How Neighborhood Crime, Built Environment, Life Events, and Attitudes Influence Vehicle Acquisitions: A Retrospective Analysis of Purchase, Body Type, and Vintage Decisions
A
A
MasoudArfaeiYazdiPour1✉Emailmarfa@uic.edu
AliMohammadi1Emailmohamadi@uic.edu
SinaAsgharpour1Emailsasgha3@uic.edu
NooshinJavidi2Emailnjavid2@uic.edu
AbolfazlKouros1Emailkouros@uic.edu
Mohammadian1
NazmulArefinKhan
Ph.D.
3Emailnazmul.arefin@anl.govYantaoHuang
Ph.D.
3Emailyantao.huang@anl.govJoshuaA.Auld
Ph.D.
4Phone(630)252-5460Emailjauld@anl.gov1
A
A
Department of Civil, Materials, and Environmental EngineeringUniversity of Illinois Chicago842 W Taylor St, ERF60607ChicagoIL2
A
Department of Urban Planning and PolicyUniversity of Illinois Chicago, CUPPA412 S Peoria St60607ChicagoIL3
A
Computational Transportation Engineer Transportation Systems and Mobility Group Vehicle and Mobility Systems DepartmentArgonne National Laboratory9700 S Cass Ave60439LemontIL4
A
Manager Transportation Systems and Mobility Group Vehicle and Mobility Systems DepartmentArgonne National Laboratory9700 S Cass Ave60439LemontILMasoud ArfaeiYazdiPour (Corresponding Author)
Department of Civil, Materials, and Environmental Engineering
University of Illinois Chicago
842 W Taylor St, ERF, Chicago, IL 60607
ORCID: 0009-0005-1711-2871
Email: marfa@uic.edu
Ali Mohammadi
Department of Civil, Materials, and Environmental Engineering
University of Illinois Chicago
842 W Taylor St, ERF, Chicago, IL 60607
ORCID: 0000-0002-5127-9609
Email: mohamadi@uic.edu
Sina Asgharpour
Department of Civil, Materials, and Environmental Engineering
University of Illinois Chicago
842 W Taylor St, ERF, Chicago, IL 60607
ORCID: 0000-0002-8192-5325
Email: sasgha3@uic.edu
Nooshin Javidi
Department of Urban Planning and Policy
University of Illinois Chicago
412 S Peoria St., CUPPA, Chicago, IL 60607
ORCID: 0009-0006-5073-7416
Email: njavid2@uic.edu
Abolfazl (Kouros) Mohammadian
Department of Civil, Materials, and Environmental Engineering
University of Illinois Chicago
842 W Taylor St, ERF, Chicago, IL 60607
ORCID: 0000-0003-3595-3664
Email: kouros@uic.edu
Nazmul Arefin Khan, Ph.D.
Computational Transportation Engineer
Transportation Systems and Mobility Group
Vehicle and Mobility Systems Department
Argonne National Laboratory
9700 S Cass Ave, Lemont, IL 60439
E-mail: nazmul.arefin@anl.gov
ORCID: 0000-0003-3175-6882
Yantao Huang, Ph.D.
Computational Transportation Engineer
Transportation Systems and Mobility Group
Vehicle and Mobility Systems Department
Argonne National Laboratory
9700 S Cass Ave, Lemont, IL 60439
E-mail: yantao.huang@anl.gov
ORCID: 0000-0001-9218-0122
Joshua A. Auld, Ph.D.
Manager
Transportation Systems and Mobility Group
Vehicle and Mobility Systems Department
Argonne National Laboratory
9700 S Cass Ave, Lemont, IL 60439
E-mail: jauld@anl.gov
Phone: (630) 252–5460
Abstract
Neighborhood crime exposure is an underexplored yet critical factor in shaping household vehicle decisions. While prior research has acknowledged the influence of crime and perceived safety on various dimensions of travel behavior, no studies to date have directly examined how crime affects vehicle ownership and type choice. This study incorporates crime indicators into a longitudinal, life course–oriented framework. Drawing on a 10-year retrospective survey in Chicago, the analysis links household vehicle purchases to socio-demographics, life events, housing changes, personal attitudes, built environment characteristics, and neighborhood crime data. A nested logit model is employed to jointly capture both the decision to purchase a vehicle and the subsequent choice of vehicle body type and vintage. Results reveal that exposure to crime, particularly violent and property crimes, increases the likelihood of vehicle purchase and shifts preferences toward used vehicles. In addition, life events such as marriage, childbirth, job changes, and residential moves significantly influence both purchase and type decisions, while built environment factors such as residential density and transit access reduce the probability of vehicle acquisition. This study advances the literature by integrating personal, spatial, psychological, and safety-related factors in a unified modeling framework. Findings highlight the importance of integrating neighborhood safety with transportation and land-use policies to effectively reduce car dependency. The results also suggest that life transitions offer critical opportunities for mobility interventions and sustainable vehicle choice programs.
Keywords:
Vehicle Purchase
Neighborhood Crime
Built Environment
Life Events
Body Type
Vintage
Retrospective Longitudinal Dataset
1. Introduction
Understanding household vehicle purchase behavior is central to transportation research due to its implications for travel demand, sustainability, and vehicle markets (Bhat & Sen, 2006). Private cars dominate travel in most U.S. urban and suburban areas. In 2021, over 282 million vehicles traveled more than 3.1 trillion miles (Bureau of Transportation Statistics, 2023). According to the 2022 National Household Travel Survey (NHTS), only 10.7 million of 129 million households lacked vehicle access, while those with three or more vehicles rose from 3 million in 1960 to 26 million in 2022 (Bricka et al., 2024). Vehicle purchase decisions also tie closely to where people live and work, as part of broader household mobility strategies (Rashidi et al., 2011). Understanding why, when, and what types of vehicles people acquire is key for forecasting travel demand and policy planning.
Traditional vehicle ownership models rely on cross-sectional data and emphasize demographic and economic factors such as income, age, gender, and household size (Cao et al., 2006; Cirillo & Liu, 2013; Karlaftis & Golias, 2002). However, these models often overlook the dynamic and context-sensitive nature of vehicle decisions. Recent research highlights the influence of life events such as marriage, childbirth, job changes, and residential moves on vehicle acquisitions and type choices (Clark et al., 2016; Khan & Habib, 2021b; Oakil, 2016). Ignoring these transitions oversimplifies behavior and limits policy insights. Spatial context also shapes ownership decisions; built environment features like density, land use mix, and transit access can reduce vehicle needs or promote smaller, efficient cars (Anowar et al., 2016; Clark et al., 2016; Rashidi et al., 2011). Attitudinal factors, including environmental concern and car dependence, are also increasingly integrated into models, recognizing the psychological side of mobility behavior.
Crime has emerged as a key factor influencing travel behavior. Studies show it significantly reduces walking and biking, particularly among vulnerable groups such as women (Bennett et al., 2007; Choobchian et al., 2024; Loukaitou-Sideris, 2005). Crime and safety concerns also affect mode choice, discouraging walking, biking, and public transit use, especially in high-crime areas or at night (Appleyard & Ferrell, 2017; Ferrell & Mathur, 2012; Heinen, 2023; Kim et al., 2007; Zhang, 2016; Lu et al., 2025). Higher neighborhood crime rates have also been linked to lower e-scooter usage (Heydari et al., 2022; Tuli et al., 2021). Overall, crime, whether experienced, perceived, or feared, shapes how, when, and whether people travel. Given its widespread influence across multiple aspects of travel behavior, crime may also play an important role in shaping vehicle purchase decisions. If this influence is ignored, there is a risk of overestimating the effects of other variables such as income, urban form, or attitudes. Accounting for crime-related factors may therefore lead to more accurate interpretations of the determinants of vehicle ownership and choice.
This study makes two main contributions. First, it examines how neighborhood crime influences households’ vehicle purchase and type decisions, a topic not addressed in prior research. Second, it jointly models a broad set of factors, including sociodemographic characteristics, housing, life events, built environment, attitudes, and neighborhood crime, within a unified longitudinal and life course-oriented framework. Few studies use such an approach to capture the combined effects of life transitions, neighborhood context, safety concerns, and personal attitudes. These contributions provide a more nuanced understanding of vehicle transaction behavior. We estimate a nested logit model using a rich retrospective dataset collected in 2024 from 228 households in Chicago. The survey includes 10 years of life history data on vehicle transactions, life events, household changes, attitudes, and perceptions. These records are spatially linked to built environment indicators from the EPA’s Smart Location Database and crime data from the City of Chicago. The model captures a two-stage decision process: vehicle acquisition, followed by the choice of body type (sedan, SUV, and truck) and vintage (new or used).
The rest of this paper is as follows. Section 2 reviews the existing literature on vehicle transactions. Section 3 describes the data sources, including the retrospective survey, neighborhood built environment indicators, and crime data from the City of Chicago. Section 4 outlines the nested logit modeling framework used to capture the sequential structure of vehicle acquisition and type choice decisions. Section 5 presents and discusses the estimation results. Section 6 concludes with key findings and suggestions for future research.
2. Literature Review
Life events are key drivers of vehicle transaction decisions. Oakil et al. (2014), using 21 years of retrospective data in the the Netherlands, found that childbirth, cohabitation, job changes, and residential relocation significantly affect car acquisition and disposal. Gu et al. (2021), used a latent class competing risks model and showed that younger households are more responsive to household composition changes, while older ones react more to employment and relocation. A recent study (Hossain & Fatmi, 2025) using a joint hazard and probit model found that childbirth accelerates first car purchase and increases preferences for SUVs and new vehicles. Clark et al. (2016), using a large UK panel, showed that gaining employment, having children, or getting a driver’s license strongly predict vehicle acquisition. Similarly, Khan and Habib (2021b), using a panel-based model with retrospective data from Halifax, found that childbirth and employment changes significantly affect both the timing and type of vehicle transactions. These studies confirm the critical role of life transitions in shaping vehicle ownership.
The influence of spatial and environmental context on vehicle ownership and choice has been widely examined. Anowar et al. (2016), using pseudo-panel data from Montreal, found that households in dense, transit-rich neighborhoods were less likely to increase vehicle holdings over time. Potoglou (2008) examined vehicle type choice in Hamilton, Canada, and found that land-use diversity and transit access reduced SUV ownership, suggesting that compact urban design discourages large vehicles preferences. In a related study, authors found that higher-density, mixed-use areas lowered the likelihood of owning multiple vehicles, reinforcing the link between urban form and car dependency (Potoglou & Kanaroglou, 2008). Chen et al. (2021), using Cincinnati data, observed that residents of denser neighborhoods tended to own smaller cars and drive less. Likewise, Caulfield (2012), analyzing Irish census data, found that better transit access and higher residential density reduced the likelihood of multiple vehicle ownership. A Czech study (Ščasný & Urban, 2011) also found lower car ownership in dense, transit-accessible areas.
Recent vehicle choice models increasingly incorporate attitudes, lifestyles, and behavioral differences. Choo and Mokhtarian (2004), using a latent variable model, found that status-oriented individuals prefer luxury vehicles, while those with environmental concerns favor smaller or efficient cars. Iogansen et al. (2025) used an Integrated Choice and Latent Variable model on panel data from during and after the COVID-19 pandemic, showing that vehicle acquisitions and disposals were shaped by novelty-seeking, life events, and remote work shifts. Khan and Habib (2021a) added to this research area by estimating latent class logit models that showed how travel context (e.g., traveling alone or with others) interacts with socio-demographics to shape vehicle type preferences. Together, these studies show growing recognition that attitudes, preferences, and behavioral context are essential to vehicle transaction and choice models.
A growing body of research shows that crime significantly influence various aspects of travel behavior. Kim et al. (2007) found that high crime near transit stations in St. Louis reduces walking, especially among women who prefer to be dropped off in high-crime areas or at night due to safety concerns. Similarly, Loukaitou-Sideris (2005) emphasized how fear of victimization leads women to avoid walking or adjust their travel times and routes. Choobchian et al. (2024) found that crime directly and indirectly reduces commute walking in Chicago. In Austin, Zhang (2016) demonstrates that crime can mediate the relationship between compact land use and transit ridership, with elevated crime levels deterring bus use. Tuli et al. (2021) and Heydari et al. (2022) reported that higher neighborhood crime significantly reduces e-scooter usage and increases crash risk. Other studies confirm that both actual and perceived crime discourage walking, biking, and shared mobility (Appleyard & Ferrell, 2017; Heinen, 2023).
Despite the growing recognition of crime as a significant factor in shaping travel behavior, to the best of our knowledge, no prior studies have examined its influence on vehicle purchase and type decisions. This study addresses that gap by investigating the impact of neighborhood crime exposure on households’ vehicle acquisition behavior, including both the decision to buy and the choice of vehicle type. Using a longitudinal, life course–oriented approach, we control for a wide range of factors, including sociodemographic characteristics, housing attributes, built environment features, and individual attitudes. The analysis draws on data from a 10-year period (2015–2024) in Chicago and employs a nested logit model, where the upper level captures the decision to purchase a vehicle and the lower level models the choice of vehicle body type and vintage (new vs. used). By incorporating crime as a contextual factor in this framework, the study offers new insights into the behavioral and spatial dimensions of vehicle transaction decisions.
3. Data
The core dataset for this study is an original survey we conducted in the Chicago metropolitan area in 2024, collecting detailed individual- and household-level information, as described in subsection 3.1. We combined this dataset with two external sources: built environment data from the U.S. Environmental Protection Agency’s Smart Location Database (SLD) (U.S. Environmental Protection Agency, n.d.) and crime data from the City of Chicago’s open data portal (City of Chicago, n.d.). The subsections that follow describe the survey instrument and these two external datasets.
3.1 Survey
This study uses data from an online survey administered via Qualtrics in the Chicago metropolitan area between March and April 2024. A total of 557 responses were initially received. The survey collected detailed information on respondents’ current conditions and changes from 2015 to 2024, covering demographics, household characteristics, housing, employment, and attitudes related to driving, and environmental concerns, among others. It also gathered data on vehicle transactions during this period, including purchase timing, body type, and model year, as well as major life events such as marriage, childbirth, job changes, and residential moves. A rigorous data cleaning process excluded incomplete responses, failed attention checks, and implausibly short completions. In addition, because the crime data only covers the City of Chicago, responses from outside city boundaries were also removed. The final sample included 228 valid responses, which were used for analysis.
2.1.1. Individual and household characteristics
To evaluate representativeness relative to the overall population, Table 1 includes statistics from the American Community Survey (ACS) for the City of Chicago. In terms of gender, 57.9% of respondents are female, compared to 51.2% in the population. The age distribution skews slightly toward middle-aged adults, with 49.6% aged 35 to 54. Household income in the sample largely mirrors the general population, except in the highest income bracket: 4.39% of sampled households report incomes of $200,000 or more, compared to 14.1% citywide. Racial composition is also diverse: 35.1% of respondents are White and 31.1% are African American, compared to 44.5% and 22.9%, respectively. Additionally, 21.1% identify as Hispanic or Latino. Educational attainment is higher in the sample, with 50.4% holding a bachelor’s degree or higher, compared to 41.1% citywide. In terms of marital status, 45.6% are married and 43.0% are single (never married). A large majority (88.2%) hold a driver’s license, closely matching the 85% rate in the general population.
This comparison highlights some discrepancies between the sample and the broader population, especially in gender, education, and income. To address these imbalances and improve generalizability, sampling weights were constructed using all variables listed in Table 1 through an iterative proportional fitting (commonly known as raking) (Kalton & Flores-Cervantes, 2003), which adjusts the sample to align with marginal population distributions from the ACS. These weights were then applied during modeling to ensure that estimates more accurately reflect citywide patterns.
Variable | Category | Unweighted Survey Sample | ACS Population | |||||
|---|---|---|---|---|---|---|---|---|
Share | Share | |||||||
Gender | Female | 57.9% | 51.2% | |||||
Male | 42.1% | 48.8% | ||||||
Age | 18 to 24 | 4.4% | 8.1% | |||||
25 to 34 | 21.5% | 20.3% | ||||||
35 to 44 | 29.4% | 18.2% | ||||||
45 to 54 | 20.2% | 16.1% | ||||||
55 to 64 | 13.1% | 15.7% | ||||||
65 and more | 11.4% | 21.6% | ||||||
Household Income | Less than $15,000 | 11.0% | 10.2% | |||||
$15,000 to $24,999 | 5.7% | 5.8% | ||||||
$25,000 to $34,999 | 8.8% | 6.5% | ||||||
$35,000 to $49,999 | 11.8% | 9.7% | ||||||
$50,000 to $74,999 | 17.5% | 14.9% | ||||||
$75,000 to $99,999 | 14.0% | 11.8% | ||||||
$100,000 to $149,999 | 19.8% | 17.5% | ||||||
$149,000 to $199,999 | 7.0% | 9.5% | ||||||
$200,000 and over | 4.4% | 14.1% | ||||||
Race | White | 35.1% | 44.5% | |||||
African American | 31.1% | 22.9% | ||||||
Asian | 7.5% | 7.8% | ||||||
Other | 26.3% | 24.8% | ||||||
Ethnicity | Hispanic or Latino | 21.1% | 26.2% | |||||
Other | 88.9% | 73.8% | ||||||
Education | Less than a high school diploma | 4.8% | 11.3% | |||||
High school graduates, no college | 13.2% | 23.2% | ||||||
Some college credit but no degree | 15.4% | 18.8% | ||||||
Associate or technical school degree | 16.2% | 5.7% | ||||||
Bachelor's or undergraduate degree | 25.0% | 24.6% | ||||||
Graduate degree | 25.4% | 16.5% | ||||||
Marital Status | Single (never married) | 43.0% | 41.9% | |||||
Married | 45.6% | 42.2% | ||||||
Divorced | 6.6% | 8.8% | ||||||
Separated but still legally married | 1.3% | 1.9% | ||||||
Widowed | 3.5% | 5.2% | ||||||
Driver’s License | Yes | 88.2% | 85.0% | |||||
No | 11.8% | 15.0% |
2.1.2. Vehicle transactions
Beyond demographic and household characteristics, the dataset includes 10 years of vehicle purchase records. Among 2,280 person-year observations, 473 involved a vehicle acquisition. For each purchase, respondents reported the body type (sedan, SUV, or truck) and model year. Vehicles were classified as new (0–1 year old) or used (more than 1 year old). Figure 1 illustrates the distribution of vehicle purchase transactions by body type and vintage. SUVs accounted for about 50% of purchases, sedans for 43.5%, and trucks for 6.5%. New vehicle purchases were more common among SUVs (39.8%) than sedans (26.7%), suggesting a preference for newer SUVs. Due to the small number of truck purchases (only 31 transactions), this category is not divided by vintage in the analysis, resulting in five total alternatives in the nested logit model: new sedan, used sedan, new SUV, used SUV, and truck.
Fig. 1
Number of vehicle acquisitions reported over the 10-year period (2015–2024)
2.1.3. Life events
Figure 2 summarizes the self-reported life events experienced by respondents between 2015 and 2024. These events fall into three main categories. The first includes household composition changes, such as moving in with a spouse or partner and the addition of children through birth or adoption. The second involves housing transitions, such as moving to a larger or smaller home. The third covers economic and educational changes, including starting or leaving a job, receiving a promotion, taking on or repaying debt, and starting or completing a level of education.
Fig. 2
Life events reported by respondents from 2015 to 2024
2.1.4. Attitudes
To examine how personal attitudes influence vehicle purchasing decisions, the survey included a series of statements rated on a 5-point Likert scale. These statements addressed views on driving, transportation, and environmental issues. Since respondents could not reliably recall their attitudes over the past decade, the items focused on relatively stable traits, such as environmental orientation, which are less likely to change over time. Exploratory Factor Analysis (EFA) (Brown, 2015) was used to simplify the data and uncover latent dimensions. The data’s suitability for EFA was confirmed by the Kaiser-Meyer-Olkin (KMO) measure and Bartlett’s Test. Factors were extracted using Principal Axis Factoring method, guided by eigenvalues greater than one and a scree plot. Since correlations among factors were expected, Promax oblique rotation with Kaiser normalization was applied to refine the factor structure (Costello & Osborne, 2005).
The analysis produced a KMO value of 0.732 and a significant Bartlett’s test result, confirming the data’s suitability for factor analysis. Table 2 reports the results based on 12 attitudinal statements, revealing four factors: Pro-car (preference for or pride in private car use), Pro-environmental (support for environmental protection and public transit), Pro-high-density (positive views toward dense urban living), and Pro-commuting (satisfaction with commuting). These factors are meaningfully correlated. For example, Pro-car and Pro-environmental attitudes are negatively related (–0.198), indicating that individuals who prioritize environmental protection or public transit are less favorable toward private car use.
State your agreement with any of the following statements | Component | ||||
|---|---|---|---|---|---|
Pro-car | Pro-environmental | Pro-high-density | Pro-commuting | ||
I take pride in owning a car | 0.833 | ||||
Owning a vehicle is necessary when you have a family | 0.621 | ||||
I perceive my driving skills as better than the average | 0.618 | ||||
Transit systems are critical for a region | 0.712 | ||||
I limit my driving because it's bad for air quality | 0.642 | ||||
I consider global warming a major concern | 0.588 | ||||
I would use transit more often if it is more convenient | 0.547 | ||||
I prefer to live in the inner city | 0.818 | ||||
Increasing residential density is good | 0.663 | ||||
Proximity to shops/services is important to me | 0.644 | ||||
My commute offers a good transition between home and work | 0.710 | ||||
My commute makes me feel relaxed | 0.684 | ||||
Component correlation (r) | |||||
Pro-car | 1.000 | ||||
Pro-environmental | -0.198 ** | 1.000 | |||
Pro-high-density | -0.121 | 0.633 ** | 1.000 | ||
Pro-commuting | 0.055 | -0.112 | -0.103 | 1.000 | |
KMO = 0.732 | Sig. of Bartlett’s test = 0.000 | ||||
| ** Correlation is significant at the 0.01 level (2-tailed) | |||||
3.2 Residential Context
Each respondent’s residential location was linked to their Census Block Group (CBG) using the 2021 EPA Smart Location Database (SLD) to capture neighborhood-level influences on vehicle purchase decisions. Due to limited historical data, the 2021 values were used for all years, based on the assumption that land use characteristics change slowly over time. Figure 3 shows spatial distributions of four key built environment variables across Chicago. Low-income workers are concentrated in the western and southern areas, while high-income workers are more common in the downtown core and northern neighborhoods. Residential density is generally higher in the eastern and northern areas. Transit coverage is extensive citywide, with the highest service frequency near major CTA and METRA lines and closer to downtown.
Fig. 3
Spatial variation of workers’ income, density, and transit characteristics in Chicago
3.3 Crime Data
This study uses publicly available crime data from the City of Chicago to incorporate neighborhood safety into the analysis of vehicle-related behavior. Figure 4 shows the total number of reported crimes from 2015 to 2024. A notable decline occurred during the COVID-19 pandemic, with total incidents falling by about 21.7% between 2019 and 2021. However, crime levels rebounded in subsequent years, returning to pre-pandemic levels by 2023.
Fig. 4
Total crimes in Chicago by year (2015–2024)
The dataset includes 31 crime types such as assault, homicide, weapons violations, gambling, and narcotics. To assess their influence on vehicle acquisition, we first counted the number of each crime type in every census tract and normalized these counts by the tract’s population in that year. Then, we applied EFA based on crime rates from 2015 to 2024 to identify broader crime categories. Crime types with extraction communalities below 0.3 were excluded, as low shared variance could undermine the factor structure’s validity.
The final factor solution retained 18 crime indicators grouped into three latent factors, as shown in Table 3. The first factor, Violent and Property Crime, includes offenses involving physical harm or property damage, such as weapons violations, assault, homicide, robbery, and motor vehicle theft. The second factor, Theft and Deception, reflects non-violent crimes aimed at financial gain and intrusion. The third, Public Disorder and Vice, includes crimes that disrupt public order or violate social norms.
Figure 5 illustrates the spatial distribution of these three crime factor scores (which are standardized scores, approximately normally distributed with a mean of zero) across Chicago in 2024. As shown, the southern and western areas experience higher levels of Violent and Property Crime and Public Disorder and Vice. In contrast, Theft and Deception offenses are concentrated in the downtown area, likely due to greater foot traffic, economic activity, and the presence of tourists, offices, and retail establishments. We linked this crime data to respondents’ residential locations over the past decade to evaluate its effect on vehicle purchase decisions.
Crime Type | Component | ||
|---|---|---|---|
Violent and Property Crime | Theft and Deception | Public Disorder and Vice | |
Weapons violation | 0.912 | ||
Assault | 0.876 | ||
Criminal damage | 0.858 | ||
Battery | 0.769 | ||
Other offense | 0.742 | ||
Motor vehicle theft | 0.736 | ||
Homicide | 0.654 | ||
Offense involving children | 0.578 | ||
Robbery | 0.533 | ||
Arson | 0.516 | ||
Burglary | 0.445 | ||
Theft | 0.971 | ||
Deceptive practice | 0.829 | ||
Criminal trespass | 0.433 | ||
Narcotics | 0.805 | ||
Gambling | 0.603 | ||
Public peace violation | 0.580 | ||
Interference with public officer | 0.546 | ||
Component correlation (r) | |||
Violent and Property Crime | 1.000 | ||
Theft and Deception | 0.407 ** | 1.000 | |
Public Disorder and Vice | 0.505 ** | 0.278 ** | 1.000 |
KMO = 0.940 | Sig. of Bartlett’s test = 0.000 | ||
** Correlation is significant at the 0.01 level (2-tailed) | |||
4. Methodology
To jointly model the decision to purchase a vehicle and, if so, the choice of vehicle type (body type and vintage), this study uses a Nested Logit (NL) model to analyze vehicle transaction behavior over a 10-year retrospective period. Specifically, the model captures two decisions: 1) whether to buy a vehicle (Buy vs. Not Buy), and 2) conditional on purchase, the choice among five alternatives: new sedan, used sedan, new SUV, used SUV, and truck. The nested structure allows for correlation in unobserved factors among alternatives within the same nest, relaxing the restrictive independence of irrelevant alternatives (IIA) assumption inherent in standard multinomial logit models. This nested structure assumes that the utility of buying a vehicle consists of both the utility of the purchase decision itself and the utility of the specific type of vehicle selected.
Let the choice set be partitioned into J nests, where each nest Bj contains alternatives
. In the nested logit model, the utility Uni that individual n associates with alternative i in nest j is:
Equation (1) | |
| Where is the observed utility component and is an unobserved random component. The model assumes that the unobserved terms follow a generalized extreme value (GEV) distribution, which allows for correlation among alternatives within the same nest. The probability that an individual 𝑛 chooses alternative 𝑖 within nest 𝑗 is given by: | |
Equation (2) |
A
Fig. 5
Spatial distribution of crime factors across Chicago census tracts in 2024
 |  |
| Where is the probability of choosing nest 𝑗 (i.e., buy vs. not buy), and is the conditional probability of choosing alternative 𝑖 within nest 𝑗 (e.g., new SUV, used sedan, etc.). The conditional probability is defined as: | |
Equation (3) |
And the marginal probability of choosing nest 𝑗 is:
Equation (4) | |
| Where is the inclusive value (IV) parameter and Inj is the inclusive value or logsum term for nest 𝑗, which summarizes the utility of all alternatives in the nest: | |
Equation (5) | |
| The IV parameters 𝜆𝑗 must satisfy 0 < 𝜆𝑗 ≤ 1 for consistency with random utility maximization. A statistically significant 𝜆𝑗 between 0 and 1 supports the appropriateness of the nested structure, indicating that alternatives within a nest share common unobserved characteristics. The log-likelihood function for the nested logit model is specified as: | |
Equation (6) |
Where N is the number of individuals in the sample, yni is an indicator variable equal to 1 if individual 𝑛 chooses alternative 𝑖, and 0 otherwise. This log-likelihood function is maximized with respect to the model parameters using maximum likelihood estimation. For more technical details, refer to (Ben-Akiva & Lerman, 1985; Train, 2009).
5. Results
Table 4 presents the estimation results of the nested logit model, which examines both the decision to purchase a vehicle and, if so, the choice of body type and vintage. The discussion below examines the influence of socioeconomic factors, life events, neighborhood characteristics, neighborhood crime, and attitudes on vehicle purchase behavior. Some variables were lagged to capture delayed effects on vehicle decisions. A lag effect refers to the delayed impact of a variable, meaning the decision to buy or change a vehicle may not occur in the same year the event takes place, but rather in the following year(s) as households adapt to new circumstances. The model was estimated using panel-structured data in NLOGIT 6 (LIMDEP). The IV parameter for the lower-level nest is 0.313, indicating moderate correlation and supporting the nested structure.
Explanatory Variable | Upper-Level Model | Lower-Level Model | |||||
|---|---|---|---|---|---|---|---|
Buy | New Sedan | Used Sedan | New SUV | Used SUV | Truck | ||
Built Env. | Moved to an area with lower gross residential density (housing units/acre), (1-year lag) | 1.144 (0.007) | |||||
Moved to an area with higher transit service frequency per capita | -1.226 (0.008) | ||||||
Number of occupied housing units in CBG | -1.797 (0.009) | ||||||
Crime | Violent and Property Crime factor in census tract | 0.833 (0.008) | -1.257 (0.000) | ||||
Moved to an area with higher Violent and Property Crime factor | 0.914 (0.002) | 0.592 (0.027) | |||||
Attitude | Pro-environmental | -0.246 (0.001) | |||||
Pro-car | 0.676 (0.003) | ||||||
Alternative Specific Constant | -2.543 (0.000) | 0.318 (0.364) | 2.344 (0.000) | 1.587 (0.002) | 2.621 (0.000) | Ref. | |
IV Parameter | 0.313 (0.024) | ||||||
Number of Observations | 2596 | ||||||
Log likelihood | -1420.81 | ||||||
McFadden Pseudo R2 | 0.437 | ||||||
* The numbers in parentheses represent p-values | |||||||
| 5.1 Sociodemographic Factors | |||||||
Sociodemographic variables significantly influence vehicle purchase decisions. Females are less likely to purchase trucks, while Black respondents show a stronger preference for trucks, possibly due to occupational needs or perceived utility. Older adults (age 55+) are less likely to purchase a vehicle, perhaps reflecting lower mobility needs, vehicle sufficiency, or retirement-related lifestyle changes. However, when they purchase, they prefer new SUVs, likely due to the comfort, safety, and visibility these vehicles offer, features often valued by older adults. In contrast, respondents under age 35 favor used sedans, reflecting budget-conscious decisions and a focus on affordability and practicality over vehicle features or condition.
Household income is another significant determinant. Households earning under $50,000 a year prefer used sedans, used SUVs, and trucks, suggesting that lower-income households are more likely to purchase second-hand or utility vehicles, possibly for cost savings or work-related needs. Conversely, households with annual incomes over $150,000 are more likely to choose new sedans and new SUVs, reflecting a greater willingness to invest in newer, higher-end vehicles. Income changes also influence purchase behavior. An income increase is associated with a higher likelihood of purchasing a new sedan, while a lagged income increase predicts a higher likelihood of choosing a truck, possibly reflecting the purchase of a second or specialized vehicle after a period of financial improvement.
5.2 Life Events
Forming a new household by moving in with a spouse or partner is strongly associated with a higher likelihood of purchasing both a new sedan and a new SUV, suggesting that household consolidation often leads to greater mobility needs and a preference for newer, more reliable vehicles. Similarly, adding children to the household (through birth or adoption) significantly increases the probability of vehicle acquisition.
Employment transitions also play a key role. Starting a new job increases the likelihood of choosing a used sedan, likely due to budget constraints during early employment. The lagged effect of job entry is associated with a preference for new sedans, which may reflect greater income stability. Leaving a job (whether due to quitting, termination, or retirement) is associated with selecting a used SUV. This may reflect a lifestyle shift toward comfort and versatility for leisure or family use after employment.
Job promotions show a dual effect. A promotion in the current year increases the probability of selecting both new and used SUVs, suggesting that career advancement encourages larger or more prestigious vehicle purchases, possibly as status symbols or for comfort. The lagged effect of a promotion is positively associated with vehicle acquisition, indicating delayed effects of financial improvement. Taking on new debt, such as a mortgage or loan, increases the likelihood of vehicle purchase, possibly reflecting new mobility needs (e.g., due to relocation to a more car-dependent area) or greater financial confidence and willingness to take on additional long-term commitments.
Housing transitions significantly affect vehicle purchase decisions. Households moving from owning to renting are more likely to buy a vehicle, which may reflect changes in residential location, parking availability, or freed-up capital from home equity. Moving from an apartment to a single-family home increases the likelihood of choosing a new SUV and, to a lesser extent, a used SUV, highlighting the influence of residential form. Moving into a larger residence, with a two-year lag, is associated with a higher likelihood of vehicle purchase, possibly reflecting growing household needs or relocation to lower-density, car-dependent neighborhoods. In contrast, downsizing to a smaller residence consistently has negative effects on vehicle behavior. In the upper-level model, it lowers the likelihood of buying a vehicle at both one- and two-year lags, with a stronger effect in the first year, suggesting that the impact of downsizing on vehicle purchase decisions is strongest shortly after the move and gradually diminishes over time. In the lower-level model, downsizing reduces the probability of choosing new SUVs, used SUVs, and trucks, suggesting that smaller homes may discourage ownership of larger vehicles due to space limits, lifestyle changes, or financial considerations. Overall, life event coefficient results align with previous studies which showed that marriage, childbirth, job changes, and residential relocation affect vehicle ownership decisions choices (Clark et al., 2016; Khan & Habib, 2021b; Oakil et al., 2016; Gu et al., 2021; Hossain & Fatmi, 2025).
5.3 Built Environment Characteristics
A higher share of low-wage workers in a CBG is associated with a lower likelihood of selecting new SUVs, possibly reflecting consumer preferences that do not prioritize newer, larger vehicles. Conversely, living in areas with a greater proportion of high-wage workers increases the probability of choosing new SUVs, likely reflecting greater neighborhood affluence, parking availability, and status-driven consumption. In addition, moving to a neighborhood with a higher share of low-wage workers reduces the likelihood of vehicle purchase with a one-year lag, possibly due to lower mobility needs, tighter travel budgets, or increased reliance on alternatives such as walking or transit.
Moving to neighborhoods with lower residential density significantly increases the likelihood of vehicle purchase, reflecting greater transportation needs in less urbanized, car-dependent areas where public transit, walking, or cycling are less feasible. In contrast, relocating to areas with higher transit service frequency per capita reduces the likelihood of vehicle acquisition, suggesting that better transit access provides viable alternatives to private vehicle ownership. Living in neighborhoods with a higher number of occupied housing units is negatively associated with truck selection, likely due to space limitations, parking challenges, and urban design that discourages large vehicles.
5.4 Crime Exposure
Crime exposure, a novel addition in this research area, shows notable effects on both vehicle acquisition and vehicle type choice. A higher Violent and Property Crime factor in a respondent’s census tract is associated with a greater likelihood of choosing used sedans and a lower likelihood of selecting new SUVs. One possible explanation is that individuals in high-crime areas may prefer less conspicuous or lower-value vehicles, such as used sedans, to reduce the risk of theft or vandalism. At the same time, they may avoid expensive or attention-drawing vehicles like new SUVs due to safety or financial concerns.
Additionally, relocating to a neighborhood with higher Violent and Property Crime significantly increases the likelihood of purchasing a vehicle. This suggests that households moving into less secure environments may become more reliant on private vehicles to reduce exposure to public or active transportation options perceived as riskier. In such contexts, private car ownership may serve as a coping strategy to enhance personal safety and mobility control. These findings highlight an important behavioral response to perceived neighborhood safety and demonstrate the value of incorporating crime metrics into transportation behavior models.
5.5 Attitudinal Factors
Attitudes also play an important role in vehicle transaction behavior. A pro-environmental orientation reduces the likelihood of vehicle purchase, suggesting that environmental concern may deter car ownership and encourage the use of more sustainable transportation modes. Conversely, individuals with strong pro-car attitudes are more likely to select new SUVs, likely reflecting the symbolic and functional value they place on private vehicles, especially those associated with power, comfort, or status.
6. Policy implications
The findings of this study highlight how personal safety, life transitions, and spatial context jointly shape vehicle acquisition, suggesting that policies seeking to reduce car dependency must extend beyond conventional economic or land-use levers. The strong influence of neighborhood crime on vehicle ownership demonstrates that safety concerns can override urban design and transit accessibility advantages. In neighborhoods where violent or property crime is prevalent, households tend to acquire vehicles (often used ones) as a form of self-protection. This pattern implies that improving neighborhood safety is not only a public security goal but also a transportation demand management strategy. Integrating crime prevention measures such as better lighting, community policing, and secure transit facilities into mobility planning could reduce residents’ perceived need for private vehicles and enhance trust in non-auto modes.
The results also reaffirm that built environment and land-use characteristics strongly affect vehicle ownership. Households in low-density and transit-poor areas are significantly more likely to purchase vehicles, underscoring the value of compact, mixed-use, and transit-oriented development. Yet, the interaction between safety and built environment is critical: even in dense or transit-rich areas, high crime rates may weaken the car-reducing effects of urban form. Therefore, transport and urban design strategies should be coupled with safety improvements to ensure that compact neighborhoods and transit corridors remain viable alternatives to auto reliance.
Life transitions emerged as another pivotal factor influencing both the decision to buy and the type of vehicle purchased. Events such as marriage, childbirth, job changes, or moving mark moments when households reassess mobility needs. Policymakers could leverage these windows of behavioral change to encourage more sustainable choices. Targeted interventions (such as mobility incentives for new parents, employer-provided transit benefits for job starters, or informational campaigns for recent movers) can help guide decisions before car-dependent habits take root. Likewise, expanding flexible mobility options such as car-sharing, on-demand services, and family-friendly public transit can address emerging needs without prompting new car acquisitions.
Finally, the heterogeneity in preferences across income and attitudinal groups calls for a balanced approach that considers both equity and environmental outcomes. Lower-income households, often constrained to used vehicles, would benefit from affordable and safe alternatives to private ownership, including reliable transit and car-sharing programs in underserved areas. Meanwhile, higher-income households, whose preferences lean toward new and larger vehicles, could be steered toward cleaner technologies through incentives for electric or low-emission models. Shaping attitudes through education and community engagement can further reinforce these structural measures.
Together, these insights suggest that reducing vehicle dependence requires an integrated framework that connects safety, land use, life-course dynamics, and social attitudes. A coordinated policy approach that simultaneously enhances neighborhood security, designs compact and connected urban environments, and targets interventions during key life transitions offers the most promising path toward sustainable, equitable, and resilient urban mobility.
7. Conclusion
This study advances the understanding of household vehicle purchase behavior by combining 10 years of retrospective survey data with contextual information on the built environment and crime exposure in the City of Chicago. By estimating a nested logit model that jointly captures vehicle acquisition decisions and subsequent vehicle type choices, the analysis provides a behaviorally rich and context-sensitive perspective on how households respond to both life transitions and external neighborhood conditions.
Two key contributions distinguish this research. First, this study is the first to explicitly incorporate neighborhood crime exposure into vehicle choice modeling. The findings reveal that both static crime levels and changes in perceived neighborhood safety significantly influence whether a household purchases a vehicle and the type of vehicle selected, offering new insights into how safety concerns shape vehicle purchase decisions. Second, the study jointly estimates vehicle purchase and type choice decisions using sociodemographic characteristics, housing attributes, life events, built environment features, individual attitudes, and crime exposure through a longitudinal dynamic approach. This integration reflects the complex and multidimensional nature of vehicle purchase behavior, which is deeply embedded in broader lifestyle and spatial contexts.
Our findings highlight that vehicle purchase behavior is strongly influenced by life events, income, age, and housing transitions. Younger and lower-income households prefer used sedans, while older and higher-income households favor new SUVs. Moving to low-density or affluent neighborhoods increases vehicle acquisition, while better transit access reduces it. Importantly, crime exposure significantly affects decisions: households in high-crime areas tend to prefer less conspicuous vehicles and are more likely to purchase vehicles after moving to unsafe neighborhoods, reflecting a safety-driven reliance on private cars. Attitudes also matter, with pro-environmental views reducing car ownership and pro-car attitudes increasing new SUV selection.
While this study focused on household and contextual factors, it did not include detailed vehicle-specific attributes such as price, capacity, or fuel economy. Including these variables in future research could enrich vehicle type choice modeling. However, collecting this information retrospectively over a 10-year period, alongside life events and attitudinal data, presents substantial challenges in a survey administration context. In addition, incorporating fuel type, especially distinguishing between electric and conventional vehicles, could improve the analysis by linking vehicle choice behavior to broader emerging mobility goals.
The results have several practical implications. Policies aimed at reducing car dependency and promoting sustainable transportation must recognize the role of neighborhood design, perceived safety, and life course events in shaping household decisions. For instance, investments in transit-oriented development may be less effective in areas where residents perceive high crime risk, which appears to push individuals toward vehicle ownership. Similarly, life transitions such as employment changes, household formation, or residential moves are key windows for intervention, where incentives or nudges could shift preferences toward more sustainable vehicle types, or away from car ownership altogether.
A
Acknowledgement
“This work was supported by the U.S. Department of Energy (DOE) Vehicle Technologies Office (VTO) under the Transportation Systems and Mobility Tools Core Maintenance, an initiative of the Energy Efficient Mobility Systems (EEMS) Program. The submitted manuscript has been created by the UChicago Argonne, LLC, Operator of Argonne National Laboratory (Argonne). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated under Contract No. DE-AC02-06CH11357. Erin Boyd, a DOE Office of Energy Efficiency and Renewable Energy (EERE) manager, played an important role in establishing the project concept, advancing implementation, and providing guidance. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the Government.”
A
Funding
This work was funded by the U.S. Department of Energy (DOE) Vehicle Technologies Office (VTO) under the Transportation Systems and Mobility Tools Core Maintenance initiative of the Energy Efficient Mobility Systems (EEMS) Program.
A
A
Author Contribution
Masoud ArfaeiYazdiPour: Investigation, Methodology, Visualization, Writing the Original Draft, Review & Editing. Ali Mohammadi: Conceptualization, Methodology, Review & Editing. Sina Asgharpour: Conceptualization. Nooshin Javidi: Visualization. Abolfazl (Kouros) Mohammadian: Conceptualization, Survey Design, Review & Editing. Nazmul Arefin Khan: Conceptualization, Review & Editing. Yantao Huang: Review & Editing. Joshua Auld: Conceptualization.
Competing interests
The authors have no relevant financial or non-financial interests to disclose.
A
Data Availability
The data that support the findings of this study were collected under a research agreement between the University of Illinois Chicago and Argonne National Laboratory. Due to privacy considerations and institutional data-sharing restrictions, the dataset is not publicly available.
References
Anowar, S., Eluru, N., Miranda-Moreno, L.F.: Analysis of vehicle ownership evolution in Montreal, Canada using pseudo panel analysis. Transportation. 43(3), 531–548 (2016). https://doi.org/10.1007/S11116-015-9588-Z
Appleyard, B.S., Ferrell, C.E.: The Influence of crime on active & sustainable travel: New geo-statistical methods and theories for understanding crime and mode choice. J. Transp. Health. 6, 516–529 (2017). https://doi.org/10.1016/J.JTH.2017.04.002
Ben-Akiva, M.E., Lerman, S.R.: Discrete Choice Analysis: Theory and Application to Travel Demand. MIT Press, Cambridge, MA (1985)
Bennett, G.G., McNeill, L.H., Wolin, K.Y., Duncan, D.T., Puleo, E., Emmons, K.M.: Safe To Walk? Neighborhood Safety and Physical Activity Among Public Housing Residents. PLoS Med. 4(10) (2007). https://doi.org/10.1371/JOURNAL.PMED.0040306
Bhat, C.R., Sen, S.: Household vehicle type holdings and usage: an application of the multiple discrete-continuous extreme value (MDCEV) model. Transp. Res. Part. B: Methodological. 40(1), 35–53 (2006). https://doi.org/10.1016/J.TRB.2005.01.003
Bricka, S., Reuscher, T., Schroeder, P., Fisher, M., Beard, J., Sun, X.L.: Summary of Travel Trends: 2022 National Household Travel Survey. United States. Federal Highway Administration. Office of Policy and Governmental Affairs (2024)
Brown, T.A.: Confirmatory Factor Analysis for Applied Research Second Edition. Basic Statistics and Epidemiology, 89–100. (2015)
Bureau of Transportation Statistics: Transportation statistics annual report 2023. U.S. Department of Transportation. (2023). https://doi.org/10.21949/1529944
Cao, X., Mokhtarian, P.L., Handy, S.L.: Neighborhood design and vehicle type choice: Evidence from Northern California. Transp. Res. Part. D: Transp. Environ. 11(2), 133–145 (2006). https://doi.org/10.1016/J.TRD.2005.10.001
Caulfield, B.: An examination of the factors that impact upon multiple vehicle ownership: The case of Dublin, Ireland. Transp. Policy. 19(1), 132–138 (2012). https://doi.org/10.1016/J.TRANPOL.2011.08.003
Chen, N., Akar, G., Gordon, S.I., Chen, S.: Where do you live and what do you drive: Built-environmental and spatial effects on vehicle type choice and vehicle use. Int. J. Sustainable Transp. 15(6), 444–455 (2021). https://doi.org/10.1080/15568318.2020.1762950
Choo, S., Mokhtarian, P.L.: What type of vehicle do people drive? The role of attitude and lifestyle in influencing vehicle type choice. Transp. Res. Part. A: Policy Pract. 38(3), 201–222 (2004). https://doi.org/10.1016/J.TRA.2003.10.005
Choobchian, P., Mohammadi, A., Zou, B., Hair, J.F., Valinejad, M., Shin, J., Sriraj, P.S.: Calibrating walkability indicators for commute walk trips: A structural equation modeling approach. Transp. Res. Part. A: Policy Pract. 179, 103896 (2024). https://doi.org/10.1016/J.TRA.2023.103896
Cirillo, C., Liu, Y.: Vehicle Ownership Modeling Framework for the State of Maryland: Analysis and Trends from 2001 and 2009 NHTS Data. J. Urban. Plan. Dev. 139(1), 1–11 (2013). https://doi.org/10.1061/(ASCE)UP.1943-5444.0000128
Clark, B., Chatterjee, K., Melia, S.: Changes in level of household car ownership: the role of life events and spatial context. Transportation. 43(4), 565–599 (2016). https://doi.org/10.1007/S11116-015-9589-Y
Costello, A.B., Osborne, J.W.: Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis. Practical Assess. Res. Evaluation. 10(1) (2005). https://doi.org/10.7275/JYJ1-4868
City of Chicago. (n.d.). Crimes – 2001 to present. Retrieved July 9: from (2025). https://data.cityofchicago.org/Public-Safety/Crimes-2001-to-Present/ijzp-q8t2/about_data
Ferrell, C.E., Mathur, S.: Neighborhood Crime and Travel Behavior: An Investigation of the Influence of Neighborhood Crime Rates on Mode Choice – Phase II. (2012). https://doi.org/10.21949/1503647
Gu, G., Feng, T., Yang, D., Timmermans, H.: Modeling dynamics in household car ownership over life courses: a latent class competing risks model. Transportation. 48(2), 809–829 (2021). https://doi.org/10.1007/S11116-019-10078-8
Heinen, E.: The impact of crime and crime-related experiences, worries, and perceptions on travel behavior. Transp. Res. Part. F: Traffic Psychol. Behav. 96, 265–284 (2023). https://doi.org/10.1016/J.TRF.2023.06.014
Heydari, S., Forrest, M., Preston, J.: Investigating the association between neighbourhood characteristics and e-scooter safety. Sustainable Cities Soc. 83, 103982 (2022). https://doi.org/10.1016/J.SCS.2022.103982
Hossain, M.S., Fatmi, M.R.: Modeling Households’ First Vehicle Purchase Timing and Vehicle Type Choices. Transp. Res. Rec. (2025). https://doi.org/10.1177/03611981251326098
Iogansen, X., Lee, Y., Malik, J.K., Bunch, D.S., Circella, G.: Household Vehicle Transactions During and after the Pandemic – a Panel Data Analysis. (2025). https://doi.org/10.2139/SSRN.5295664
Kalton, G., Flores-Cervantes, I.: Weighting Methods. Journal of Official Statistics, 19, 81–97. (2003)
Karlaftis, M., Golias, J.: Automobile ownership, households without automobiles, and urban traffic parameters: Are they related? Transp. Res. Rec. 1792, 29–35 (2002). https://doi.org/10.3141/1792-04
Khan, N.A., Habib, M.A.: Development of households’ vehicle choice models for travel accompanying arrangements. Transp. Eng. 4, 100071 (2021a). https://doi.org/10.1016/J.TRENG.2021.100071
Khan, N.A., Habib, M.A.: When do households make vehicle transactions? What type of vehicle households choose after transaction events? Travel Behav. Soc. 22, 32–47 (2021b). https://doi.org/10.1016/J.TBS.2020.08.002
Kim, S., Ulfarsson, G.F., Hennessy, J.T.: Analysis of light rail rider travel behavior: Impacts of individual, built environment, and crime characteristics on transit access. Transp. Res. Part. A: Policy Pract. 41(6), 511–522 (2007). https://doi.org/10.1016/J.TRA.2006.11.001
Loukaitou-Sideris, A.: Is it safe to walk here. Res. Women’s Issues Transp., 102. (2005)
Lu, D., Yazdiani, A., Fraser, T., Tayarani, M., Gao, H.O.: Machine learning in spatiotemporal PM2.5 hotspots analysis for better targeted mitigation strategies. Transp. Res. Part. D: Transp. Environ. 148, 105017 (2025). https://doi.org/10.1016/J.TRD.2025.105017
Oakil, A.T.M.: Securing or sacrificing access to a car: Gender difference in the effects of life events. Travel Behav. Soc. 3, 1–7 (2016). https://doi.org/10.1016/J.TBS.2015.03.004
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, 889–904 (2014). https://doi.org/10.1007/s11116-013-9507-0
A
Potoglou, D.: Vehicle-type choice and neighbourhood characteristics: An empirical study of Hamilton, Canada. Transp. Res. Part. D: Transp. Environ. 13(3), 177–186 (2008). https://doi.org/10.1016/J.TRD.2008.02.002Potoglou, D., Kanaroglou, P.S.: Modelling car ownership in urban areas: a case study of Hamilton, Canada. J. Transp. Geogr. 16(1), 42–54 (2008). https://doi.org/10.1016/J.JTRANGEO.2007.01.006
Rashidi, T.H., Mohammadian, A., Koppelman, F.S.: Modeling interdependencies between vehicle transaction, residential relocation and job change. Transportation. 38(6), 909–932 (2011). https://doi.org/10.1007/S11116-011-9359-4
Ščasný, M., Urban, J.: Passenger Car Ownership in the Czech Republic. International Days of Statistics and Economics, Prague, September 22, 2011. (2011)
Train, K.E.: Discrete Choice Methods with Simulation. Cambridge University Press (2009). https://doi.org/10.1017/CBO9780511805271
Tuli, F.M., Mitra, S., Crews, M.B.: Factors influencing the usage of shared E-scooters in Chicago. Transp. Res. Part. A: Policy Pract. 154, 164–185 (2021). https://doi.org/10.1016/J.TRA.2021.10.008
U.S. Environmental Protection Agency. (n.d.). Smart location mapping. Retrieved July 6, from (2025). https://www.epa.gov/smartgrowth/smart-location-mapping#SLD
Zhang, W.: Does compact land use trigger a rise in crime and a fall in ridership? A role for crime in the land use–travel connection. Urban Stud. 53(14), 3007–3026 (2016). https://doi.org/10.1177/0042098015605222