and prospects; (c) impacts of the
market introduction and penetration of new types of vehicles, including flexible fuel,
electric, and hybrid vehicles, as well as the resulting changes in the composition and
use of household vehicle fleets; (d) the new development and addition of roadway and
transit infrastructure components; (e) the introduction of new technology including
autonomous vehicles and other robotic innovations; (f) changes to the pricing of
services (including parking and toll roads) and access restrictions to congested parts of
the city; and (g) policy interventions that lead to spatial and temporal changes in
patterns of housing, school, and employment.
These new policy questions place more complex issues in the domain of regional
policy analysis and forecasting, and amplify the need for methods that produce
forecasts at the level of the individual traveler and her/his household, instead of the
level of the traffic analysis zone. In addition to the long range planning activities and
the typical traffic operations and management activities, analysts need to develop
plans to manage traveler and transportation system information provision and use (e.g.,
location based services, smart environments providing real time information to
travelers, vehicles, and operators), compare combinations of transportation
management actions and their impacts (e.g., parking fee structures and city center
restrictions, congestion pricing, locations of bicycle stations), and create scenarios for
combinations of environmental policy actions (e.g., carbon taxes and information
campaigns about health effects of pollutants).
A substantive implication of all these considerations is an expanded scope of modeling
and simulation. New processes that in the past were considered to be exogenous to
transportation modeling and simulation need to be incorporated into model systems.
For example, residential location choice, work location choice, and school location
choice must be included to capture the spatial distribution and relative location of
6!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
6
important anchoring points for travel behavior. Similarly, attention is needed to car
ownership and car-type choices for households, or fleet sizes and composition for
firms. These choices influence parking availability, energy, and other costs, and the
level of service offered by the transportation system (highway and transit). To account
for other resources and facilities available for household travel, we also need to
consider processes for driver’s licensing, acquiring of public transportation
subscriptions (passes), and participation in car-sharing programs. In this way,
variables concerning car and public transportation availability for households can be
used as determinants of travel behavior. Many variables that were once considered
determinants of travel behavior have now become variables to be explained by other
variables, such as attitudes. In many studies, similar treatment is also required to
understand changes in attitudes, perceptions, and knowledge about travel options as
new services are created.
As a result, a new approach is emerging in which behavioral models are applied to
individual decision makers, which are then used to (micro-)simulate behavior during a
given day. The result is, in essence, a synthetic generation of detailed travel for the
entire resident population of a study. This microsimulation also includes activity types
and durations at particular locations, producing a synthetic detailed activity schedule.
For inventory and forecasting purposes, a synthetic population is created for each land
subdivision, with all the relevant characteristics of persons and a schematic
representation of their living environment (e.g., a network of connectivity among
different means of travel, and summary indicators of access to facilities and
opportunities). These models are then applied to the residents of each subdivision to
recreate detailed area-wide behavior. Changes can then be imposed on each synthetic
individual in response to policies; predictive scenarios of policy impacts are thus
developed. The evolution of individuals, their groups, and the entire study area can be
used for trend and scenario analysis that includes details at the level of individual
decision makers. In addition, progression in time can happen in the models, from the
present to the future, allowing one to identify paths of change for individuals and
groups, while keeping detailed accounting of individuals as they move in time.
Forecasting is done using calendar time, most often yearly, and behavioral dynamics
are represented as a causal stream into the future.
2. Behavior Modeling
Here, I examine travel behavior analysis and synthesis along the three dimensions of
geographic space, time, and social space. In a few cases, the models reviewed
integrate time and space as conceived in physical science with perceptions of time and
space by humans in their everyday life. Research that includes theory formation, data
collection, modeling, inference, and simulation methods aims to create decision
support systems for policy assessment and evaluation combining different views of
time and space. Geographic space here is intended as the physical space in which
human action occurs. This dimension has played an important role, because
transportation system designers aim to overcome spatial separation (distance in the
broad sense). Early applications divided territory into large areas (traffic analysis
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
7
zones—TAZ), represented by a virtual center (centroid) and connected by facilities
(higher level roadways). The centroids were connected to major facilities using a
virtual connector, which summarized the characteristics of all the local roads within a
zone. Increased computational power offered the opportunity to consider policy
actions at an increased spatial resolution. Today we expect computers to handle zones
that are as small as a parcel of land connected to the network by a local road (the
centroid becomes the building on a parcel, and the centroid connector is the driveway
of the unit). Geographic space considers more than just physical features—it includes
place and social space (Golledge and Stimpson, 1997).
The second dimension is time, considered here as the continuity of time, the
irreversibility of the temporal path, and the time period considered in many models.
Models in long-range planning applications use typical days (e.g., a summer day for
air pollution). In many regional long-range models, the implied assumption is that we
model a typical work weekday. Households and their members, however, may not
always (if ever) obey this strict definition of a typical weekday when they schedule
their activities; they may follow very different decision-making horizons in allocating
time to activities within a day, spreading activities among many days, including
weekends, substituting out-of-home with in-home activities on some days but doing
exactly the opposite on others, and using telecommunications only selectively (e.g., on
Fridays and Mondays more than on other days).
Finally, the dimension of social space takes the place of jurisdictions for our analyses.
It is worth mentioning that the actions of each person are “regulated” by jurisdictions
with overlapping domains, such as federal agencies, state agencies, regional
authorities, municipal governments, neighborhood associations, trade associations and
societies, religious groups, and formal and informal networks of families and friends.
Everybody belongs to one or more social networks, and for this reason, a dimension
named social space and the relationships among persons within this space are key to
understanding travel behavior. For example, individuals from the same household
living in a neighborhood may change their daily time-allocation patterns, visiting
locations to accommodate changes in their neighborhood or the actions of another
household member engaging in another social network. Changes to the infrastructure
and its management may motivate mutually cancelling impacts because of the
different networks. This may lead to the unintended consequences of policy failures,
which models should be able to account for.
An especially important domain and entity within this social space is the household.
This has been the fundamental unit of analysis in transportation planning since the
1950s, as reviewed above, in recognition that strong relationships within a household
can be used to capture behavioral variation (e.g., predicting car-sharing probability
using household size and number of children as explanatory variables). In this way,
any changes in a household’s characteristics (e.g., changes in the composition due to
birth, death, divorce, children leaving the nest, or adults moving into the household)
can be used to predict changes in travel behavior. More recent models study this
interaction within a household by looking at patterns of time use in a day, and changes
8!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
8
across days and years. It is therefore very important in modeling and simulation used
for policy analysis to incorporate these interactions.
Across the three dimensions of geographic space, time, and social space, scale
emerges as very important. Typical long-range planning analysis is defined for larger
geographical areas (region, states, and countries) and addresses issues within time
horizons from 10 to 50 years. In many instances, we find that large geographic scale
also means longer time frames applied to wider mosaics of social entities and
including more diverse jurisdictions. On the other side of the spectrum, issues that are
relevant to smaller geographic scales are most likely to be accompanied by shorter-
term time frames applied to a few social entities that are relatively homogeneous and
subject to the rule of local jurisdictions. This is one important organizing principle but
also an indicator of the complex relationships we attempt to recreate in our computer
models for decision support. In developing the blueprints of these models, a variety of
theories (e.g., neoclassical microeconomics) and conceptual representations of the real
world can help us. At the heart of our understanding of how the world works (as an
organization, a household, a formal or informal group, or an individual human being)
are models of decision making and conceptual representations of relationships among
entities making up the world. In this chapter I focus on one group of actors that are
travelers moving around the transportation network and visiting locations where they
can participate in activities. Behavioral models created to represent them are
simplified versions of strategies used by travelers when they select among options that
are directly related to their desired activities. In these models, we make assumptions
about hierarchies of motivations, plans, actions, and consequences. Some of these
assumptions are explicit within models (e.g., when deriving the functional forms of
choice models or the rules of an algorithm) and some are implicit (e.g., by the
sequence of model application in software).
The two methodological developments that form the basis for today’s travel-behavior
analysis are choice models and the activity-based approach. Choice models enable
analysis at the individual level, encompass a variety of behavioral process types, and
provide the mathematical/statistical apparatus used to analyze many facets of behavior.
On the other hand, the activity-based approach expands the conceptual framework of
travel behavior modeling with a more realistic representation of the context within
which travel behavior actually takes place. This approach has moved travel-behavior
analysts away from analyzing trips to analyzing everyday life in its entirety; this
approach today provides guiding principles for model development. A separate section
below on behavioral dynamics points out that considering dynamics expands the
analytical envelope even further than the activity-based approach to consider life-long
processes in travel behavior.
2.1 Choice and Related Models
A decision making paradigm that has been used extensively in travel behavior
research and is still the frame of reference for many research directions is the random
utility model and its basic formulation, reviewed here briefly. Rational decision
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
9
making is a label associated with human behavior that follows a strategy in identifying
the best course of action. In summary, a decision maker solves an optimization
problem and identifies the best existing solution to this problem. Within this general
strategy, when an operational model is needed and this operational model provides
quantitative predictions about human behavior, a mathematical apparatus is needed to
produce the predictions. One such model is the subjective expected utility (SEU)
formulation of human behavior. Describing SEU, Simon (1983) defined four
theoretical components: (a) existence and consideration of a cardinal utility
function—a person’s decision is based on a utility function which assigns a numerical
value to each option; (b) ability to enumerate all strategies and their consequences—
the person defines an exhaustive set of alternative strategies among which just one will
be selected; (c) infinite computational ability—the person builds a probability
distribution of all possible events and outcome for each alternate option; and (d)
maximizing utility behavior—the person selects the alternative that has the maximum
utility.
Application of these behavioral models (dubbed Discrete Choice Models, Train 2009)
commenced with the random utility models (RUM) for the San Francisco Bay Area
Rapid Transit (BART) based on the foundation of work by McFadden (1974). RUMs
offered the possibility to predict mode choices more accurately than ever before.
Although today the original RUMs are considered to be very restrictive in their
assumptions, they are still continuously being improved and have become the standard
tool in evaluating discrete choices of many kinds. They also serve as the theoretical
framework for consumer-choice models and attempts to develop models for other
hypothetical situations (Louviere et al., 2000). In the past 15 years a tremendous
advancement has been accomplished by making discrete-choice models more
behaviorally realistic, creating flexible functional forms, and alleviating the
detrimental effects of basic model assumptions (Bhat, 2003). Moreover, methods to
combine observed choices with answers to hypothetical scenarios are also now
feasible (Ben-Akiva et al., 1994). The development of hybrid-choice models that
explicitly incorporate psychometric variables in order to enhance the behavioral
representation of the underlying choice process allows the inclusion of attitudes and
traveler perceptions in a system of structural equations that include latent variables
and a measurement component that regresses latent variables on observable indicators
(Ben-Akiva et al., 2002). Related to this development is the use of latent class models
that essentially allow partitioning a sample into data-driven groups derived from
heterogeneous behaviors (Greene and Hensher, 2003), enhancing our ability to detect
and understand heterogeneity in choice processes.
People in their everyday lives may never behave exactly according to an SEU or other
maximizing and infinite computational capability models. Based on this argument,
researchers have proposed a variety of decision-making strategies or heuristics that
would more realistically model human behavior. This includes: Simon’s bounded
rationality—the limited extent to which rational calculation can direct human behavior
(Simon, 1983) and the prospect theory (Kahneman and Tversky 1979; Tversky and
Kahneman, 1992) - in which the decision maker uses a simplification step to edit the
10!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
10
alternatives. Then a value is assigned to each outcome and a decision is made based on
the sum of values, multiplying each by a decision weight. Losses and gains are treated
differently. In travel behavior one of the earlier examples that used the Simon
invention of satisficing behavior—acceptance of viable choices that may not be
optimal—is a series of applications that appeared in Mahmassani and Herman (1990).
Cumulative prospect theory was also used as a more behaviorally realistic model for
choosing routes by Avineri and Prashker (2003) and in the context of unreliable
situations (Schwanen and Ettema, 2009).
Other choice methods (using choice loosely), however, also exist, and they may
provide additional information about decision-making processes. One such example
that is more comprehensive is the situational approach (Brög and Erl, 1989). This
method uses in-depth information from surveys to derive sets of reasons for the
alternatives that are not considered for specific choices (i.e., specific individual trips).
It allows separating analyst-observed system availability from user-perceived system
availability (e.g., caused by misinformation or unwillingness to consider information).
Modeling the process of perceived constraints may be far more complex when one
considers the influence of the context within which decisions are made. Golledge and
Stimpson (1997, pp. 33–34) describe this within a conceptual model of decision
making that has a cognitive foundation to it. In the context of behavioral modification
and, particularly, active transportation, many other theories have been considered and
modified to fit the transportation context(s) by integrating habits, attitudes, intentions,
and hierarchy of needs (see the review by van Acker et al., 2010).
2.2 Activity-based Models
Chapin’s (1974) research, providing one of the first comprehensive studies about time
allocated to activity in space and time, is credited for motivating the foundations of
activity-based approaches to travel-demand analysis. His focus was on linking patterns
of individual’s participation in activities and travel to urban planning. At about the
same time, Becker (1976) also developed his theory of time allocation from a
household production viewpoint, applying economic theory to a non-market sector and
demonstrating the possibility of formulating time-allocation models using economic
reasoning (i.e., activity choice). In parallel, another approach developed in geography,
with Hägerstrand’s (1970) seminal publication on time-space geography presenting
the foundations of the approach. The idea of constraints on the movement of persons
was taken a step further by the time-geography school in Lund, Sweden. In that
framework, the movement of persons among locations was viewed as movement in
space and time under external constraints. Movement in time was viewed as the one-
way (irreversible) movement along the path while space was viewed as a three-
dimensional domain. One important consideration of this “theory” is the idea of
constraints on human paths in time and space for a variety of planning horizons. These
are capability constraints (e.g., physical limitations such as speed), coupling
constraints (e.g., requirements to be with other persons at the same time and place),
and authority constraints (e.g., restrictions caused by institutional and regulatory
contexts such as the opening and closing hours of stores).
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
11
Cullen and Godson (1975) attempted to bridge the gap between the motivational
approach (Chapin) to activity participation and the constraints approach (Hägerstrand)
by creating a model depicting a routine and deliberate approach to activity analysis.
The Cullen and Godson study also defined many terms often used today in activity-
based approaches. For example, each activity (stay home, work, leisure, shopping) is
an episode characterized by start time, duration, and end time. Activities are also
classified into fixed and flexible, and they can be engaged alone or with others (joint).
Moreover, these researchers also analyzed sequencing of activities as well as pre-
planned, routine, and spur-of-the-moment activities. Subsequent contributions to the
activity-based approach follow these basic principles (for reviews, see Kitamura, 1988,
Bhat and Koppelman, 1999). Three basic ingredients of an activity-based approach are
particularly important to highlight. Travel demand is derived demand. Participation in
activities such as work, shop, and leisure motivate travel, but travel could also be an
activity as well (e.g., taking a drive). These activities are viewed as episodes
(characterized by starting time, duration, and ending time), and they are arranged in a
sequence forming a pattern of behavior that can be distinguished from other patterns (a
sequence of activities in a chain of episodes). In addition, these events are not
independent, and their interdependency is accounted for in the theoretical framework
(e.g., by building chains of activities and trips called tours). The household is again
considered the fundamental decision-making unit, as in the very first travel-behavior
models, but the interactions among household members are explicitly modeled to
capture task allocation and roles within the household, relationships at a given point in
time, sequences of relationships, and changes in these relationships as households
move along their life-cycle stages and the individual’s commitments and constraints
change; these are depicted in the activity-based model explicitly. In fact, constraints
and commitments can be explicit modeled in time-space prisms (Pendyala, 2003) or
reflections of these constraints in the form of model parameters and/or rules in an
algorithmic production-system format (Arentze and Timmermans, 2000). Human
interaction in most models is limited to within-household interaction and is
incorporated by relating the day pattern of one person to the day patterns of other
persons within a household. Their joint activities and trip making are explicitly
modeled (joint recreation), and activity-roles are allocated (Bhat et al., 2013). Inputs
to activity-based travel demand models are the typical regional model data of social,
economic, and demographic information about potential travelers and land-use
information, allowing the creation of schedules followed by people in their everyday
life. The outputs are detailed lists of activities pursued, times spent in each activity,
and travel information from activity to activity (including travel time, mode used, and
so forth). Activity participation and travel take place in many social fields (or
networks) that involve many persons outside the household who exert social influence
(Dugundji and Walker 2005; Carrasco and Miller, 2006; Farber and Páez, 2009).
An aspect of travel behavior that is closely related to activity participation is the
relationship between telecommunications and travel. The study of telecommunications
and travel behavior became popular sometime in the 1980s, motivated by the hope that
telecommuting would decrease congestion (Salomon and Salomon, 1984; Pendyala et
al., 1991). Considerable analysis attempted to understand the relationships between
12!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
12
telecommunication and travel behavior, expanding telecommuting to include the
effects of all information and communication technologies (ICT) on travel. Salomon
(1986) sketched a framework where he recognized possible effects of ICT on travel:
substitution, modification, enhancement, and neutrality. Substitution means that ICT
actually eliminates trips. Telecommuting, teleshopping, and teleconferencing are some
examples. Modification indicates when ICT alters the travel behavior of individuals,
changing the order of trips (sequencing), the travel mode, or the timing of trips (e.g.,
departure time). From an operations standpoint, this is particularly important when a
shift of commuting trips to off-peak hours occurs or a switch to public transportation
and/or car-pooling happens because of ICT use. The third category, enhancement,
reflects those trips that would not have been generated without ICT. For example,
when there is more information available for particular activities, one expects an
increase in the desire to travel and participate in these activities. Also, people are able
to save time by better planning of their schedules (thanks to ICT) and by
communicating while traveling. The saved time is often used to make other trips. The
last category, neutrality, reflects those instances of ICT that have no marked effect on
travel behavior (see the reviews by Mokhtarian, 1991 and Mokhtarian, 2009) with
important explorations in activity fragmentation, polychronic time use (multi-tasking),
and sociality (Lenz and Nobis, 2007 and Kenyon and Lyons, 2007).
2.3 Behavioral Dynamics
In transportation policy analysis we need to measure the relationships among
behavioral indicators that change over time, including the timing, sequencing, and
staging of these changes (Goodwin, 1992). Individuals may change their behavior in a
variety of ways as a consequence of events. Behavioral change can be gradual, following a
smooth trend over time, or sudden. These changes can happen based on past events or in
anticipation of a future event. Behavioral changes at the household level are combinations of
events experienced by individuals and their households. The combination of events (or triggers)
determines the choice context and should be taken into account in estimating models,
particularly when these models are about change in behavior. Goulias and Pendyala (2014)
describe many different types of processes and events that can alter the individual and the choice
context. Types include physiological alterations (e.g., hormonal changes that alter a person’s
physical and social selves), transitions (e.g., age-related movement into and out of social roles,
such as changing school grades or loss of a parent), and turning points (e.g., events that cause
reorientation of priorities and lasting alterations of a person’s developmental trajectory).
The exploration of behavioral dynamics has progressed in a few directions. First, data
are collected with panel surveys, which are repeated observations of the same persons
over time; they now give us new ideas about data collection but also about data
analysis (Kitamura, 1990, Golob et al., 1997), including interactive and laboratory
data-collection techniques (Doherty, 2003) that allow for more in-depth examination
of behavioral processes. The second direction is in the development of dynamic
formulations for travel behavior that challenge conventional assumptions about travel
behavior. These formulations recognize that urban environments transform in a
cyclical manner, offering opportunities for activity engagement that change by time of
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
13
day, day of week, month of year, and so on. This variability results in multiple layers
of dynamics that should be included in data collection and in activity-based travel-
demand simulators (Goulias 2003). At the heart of this cyclic activity engagement and
travel is the interaction between a person and the built environment, and the multiple
interactions among people. One way to take into account the influence of the built
environment on behavior is to develop models of opportunities and constraints that
shape individual behavior, and conceptualize action within the boundaries of
Hagerstrand’s time-space prism. In this context, new models are emerging in which
the process of scheduling activities is tracked and then scheduling and executing
activities is computationally recreated in the software (Auld and Mohammadian, 2009).
This type of model uses a prompted recall survey to identify parameters of people’s
activity planning. Finally, a third direction conceptualizes behavior as a process which
develops in stages and as the outcome of multiple processes operating at different
levels. To this end, studies experimenting with theories from social psychology that
emphasize developmental dynamics is a potentially rich area that is just beginning to
emerge (Lazendorf, 2003, 2010, Goulias and Pendyala, 2014).
2.4. Stochastic Simulation and Production Systems
In the context of travel-behavior models, stochastic microsimulation is an evolutionary
engine implemented in software, used to replicate relationships among social,
economic, and demographic factors concerning land use, time use, and travel. The
causal links among these groups of entities are extremely complex, nonlinear, and in
many instances, unknown or incompletely specified. This is the reason that no closed
form solution can be created for a travel-demand forecasting system. An evolutionary
engine aims to realistically represent individual and household life histories (e.g., birth,
death, marriage, divorce, birth of children, etc.); the evolution of spatio-temporal
activity opportunities; and uncertainties in data, models, and behavioral variation. Life
simulation with people as agents can also be done using production systems to
explicitly depict the way humans solve problems (Golledge and Stimson, 1997). These
are a series of condition-action statements in a sequence. Production systems are
search processes that attempt to replicate human thought and action. Models of this
kind are called computational process models (CPM). Through the use of IF-THEN
rules, they have made possible a variety of new models. In travel behavior applications
stochastic microsimulation and CPMs are used to explore relationships, and to predict
and forecast travel demand. In some examples, combinations of CPMs with stochastic
microsimulation are also used. All models share a similar approach: (a) derive
relationships among behavioral facets using samples and statistics; (b) develop an
algorithm that uses these relationships; and (c) run simulations and test their
conceptual and empirical validity. Reviews of travel behavior and related models can
be found in Golledge and Gärling (2003), Henson, Goulias, and Golledge (2009), and
Ravulaparthy and Goulias (2011).
There are many similarities and differences among various modeling ideas. A
hierarchy of decisions by households is assumed that identifies longer-term choices
that determine shorter-term choices. In this way, different blocks of variables can be
14!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
14
identified and their mutual correlation used to derive equations useful in forecasting.
Most models use space and time fixity around which scheduling of activities can be
simulated. These are often spatial anchor points (home, work, school locations) that
are inserted at the first choice level. They define the overall spatial structure of activity
scheduling in a day. Temporal anchors include arrival and departure of children from
school. More recent models have an expanded list of out-of-home activities/trip
purposes, such as work, school, shopping, meals, personal business, recreation, and
escort passengers. In addition, in-home activities are explicitly modeled or allowed to
enter the model structure as a “stay-at-home” choice, with some models allowing for
activity choice at home (work, maintenance, discretionary). In this way, (limited)
substitution between at-home and outside-home can be reflected in the model.
Activities of a household are divided into independent activities and joint activities
with other household members. In this way, the intra-household interactions can be
modeled explicitly, and the impact of policies on their propensity captured in
simulation. Activities and trips are viewed as chains of events (tours), and the choice
of a mode is at the tour level. Each stop in a tour contains activities with beginning
and ending times. Moreover, stop frequencies and activities at stops are modeled
according to a day pattern or tour level to distinguish between activities and trips that
can be rescheduled with little additional effort, versus activities and trips that cannot
be rescheduled (e.g., school trips, doctor appointments). There is also an increasing
trend to model different behavioral facets (e.g., modes and destinations) jointly. In this
way, the mutual influence, sequential, and/or simultaneous relationships can be
reflected in the model structure, and tested. Another aspect that is now integral to the
models systems is time. For example, departure time for trips and tour time-of-day
choice are modeled explicitly. Model time periods are anywhere between 30 minutes
and second-by-second; time windows are used to account for scheduling. This
modeling component allows the incorporation of time-of-day in the models. It also
allows the identification of windows of activity and travel opportunities, and enables
accounting for “feedback” between congestion and trip-related decisions.
Space is also treated differently than in the first applications of behavioral models. The
population of a region more accurately varies in space based on methods that produce
spatially distributed synthetic populations, using as external control the totals,
averages, and relative frequencies of population characteristics. In addition,
accessibility measures are used to capture spatial interaction among activity locations
and the level of service offered by the transportation system. These are also the
indicators used to account for feedback among lower-level decisions (e.g., activity
location choices, routes followed, congestion) and higher-level decisions (e.g.,
residence location choice). Finally, spatial resolution, which is heavily dependent on
data availability, has already reached the level of a parcel and/or building, at its most
disaggregate level. Outputs of models are then aggregated to whatever level is
required by the limitations or needs of other models.
Unlike the 1950s models, these new formulations include increasing details, as Tables
1 and 2 show. The two tables provide a sample of behavioral facets that are now used
in a large scale simulation model developed for the Southern California Association of
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
15
Governments and called Simulator of Activities, Greenhouse Gas Emissions,
Networks, and Travel – SimAGENT (Bhat et al., 2013; Goulias et al., 2011; Pendyala
et al., 2012).
<Insert Tables 1 & 2 Here>
3. Data Collection
In travel behavior research, we find at least five different data collection approaches.
The first and most popular is a natural evolution of the household-based questionnaire
and diary of activities and trips in a day. Data from this source have been used to
estimate many revealed preference discrete-choice models and the majority of
activity-based models. This method with its focus on travel became the national
standard for monitoring changes of behavior with a repeated cross-sectional approach
(see http://nhts.ornl.gov/index.shtml). Modern travel-behavior surveys are also being
designed to be more interactive with respondents, by requesting scheduling details via
the internet and applying geospatial technologies in order to build complex dynamic
interviews (Doherty, 2003, Auld and Mohammadian, 2009, 2012). A very important
development is the emergence of smart phones to record richer data (Cottrill et al.,
2013, Reinau et al., 2015). The second is the panel survey, designed to improve our
understanding of changes in behavior (see Golob et al. 1997). The third asks
hypothetical questions of respondents by varying attributes of choices (Louviere, et al.,
2000) to study options that we cannot implement in real life. The fourth is real-life
interactive experimentation. In an interactive experiment, the aim is to change
behavior by providing feedback to participants about their past behavior (Roth et al.,
2003). A fifth data collection approach is based on qualitative research methods and
can be used for any behavioral issue (Goulias, 2003). Online sources such as social
media (e.g. Facebook, Foursquare, Twitter) are also used to obtain data on travel
satisfaction, weekly activities, traditional origin-destination patterns, and place-
specific sentiments (Collins et al., 2013, Coffey and Pozdnoukhov, 2013, Hassan and
Ukkusuri, 2014).
Based on a review of model needs, an ideal data-collection design can be sketched
(Goulias et al., 2013). This design contains a main household survey (the Core) that
collects the base data elements needed for an activity-based model system but can also
serve other simplified travel-behavior models to test a variety of other hypotheses
often tested in travel-demand forecasting. This is a household, individual, and a record
of movements (in the form of a daily diary) with the aim of developing models in
Tables 1 and 2. Today, diaries are often collected by one of three administration
methods: internet, mail, or telephone. This design aims to collect in-depth data while
minimizing the overall cost and time to implement the survey, at the same time
providing the flexibility to accept relatively independent survey modules (satellites).
The minimum data elements required are household composition information, person
characteristics (age, gender, education, employment, driver’s license, marital status),
and vehicle data. A two-day base activity-travel diary is desirable for studying day-to-
day variation, but a single-day diary is current practice in the U.S. The diary includes
16!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
16
a complete record of each person’s daily schedule, including all activities engaged in
and all trips made (with information on their assembly into tours), locations visited,
persons with whom each activity and trip were made, and activities carried out at
home and at other places. Ideally, this diary will cover a pre-specified pair of days for
all persons in the household and will spread over a 12–month period (to mimic the
American Community Survey of the U.S. Census) with a uniform distribution of
interviews throughout the survey period. Development of the next generation activity-
based model(s) requires a greatly enhanced activity diary of complete households. The
usual design in this setting is a household questionnaire that includes social and
demographic information for each household member, housing characteristics, and
automobile ownership. The activity diary portion of the questionnaire includes but is
not limited to respondents’ activities and travel, and characteristics of the surrounding
environment, including parking locations and parking rates.
For a travel behavior survey to serve general purposes of analysis, one could stop here.
However, we often face questions about specific policies or need to create behavioral
models for specific issues. For this reason, and to avoid overburdening respondents,
smaller subsamples can be selected to participate in additional in-depth surveys. We
call this design a core-satellite design. For example, satellite surveys can be: (a) a one-
week diary to capture day-to-day variation and substitution of activities between
weekdays and weekends; (b) an in-depth survey to identify what determines decisions
about car ownership, car type (e.g., new/used, model, make, fuel type), and car
assignment to household members; (c) a destination choice and satisfaction survey to
quantify contribution to subjective well-being experienced from each activity and
travel episode; (d) a retrospective and prospective survey of residential, workplace,
and school-location choices; (e) a component on attitudes and willingness to pay for
tolls on highways; (f) a component to understand interregional and long-distance
travel whether related to business, leisure, or commuting; (g) a housing, energy, and
transportation expenditures and household budgeting survey; (h) a survey on
situational constraints, attitudes, and predispositions for or against travel modes, such
as walking, biking, and public transportation; (i) a tracking survey for persons and
vehicles (e.g., using GPS and smart phones), and on-board diagnostics (OBD) to
identify driving patterns and correlate/link them with emissions models; and (j) a
households component with a joint multi-day diary to identify weekly rhythms in
activity scheduling, and travel and household panel surveys that enable disentangling
temporal order and causality of behavior. The overall survey design also enables the
study of toll modes and willingness to pay, decisions on walk/bike modes,
participation in the Transportation Demand Management program, auto ownership
(including fuel efficiency and use), and the study of equity and environmental justice
issues. The California Household Travel Survey (CHTS,
http://www.dot.ca.gov/hq/tpp/offices/omsp/statewide_travel_analysis/chts.html) is the
latest example of a design based on the above with modifications to accommodate
constraints due to budgetary reasons.
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
17
References
Arentze, T. and H. Timmermans (2000). ALBATROSS—A Learning Based
Transportation Oriented Simulation System. European Institute of Retailing and
Services Studies (EIRASS), Technical University of Eindhoven, Eindhoven, NL.
Auld, J., & Mohammadian, A. (2009). Framework for the development of the agent-
based dynamic activity planning and travel scheduling (ADAPTS) model.
Transportation Letters, 1(3), 245-255.
Auld, J., & Mohammadian, A. K. (2012). Activity planning processes in the Agent-
based Dynamic Activity Planning and Travel Scheduling (ADAPTS) model.
Transportation Research Part A: Policy and Practice, 46(8), 1386-1403.
Avineri, E., & Prashker, J. (2003). Sensitivity to uncertainty: need for a paradigm shift.
Transportation Research Record: Journal of the Transportation Research Board,
(1854), 90-98.
Becker, G. S. (1976). The Economic Approach to Human Behavior. Chicago: The
University of Chicago Press.
Ben-Akiva, M., Bradley, M., Morikawa, T., Benjamin, J., Novak, T., Oppewal, H., &
Rao, V. (1994). Combining revealed and stated preferences data. Marketing Letters,
5(4), 335-349.
Ben-Akiva, M., McFadden, D., Train, K., Walker, J., Bhat, C., Bierlaire, M., ... &
Daly, A. (2002). Hybrid choice models: progress and challenges. Marketing Letters,
13(3), 163-175.
Bhat C. R. (2003). “Random Utility-Based Discrete Choice Models for Travel
Demand Analysis.” In Transportation Systems Planning: Methods and Applications
(K.G. Goulias, Ed.), Boca Raton, FL: CRC Press, pp. 10-1 to 10-30.
Bhat, C. R., Goulias, K. G., Pendyala, R. M., Paleti, R., Sidharthan, R., Schmitt, L., &
Hu, H. H. (2013). A household-level activity pattern generation model with an
application for Southern California. Transportation, 40(5), 1063-1086.
Bhat C. R. and F. Koppelman (1999). “A Retrospective and Prospective Survey of
Time-Use Research.” Transportation, 26(2):119-139.
Brög W. and E. Erl (1989). “Interactive Measurement Methods—Theoretical Bases
and Practical Applications.” Transportation Research Record, 765, 1-11.
Carrasco, J. A., & Miller, E. J. (2006). Exploring the propensity to perform social
activities: a social network approach. Transportation, 33(5), 463–480.
18!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
18
Chapin F. S. Jr. (1974). Human Activity Patterns In The City: Things People Do in
Time and Space. New York: Wiley.
Coffey, C., & Pozdnoukhov, A. (2013, November). Temporal decomposition and
semantic enrichment of mobility flows. In Proceedings of the 6th ACM SIGSPATIAL
International Workshop on Location-Based Social Networks (pp. 34-43). ACM.
Collins, C. H, Samiul, S. V. Ukkusuri (2013) A novel transit rider satisfaction metric:
rider sentiments measured from online social media data. J. Public Transport. 16 (2).
Cottrill, C., Pereira, F., Zhao, F., Dias, I., Lim, H., Ben-Akiva, M., & Zegras, P.
(2013). Future mobility survey: Experience in developing a smartphone-based travel
survey in Singapore. Transportation Research Record: Journal of the Transportation
Research Board, (2354), 59-67.
Cullen, I., & Godson, V. (1975). Urban networks: the structure of activity patterns.
Progress in planning, 4, 1-96.
Doherty S. (2003). Interactive Methods for Activity Scheduling Processes. In
Transportation Systems Planning: Methods and Applications (K.G. Goulias, Ed.),
Boca Raton, FL: CRC Press, pp. 7-1 to 7-25.
Dugundji, E., & Walker, J. (2005). Discrete choice with social and spatial network
interdependencies: an empirical example using mixed generalized extreme value
models with field and panel effects. Transportation Research Record: Journal of the
Transportation Research Board, (1921), 70-78.
Farber, S., & Páez, A. (2009). My car, my friends, and me: a preliminary analysis of
automobility and social activity participation. Journal of Transport Geography, 17(3),
216-225.
Golledge, R. G. and T. Gärling (2003). “Spatial Behavior in Transportation Modeling
and Planning.” In Transportation Systems Planning: Methods and Applications (K. G.
Goulias, Ed.), Boca Raton, FL: CRC Press, Chapter 3, 1-27.
Golledge, R. G. and R.J. Stimson (1997). Spatial behavior: A geographic perspective.
Guilford Press.
Golob, T. F., Kitamura, R., & Long, L. (Eds.). (1997). Panels for transportation
planning: methods and applications. Springer Science & Business Media.
Goodwin, P. B. (1992). A review of new demand elasticities with special reference to
short and long run effects of price changes. Journal of transport economics and policy,
155-169.
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
19
Goulias, K. G. (2003). Transportation planning and modeling. Handbook of
Transportation Engineering (Editor M. Kutz), 3-25.
Goulias K.G. and R.M. Pendyala (2014) Choice Context. Handbook of Choice
Modelling (eds Stephane Hess and Andrew Daly), Emerald.
Goulias, K. G., Pendyala, R. M., & Bhat, C. R. (2013). Keynote—total design data
needs for the new generation large-scale activity microsimulation models. Transport
Survey Methods: Best Practice for Decision Making, 21.
Goulias K.G., C. R. Bhat, R. M. Pendyala, Y. Chen, R. Paleti, K. C. Konduri, G.
Huang, and H. Hu (2011) Simulator of Activities, Greenhouse Emissions, Networks,
and Travel (SimAGENT) in Southern California: Design, Implementation,
Preliminary Findings, and Integration Plans. Paper in proceedings of the Institute of
Electrical and Electronics Engineers (IEEE) Forum on Integrated and Sustainable
Transportation System, Vienna, Austria, June 29 to July 1, 2011, pp. 164-169.
Greene, W. H., & Hensher, D. A. (2003). A latent class model for discrete choice
analysis: contrasts with mixed logit. Transportation Research Part B: Methodological,
37(8), 681-698.
Hägerstrand T. (1970). “What about People in Regional Science?” Papers of the
Regional Science Association, 10, pp 7-21.
Henson, K., Goulias, K., & Golledge, R. (2009). An assessment of activity-based
modeling and simulation for applications in operational studies, disaster preparedness,
and homeland security. Transportation Letters, 1(1), 19-39.
Jones, P. M. (2012). The role of an evolving paradigm in shaping international
transport research and policy agendas over the last 50 years. In Travel Behaviour
Research in an Evolving World: Selected Papers from the 12th International
Conference on Travel Behaviour Research (pp. 13-34).
Kahneman, D., and A. Tversky (1979). “Prospect Theory: An Analysis of Decisions
under Risk.” Econometrica, 47(2): 263-291.
Kenyon, S., & Lyons, G. (2007). Introducing multitasking to the study of travel and
ICT: Examining its extent and assessing its potential importance. Transportation
Research Part A: Policy and Practice, 41(2), 161-175.
Kitamura R. (1988). An Evaluation of Activity-Based Travel Analysis. Transportation,
15, 9-34.
Kitamura, R. (1990). Panel analysis in transportation planning: An overview.
Transportation Research, 24A( 6), 401-415.
20!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
20
Lenz, B., & Nobis, C. (2007). The changing allocation of activities in space and time
by the use of ICT—“Fragmentation” as a new concept and empirical results.
Transportation Research Part A: Policy and Practice, 41(2), 190-204.
Louviere, J. J., Hensher, D. A., & Swait, J. D. (2000). Stated choice methods: analysis
and applications. Cambridge University Press.
Mahmassani H.S. and R. Herman (1990). “Interactive Experiments for the Study of
Tripmaker Behaviour Dynamics in Congested Commuting Systems.” In Developments
in Dynamic and Activity-Based Approaches to Travel Analysis. A compendium of
papers from the 1989 Oxford Conference. Avebury, UK. 272-298.
McFadden, D. (1974). The measurement of urban travel demand. Journal of public
economics, 3(4), 303-328.
Mokhtarian, P. L. (1991). Telecommuting and travel: state of the practice, state of the
art. Transportation, 18(4), 319-342.
Mokhtarian, P. (2009). If telecommunication is such a good substitute for travel, why
does congestion continue to get worse?. Transportation Letters, 1(1), 1-17.
Pendyala R. (2003). “Time Use and Travel Behavior in Space and Time.” In
Transportation Systems Planning: Methods and Applications. (K.G. Goulias, Ed.).
Boca Raton, FL: CRC Press, pp. 2-1 to 2-37.
Pendyala, R., Bhat, C., Goulias, K., Paleti, R., Konduri, K., Sidharthan, R., ... &
Christian, K. (2012). Application of Socioeconomic Model System for Activity-Based
Modeling: Experience from Southern California. Transportation Research Record:
Journal of the Transportation Research Board, (2303), 71-80.
Pendyala, R. M., Goulias, K. G., & Kitamura, R. (1991). Impact of telecommuting on
spatial and temporal patterns of household travel. Transportation, 18(4), 383-409.
Ravulaparthy S. and K.G. Goulias (2011) Forecasting with Dynamic Microsimulation:
Design, Implementation, and Demonstration. Final Report on Review, Model
Guidelines, and a Pilot Test for a Santa Barbara County Application. University of
California Transportation Center (UCTC) Research Project. Geotrans Research Report
0511-01, May, Santa Barbara, CA.
Reinau, K. H., Harder, H., & Weber, M. (2015). The SMS-GPS-Trip method: A new
method for collecting trip information in travel behavior research.
Telecommunications Policy, 39(3), 363-373.
Roth, M., Ker, I., James, B., Brög, W., Ashton-Graham, C., Ryle, J., ... & Richardson,
E. (2003, October). A dialogue on individualised marketing: Addressing
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
21
misperceptions. In Proc. 26th Australasian Transport Research Forum, Wellington,
New Zealand.
Salomon, I. (1986). Telecommunications and travel relationships: a review.
Transportation Research Part A: General, 20(3), 223-238.
Salomon, I., & Salomon, M. (1984). Telecommuting: The employee’s perspective.
Technological Forecasting and Social Change, 25(1), 15-28.
Schwanen, T., & Ettema, D. (2009). Coping with unreliable transportation when
collecting children: examining parents’ behavior with cumulative prospect theory.
Transportation research part A: Policy and Practice, 43(5), 511-525.
Simon H. A. (1983). “Alternate Visions of Rationality.” In Reason in Human Affairs
(H.A. Simon, Ed.), Stanford, CA: Stanford University Press, 3-35.
Train, K. E. (2009). Discrete choice methods with simulation. Cambridge university
press.
Tversky, A. and D. Kahneman (1992). “Advances in Prospect Theory: Cumulative
Representation of Uncertainty.” Journal of Risk and Uncertainty, 9, 195-230.
van Acker, V., van Wee, B., & Witlox, F. (2010). When transport geography meets
social psychology: toward a conceptual model of travel behaviour. Transport Reviews,
30(2), 219-240.
Weiner, E. (2012). Urban transportation planning in the United States: history, policy,
and practice. Springer Science & Business Media.
22!
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
© Konstadinos G. Goulias
22
Table 1 Sample of Behavioral Facets Modeled in SimAGENT Part 1
Adults and Children
Households
Employed Persons
Child’s decision to go to
school
Allocation of drop off
episode to parent
Commute mode
Child’s school start time
Allocation of pick up
episode to parent
Number of before-work
tours
Child’s school end time
Determination households
with non-zero out of home
duration
Number of work-based
tours
Decision to go to work
Determination of total out-
of home time of a
household
Number of after-work
tours
Work start and end times
Independent and Joint
Activity participation for
households
Before-work tour mode
Adult’s decision to go to
school
Independent Activity
participation for
households
Work-based tour mode
Adult’s school start time
Decision of an adult to
undertake other serve-
passenger activities
After-work tour mode
Adult’s school end time
Number of stops in a tour
Child’s travel mode to
school
Home or work stay
duration before the tour
Child’s travel mode from
school
Activity type at a stop
Activity duration at stop
Travel time to a stop
Location of a stop
Draft of a Chapter for Handbook of Behavioral and Cognitive Geography (ed. Montello)
23
Table 2 Sample of Behavioral Facets Modeled in SimAGENT Part 2
Non-Worker
Household Joint
Discretionary Travel
Children Scheduling
Number of independent
tours
Decision of Joint or
Separate Travel
School to home commute
time
Decision to take an
independent tour before
pick-up or joint
discretionary tour
Joint Activity Start time
Home to school commute
time
Decision to take an
independent tour after
pick-up or joint
discretionary tour
Joint Activity travel time
to stop
Mode for independent
discretionary tour
Tour mode
Joint Activity location
Departure time from home
for independent
discretionary tour
Number of stops in a tour
Vehicle Used For Joint
Home-Based Tour
Activity duration at
independent discretionary
stop
Number of stops following
a pick-up/drop-off stop in
a tour
Location of independent
discretionary stop
Home stay duration
Activity type at stop
Stop location