concept of travel behavior

Travel Behavior and Urban Land Use

Chapter 2 travel behavior theories.

Two different perspectives, individual and collective, can explain travel behavior. When people contextualizing travel as a personal choice or decision-making, the traveler as a subject make mode choices, driving or not. When travel behavior is understood as a social phenomenon, researcher observe and understand the trips as a whole. The two perspectives derived two schools of theory, mode choice and human mobility. In the school of mode choice, travel distance could be treat as an independent variable, a part of travel cost, or could be decided in the next step after mode choice , such as route choice. In the school of human mobility, driving distance grab more attentions.

2.1 Mode Choices (Travel as a subject)

Are ‘decision’ and ‘choice’ the same when discussing travel modes? Literally, a ‘choice’ is one decision given all available options at the same time. While ‘decision’ is a broader concept. A decision could be a schedule with a combination of many choices, such as modes, destination, and activities. A decision related to travel behavior could even include bicycle or car purchase, and relocation. This section will start from the theories of mode choice, then extend to a broader discussion of decision processes.

2.1.1 Rational Choice Theory

For prescriptive, analytical everyday decision-making, rationality is a basic assumption in reasoned behavior or rational choice theories ( Edwards 1954 ; Von Neumann and Morgenstern 1944 ) .

This category is also called ‘Normative Decision Theory,’ which assume people a traveler is an ideal decision maker who are full rational. It requires three necessary steps including information collection, utility evaluation, and choice making.

• Expected Utility Theory (EUT)

Traditional economics focus on the utility evaluation and come up with the Expected Utility Theory (EUT) which is also called Consumer Choice Theory. The rule of EUT is Random Utility Maximization (RUM) ( Ben-Akiva and Lerman 1985 ; McFadden 1973 ) . This classical theory claims that customer always choose the one most appropriate by comparing the advantages and disadvantages of a range of alternatives, evaluating the benefits and costs of each possible outcome. Eventually travelers will select the optimal solution with the maximum ‘utility’ from the choice set.

In real life, Rational Choice Theory can not accurately describe the actual human behavior. Individuals do not often collect and analyse all the relevant information. They are not ‘ideal’ and are not able to calculate the utility for all possible alternatives with perfect accuracy. In many cases, the travel decision is not regarded as the ‘best’ one to achieve travelers’ desired objective. Many other theories were developmed to fix these issues.

2.1.2 Bounded Rational Behavior

Bounded rationality focused on the limitation of self-control ( March and Simon 2005 ) . In reality, individuals are behaving under many constraints including incomplete information, limited time, and cognitive capacity. The observed behaviors often are not optimal and are inconsistent with ‘pure’ rationality. Bounded rationality claims that, when people make decisions under constraints, heuristics and rules of thumb are more common than statistical inference. People are satisfied with a ‘good enough’ decision unless there is a definitively better alternative. The recently witnessed events would have stronger effects on an individual’s decision than others ( Camerer, Loewenstein, and Rabin 2004 ) .

2.1.3 Theory of Planned Behavior

In psychology, many theories and models are developed to explain people’s decision-making processes. 4

Ajzen and Fishbein ( 1977 ) proposed the Theory of Reasoned Action (TRA) to understand people’s behavioral intentions and actual behaviors. They found two deciding psychological elements as attitudes and subjective norms . Ajzen ( 1991 ) adds a new part of Perceived Behavioral Control (PBC) and renames TRA as Theory of Planned Behavior (TPB).

Attitudes are personal evaluation and it means how people prefer or are against performing an activity. For example, a commuter might choose transit in spite of the longer travel time because this person believes that transit is an environment-friendly transport mode.

Subjective norm is the social pressure from others. In the example above, choosing transit is because of other people’s normative expectations rather than personal desirability.

PBC represents some nonvolitional factors such as time, budget, and resources. PBC is assessed by the individual’s perception of ease or difficulty of the behavior. PBC is one reason of the different between intentions and actual behaviors, which is called attitude-behavior gap ( Kollmuss and Agyeman 2002 ; Lane and Potter 2007 ) . In this case, a commuter might choose transit because this person is confident in catching the bus every day.

Based on RUM models, McFadden ( 2001 ) proposes a similar framework called the choice process including attitudes, perception, and preference. This framework is further developed to hybrid choice model (HCM) and non-RUM decision protocols ( Ben-Akiva et al. 2002 ) .

Two meta-analyses found that intentions to drive, perceived behavioral control, habits and past behavior play the primary roles in travel mode choice. Among these factors, PBC have the strongest effects on private car use. People don’t want to reduce the car use because they think it is very inconvenient. The effect of attitudes is modest while subjective norms have weak effect on car use ( Lanzini and Khan 2017 ; Gardner and Abraham 2008 ) .

2.1.4 Prospect Theory

Kahneman and Tversky ( 1979 ) introduced the ProspectTheory 5 to study the impacts of biases. Prospect Theory is a descriptive theory with three main components: First, people are more sensitive to the sure things (e.g., the probability between 0.9 and 1.0, or between 0.0 and 0.1 ), while being indifferent to the middle range (e.g., from 0.45 to 0.55). Second, people care more about the change of overall proportion than the absolute values regardless of gains or losses. Third, people make choice based on a reference point, rather than the overall situation or worth. Economist also extend the theory of expected utility maximization to Behavioral Economics by address the influence of psychology on human behavior.

• Regret Theory

Regret Theory introduces the notions of risk or uncertainty in decisions ( Loomes and Sugden 1982 ) . Psychological studies found that individuals will not only try to maximize the utility but to minimize the anticipation of regret. The fear of regret could affect people’s rational behavior. For example, A high risk of congestion in peak hours could encourage a commuter to choose transit mode. Likewise, a good reputation for punctuality can give traveler confidence in the rail system.

In addition to the traditional utility framework, a regret term is added to address the uncertainty resolution. The utility function on the best alternative outcome will be smaller after subtracting the regret term, which is an increasing, continuous and non-negative function.

• Cognitive Bias

Another psychological factor, cognitive bias can result in judgement errors. For example, people treat potential gains and losses differently, that is called Loss Aversion. Loss Aversion suggests that the negative feeling about losses is greater than the positive response to gains ( Tversky and Kahneman 1992 ) . As a result, individual’s decisions may not be consistent with evidence and tend to pay additional costs to avoid losses.

2.2 Human Mobility (Travel as an object)

2.2.1 trip distribution laws.

There is a long history of human mobility studies. In Physics and Geography, travel distance and pattern are treated as an objective phenomenon. The related theories try to use some statistical expressions to fit the aggregated trip distributions.

Gravity Law is a dominant theory in this field. Scholars have developed some more delicate forms of Gravity Law and found some mathematical relationship to other famous distribution laws. Some theories from different perspectives, like intervening opportunities also show strong ability for explaining travel patterns and regularities.

2.2.1.1 Distance Based Theories

• Law of Migration

An early theory called Law of Migration by Ravenstein ( 1885 ) tried to explain the regional migration patterns. This found is based on observation rather than quantitative analysis. But it capture the fact that the direction of migration is toward the regional center with great commerce and industry. It also pointed out that distance is a primary factor for migrant. This theory inspired many studies on population movement consequently. Even today, socio-economic factors and distance-constraints are the essential parts in the relevant models and frameworks.

Zip’s law is also called discrete Pareto distribution . It is found in linguistics to explain the inverse relationship between the frequency and rank of a word. The charm is that this rank-frequency distribution disclosed a universal law in many realms of society and physics, such as urban size, corporation sizes, cells’ transcriptomes and so on. Zipf interpreted the two competing factors as force of diversification and unification . The former produces larger amount of cases and the later tries to upgrade the rank. An equilibrium of the rank-frequency balance is controlled through a parameter \(\alpha\) in the exponent. For example, a city’s population size \(m\) has a negative power relationship to its rank \(r\) as below. 6

\[ m \sim 1 / r^{\alpha} \]

Zipf ( 1946 ) extended this expression to describe the traffic in both directions between two cities:

\[ t_{ij}\propto \frac{m_i m_j} {d_{ij}} \]

where \(t_{ij}\) represent the traffic flow of goods between two centers \(i\) and \(j\) with population sizes \(m_i\) and \(m_j\) . \(d_{ij}\) is the distance from \(i\) to \(j\) . Because Zipf’s formula has a same form with Newtonian mechanics ( Newton 1848 ) , people call this expression as Gravity Law.

• Gravity Law

As the most influential theory, Gravity Law asserts that the amount of traffic flow between two centers is proportional to the product of their mass and inverse to their distance. The mass is often measured by population size.

\[ p_{ij}\propto m_i m_j f(d_{ij}), \qquad i\ne j \]

where \(p_{ij}\) is the probability of commuting between origin \(i\) and destination \(j\) , satisfying \(\sum_{i,j=1}^n p_{ij}=1\) . \(m_i\) and \(m_j\) are the population of two census units. The travel cost between the two places is represented as a distance decay function of \(d_{ij}\) .

Exponential and power are the two forms of the distance decay function with a parameter \(\lambda\) showed as below:

\[ f(d_{ij})=\exp(-\lambda d_{ij}) \] and

\[ f(d_{ij})={d_{ij}}^{-\lambda} \] The function implies that the movements between the origin and destination decays with their distance. In transportation modeling, a common form of gravity model is :

\[ T_{ij}= \alpha_i O_i \cdot \beta_j D_j \cdot f(d_{ij}) \]

where \(T_{ij}\) is the flow between \(i\) and \(j\) . the two population are replaced by total tirp generation of origin \(O_i\) and total trip attraction of destination \(D_i\) . \(\alpha_i\) and \(\beta_j\) are two constraining parameters to satisfy \(\sum_{i}^{n_i}T_{ij} = D_j\) and \(\sum_{j}^{n_j}T_{ij} = O_i\) . It means that \(\alpha_i = [\sum_{j}^{n_j} \beta_j D_j \cdot f(d_{ij})]^{-1}\) and \(\beta_j = [\sum_{i}^{n_i} \alpha_i O_i \cdot f(d_{ij})]^{-1}\) . Thus, this model is called as doubly constrained gravity model.

If it relieves the two constrains. this model will be simplified to single-constrained and unconstrained gravity model. By assuming \(\alpha\beta\) is an adjustment parameter irrelevant to locations \(i\) and \(j\) for controlling the total flows, this model will not guarantee that the attraction of a destination equals the sum of flow from all origins, and the generation of a origin equals the sum of flow to all destinations.

Broadly speaking, Zipf’s law and Gravity Law have a common essence of power law, or scaling pattern. The Zipfian distribution is one of a family of power-law probability distributions. The power-law distribution also holds in many realms: urban size, population density, street blocks, building heights, etc.

The state-of-the-art studies of human mobility agree that travel behavior follows a power-law distribution at the population level ( Barbosa et al. 2018 ) . An example is Brockmann, Hufnagel, and Geisel ( 2006 ) use dollar bills to track travel habits and confirm this theory. It reflects the fact that both trip and land use, as two geographic variables, follow some Paretian-like distribution. Apparently, it conflicts with Gaussian thinking, the foundation frame of linear models based on the location and scale parameters. 7

Meanwhile, the log-normal distribution may be asymptotically equivalent to a special case of Zipf’s law, which could support the logarithm transform in current VMT-density models ( Saichev, Malevergne, and Sornette 2010 ) .

2.2.1.2 Opportunity Based Theories

• Law of Intervening Opportunities

Law of Intervening Opportunities by Stouffer ( 1940 ) developed the migration theory in a different direction. Stouffer proposed that “the number of people going a given distance is directly proportional to the number of opportunities at that distance and inversely proportional to the number of intervening opportunities.”

Comparing with gravity law, the number of intervening opportunities \(s_{ij}\) replaces the distance between origin and destination. For example, a resident living in location \(i\) is attracted to location \(j\) with \(s_{ij}\) job opportunities in between.

\[ p_{ij}\propto m_i \frac{P(1|m_i,m_j,s_{ij})}{\sum_{k=1}^n P(1|m_i,m_j,s_{ij})}, \qquad i\ne j \]

where the conditional probability \(P(1|m_i,m_j,s_{ij})\) can be expressed by Schneider (1959) as:

\[ P(1|m_i,m_j,s_{ij})=\exp[-\gamma s_{ij}] - \exp[-\gamma (m_j + s_{ij})] \]

• Radiation Law

Simini et al. ( 2012 ) propose a radiation model express the probability of the destination \(j\) absorbing a person living in location \(i\) as below:

\[ P(1|m_i,m_j,s_{ij})= \frac{m_i m_j}{(m_i + s_{ij})(m_i + m_j + s_{ij})} \]

Or in transportation model it is expressed as:

\[ T_{ij}= O_i\cdot\frac{m_i m_j}{(m_i + s_{ij})(m_i + m_j + s_{ij})} \] To approximating the number of opportunities, \(s_{ij}\) is from the population within a circle centered at origin. The radius is the distance between \(i\) and \(j\) . Then \(m_i + m_j + s_{ij}\) represents the total population within the circle, and \(m_i + s_{ij}\) is the total population within the circle but excluding \(j\) , that is:

\[ T_{ij}= O_i\cdot\frac{m_i }{m_i + s_{ij}}\cdot\frac{m_j}{m_i + m_j + s_{ij}} \] The part of fraction converts to the product of two weights, the weights of origin and destination in the whole region. Although distance \(d_{ij}\) doesn’t appear in the expression of radiation model, it is still a determinant as in gravity model.

• Distance Decay (hazard models)

Using the survival analysis framework, Yang et al. ( 2014 ) further extended this model by assuming a trip from origin to destination as a time-to-event process. Here time variable is replaced by the number of opportunities.

The survival function \(S(t)=Pr(T>t)\) represents the cumulative probability of the event not happened within a certain amount of opportunities. Choosing Weibull distribution as the survival function, \(S(t)=\exp[-\lambda t^\alpha]\) with scale parameter \(\lambda \in (0, +\infty)\) . By assuming \(f(\lambda)=\exp[-\lambda]\) and integral on \(\lambda\) , the derivation is:

By replacing \(t\) with \(m_i+s_{ij}\) , the conditional probability is:

\[ \begin{aligned} P(1|m_i,m_j,s_{ij})= &\frac{P(T>m_i+s_{ij})-P(T> m_i+s_{ij}+m_j)}{P(T>m_i)} \\ =&\frac{[(m_i + s_{ij} + m_j)^{\alpha}-(m_i + s_{ij})^{\alpha}](m_i^{\alpha}+1)}{[(m_i + s_{ij} + m_j)^{\alpha}+1][(m_i + s_{ij})^{\alpha}+1]}\\ \end{aligned} \]

where \(\alpha\) is a parameter adjusting the effect of the number of job opportunities between origins and destinations.

A similar method can be found in Ding, Mishra, et al. ( 2017 ) ’s study. They use a multilevel hazard model to examine the effects of TAZ level and individual level factors with respect to commuting distance using the data of Washington metropolitan area.

Based on commuting data from six countries, Lenormand, Bassolas, and Ramasco ( 2016 ) found gravity law performs better than the intervening opportunities law. The reasons could be the circle with radius \(d_{ij}\) can not accurately represent the real influencing area, and the different between population and opportunities is not captured in this way.

2.2.2 Time Geography

In contrast to overall trip distribution, the movements of individuals are always research interest in geography. Hägerstraand ( 1970 ) proposed some concepts and tools in space and time to measure and understand the individual trajectories. This branch is called time geography. The famous “space-time aquarium/prism” is a 3D cube by adding temporal scales on the geographic space. It can capture the detailed structure and behavior of traveler.

A daily travel could include multiple trips and form a travel chain. The traveler may switch the sequence or adjust the routes to optimize the chain and minimize the travel costs. The daily total travel distance is the summation of every trip distances. The number of trips denotes as trip count. It exists but not so common that driving itself is the travel purpose, especially in daily life.

At individual level, time geography borrows some physical and mathematical concept and methods such as random walk, Brownian motion, and Levy flight

Along with the wide usage of Global Positioning System (GPS), high performance computer, and sophisticated algorithms, the high-resolution data being collected. The relevant studies also have a dramatic increase after 2005.

2.3 Discussion

The theories of travel behavior follow a positivism tradition for a long time. Economics and geography give some strong explanations for both macro and micro travel patterns. In order to remove the limitation of ideal rationality, more sociological and psychological theories and methods are introduced into this field. Gradually, people realized the normative concept is not sufficient for real world applications. More descriptive and narrative arguments appear in transportation and land use planning. An example is the shift from mobility to accessibility.

A primary trend in urban and transportation fields recently is the transition from techno-centric to socio-centric (Lanzini & Stochetti, 2020) 8 The socio-centric methods claim that accessibility is the key concept for evaluating urban sustainable mobility. This trend emphasizes the interpretations of travel behavior are context dependent and avoids generalizations.

Research in human mobility insist the positivism methodology and has some significant contributions because the individuals differences are confounded at the macro level. Under this framework, geographic distance always plays a prominent role in all human mobility theories. In adding to travel distance and Origin-Destination Matrices, Some primary metrics such as Mean Square Displacement and Radius of gyration are defined to quantitatively describe travel behaviors

A vital insight is that human behavior has two mobility roles: explorers and returners. It might be an inherent property of society, the instinct of exploring more territory and keeping together for division of labor. The explorers’ behavior is consistent with the theory of utility maximization. People are always looking for more benefit. The concept of habit also match the behavior of ‘preferential return,’ which means people are natural or nurtured likely to return to frequently visited locations or recently-visited locations.

Both gravity and opportunities theory choose population size as the source of travel demand. This is a rough assumption and is not enough to get more accurate predictions. One solution is to use empirical observed demand to calibrate the model case by case. Another way is to find more suitable variables such as residential, employment, or activity size to improve the model.

When area of interest is intra-urban, the O-D matrix records the trip connections among all paired locations. The matrix contain plenty of information including urban spatial structure, opportunities, activities and other socio-economic characteristics. The theories imply that O-D matrix have some strong connections to travel behavior in some ways. The first challenge is how to mine the information and extract some explainable elements. A limitation is that the empirical O-D matrix may only reflect the particular characteristics in that city and can not be applied to others. The second challenge is how to get a generalized interpretation,

Once choosing the individual perspective, current theories and methods are still insufficient. For example, the physical transportation network is only a part of travel decisions. Social networks with a ‘hub-and-spoke structure’ play a prominent role in finding a job. Using social media data, some studies provide valuable insight but still have a gap to form new theories.

An interdisciplinary perspective could provide a theoretical explanation for model selection. Existing mobility theories can play an anchor to identify the key variables’ property and confirm the additive and linear relation among the factors.

CMDT=Cognitive moral development theory (Kohlberg, 1984),

ITB=Ipsative theory of behavior (Frey, 1988),

NAM=Norm activation model (Schwartz, 1977,Schwartz and Howard, 1981),

SDT=Self-determination theory(Deci & Ryan, 1985),

TAM=Technology acceptance model(Davis, 1989),

TDM=Travel demand management measures,

TNC=Theory of normative conduct (Cialdini et al., 1990,Cialdini et al., 1991),

TPB=Theory of planned behavior(Ajzen, 1985,Ajzen, 1991),

VBN=Value-belief-norm (Stern, 2000,Stern et al., 1999),

MGB=Model of goal-directed behavior(Perugini & Bagozzi, 2001), ↩︎

“Prospect Theory - an Overview ScienceDirect Topics ” ( n.d. ) ↩︎

Visser ( 2013 ) ; Jiang, Yin, and Liu ( 2015 ) ; Rozenfeld et al. ( 2011 ) ; Gomez-Lievano, Youn, and Bettencourt ( 2012 ) ; Hackmann and Klarl ( 2020 ) ↩︎

Jiang and Jia ( 2011 ) ; Chen and Jiang ( 2018 ) ; Jiang ( 2018a ) ; Jiang ( 2018b ) ↩︎

Lanzini P., Stochetti A.. From Techno-Centrism toSocio-Centrism: The Evolution of Principles for Urban Sustainable Mobility [J].International Journal of Sustainable Transportation, 2020, in Press. ↩︎

To read this content please select one of the options below:

Please note you do not have access to teaching notes, concepts of travel behaviour research.

Threats from Car Traffic to the Quality of Urban Life

ISBN : 978-0-08-044853-4 , eISBN : 978-0-08-048144-9

Publication date: 12 April 2007

Acknowledgements

This chapter draws on Axhausen (2006) and discussions during my lecturing and supervision. I gratefully acknowledge the input from those taking part.

Axhausen, K.W. (2007), "Concepts of Travel Behaviour Research", Gärling, T. and Steg, L. (Ed.) Threats from Car Traffic to the Quality of Urban Life , Emerald Group Publishing Limited, Leeds, pp. 165-185. https://doi.org/10.1108/9780080481449-009

Emerald Group Publishing Limited

Copyright © 2007 Emerald Group Publishing Limited

All feedback is valuable

Please share your general feedback

Report an issue or find answers to frequently asked questions

Contact Customer Support

• View Record

TRID the TRIS and ITRD database

Concepts of Travel Behavior Research

In the first part this chapter proposes a conceptual framework for travel behavior research through a definition of the scope of the research topic, essentially human activity schedules, and a conceptualization of the traveler as a network actor negotiating infrastructure and human networks and dealing with the social content of the activities undertaken. In the second part of the chapter an operationalization of this framework through the dynamic microsimulation of daily life nested within the microsimulation of longer-term projects and choices.

• Find a library where document is available. Order URL: http://worldcat.org/isbn/9780080448534
• Abstract reprinted with permission from Elsevier.
• Axhausen, Kay W
• Publication Date: 2007
• Media Type: Print
• Edition: First
• Features: Figures; Maps; References;
• Pagination: pp 165-185
• Monograph Title: Threats from Car Traffic to the Quality of Urban Life: Problems, Causes, and Solutions

Subject/Index Terms

• TRT Terms: Activity choices ; Behavior ; Human factors ; Microsimulation ; Research ; Schedules ; Travel behavior ; Travelers
• Uncontrolled Terms: Daily activity plans
• Subject Areas: Highways; Planning and Forecasting; Public Transportation; Research; Safety and Human Factors; Society; I72: Traffic and Transport Planning;

Filing Info

• Accession Number: 01076811
• Record Type: Publication
• ISBN: 9780080448534
• Files: TRIS
• Created Date: Sep 21 2007 1:55PM

Travel Behavior

Introduction

Topics in Travel Behavior

Representations of Decision Making Behavior in Travel Modeling

Activity-Travel Planning and Decision Making Behaviors

Joint Travel Behavior

Other Resources

Page categories

More pages in this category:

# introduction.

(opens new window) . When individuals travel to specific locations at certain times, they are engaging in activities like work (satisfying the need for resources), socializing (belonging and esteem), recreation and leisure (self-actualization), etc. Travel behavior is the process people use to organize their time to meet these needs under a variety of constraints (e.g., time, cost, social obligations).

Travel behavior is a broad topic that touches many aspects of travel demand modeling and forecasting. Kostas Goulias has stated that travel behavior modeling:

...refers primarily to the modeling and analysis of travel demand on the basis of theories and analytical methods from a variety of scientific fields. They include, but are not limited to, the use of time and its allocation to travel and activities, the use of time in a variety of time contexts and stages in the life of people, and the organization and use of space at any level of social organization, such as the individual, the household, the community, and other formal or informal groups. [2]

Some basic assumptions are often used when analyzing travel behavior. In some cases, these are generalities built from observed behavior or simplifications of complexities of human behavior. Common assumptions include the following:

• An individual with a higher income will tend to be less sensitive to travel cost than an individual with lower income.
• Travelers are attracted to locations with a large amount of households and jobs and will tend to travel further distances to these locations than lower density areas.
• Single occupancy vehicle travel is generally preferred over transit, even after accounting for time and cost of travel.

(opens new window) is collected on a periodic basis to understand the local and temporal context of travel behavior. Modelers must be careful not to imbed behavioral assumptions into models that may change over time when forecasting.

To the extent that is possible, travel behaviors should be treated often as sensitivities in the model so that a range of scenarios are predicted, realizing that certain behaviors are not precise mechanical operations, but malleable based on many possibly unknowable factors. One role of the modeler, then, is to identify how travel will be impacted if certain behavioral trends continue or cease.

(opens new window) depict the time tradeoffs among activities and travel in the day.

(opens new window) but requires a subscription to access. These reviews taken together provide a solid foundation to understanding the current state of travel behavior research

# Topics in Travel Behavior

# representations of decision making behavior in travel modeling (opens new window).

Decision-making is a central element of traveler behavior modeling. Many of the behaviors represented in travel demand model are directly related to a decision making process - from choosing travel modes to selecting routes to reach a destination. The way in which these decisions are modeled, however, varies greatly for different decisions and in different model frameworks. The most common form that decision making behavior takes is that of the discrete choice model . However, many other decision making behaviors are utilized in travel models, including heuristic methods, or processes derived from artificial intelligence such as the computational process model.

# Activity-Travel Planning and Decision Making Behaviors (opens new window)

Activity-travel planning behaviors generally refer to the behavioral processes by which individuals plan, schedule and implement their day-to-day activity and travel plans. The topic of activity-travel planning encompasses many different traveler behaviors, from activity conception/generation, to the decision-making processes by which the activity plan is carried out. These behaviors are represented in various conceptional models of activity planning behavior, and are implemented in many activity-based models.

# Joint Travel Behavior (opens new window)

Joint Travel Behavior refers to how people choose to travel when considering the travel of other household members or persons in their social networks. For instance, children under driving age need to be driven to activities by parents, which requires coordination of household schedules. Household members often need to decide how to share vehicles to conduct daily activities. Some activities are more likely to be conducted with other people, like eating a meal or socializing, whereas other activities may often be conducted alone, such as commuting to work.

# Other Resources

(opens new window)

(opens new window) .

Other pages on this website include:

• Travel Behavior of Diverse Populations
• Using policy to affect travel behavior

# References

Maslow, A (1954). Motivation and personality. New York, NY: Harper. p. 236. ISBN 0-06-041987-3. ↩︎

(opens new window) ↩︎

← Index Accessibility →

This site uses cookies to learn which topics interest our readers.

Travel attitudes or behaviours: Which one changes when they conflict?

• Published: 16 October 2021
• Volume 50 , pages 25–42, ( 2023 )

Cite this article

• Laura McCarthy 1 ,
• Alexa Delbosc   ORCID: orcid.org/0000-0001-8744-3469 1 ,
• Maarten Kroesen 2 &
• Mathijs de Haas 3  

13k Accesses

16 Citations

506 Altmetric

Explore all metrics

In recent years, there has been a resurgence of interest in the ‘disagreement’ or dissonance between travel attitudes and behaviours. This has shown that when people experience travel-related dissonance they are less satisfied with their travel experience. However, what remains unclear is whether people experiencing dissonance are more likely to change their travel attitude or their behaviour, so that they are more closely aligned. Moreover, it is unclear whether and how life events, such as having a child, interact with creating or reducing travel-related dissonance. Using data from a large and well-designed longitudinal study, this paper addresses these two gaps in the literature on travel-related cognitive dissonance through an exploratory study. The findings suggest that dissonant travellers are more likely to change their segment membership than consonant travellers. Furthermore, in line with the theory of cognitive dissonance, people may adjust either their attitudes or behaviours to achieve a state of consonance. This suggests that policymakers should not only focus on subtle nudges aimed at changing attitudes (and subsequently behaviours) in desirable directions but also on implementing policies aimed at directly influencing behaviours, assuming that attitudes will follow.

Similar content being viewed by others

The Need to Change How People Think About the Consequences of Travel

Normative beliefs and modality styles: a latent class and latent variable model of travel behaviour.

The role of culture and evolving attitudes in travel behaviour assimilation among south asian immigrants in Melbourne, Australia

Avoid common mistakes on your manuscript.

Introduction

Transportation researchers have long been interested in the relationship between attitudes and travel behaviour. Most of this research focuses on the congruent relationship between attitudes and behaviour, showing that positive mode-specific attitudes predict the use of that mode (Kroesen and Chorus 2018 ) or conversely that travel behaviours shape attitudes (Tardiff 1977 ; Dobson et al. 1978 ; Kroesen et al. 2017 ). More recently, researchers have become interested in the state of ‘disagreement’ between attitudes and behaviours. When actions and attitudes do not match, people experience an unpleasant psychological state. In psychology, this state is referred to as cognitive dissonance and it has been a subject of research since the 1950s (Festinger 1957 ). When people experience cognitive dissonance, they tend to change either their attitude or their behaviour so that the two are more closely aligned.

In recent years transportation researchers have found that when people experience travel-related dissonance they are less satisfied with their travel experience (De Vos 2018 ). However, a recent review highlighted that to date no studies have been able to examine whether people change their travel behaviour to reduce travel-related dissonance (De Vos and Singleton 2020 ).

Furthermore, to date, no studies on travel-related dissonance have examined the role that life events might play in inducing (or reducing) a state of cognitive dissonance. The mobility biographies approach highlights the important role of life events, such as moving houses or changing jobs, in triggering travel behaviour change (Clark et al. 2016a ; de Haas et al. 2018 ). For instance, moving house may allow an individual with negative attitudes towards the car but frequent car-use to move to a less car-dependent location. However, if the choice of home location is constrained, it may also induce dissonance, for example, if an individual who prefers transit must move to a car-dependent area for work (see for example the ‘reverse causality hypothesis’ between the built environment and travel, (van Wee et al. 2019 )).

This paper attempts to address these two gaps in the literature on travel-related cognitive dissonance through an exploratory study. It does this through three research questions:

Over time, how stable is the state of travel-related cognitive dissonance?

Are people who experience travel-related cognitive dissonance more likely to change their travel behaviour or attitudes?

What is the impact of life events on inducing or reducing travel-related cognitive dissonance?

We explore these research questions using longitudinal panel data to examine whether mode-dissonant or mode-consonant individuals are more likely to remain in the same category in subsequent waves. Secondly, we explore whether mode-dissonant travellers are more likely to adapt their behaviour or attitudes to reduce dissonance. Finally, we explore the effect of life events on creating or reducing dissonance between travel attitudes and behaviours. Insights from these research questions may have implications for strategies encouraging travel behaviour change, identifying potential segments of the population who are more open to changing their travel behaviour.

Cognitive dissonance theory and travel behaviour

Cognitive Dissonance Theory (CDT) posits that people have an internal drive to resolve inconsistencies between attitudes and behaviours (Festinger 1957 ). If an inconsistency exists or arises between a particular attitude and behaviour, people will typically experience some level of psychological discomfort. In turn, this motivates people to adopt strategies to reduce this discomfort. Festinger ( 1957 ) outlined typical strategies to alleviate the discomfort: changing one of the dissonance conditions, changing the cognition through acquiring new knowledge that assists to alleviate the dissonance or trivialising the dissonance. For example, a person experiencing dissonance between his/her smoking behaviour and the cognition that smoking is bad for health may alter the behaviour (quit smoking) or the cognition (downplay the negative health effects of smoking). A priori, the theory of cognitive consistency assumes no dominance over which strategy is more likely to occur. Hence, based on this theory, an influence from behaviour towards attitudes is as likely as an influence from attitudes towards behaviour.

In the early days of social-psychological research, CDT was among the most heavily researched theories in the field. Literally hundreds of experiments have been conducted (Aronson 1992 ). Interestingly, however, little efforts have been made to operationalise the theory in a field setting. With respect to a particular target behaviour, one may identify four groups, two consonant ones and two dissonant ones, as depicted in Table 1 below. This approach to classifying people into categories of dissonant vs consonant has been used in studies of travel behaviour and residential location choice (Kroesen et al. 2017 ; Kamruzzaman et al. 2021 ). By assessing how membership of these four groups evolves, the main premise of CDT may already be tested in an observational setting, namely the notion that members of consonant groups are more likely to stay in their respective groups compared to members of the dissonant groups, who may be expected to either change their behaviour or their attitude towards the behaviour or adjust their circumstances. This approach assumes the adjustment of attitudes or behaviours is the only mechanism adopted to alleviate dissonance. Moreover, it does not account for the scale or variations in other background factors that may influence the extent and type of reduction strategies adopted. Nonetheless, this simplified approach allows for the fundamental concept of CDT to be tested in a field setting.

In this paper, we aim to return to the exploration of changes in attitude-behaviour inconsistencies over time, focusing on travel behaviour in particular. A few recent studies have explored this topic. Kroesen et al. ( 2017 ) used latent class analysis to identify groups of consonant and dissonant travellers and found, in line with CDT, that consonant travellers were less likely to change travel behaviour over time than dissonant travellers. De Vos ( 2018 ) also identified groups of consonant/dissonant travellers for different modes, revealing sizeable groups of dissonant travellers, especially regarding the use of the bicycle. The contribution of that study rests in the link that was made with travel satisfaction. In line with expectations, respondents travelling with their preferred travel mode (i.e., consonant travellers) seemed to experience their trip more positively compared to people travelling with a non-preferred travel mode (i.e., dissonant travellers). This result is well in line with CDT, in which inconsistency is assumed to create a state of stress/arousal, which can also be regarded as a state of dissatisfaction.

The present study also begins by classifying and describing consonant/dissonant travellers. However, this study extends the analysis to a second point in time to assess how stable these states are over time, and whether people are more likely to change their attitude or their behaviour to reduce dissonance. Given this is a major gap in the field, (identified in a recent review of CDT research (De Vos and Singleton 2020 )), it represents the first contribution of this paper.

Life events and cognitive dissonance

The second contribution of this paper is to explore the relationship between life events and travel-related dissonance. The mobility biographies approach, first introduced by Lanzendorf ( 2003 ), provides a framework for understanding travel behaviour changes throughout the life course. The framework introduces the theory of multiple life domains and recognises that a change in one domain or area will have ramifications for other areas (Lanzendorf 2003 ). As comprehensive reviews of the state of the field and the development of the theoretical framework have recently been conducted (see Müggenburg et al. ( 2015 ) and Scheiner ( 2018 )), we will instead focus this discussion on aspects of the framework relevant to our study.

A key concept emerging from the mobility biographies approach is the notion of stability and change in travel behaviour throughout the life course. Drawing on the role of habit as an important determinant of travel mode choice (Banister 1978 ; Aarts et al. 1997 ), the approach posits that life events, such as moving house or changing jobs, can disrupt stable travel routines. A life event can prompt changes to circumstances or context, which, in turn, requires an individual to switch from habitual to reflective thinking about their travel mode choices. It is these periods, in which travel routines are disrupted and habits are weakened, that provide a valuable opportunity to intervene and encourage the adoption of more sustainable travel choices (Verplanken et al. 2008 ).

Studies in this field have tended to focus on how behaviour, specifically mode use, changes following (or in anticipation of) a life event. Life events associated with an increase in car-based mobility include childbirth (Oakil et al. 2016 ; de Haas et al. 2018 ) and entering the workforce for the first time (Busch-Geertsema and Lanzendorf 2017 ). In contrast, life events associated with a decrease in car-based mobility are often associated with changes resulting in a reduction of income or change in activity patterns, such as unemployment or retirement (Oakil et al. 2014 ).

Overall, research in this field has shown that individuals are more likely to change their travel behaviour as a result of a life event than during stable circumstances (Clark et al. 2016b ). A ‘dissonant’ life event may induce an inconsistency between the frequency of using a travel mode and the attitudes towards that travel mode. For instance, in the case of childbirth, cycling frequency may reduce but positive attitudes towards cycling may remain. Conversely, a dissonance-reducing life event may create an opportunity for an individual to align their travel mode use with their attitudes to that travel mode. For instance, an individual with positive attitudes towards public transport moving home from a car-dependent location to a transit orientated location may reduce dissonance. It should be noted here that there are differing views regarding the hypothesized role of life events and inconsistencies between travel attitudes and behaviours. While Verplanken et al. ( 2008 ) posits that life events can prompt self-activation, enabling an individual to actively reflect on their optimal travel choices, applying cognitive dissonance theory, we are proposing that the life event may provide an opportunity to reconcile mismatches between travel attitudes and behaviours (rather than the self).

Using the four profiles of dissonant/consonant travellers described in Table 1 , we explore whether respondents who have experienced a life event between the two waves are more likely to transition into a different profile in the second wave. This analysis, which forms the second contribution of this paper, is largely exploratory.

Data source

The Netherlands Mobility Panel (knowns as the MPN) is an annual household panel survey that was set up to study the short-run and long-run dynamics in the travel behaviour of Dutch individuals and households, and to determine how changes in personal- and household characteristics, and in other travel-related factors, correlate with changes in travel behaviour (Hoogendoorn-Lanser et al. 2015 ). The first wave of data collection started in 2013 and the panel consists of approximately 2,000 complete households. Respondents are recruited by means of a screening questionnaire in which an adult household member is asked whether the whole household wants to participate in the MPN for several years. Yearly, after entering the panel, the same adult household member fills out a household questionnaire to gather basic information about the household. Furthermore, all household members of 12 years and older are asked to fill out an individual questionnaire and complete a three-day online travel diary. The individual questionnaire consists largely of questions that are repeated yearly. This includes questions regarding any life events that respondents have experienced in the previous 12 months. However, a special topic is repeated bi-annually. In the second and fourth waves of the MPN the special topic focused on attitudes and perceptions with regard to various modes of transportation. Therefore, the second (2014) and fourth (2016) waves of the MPN are used in this study. Table 2 presents basic demographics and life events among the survey sample.

In order to account for attrition and to maintain a representative sample, additional households are recruited yearly. In the second wave, additional focus was placed on recruiting certain groups (such as young and low educated people) since they were somewhat underrepresented in the first wave and had higher nonresponse levels. Due to attrition and recruitment of new households, there are some slight variations between waves in terms of sample composition. However, based on the so-called Gold Standard (which reflects the composition of the Dutch population on several personal- and household characteristics) it can be concluded that the sample is fairly representative for the Dutch population. The largest deviation is found on educational level, with an underrepresentation of low-educated people and an overrepresentation of high-educated people.

Defining segment membership

Table 3 presents the two measures used to determine segment membership: a composite measure of mode specific attitudes and the frequency of mode use. The measures derived from two sets of questions in the MPN survey. The MPN asks about attitudes towards travel by car, cycling and public transit. Six items are asked (Comfort/Relaxing/Saves time/Safe/Flexible/Pleasurable) measured on a five-point Likert scale ranging from “Strongly Disagree” to “Strongly Agree”. Cronbach’s Alpha for all modes exceeded 0.8 (Table 3 ), indicating a sufficient level of internal consistency to allow for a composite score to be generated. Scores for each of the six items were combined, creating a scale ranging from 6 to 30. An average of the six items was then generated and rounded to the nearest integer, providing a value of between 1 and 5. The measure regarding mode use frequency derives from a question asking about frequency of car, public transport and bicycle use over the previous year. Responses were recoded to a five-point scale to align with the attitude measure.

The attitudinal measure does not directly align with more common psychological literature typically operationalising dissonance as an unpleasant feeling or tension. Nonetheless, cognitive dissonance theory explicitly states that people attempt to align their attitudes with behaviour. As such, the attitudinal measure we use in our study is still largely aligned with what CDT posits.

We used the two measures to categorise respondents into four groups based on their behaviour (frequent or infrequent) and attitudes (negative or positive). Each respondent was categorised into one of the four groups (Frequent-Positive; Frequent-Negative; Infrequent-Positive; Infrequent-Negative) for each mode (car, cycling and public transport). Respondents were categorised as “Infrequent” if they used the travel mode three days per month or less and “Frequent” if they used the travel mode at least weekly.

To allow a more equitable distribution of the four categories, respondents whose attitude towards a particular mode was a value of between 1 and 3 were classified as “Negative” and 4 or 5 as “Positive”. Classifying respondents with neutral attitudes towards cycling and car in either the positive or negative category had a minimal bearing on the results. However, as a large number of respondents held neutral attitudes towards public transport, changing whether they were grouped in the positive or negative category lead to a significant change in the size of the consonant and dissonant segments for this mode. However, if “neutral” attitudes were re-classified as “positive” for public transport use, this resulted in multiple segments with very few respondents. Therefore we continued the analysis with all “neutral” responses coded as negative. While this approach is imperfect, and the public transport results must be treated with some caution, it ensures all segments are of a sufficient size to allow for meaningful analysis.

Table 4 shows the size of each segment and the percentage of respondents in each behaviour or attitudinal category. Overall, the largest segments among each of the three modes are consonant. Given the considerably lower frequency of public transport use compared with cycling and car use among respondents, the largest segment size in the Car and Cycle mode is Frequent-Positive, whereas among the Public Transport mode it is the Infrequent-Negative group. The Car model has the largest share of consonant respondents (81%), followed by Cycling (76%) and Public Transport (68%).

Once the travellers were classified into consonant and dissonant groups, we conducted a logistic regression model that included initial segment membership for each of the modes. Seven life events were included and four exogenous variables (gender, employment status, age and location) as active covariates. The seven life events included: childbirth, partnership formation and dissolution, starting an education course, starting or stopping working and changing employment.

Segment profiles

Table 5 outlines the demographic characteristics of each of the twelve groups. As the characteristics of each segment is not the focus of this paper, these will not be discussed in detail. However, the profile of each segment is plausible. For example, those travellers typically comprising ‘captive’ transport markets, such as students (who are entitled to free public transit in The Netherlands), are underrepresented in the Frequent-Positive Car group but overrepresented in the Frequent-Negative Cycling and Public Transport Groups. This suggests that while members of these groups have a preference for car-based mobility, their current financial circumstances make car ownership unattainable and instead they use public transport or cycle. Similarly, Frequent Public Transport users, both negative and positive, are overrepresented in highly urbanised locations and underrepresented in rural locations, likely reflecting the different level of public transport service in the respective locations.

Stability of travel-related cognitive dissonance

Table 6 shows the movements between groups between Wave 1 (2014) and 2 (2016) for car, cycling and public transit modes respectively. Given earlier research (Kroesen et al. 2017 ), we would expect the consonant segments to be the most stable (and have the highest probability of remaining in the same segment in the second wave) and the dissonant segments to be the least stable. This is the case for the largest consonant segments within each mode. Approximately 89% of the FP Car segment remained in the same segment in Wave 2 with similar proportions for the FP Cycling (83%) and IN Public Transit segments (74%). However, the smallest consonant segments (IN Car, IN Cycling and FP Public Transit) were less likely to remain in the same segment in the second wave. This is likely due to forming a much smaller proportion of the sample (5% car; 15% cycling; 7% PT) and, as such, slight changes (which may in part be random, e.g. when resulting from random from measurement errors) may appear more pronounced.

Interestingly, across all modes, dissonant travellers were more likely to remain in the Infrequent-Positive segment between waves than their counterparts in the Frequent-Negative dissonant segment. For instance, approximately half of the 2014 IP segment for car (53%), cycling (51%) and public transport (48%) remained in this state in 2016. In contrast, the proportion of respondents remaining in the Frequent-Negative segments between waves was 34% (car), 32% (cycling) and 42% (public transport). This suggests infrequent use and positive attitudes are perhaps less of a motivator to reduce dissonance than frequent use and negative attitudes.

These findings address the first research question of this paper. The next section explores whether people who experience travel-related cognitive dissonance are more likely to change their travel behaviour or their attitudes.

Change in behaviour and attitudes among formerly dissonant travellers

A core aim of this paper was to explore whether respondents experiencing travel-related cognitive dissonance more likely to change their travel behaviour or attitudes. To explore this, we assessed the movements between segments, between waves. Respondents in one of the dissonant segments in 2014 (either ‘Frequent-Negative’ or ‘Infrequent-Positive’) could transition into a consonant segment in 2016 by adjusting either their attitudes or their behaviour.

We would expect Frequent Negative travellers to be travelling against their attitudes due to external factors, such as poor transport infrastructure or conceivably time, income or family limitations. As such, we would expect an adjustment in attitudes in order to move to a consonant group (FP). Interestingly, however, whether respondents in a dissonant segment changed their attitude or behaviour varied depending on the mode and their original segment (see Table 7 )

For both car travel and cycling, people who were IP in 2014 were more likely to change their behaviour than their attitude to become consonant travellers. Over 80% of IP drivers in 2014 who became consonant in 2016 did so because they drove more in 2016; 54% of IP cyclists did the same. In contrast, FN drivers and cyclists in 2014 who became consonant in 2016 were more likely to change their attitudes; 87% of FN drivers and 70% of cyclists who changed classification did so because they had a more positive attitude to the respective modes in 2016.

The pattern was reversed for public transport. Here, 73% of IP transit users who became consonant between 2014 and 2016 did this by reducing their attitude to public transport, becoming IN users. Among the FN public transport riders who became consonant in 2016, 80% did this by reducing their ridership (becoming IN).

These contrasting patterns may be explained by the baseline attitudes toward the three modes and considerably lower use of public transport among the sample. Overall attitudes toward cars and bicycles are more positive than public transport and only 20% of respondents are categorised as ‘frequent’ public transport users compared with 83% of car travellers and 70% of cyclists.

The final section of the findings addresses the third research question, exploring whether life events induce or reduce travel-related cognitive dissonance.

Life events and changes in dissonance

Table 8 presents the results of paired sample t-tests showing the changes in mean attitudes and behaviour between 2014 and 2016. The changes are presented overall and by respondents who have experienced a life event. It should be noted that these are derived from ordinal scales and no correction was made for multiple comparisons; however these initial results provide an indication of the direction of change between survey waves. Among the overall sample, there are marginally significant declines in cycling and increases in public transit use. There were no significant changes to attitudes towards each mode between the survey years. Interestingly, among the life events examined, only behaviour changes are significant suggesting that when experiencing a life event, behavioural rather than attitudinal changes create dissonance. However, as we are examining the aggregate changes this may mask changes apparent within individual groups.

Most life events analysed were associated with some behaviour change between the two survey years. Echoing findings from previous studies examining the effects of different life events on travel mode use, childbirth is associated with a slight decline in cycling (de Haas et al. 2018 ). Of note, partnership formation is associated with an increase in car-use and decrease in cycling. This is likely attributed to the new partner owning a vehicle and, as a result, more joint trips being conducted by car rather than cycling. An alternative interpretation could also be, however, that it is more cost-efficient to own a vehicle as couple, and, as a result respondents may have been more likely to purchase a vehicle in the intervening two years between the survey waves. Starting and stopping working is associated with an increase and decrease in car use, respectively. Stopping working is also associated with an increase in public transit use. These mode changes are likely a result of income or time-budget changes, as a result of changing employment status.

Starting an education course is associated with a decrease in cycling. This may reflect the local context, as in The Netherlands students can travel for free on public transport to and from their education institution. Finally, changing jobs (but not changing job location) is associated with an increase in car use and decrease in public transport use. This could reflect gaining a promotion and a subsequent increase in salary or obtaining a company car.

Table 9 presents the results of a logistic regression model that included initial segment membership for each of the modes and shows the probabilities of changing segments between 2014 and 2016 by type of life event. The results have been presented in this format rather than logit parameters for easier interpretation. Statistically significant movements between segments were evident among respondents who have experienced childbirth, partnership formation and dissolution, starting an education course and changing employment or employment location. However, across all life events and modes, respondents were equally likely to become either more consonant or more dissonant following a life event.

Of note, childbirth was associated with a greater likelihood of forming part of the IP Cycling segment and a lower likelihood of forming part of the FP Cycling segment. This reflects the decline in cycling evident in Table 6 and echoes previous research showing new parenthood is associated with a decline in cycling (Scheiner and Holz-Rau 2013 ; de Haas et al. 2018 ). Interestingly, childbirth is also associated with an increasing likelihood of forming part of the IN Public Transit segment. As public transit is commonly used for commuting, this may reflect the changes in activity patterns as result of taking parental leave and reduced use of this mode for work travel.

Partnership formation and dissolution were associated with slight movements between the segments. Partnership dissolution was associated with an increased likelihood of forming part of the IP Car and IP PT segment. In contrast, partnership formation increased the likelihood of forming part of the IN Cycling and PT segments. These changes could be attributed to the new partner bringing a car to the household (or, in the case of dissolution, taking away a car). Among respondents who experienced a partnership formation, vehicle ownership increased from 0.8 vehicles per household to 1.0 vehicles per household (compared to a slight decline in vehicle ownership across the entire survey sample). This suggests the new partner may bring a vehicle to the relationship or they may be more likely to purchase a new vehicle as a couple, with subsequent implications for the respondents’ daily travel behaviour.

Starting an education course is associated with a decreased likelihood of forming part of the Infrequent-Negative PT segment and increased likelihood of forming part of either Frequent Positive or Frequent Negative PT segment (although the latter two changes are not significant). This likely reflects the growth of public transit use during tertiary study and provision of reduced fares available to students.

Finally, the two employment life events were associated with only marginal movements between segments. Changing job was associated with a greater likelihood of forming part of the FP Car segment while changing job location was associated with a greater likelihood of forming part of the IN PT segment and a less likely to form part of the FP PT segment.

For some life events the results are difficult to interpret. Perhaps this is because the life event is not captured at a sufficiently granular level. For instance, life events resulting in location changes (such as moving house, changing jobs or changing job locations) will likely have varying effects depending on the spatial and temporal context of the change. Surprisingly, no significant effects were evident for starting or stopping employment.

Discussion and conclusion

This paper explored stability among dissonant and consonant travellers and whether dissonant travellers are more likely to adjust their behaviour or attitudes to reach a state of consonance. Moreover, it explored the role of life events in creating or reducing dissonance between travel attitudes and behaviours. Differences between travel attitudes and behaviour were explored by public transit, cycling and car use, using panel data. It should be noted there are limitations regarding the measure we used to assess dissonance and the dissonance strategies we tested (only behaviour or attitude change). The psychology literature has identified a range of other mechanisms to reduce dissonance, such as self-forgiving, denial of responsibility or forgetting about the dissonance (Kruglanski et al. 2018 ), that depend on the intensity of negative feeling and familiarity of the situation (Cancino-Montecinos et al. 2020 ). In a travel behaviour context, it is plausible people could adopt a range of these and other strategies to reduce dissonance. The implications of this limitation are discussed below.

Addressing our first research question, consistent with earlier research (Kroesen et al. 2017 ), between the survey years, dissonant travellers were more likely to change their segment membership than consonant travellers. Interestingly, dissonant travellers were more likely to remain in the Infrequent-Positive segment between waves than travellers in the Frequent-Negative segment. This suggests that infrequent use and positive attitudes are perhaps less cognitively distressing than frequent use and negative attitudes. This may be evidence that people are practicing a range of mental mechanisms to reduce this dissonance, such as self-forgiving or denial of responsibility (e.g. ‘I’d love to cycle but it’s not my fault it isn’t safe) (Kruglanski et al. 2018 ).

Should this indeed be true, then a straightforward implication for policy would be that positive attitudes, e.g. towards bicycle use, are not enough to increase the uptake of cycling. Instead, policies that negatively influence the attitudes towards alternative modes (e.g. the car) may potentially be more effective. Similarly, encouraging an increase in behavioural frequency for certain modes (e.g. encouraging infrequent-negative cyclists to cycle more frequently), may lead to more positive attitudes towards this mode.

In addition, from a scientific perspective, it would be interesting to explore whether dissonance resulting from infrequent use and positive attitudes is indeed less distressing than dissonance resulting from frequent use and negative attitudes. As shown by De Vos ( 2018 ) dissonance between the chosen and preferred mode also translates itself into travel dissatisfaction. This concept could be used to assess whether those belonging to the segment “frequent use and negative attitudes” are (even) more dissatisfied with their travel behaviour than those belonging to the segment “infrequent use and positive attitudes”.

Our second research question explored whether dissonant travellers were more likely to adjust their behaviour or attitudes. The results depended on the mode and their original class. A similar pattern was evident between car and cycling modes. Infrequent car travellers and cyclists with positive attitudes were more likely to increase their use than adjust their attitudes to reach a state of consonance in the following wave. Conversely, among frequent car-users and cyclists with negative attitudes, their attitudes were likely to become more favourable in the following wave. The reverse picture was evident for public transport; this is likely to be attributed to the baseline use and attitudes towards this mode being considerably lower than for car and cycling.

A general take-way from the results above is that, in line with the theory of cognitive dissonance, people may adjust both their attitudes and behaviours to achieve a state of consonance. From a policy perspective, this means that policymakers should not only focus on subtle nudges aimed at changing attitudes (and subsequently behaviours) in desirable directions but may also implement policies that aim to directly influence behaviours, e.g. by setting rules or pricing policies, assuming that attitudes will follow.

Finally, we explored whether life events induced or reduced dissonance. Nearly all of the life events included in the analysis were associated with changing levels of cognitive dissonance across at least one travel mode. However, across all life events participants were equally likely to become more or less consonant following a life event. And while certain life events were associated with behavioural changes between the two survey years, no life events were associated with attitudinal changes towards the travel mode. This suggests that following life events, behavioural rather than attitudinal changes tend to create dissonance. Future analysis, using data from subsequent waves, could explore whether there is a lag in attitude change following the behavioural change.

Life events associated with an increase in car-based mobility (either as a result of increasing car-use or decreasing use of cycling or public transit) included: childbirth, moving in with a partner, starting work and changing jobs. An increase in car-use, over time, is likely to contribute to the adoption of more favourable attitudes towards car-use. This, in turn, makes it less likely individuals will return to their former sustainable travel patterns, if, or when, their circumstances change. The adoption of more favourable attitudes towards car-use poses problems for transportation planners and policymakers seeking to encourage more sustainable travel modes. Further qualitative research may assist in unpicking the processes by which travel attitudes are adjusted (or not) to align with new travel behaviours. This would provide important insights to contribute to policies tasked at curbing the adoption of car-based mobility associated with certain life events.

Although the panel data provides a rich and detailed source of life events, several limitations exist. Due to the small number of respondents experiencing certain life events, some of the results were difficult to interpret. Furthermore, the classification of life events, especially those that instigated a relocation of work or home location, could be analysed at a more granular level. As data from future years becomes available, and more participants experience these life events, this will assist with interpreting the changing levels of cognitive dissonance associated with experiencing less common life events.

Further limitations regard the measures and classification process we adopted. As this was an exploratory analysis, we conducted an a priori classification. However, future analyses could adopt a post-hoc classification process which may yield more nuanced segments of behavioural and attitudinal adjustments made by dissonant travellers. Moreover, we assumed travellers were only to reach a consonant state through adjusting their attitudes or behaviours where we have acknowledged that multiple strategies (such as forgetting or trivialising) may also be adopted (Kruglanski et al. 2018 ). In addition, we did not account for the magnitude of the dissonance or familiarity of the situation in influencing the dissonance reduction strategy adopted (Cancino-Montecinos et al. 2020 ). As this was an exploratory study, aiming to test fundamental concepts of CDT, these topics are proposed for future research.

The panel data used in this analysis was conducted before the COVID-19 pandemic. Given the anticipated structural changes to employment practices prompted by the pandemic, it may provide an opportunity for respondents to reconcile disparities between their travel attitudes and behaviour. For instance, respondents in the ‘Frequent-Negative’ categories for car or public transport may have the opportunity to work from home, reducing the need for commuting. Initial analysis of MPN respondents suggests that, of those respondents working at home during the pandemic, over a quarter expect to work from home more in the future (de Haas et al. 2020 ). Data from future waves will reveal the extent that this has transpired into changing travel practices and the subsequent impact this may have on reducing travel related cognitive dissonance.

Aarts, H., Verplanken, B., Van Knippenberg, A.: Habit and information use in travel mode choices. Acta Physiol. (oxf.) 96 (1–2), 1–14 (1997)

Google Scholar  

Aronson, E.: The return of the repressed: dissonance theory makes a comeback. Psychol. Inq. 3 (4), 303–311 (1992)

Article   Google Scholar  

Banister, D.: Influence of habit formation on modal choice—a heuristic model. Transportation 7 (1), 5–18 (1978)

Busch-Geertsema, A., Lanzendorf, M.: From university to work life—jumping behind the wheel? Explaining mode change of students making the transition to professional life. Transp. Res. Part a: Policy Pract. 106 , 181–196 (2017)

Cancino-Montecinos, S., Björklund, F., Lindholm, T.: A general model of dissonance reduction: unifying past accounts via an emotion regulation perspective. Front. Psychol. 11 , 540081 (2020)

Clark, B., Chatterjee, K., Melia, S.: Changes in level of household car ownership: the role of life events and spatial context. Transportation 43 , 565–599 (2016a)

Clark, B., Chatterjee, K., Melia, S.: Changes to commute mode: the role of life events, spatial context and environmental attitude. Transp. Res. Part a: Policy Pract. 89 , 89–105 (2016b)

de Haas, M.C., Scheepers, C.E., Harms, L.W.J., Kroesen, M.: Travel pattern transitions: applying latent transition analysis within the mobility biographies framework. Transp. Res. Part a: Policy Pract. 107 , 140–151 (2018)

de Haas, M., Faber, R., Hamersma, M.: How COVID-19 and the Dutch ‘intelligent lockdown’ change activities, work and travel behaviour: evidence from longitudinal data in the Netherlands. Transp. Res. Interdiscip. Perspect. 6 , 100150 (2020)

De Vos, J.: Do people travel with their preferred travel mode? Analysing the extent of travel mode dissonance and its effect on travel satisfaction. Transp. Res. Part a: Policy Pract. 117 , 261–274 (2018)

De Vos, J., Singleton, P.A.: Travel and cognitive dissonance. Transp. Res. Part a: Policy Pract. 138 , 525–536 (2020)

Dobson, R., Dunbar, F., Smith, C.J., Reibstein, D., Lovelock, C.: Structural models for the analysis of traveler attitude-behavior relationships. Transportation 7 (4), 351–363 (1978)

Festinger, L.: A theory of cognitive dissonance. Stanford University Press, Stanford, CA (1957)

Book   Google Scholar  

Hoogendoorn-Lanser, S., Schaap, N.T.W., OldeKalter, M.-J.: The Netherlands mobility panel: an innovative design approach for web-based longitudinal travel data collection. Transp. Res. Procedia 11 , 311–329 (2015)

Kamruzzaman, M., Giles-Corti, B., De Vos, J., Witlox, F., Shatu, F., Turrell, G.: The life and death of residential dissonants in transit-oriented development: a discrete time survival analysis. J. Transp. Geogr. 90 , 102921 (2021)

Kroesen, M., Chorus, C.: The role of general and specific attitudes in predicting travel behaviour—a fatal dilemma? Travel Behav. Soc. 10 , 33–41 (2018)

Kroesen, M., Handy, S., Chorus, C.: Do attitudes cause behavior or vice versa? An alternative conceptualization of the attitude-behavior relationship in travel behavior modeling. Transp. Res. Part a: Policy Pract. 101 , 190–202 (2017)

Kruglanski, A.W., Jasko, K., Milyavsky, M., Chernikova, M., Webber, D., Pierro, A., Di Santo, D.: Cognitive consistency theory in social psychology: a paradigm reconsidered. Psychol. Inq. 29 (2), 45–59 (2018)

Lanzendorf, M.: Mobility biographies. A new perspective for understanding travel behaviour. In: 10th International Conference on Travel Behaviour Research (IATBR). Lucerne (2003)

Müggenburg, H., Busch-Geertsema, A., Lanzendorf, M.: Mobility biographies: a review of achievements and challenges of the mobility biographies approach and a framework for further research. J. Transp. Geogr. 46 , 151–163 (2015)

Oakil, A.T.M., Ettema, D., Arentze, T., Timmermans, H.: Changing household car ownership level and life cycle events: an action in anticipation or an action on occurrence. Transportation 41 (4), 889–904 (2014)

Oakil, A.T.M., Manting, D., Nijland, H.: Dynamics in car ownership: the role of entry into parenthood. Eur. J. Transp. Infrastruct. Res. 16 (4), 661–673 (2016)

Scheiner, J.: Why is there change in travel behaviour? In search of a theoretical framework for mobility biographies. Erdkunde 72 (1), 41–62 (2018)

Scheiner, J., Holz-Rau, C.: A comprehensive study of life course, cohort, and period effects on changes in travel mode use. Transp. Res. Part a: Policy Pract. 47 , 167–181 (2013)

Tardiff, T.J.: Causal inferences involving transportation attitudes and behavior. Transp. Res. 11 (6), 397–404 (1977)

van Wee, B., De Vos, J., Maat, K.: Impacts of the built environment and travel behaviour on attitudes: theories underpinning the reverse causality hypothesis. J. Transp. Geogr. 80 , 102540 (2019)

Verplanken, B., Walker, I., Davis, A., Jurasek, M.: Context change and travel mode choice: Combining the habit discontinuity and self-activation hypotheses. J. Environ. Psychol. 28 (2), 121–127 (2008)

Download references

Author information

Authors and affiliations.

Department of Civil Engineering, Institute of Transport Studies, Monash University, 23 College Walk, Clayton, VIC, 3800, Australia

Laura McCarthy & Alexa Delbosc

Faculty of Technology, Policy and Management, TU Delft, Building 31. Jafalaan 5, 2628 BX, Delft, The Netherlands

Maarten Kroesen

KiM Netherlands Institute for Transport Policy Analysis, Bezuidenhoutseweg 20, 2594 AV, The Hague, The Netherlands

Mathijs de Haas

You can also search for this author in PubMed   Google Scholar

Contributions

Laura McCarthy: Research conception, research design, data analysis, results interpretation, lead paper writing. Alexa Delbosc: Research design, results interpretation, paper writing. Maarten Kroesen: Research design, results interpretation, paper writing. Mathijs de Haas: Data analysis, paper writing.

Corresponding author

Correspondence to Alexa Delbosc .

Ethics declarations

Conflict of interest.

On behalf of all authors, the corresponding author states that there is no conflict of interest.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

McCarthy, L., Delbosc, A., Kroesen, M. et al. Travel attitudes or behaviours: Which one changes when they conflict?. Transportation 50 , 25–42 (2023). https://doi.org/10.1007/s11116-021-10236-x

Download citation

Accepted : 21 September 2021

Published : 16 October 2021

Issue Date : February 2023

DOI : https://doi.org/10.1007/s11116-021-10236-x

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Cognitive dissonance
• Travel behaviour
• Life events
• Find a journal
• Publish with us
• Track your research

Advertisement

News Analysis

An Assassination Attempt That Seems Likely to Tear America Further Apart

The attack on former President Donald J. Trump comes at a time when the United States is already polarized along ideological and cultural lines and is split, it often seems, into two realities.

• Share full article

By Peter Baker

Peter Baker has covered the past five presidents.

• Published July 14, 2024 Updated July 15, 2024

Follow the latest news on the Trump assassination attempt.

When President Ronald Reagan was shot by an attention-seeking drifter in 1981, the country united behind its injured leader. The teary-eyed Democratic speaker of the House, Thomas P. O’Neill Jr., went to the hospital room of the Republican president, held his hands, kissed his head and got on his knees to pray for him.

But the assassination attempt against former President Donald J. Trump seems more likely to tear America further apart than to bring it together. Within minutes of the shooting, the air was filled with anger, bitterness, suspicion and recrimination. Fingers were pointed, conspiracy theories advanced and a country already bristling with animosity fractured even more.

The fact that the shooting in Butler, Pa., on Saturday night was two days before Republicans were set to gather in Milwaukee for their nominating convention inevitably put the event in a partisan context. While Democrats bemoaned political violence, which they have long faulted Mr. Trump for encouraging, Republicans instantly blamed President Biden and his allies for the attack, which they argued stemmed from incendiary language labeling the former president a proto-fascist who would destroy democracy.

Mr. Trump’s eldest son, his campaign strategist and a running mate finalist all attacked the political left within hours of the shooting even before the gunman was identified or his motive determined. “Well of course they tried to keep him off the ballot, they tried to put him in jail and now you see this,” wrote Chris LaCivita, a senior adviser to the former president.

But the Trump campaign seemed to think better of it, and the post was deleted. A memo sent out on Sunday by Mr. LaCivita and Susie Wiles, another senior adviser, instructed Trump team members not to comment on the shooting.

Either way, the episode could fuel Mr. Trump’s narrative about being the victim of persecution by Democrats. Impeached, indicted, sued and convicted, Mr. Trump even before Saturday had accused Democrats of seeking to have him shot by F.B.I. agents or even executed for crimes that do not carry the death penalty.

After being wounded at the rally, Mr. Trump, with blood staining his face, pumped his fist at the crowd and shouted, “Fight! Fight! Fight!”

What exactly drove the gunman, who was quickly killed by Secret Service counter snipers, remained a matter of speculation. Identified as Thomas Matthew Crooks , 20, from Bethel Park, Pa., he was a registered Republican but had also given $15 to a progressive group on Mr. Biden’s Inauguration Day, more than three years ago. The authorities said they were still investigating his motive.

The shooting came at a time when the United States was already deeply polarized along ideological, cultural and partisan lines — split, it often seems, into two countries, even two realities. More than at any time in generations, Americans do not see themselves in a collective enterprise but perceive themselves on opposite sides of modern ramparts.

The divisions have grown so stark that a Marist poll in May found that 47 percent of Americans considered a second civil war likely or very likely in their lifetime, a notion that prompted Hollywood to release a movie imagining what that could look like.

The propulsive crescendo of disruptive events lately has led many to compare 2024 to 1968, a year of racial strife, riots in the cities and the assassinations of the Rev. Dr. Martin Luther King Jr. and Robert F. Kennedy. Protests over the Vietnam War helped prompt President Lyndon B. Johnson to drop out of his race for re-election that year.

Until now, there had been one important difference. “Of all the similarities between 1968 and 2024, the lack of political violence this year has been one of the key areas where the years diverge,” said Luke A. Nichter, a historian at Chapman University and the author of “The Year That Broke Politics,” a history of 1968. “That is no more.”

Michael Kazin, a historian at Georgetown University, said political violence had a long history in America. “As in 1968 — or 1919 or 1886 or 1861 — the violence that just occurred is rather inevitable in a society as bitterly divided as ours,” he said. “And of course there’s actually less violence in politics now than there was in those other years.”

Yet not since President Abraham Lincoln was shot by a Confederate sympathizer at Ford’s Theater has an assassination attempt against a president or major presidential candidate so sharply exacerbated the partisan divide.

Presidents James A. Garfield, William McKinley and John F. Kennedy were shot to death by lone gunmen who were upset with them for one reason or another, but the killings did not become sources of schism between the Republican and Democratic Parties. The same was true with Dr. King and Robert Kennedy’s assassinations, as well as shootings that missed President-elect Franklin D. Roosevelt and President Gerald R. Ford.

Gov. George C. Wallace, Democrat of Alabama, was shot at a campaign event during his 1972 presidential run by a man who wanted to be famous. The attack left the segregationist governor paralyzed but eventually contributed to his evolution and disavowal of past racism. John Hinckley attacked Mr. Reagan out of an obsession to impress the movie star Jodie Foster.

In recent years, political violence in America at levels below the presidency has become increasingly partisan. Representative Gabrielle Giffords, Democrat of Arizona, was critically wounded in a mass shooting in 2011, prompting angry criticism of Republicans for fomenting hate. Representative Steve Scalise of Louisiana, now the Republican majority leader, was shot and injured during a congressional baseball game practice in 2017 by a supporter of Senator Bernie Sanders, Independent of Vermont.

An armed man was arrested outside the home of Justice Brett Kavanaugh in 2022 and told the authorities that he wanted to kill the conservative Supreme Court justice because of his positions against abortion and gun control. Later that year, a man wielding a hammer broke into the San Francisco house of Representative Nancy Pelosi, then the Democratic speaker, and beat her husband, Paul Pelosi .

The most famous recent case of political violence before this weekend was the attack on the Capitol on Jan. 6, 2021, by supporters of Mr. Trump trying to block the certification of Mr. Biden’s election victory. The Capitol Police investigated 8,008 cases of threats involving members of Congress last year. While most of them were not serious, it was the second-highest total in the department’s history and has prompted the hiring of more prosecutors.

Many of these recent cases have led to not so much soul-searching as blame-setting. After Ms. Giffords was shot, Democrats assailed Sarah Palin, the former Republican vice-presidential candidate, because Ms. Giffords’s district had been among 20 singled out underneath digitized cross hairs on a map circulated by Ms. Palin’s political action committee, although there was no evidence the gunman knew about or was driven by the map.

House Democrats impeached Mr. Trump for instigating the Capitol attack with his inflammatory language at a rally beforehand. The former president has a long history of encouraging violence . He urged supporters to beat up protesters at rallies, cheered a Republican congressman for body-slamming a reporter , called for looters and shoplifters to be shot, made light of the attack on Mr. Pelosi and promised pardons to Jan. 6 rioters. When some of his supporters chanted “Hang Mike Pence!” on Jan. 6, Mr. Trump told aides that maybe the vice president deserved it because he had defied efforts to overturn the 2020 election.

Republicans turned the tables on Democrats this weekend, arguing that if Mr. Trump was responsible for provocative rhetoric, then Mr. Biden should be as well. Speaking with donors on Monday, the president said he wanted to stop talking about his poor debate performance and instead “put Trump in a bull’s-eye.” He described his strategy as “attack, attack, attack.”

“The central premise of the Biden campaign is that President Donald Trump is an authoritarian fascist who must be stopped at all costs,” Senator J.D. Vance, Republican of Ohio and a front-runner to be named Mr. Trump’s running mate, wrote on social media two hours after the attack on Saturday. “That rhetoric directly led to President Trump’s attempted assassination.”

Mr. Scalise, also the victim of a political attack, agreed. “For weeks, Democrat leaders have been fueling ludicrous hysteria that Donald Trump winning re-election would be the end of democracy in America,” he said. “Clearly, we’ve seen far-left lunatics act on violent rhetoric in the past. This incendiary rhetoric must stop.”

Representative Mike Collins, Republican of Georgia, wrote on social media that “ Joe Biden sent the orders ” and urged the local prosecutor to “immediately file charges against Joseph R. Biden for inciting an assassination.” But not all hands are clean. Mr. Collins once ran a campaign ad in which he fired a rifle at Ms. Pelosi’s agenda and shot a cardboard cutout of so-called RINO Republicans.

Some Republican leaders took a more measured approach. Speaker Mike Johnson, speaking on “Today” on NBC, said on Sunday that Mr. Trump had “been so vilified and really persecuted by media, Hollywood elites, political figures, even the legal system” and cited Mr. Biden’s “bull’s-eye” comment.

“I know he didn’t mean what is being implied there, but that kind of language on either side should be called out,” Mr. Johnson said . But he emphasized that “both sides” have “got to turn the temperature down in this country.”

Mr. Biden did not directly respond to criticism of his language during three televised appearances since the shooting, but he flatly condemned the attack and called Mr. Trump to wish him well. Like Mr. Johnson, he said that Americans must “lower the temperature” and that “it’s time to cool it down.” During a rare Oval Office address, he added: “Politics must never be a literal battlefield, and God forbid a killing field.”

The danger is if political violence becomes normalized, just another form of the endless partisan wars. A study published in May found that 11 percent of Americans said violence was sometimes or always justified to return Mr. Trump to the presidency, and 21 percent said it was justified to advance an important political objective.

But Garen J. Wintemute, the director of the Violence Prevention Program at the University of California, Davis, and the lead author of the study, said it was important to remember that most Americans still rejected political violence.

“It’s the job of that majority to make their views known, over and over again, and as publicly as possible,” Dr. Wintemute said. “A climate of intolerance for violence reduces the chance that violence will occur. The question before us as a nation is, ‘Will violence become part of American politics?’ Each of us as an individual needs to answer that question, ‘Not if I can help it.’”

Peter Baker is the chief White House correspondent for The Times. He has covered the last five presidents and sometimes writes analytical pieces that place presidents and their administrations in a larger context and historical framework. More about Peter Baker

Our Coverage of the Trump Rally Shooting

The Investigation : F.B.I. officials told Congress that the 20-year-old gunman who tried to kill Donald Trump used his cellphone and other devices to search for images of Trump and President Biden .

Security Blind Spots : Even as investigators continue to examine what happened at the Trump rally, it is already clear that there were multiple missed opportunities to stop the gunman  before the situation turned deadly.

The Gunman : In interviews, former classmates of the suspect described him as intelligent but solitary , someone who tried to avoid teasing by fellow students.

Secret Service Director : Kimberly Cheatle  returned in 2022 to lead the agency she had served for nearly 30 years. Now, the assassination attempt on Trump has thrown her tenure into uncertainty .

Fears of What’s Next : Among voters, there is growing anxiety that America’s political divide is nearly beyond repair, and the shooting only made things worse .

Pops, screams and then blood: On the scene at the Trump rally shooting

A view from the press riser of the chaos surrounding what authorities are investigating as an assassination attempt against Donald Trump.

BUTLER, Pa. — The gunshots were high-pitched pops, slight and hollow in the open air.

Donald Trump , the former president set to accept the Republican nomination in five days, was less than 10 minutes into his speech here to a crowd of tens of thousands. A miles-long line of cars crawled for hours to pass through metal detectors and bag inspections, just like any Trump event, until these green fairgrounds became a sea of red hats.

Trump was almost an hour late, and his supporters waited impatiently under the blazing sun and thumping music. In the middle of the crowd, opposite the stage, a platform of TV cameras pointed at the stage, with reporters huddled underneath for shade.

Finally Trump walked out, as usual, to chants of “USA” and marveled: “This is a big crowd. This is a big, big, beautiful crowd.” A bright red MAGA cap shaded his eyes, and his white shirt was open-collared in the heat as he leaned his arms on the lectern.

He launched into his stump speech but quickly got bored with the prepared script. He offered to invite the Republican Senate candidate, Dave McCormick, to speak, but McCormick wasn’t ready.

2024 presidential election

“You don’t mind if I go off teleprompter, do you?” Trump teased. “Because these teleprompters are so damn boring.” He asked to show “that chart that I love so much,” showing border crossings across his and Joe Biden’s presidencies, and acted amazed that his producers obliged, projecting it onto the giant screens to either side. “Wow, you guys are getting better with time.”

He was pointing to one of the screens, narrating the increase in immigration since he left office in 2021. “Look what happened to our country!”

The pops came in pairs, a burst of five or six total. Trump swatted his ear, as if he heard a mosquito. Then he hunched his shoulders and ducked.

Podcast episode

“Get down, get down, get down!” Secret Service agents shouted as they rushed up onto the stage and surrounded him. The crowd screamed. Another burst of popping noises. More screaming. The people in the bleachers behind Trump shuffled, unsure about where to go. The people in chairs or standing crouched or fell to the ground. A dense cloud of smoke hung to the right of the stage, then dispersed quickly.

One more solitary shot.

More suited Secret Service agents rushed the stage, then black-clad men wearing body armor and helmets, and carrying assault rifles. The crowd shouted in confusion.

“Are we good?” one of the officers said, audible from the podium microphone.

“Shooter’s down,” another answered.

“We’re good to move.”

“Are we clear?”

“We’re clear!”

“Let me get my shoes on,” Trump said, as the agents lifted him.

“I got you, sir.”

“Hold on, your head is bloody.”

“Let me get my shoes on,” he said again, as the agents formed a ring around him.

The crowd, seeing him standing, started to cheer.

Photos from the Trump rally shooting

“Wait,” Trump said, and thrust up a fist. “Fight!” he said. “Fight!”

Then the people roared and chanted again: “USA!”

“We gotta move,” an agent said. Leaning on the agents for assistance, Trump kept his fist raised as he hobbled off the stage, down the stairs and into his black SUV. One black dress shoe remained on the red-carpeted stage.

Officers — Secret Service, county sheriff’s deputies, state troopers, U.S. Department of Homeland Security — started telling the crowd to evacuate, calling the site an active crime scene. The rallygoers walked out, calling and texting family and friends and recording videos. People were shocked but calm.

As people passed the press risers elevating the cameras, some took out their anger on the media.

“You’re not safe. It’s your fault.”

“You wanted political violence, now you got it. Hope you’re all f---ing happy.”

“The shot heard ’round the world.”

“The liberal media is responsible!”

“Every f---ing one of y’all!”

Others sought out the cameras to offer eyewitness accounts, but they were jumbled and sometimes contradictory amid the panic.

The crowd trudged glumly to the parking lot, a few stopping for a last-minute hot dog or snow cone.

A man with a cane cowered behind the bathrooms, vomiting.

They walked to their cars past Trump flags streaming in the wind over a long row of vendors selling MAGA hats and mug shot T-shirts and Trump keychains and vulgar bumper stickers and Trump visors topped with bright orange fake hair.

A man with a bullhorn wearing a homemade “JAN 6 SURVIVOR” shirt called on people to march on Main Street, “peacefully and patriotically,” echoing Trump’s speech on the Ellipse on Jan. 6 , 2021. Most everyone ignored him. One young man accused him of being an undercover federal agent and told him to shut up.

They left behind a field strewn with empty plastic water bottles. A giant American flag hoisted from two cranes flapped high above the empty white bleachers bordered with red, white and blue bunting.

Election 2024

Catch up on key takeaways from the final day of the Republican National Convention in Milwaukee, where Donald Trump dramatically recounted the assassination attempt at his rally.

Biden pressure: President Biden is facing the most concerted effort yet by leading Democrats seeking to force him out of the presidential race amid concerns over his advanced age and sluggish poll numbers. Here’s what would happen next if Biden dropped out .

Trump VP pick: Trump has chosen Sen. J.D. Vance of Ohio as his running mate , selecting a rising star in the party and previously outspoken Trump critic who in recent years has closely aligned himself with the former president.

Presidential election polls: Check out The Post’s presidential polling averages of the seven battleground states most likely to determine the outcome of the election.

Key dates and events: Voters in all states and U.S. territories have been choosing their party’s nominee for president ahead of the summer conventions. Here are key dates and events on the 2024 election calendar .