TRB Annual Meeting

Beyond Vehicle Counts: Developing an Individual-Level Car Access Index for Improved
Travel Behavior Modeling
--Manuscript Draft--

Manuscript Number: TRBAM-26-01273

Full Title: Beyond Vehicle Counts: Developing an Individual-Level Car Access Index for
Improved Travel Behavior Modeling

Order of Authors: Baldev Giri

Suman Kumar Mitra, Ph.D

Arna Nishita Nithila

Abstract: A substantial body of research has examined household vehicle ownership and its
influence on travel behavior. Common measures such as the number of household
vehicles, vehicles per person, and vehicles per licensed driver are calculated at the
household level and assume equal access for all members. However, this assumption
often fails. For example, in a two-person household with one vehicle and only one
licensed driver, the driver has significantly greater access to the vehicle than the non-
driver. Traditional measures do not capture these differences and may introduce bias
when applied to models focused on individual travel behavior. To address this
limitation, this study develops the Car Access Index (CAI), a person-level measure that
reflects individual access to vehicles based on household vehicle availability, driver’s
licensing status, and support from licensed household members. The CAI is validated
through scenario analysis, descriptive statistics, and statistical modeling using data
from the 2022 National Household Travel Survey. Results show that using CAI in place
of traditional indicators improves the accuracy of models estimating car trip frequency
and vehicle miles traveled, offering a more realistic view of individual mobility within
households. CAI provides researchers and practitioners with a strong foundation for
developing more targeted and effective transportation strategies.

Manuscript Classifications: Transportation Planning; Travel Behavior; Transportation System User Communities;
Accessibility and/or Mobility; Travel Behavior

Additional Information:

Question Response

The total word count limit is 7500 words 7467

including tables. Each table equals 250
words and must be included in your count.
Papers exceeding the word limit may be
rejected. My word count is:

Article Type: Presentation

Powered by Editorial Manager® and ProduXion Manager® from Aries Systems Corporation
Manuscript

Baldev, Mitra, and Nithila

1 Beyond Vehicle Counts: Developing an Individual-Level Car Access Index for

2 Improved Travel Behavior Modeling
3
4
5
6 Baldev Giri
7 Department of Civil Engineering
8 University of Arkansas, Fayetteville, Arkansas, 72701
9 Email: [email protected]
10
11 Suman Mitra, Ph.D., Corresponding Author
12 Associate Professor
13 Department of Civil Engineering
14 University of Arkansas, Fayetteville, Arkansas, 72701
15 Email: [email protected]
16
17 Arna Nishita Nithila
18 Department of Civil Engineering
19 University of Arkansas, Fayetteville, Arkansas, 72701
20 Email: [email protected]
21
22 Word Count: 6,717 + 3 table (250 words per table) = 7,467 words
23
24
25 Submission Date: July 29th, 2025
26
 Baldev, Mitra, and Nithila

1 ABSTRACT
2 A substantial body of research has examined household vehicle ownership and its influence on
3 travel behavior. Common measures such as the number of household vehicles, vehicles per person, and
4 vehicles per licensed driver are calculated at the household level and assume equal access for all
5 members. However, this assumption often fails. For example, in a two-person household with one vehicle
6 and only one licensed driver, the driver has significantly greater access to the vehicle than the non-driver.
7 Traditional measures do not capture these differences and may introduce bias when applied to models
8 focused on individual travel behavior. To address this limitation, this study develops the Car Access
9 Index (CAI), a person-level measure that reflects individual access to vehicles based on household vehicle
10 availability, driver’s licensing status, and support from licensed household members. The CAI is
11 validated through scenario analysis, descriptive statistics, and statistical modeling using data from the
12 2022 National Household Travel Survey. Results show that using CAI in place of traditional indicators
13 improves the accuracy of models estimating car trip frequency and vehicle miles traveled, offering a more
14 realistic view of individual mobility within households. CAI provides researchers and practitioners with a
15 strong foundation for developing more targeted and effective transportation strategies.
16
17 Keywords: Car access, household vehicle count, car-sufficiency, driving status, vehicle miles traveled

2
 Baldev, Mitra, and Nithila

1 INTRODUCTION
2 Over the past five decades, an extensive body of research has examined car ownership, car
3 mobility, and carlessness, highlighting their wide-ranging impacts on economic opportunity, social
4 inclusion, and travel behavior (1–3). Despite these efforts, a critical gap remains: much of this research
5 focuses on household-level metrics, overlooking the complex, unequal dynamics of individual car access
6 within households. Understanding how individuals access and utilize vehicles within a household is
7 essential, as travel decisions are shaped not only by personal preferences or needs but also by shared
8 constraints and opportunities related to intra-household resource distribution (4).
9 Traditionally, researchers have relied on household-level indicators such as the absolute number
10 of vehicles, the vehicle-to-household size ratio, and the vehicle-to-driver ratio as proxies for individual
11 private vehicle access within a household (5). While convenient, these aggregate measures can obscure
12 meaningful disparities in car access. For example, consider two households: Household A has one vehicle
13 and two adults, only one of whom holds a driver’s license; Household B also has one vehicle and two
14 adults, both licensed drivers. The vehicle-to-household-member ratio is the same in both cases (0.5),
15 which underestimates the advantage of having two licensed drivers in Household B who can operate a
16 vehicle independently. Meanwhile, the vehicle-to-driver ratio is 1.0 for Household A and 0.5 for
17 Household B, suggesting that Household A is better off; however, this is not the case. This metric
18 overestimates access in Household A by failing to account for the burden placed on the sole driver, who
19 must meet the transportation needs of both individuals. In contrast, the drivers in Household B can share
20 the vehicle more flexibly and independently; if needed, one can rent or borrow a car while the other is
21 using the household vehicle. Thus, both indicators, vehicle-to-member and vehicle-to-driver, can be
22 misleading: the former underestimates the role of driving eligibility, while the latter overestimates
23 mobility in households with only one licensed driver.
24 Further complicating matters, such metrics typically ignore driver eligibility, license restrictions,
25 and functional ability. A person may hold a license with restrictions, such as limits on driving distance or
26 daylight-only use, due to age, vision, or medical conditions. These restricted licenses are especially
27 common among older adults, teenagers, and individuals with disabilities, and they significantly affect
28 one’s ability to use a car independently. Similarly, individuals without a household vehicle may still
29 maintain regular car access through borrowing, employment-based arrangements, or short-term rentals,
30 scenarios missed by traditional indicators.
31 These limitations underscore the need for a more nuanced approach to measuring car access, one
32 that centers on the individual rather than the household and considers both the availability of vehicles and
33 the functional ability to use them. To fill this gap, we propose the development of a Car Access Index
34 (CAI): a continuous, person-level measure ranging from 0 to 1 that reflects an individual’s practical
35 access to private vehicles. The CAI accounts for driver’s license status (including restrictions), household
36 vehicle resources, and intra-household sharing dynamics to offer a more accurate picture of personal car
37 mobility.
38 This paper advances two key hypotheses: (i) individual car access can be meaningfully captured
39 through a single composite measure, and (ii) the proposed CAI outperforms traditional household-level
40 metrics in predicting individual- and household-level travel behavior, such as car-trip frequency and
41 vehicle miles traveled (VMT). To test these hypotheses, we first developed the CAI through a scenario-
42 based methodology guided by clear validation criteria. We then applied the CAI to the 2022 National
43 Household Travel Survey (NHTS) (6) dataset to explore its association with travel behavior and socio-
44 demographic disparities. Finally, we compared the performance of the CAI with conventional metrics in
45 modeling individual and household travel outcomes.
46 The remainder of this paper is structured as follows. The next section discusses the conceptual
47 framework and reviews the existing literature. The formulation and calculation of the CAI are discussed
48 in the following section, which is then followed by a case study and results. Finally, the conclusion
49 section provides a summary, discusses limitations, and outlines future research opportunities.
50

3
 Baldev, Mitra, and Nithila

1 CONCEPTUAL FRAMEWORK
2 We have defined car access as an individual’s ability to use a private motor vehicle when desired
3 or needed for travel, reflecting both the availability of vehicles within the household and the individual’s
4 ability and opportunity to utilize them. The index designed to capture this concept of car access is referred
5 to as the Car Access Index (CAI). At its core, an individual’s car access is primarily determined by two
6 key factors: the availability of household vehicles and the individual’s driving capability. These two
7 components serve as the foundational elements in the construction of the Car Access Index (CAI), as they
8 directly influence a person’s potential to use a private vehicle when needed or desired. However, to
9 comprehensively capture access dynamics at the individual level, it is essential to consider the car access
10 of those who don’t have driving capability or who lack a valid driver’s license. For non-drivers, the
11 ability to utilize a household vehicle often depends on support from other licensed household members.
12 Therefore, we introduce the concept of a support resource, which can be provided by licensed drivers
13 within the household who can facilitate car-based mobility for non-drivers. This inclusion acknowledges
14 that even without direct driving capability, individuals may still fulfill their car mobility through the
15 assistance of others. Using trip frequency data from the 2022 NHTS, we observe that non-drivers
16 undertake approximately 30 percent as many car trips and person-miles traveled as drivers. Consequently,
17 the maximum attainable CAI value for non-drivers is capped at 0.3.
18 While driving capability and vehicle availability are objective and straightforward to measure, the
19 notion of support introduces an element of subjectivity and interdependence into the framework. A non-
20 driver’s access is contingent not only on the presence of vehicles but also on the presence and capability
21 of others in the household to provide transportation. It is assumed that licensed drivers maintain full
22 access even when supporting others, as they typically accommodate non-drivers without sacrificing their
23 own travel.
24 Drivers with restrictions are individuals whose driver’s licenses include specific limitations, such
25 as requiring corrective lenses, daytime-only driving, or the use of adaptive equipment, due to age,
26 medical, or testing-related factors. The car access of drivers with restrictions is limited because they
27 might not be able to access the vehicle due to their specific driving-related limitations. A study of older
28 drivers found that those with license restrictions reduced their average weekly mileage by 36% suggesting
29 that restricted drivers have substantially lower access or usage (7). Therefore, in this study, we have
30 assumed that the car access of drivers with restrictions is less than that of the drivers who drive without
31 restrictions.
32 Thus, the operationalization of the Car Access Index (CAI) rests on three core components:
33 vehicle availability, driving capability, and support from other household members (Figure 1). Among
34 these, vehicle availability is considered the most critical determinant of car access. The second most
35 influential factor is an individual’s driving capability, as captured by the possession of a valid driver’s
36 license. The third component of the CAI accounts for support-dependent access among non-drivers.
37 However, the value of this support is conditional on household vehicle ownership. In households without
38 vehicles, the likelihood of support for non-drivers is significantly reduced, primarily due to logistical and
39 financial barriers to renting or borrowing vehicles for the sole purpose of supporting others. To capture
40 this distinction, the support component for non-drivers is weighed more heavily in vehicle-owning
41 households and reduced in vehicle-less households. The reduction reflects an assumed, minimal
42 likelihood of car-based support being realized through alternative means such as rentals.

4
 Baldev, Mitra, and Nithila

1
2 Figure 1: Theoretical Framework for Car Access Index
3
4 LITERATURE REVIEW
5 This review synthesizes existing literature focusing on four areas: the impacts of limited vehicle
6 access within a household, intra-household car access allocation, the use of vehicle counts as a measure of
7 individual car access, and the literature on individual car access. Although many studies have focused on
8 car-ownership and carlessness, no study has specifically attempted to develop an intrahousehold car
9 access index.
10 A large body of work has examined the determinants and implications of household car
11 ownership, highlighting the influence of socio-economic factors, household composition, and life events
12 (2, 8, 9). Several studies have shown that limited vehicle access disproportionately affects low-income
13 households, women, youth, and older adults. Blumenberg et al. (6) introduced the concept of “car-
14 deficient households”, those with fewer vehicles than licensed drivers, and demonstrated how such
15 deficits constrain trip-making and increase reliance on slower modes. Jeon et al. (12) linked limited
16 access to reduced residential mobility. Vandersmissen et al. (12) emphasized disparities by gender and
17 age, while Scheiner and Holz-Rau (13), along with Morris et al. (14), demonstrated how restricted access
18 reduces activity participation and the time spent outside the home. Qin et al. (15) further linked driving
19 cessation among older adults to increased social isolation. Together, these studies underscore the need to
20 measure car access at the individual level to inform equitable transportation policy.
21 Research has shown that car access is often negotiated within households and varies by role,
22 employment status, and gender. Beige and Cyganski (13) found that vehicles are more likely to be
23 allocated to males, retirees, and rural residents. Subsequent modeling by Beige et al. (14) simulated
24 household-level car assignment but did not propose a person-level access metric. Olde Kalter and Geurs
25 (4) demonstrated that intra-household interactions account for a substantial share of variance in car use.
26 Scheiner and Holz‐Rau (15) emphasized that employment more strongly predicts car access than
27 domestic responsibility. Oakil (18), Vance, and Iovanna (19) found that life events and car scarcity

5
 Baldev, Mitra, and Nithila

1 disproportionately reduce women’s access to cars. Blumenberg et al. (16) similarly reported gendered and
2 age-based disadvantages in vehicle use and driving continuity.
3 Vehicle ownership remains a foundational variable in travel behavior research and is frequently
4 used to categorize households in major surveys, such as the NHTS. While these counts correlate with
5 travel behavior (e.g., trip frequency, mode choice), they offer a coarse view of access. Ratios such as
6 vehicles per household member or per licensed driver have been used to adjust for household composition
7 (17). To adjust for household composition, researchers have proposed refined metrics, such as vehicles
8 per household member and vehicles per licensed driver (20, 21). However, these measures still assume
9 uniform access within households. They overlook individual variation, support dynamics, and sharing
10 burdens, leading to biased estimates in mode choice and trip generation models (22, 23).
11 Recognizing the inadequacies of ownership-based proxies, researchers have called for multi-
12 dimensional and person-centered approaches to measuring mobility. Several frameworks have emerged
13 that assess transportation disadvantage, accessibility, and individual-level inclusion by incorporating not
14 just vehicle counts but also mode availability, infrastructure access, and user capabilities. For example,
15 accessibility indices evaluate the number of destinations reachable within a given time frame or mode,
16 while transportation disadvantage indices identify vulnerable populations based on combined socio-
17 economic and mobility criteria (19). The capability approach has also gained traction in recent years.
18 Rooted in Amartya Sen’s framework (20), this approach assesses individuals’ ability to convert available
19 resources, such as vehicles, transit, or social support, into actual mobility opportunities. Applied to
20 suburban contexts and housing developments, this framework emphasizes functional access over mere
21 ownership or trip counts (11). Despite these advances, there remains no simple, standardized, or scalable
22 metric that directly captures individual car access within. This gap motivates the development of the Car
23 Access Index (CAI), a continuous, person-level metric designed to fill this void and provide a more
24 equitable and actionable tool for both research and policy.
25
26 FORMULATION OF THE INDEX
27 Calculation Method
28 The CAI captures intra-household variability by incorporating three key dimensions: (1) household
29 vehicle sufficiency, (2) individual driving capability, and (3) the availability of driving support to non-
30 drivers from other licensed household members. The CAI score for each ranges from 0 (no access) to 1
31 (maximum access), reflecting a composite estimate of car access potential. For a household with H
32 members, the CAI is calculated separately for each individual as a weighted combination of the three
33 components. The weights vary based on driving status and license restrictions:
34
35 For drivers without restrictions,
36 𝐶𝐶𝐶𝐶𝐶𝐶 = 0.7𝐶𝐶 + 0.3𝑅𝑅
37
38 For drivers with restrictions,
39 𝐶𝐶𝐶𝐶𝐶𝐶 = 0.7𝐶𝐶 + 0.75 × 0.3𝑅𝑅
40
41 For non-drivers,
42 𝐶𝐶𝐶𝐶𝐶𝐶 = 0.15𝐶𝐶 + 0.15𝑆𝑆
43
44 Alternatively, a unified formula expresses all cases as:
45 𝐶𝐶𝐶𝐶𝐶𝐶 = 𝑅𝑅 (0.7𝐶𝐶 + 0.3 − 0.075𝑄𝑄) + (1 − 𝑅𝑅)(0.15𝐶𝐶 + 0.15𝑆𝑆)
46
47 Where:
48 • C: Car sufficiency (vehicle availability relative to number of drivers)
49 • R: Binary indicator for whether the individual is a licensed driver (1 = yes; 0 = no)

6
 Baldev, Mitra, and Nithila

1 • Q: Restriction flag for licensed drivers (Q = 1 if restricted; Q = 0 if unrestricted)

2 • S: Support from other licensed household members
3
4 Component Definitions
5 Car Sufficiency (C)
6 This component captures the vehicle sufficiency in the household, relative to the number of household
7 drivers. This component considers the fact that only the household drivers drive the available car in a
8 household.
𝑉𝑉
9 𝐶𝐶 = � min � 𝐿𝐿 � , 1), 𝑤𝑤ℎ𝑒𝑒𝑒𝑒 𝐿𝐿 > 0
0 𝑤𝑤ℎ𝑒𝑒𝑒𝑒 𝐿𝐿 = 0
10
11 Where:
12 V = number of vehicles in the household
13 L = number of licensed drivers in the household
14 This formula caps sufficiency at 1 and assigns a value of 0 if there are no licensed drivers in the
15 household.
16
17 Individual Driving Capability (R)
18 The variable captures whether an individual has a driving license, assuming that a person with a driving
19 license is capable of driving.
1 𝑖𝑖𝑖𝑖 𝑡𝑡ℎ𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
20 𝑅𝑅 = �
0 𝑜𝑜𝑜𝑜ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
21
22 Support from Licensed Household Members (S)
23 This component approximates the potential for driving assistance from other household members. It is
24 only defined for multi-person households (H ≥ 2).
𝐿𝐿
𝑚𝑚𝑚𝑚𝑚𝑚 � , 1� 𝑖𝑖𝑖𝑖𝑖𝑖 > 𝐿𝐿
25 𝑆𝑆 = � 𝐻𝐻 − 𝐿𝐿
0 𝑖𝑖𝑖𝑖 𝐻𝐻 = 𝐿𝐿
26 Where:
27 H = Total number of household members
28 This reflects the likelihood that a non-driver has access to support from other drivers in the household.
29
30 Weighting Schemes and Scenario Analysis
31 The weighting scheme for the Car Access Index (CAI) was developed through a two-stage,
32 theory-informed calibration process. In the first stage, a conceptual hierarchy was established based on
33 the literature and the proposed framework. This hierarchy emphasized the greater importance of
34 household vehicle sufficiency (C) over individual licensing status (R) in facilitating car use, while also
35 recognizing the supplemental but essential role of support from co-resident drivers (S) for non-licensed
36 individuals. Based on this foundation, provisional weights were assigned: for licensed drivers, the CAI
37 was calculated as 0.7 × C + 0.3 × R, giving vehicle sufficiency twice the weight of individual licensing.
38 For non-drivers, equal weights of 0.15 were assigned to both C and S to reflect their partial yet
39 meaningful dependence on both factors. Drivers with restricted licenses were further discounted by 25%
40 (0.75 × 0.3 × R), in alignment with U.S. licensing laws that limit night and highway driving. In the second
41 stage, these provisional weights were empirically validated using a scenario-based analysis guided by
42 predefined validation criteria (Table 1).
43 To verify the robustness and internal consistency of the Car Access Index (CAI), a
44 comprehensive set of synthetic household scenarios was generated. These scenarios captured all
45 combinations of household sizes ranging from 1 to 10 members, vehicle ownership from 0 to 10 vehicles,

7
 Baldev, Mitra, and Nithila

1 and varying configurations of licensed and unlicensed individuals, including distinctions between drivers
2 with and without driving restrictions. Using Python, these combinations were systematically constructed,
3 resulting in 1,815 unique individual-level scenarios. For each scenario, the CAI was computed using the
4 proposed methodology and evaluated against the nine validation criteria outlined in Table 1 . The
5 analysis confirmed that all scenarios met the validation criteria, thereby establishing the theoretical
6 soundness and practical reliability of the CAI across a wide range of household structures. Given the
7 consistency observed across households of up to 10 individuals, it can be reasonably inferred that the
8 index will remain valid in larger and more complex configurations as well.
9
10 Table 1: Validation Criteria for CAI Calculation
No Criteria Description
As the number of household vehicles (V) increases, CAI should not decrease
Monotonicity with respect
1 for any household member. For unrestricted drivers, CAI should saturate at
to the Vehicle Ownership
1.0 once vehicle sufficiency (V ≥ L) is met.
Dominance of Driver Given fixed household size (H), vehicle count (V), and licensed drivers (L),
2
Status a licensed driver should have a higher CAI than a non-driver.

Monotonicity with respect With fixed household size (H) and vehicle count (V), CAI should increase
3
to the Licensed Drivers or remain the same as more licensed drivers (L) are added, provided V ≥ L.

Premium for Vehicle CAI values should be strictly higher in households with at least one vehicle
4
Ownership (V > 0) than in those with none (V = 0), assuming L > 0.
Penalty for Driving Drivers with restrictions should receive a lower CAI than unrestricted
5
Restrictions drivers in the same household.
6 Boundedness of CAI All CAI scores must lie in the closed interval [0, 1].
Upper-Bound Consistency When V ≥ L, unrestricted drivers should have CAI = 1.0. Restricted drivers
7
under Sufficiency should not exceed CAI = 0.925.
Recognition of Complete In households with no vehicles (V = 0) and no licensed drivers (L = 0), CAI
8
Deprivation = 0 for all individuals.
Saturation of Support for For non-drivers, CAI increases with L until support term (S) = 1. After that,
9
Non-Drivers more licensed drivers do not improve CAI unless more vehicles are added.
11
12 Figures 2 to 5 present heat maps illustrating how key household structure elements such as
13 household size, driver count, and vehicle count influence car access differently for drivers and non-
14 drivers. Figure 2 indicates that for licensed drivers, CAI is effectively a function of the number of drivers
15 in the household, rather than the total number of residents. With exactly one driver, the index remains
16 stable at 0.90 across all household sizes. However, access diminishes slightly with each additional driver,
17 dropping to 0.87 with two drivers, 0.84 with three, and down to approximately 0.61 when ten drivers
18 share the same vehicle fleet.
19 As shown in Figure 3, for non-drivers in households without a licensed driver, the CAI is zero
20 across all scenarios. The introduction of a single licensed driver raises the index to 0.29 in a two-person
21 household, although this benefit diminishes with household size, falling to around 0.15–0.17 when ten
22 individuals rely on one driver. Further, adding more drivers yields diminishing returns: the CAI peaks
23 around 0.27–0.25 for households with three to six drivers and levels off at ~0.23 even with nine. This
24 plateau reflects constraints introduced by shared vehicle use, where growing driver counts are not
25 matched by additional vehicles.
26 Figure 4 displays that when vehicle supply is ample, car access is near optimal. As for drivers, a
27 single-person household with one vehicle achieves an average CAI of 0.96. Any scenario in which the
28 number of vehicles matches or exceeds household size maintains similarly high access levels. In contrast,

8
 Baldev, Mitra, and Nithila

1 access deteriorates quickly when vehicles are scarce: one vehicle for two residents lowers the CAI to
2 0.79, while the same vehicle supporting ten individuals reduces it to 0.47. In the complete absence of
3 vehicles, the CAI bottoms out at 0.26, indicating that holding a driver’s license alone confers roughly a
4 quarter of full access.
5

6
7 Figure 2: Variation in CAI for Drivers by Household Size And Number of Licensed Drivers
8

9
10 Figure 3: Variation in CAI for Non-Drivers by Household Size and Number of Licensed Drivers
11

9
 Baldev, Mitra, and Nithila

1
2 Figure 4: Variation in CAI for Drivers by Household Size and Household Vehicle Count
3

4 Figure 5 presents the Car Access Index (CAI) for non-drivers as a function of household size and
5 household vehicle count. The values remain low overall, reflecting limited direct mobility access for
6 individuals without a license. However, CAI increases with both household size and vehicle count,
7 indicating that non-drivers in larger, vehicle-rich households benefit more from intra-household support.
8 Notably, CAI values plateau around 0.22–0.23 in high-vehicle and high-household-size scenarios,
9 showing diminishing returns beyond a certain threshold of household resources. In contrast, non-drivers
10 in one-person or zero-vehicle households have no access, with CAI values of 0.
11 Taken together, the heat maps reflect the conceptual framework and confirm that the CAI
12 behaves in accordance with its underlying validation criteria.
13

14
15 Figure 5: Variation in CAI for Non-Drivers by Household Size and Household Vehicle Count

10
 Baldev, Mitra, and Nithila

1 CASE STUDY
2 To assess the empirical validity and practical relevance of the Car Access Index (CAI), this case
3 study applies the index to a nationally representative dataset and examines its relationship with individual
4 travel behavior across diverse socio-demographic and geographic contexts.
5
6 Data
7 This study utilizes data from the 2022 National Household Travel Survey (6), a nationally
8 representative dataset that captures the daily travel behavior of individuals and households across the
9 United States, to validate the application of the developed car access index. The selection of independent
10 variables for modeling reflects established determinants of car mobility access, as identified in the
11 previous literature. The dependent variables are the number of car trips and vehicle miles traveled at the
12 person level. Demographic characteristics include binary indicators for gender (male, female) and
13 categorical indicators for age groups: 16–24, 25–44, 45–64, and 65 years or older. Race and ethnicity are
14 represented through mutually exclusive categories: White, Black, and other racial groups. Socio-
15 economic characteristics include educational attainment, categorized as high school or less, some college
16 or associate’s degree, and bachelor’s degree or higher. Household income is divided into three brackets:
17 lower, middle, and higher, following the categorization of the Pew Research Center (21). Mobility-related
18 characteristics incorporate driver’s license status (driver vs. non-driver), employment status (worker vs.
19 non-worker), and household vehicle availability. Household structure is captured using life-cycle-based
20 classifications: younger adults without children, retired adults without children, and households with
21 children. Land use variables include an urban–rural binary indicator and four metropolitan area
22 classifications: large metropolitan areas with rail service, large metropolitan areas without rail, smaller
23 metropolitan areas, and non-metropolitan areas.
24
25 Results and Discussion
26 Descriptive Statistics
27 The analysis began with the calculation of the Car Access Index (CAI) for all individual
28 respondents in the NHTS 2022 dataset. Before computing the index, relevant variables were extracted
29 from the person-level files, specifically household size, number of drivers in the household, and the
30 respondent’s driving status. The CAI was computed for each individual using the proposed formula.
31 Figure 6 displays the resulting distribution, which reveals a highly right-skewed pattern with a substantial
32 concentration at CAI = 1, indicating full vehicle access. Smaller clusters are observed at mid- and low-
33 range CAI values, though these are relatively sparse. This distribution suggests that while most
34 individuals in the U.S. enjoy full vehicle access, a notable segment experiences limited or no access,
35 highlighting disparities in mobility resources across households. To facilitate further analysis, individuals
36 were grouped into three equal-width CAI categories: Low (0–0.33), Medium (0.34–0.66), and High
37 (0.67–1.0). Descriptive statistics were calculated within each group (Table 2), revealing variation in
38 travel behavior and socio-demographic characteristics across CAI levels.

11
 Baldev, Mitra, and Nithila

1
2 Figure 6: Distribution of the Car Access Index (CAI)
3
4 Table 2: Descriptive Statistics by CAI Groups with equal intervals
Variables CAI = 0 0 < CAI <= 0.33 0.33 < CAI < 0.66 < CAI CAI = 1
(N=339) (N=1146) =0.66 (N=1144) <=0.99(N=637) (N=10434)
Continuous Variables
CAI 0.0 0.27 0.64 0.79 1.0
Total number of trips 0.97 1.01 1.71 1.57 1.96
Number of car trips 0.09 0.61 1.39 1.43 1.82
Person miles traveled 2.86 9.82 21.95 20.57 30.03
Household size 1.81 3.13 2.81 4.45 2.52
Household vehicle count 0.1 1.32 1.05 2.51 2.33
Age 50.61 44.42 50.24 42.53 52.36
Working Status
Worker 0.38 0.28 0.45 0.56 0.58
Non-worker 0.62 0.72 0.55 0.44 0.42
Driving Status
Driver 0.0 0.2 1.0 1.0 1.0
Non-driver 1.0 0.8 0.0 0.0 0.0
Race
Race -White 0.59 0.71 0.79 0.74 0.86
Race - Black 0.21 0.14 0.07 0.07 0.06
Race - Others 0.2 0.14 0.13 0.19 0.08
Residential Location
Urban 0.94 0.85 0.85 0.82 0.77
Rural 0.06 0.15 0.15 0.18 0.23
Msa_1m_rail 0.37 0.31 0.26 0.28 0.2
Msa_1m_no_rail 0.23 0.29 0.32 0.31 0.32

12
 Baldev, Mitra, and Nithila

Msa_less_than_1m 0.36 0.37 0.38 0.37 0.42

Not_in_msa 0.04 0.04 0.05 0.03 0.06
Income
Lower income 0.79 0.43 0.33 0.2 0.21
Middle income 0.15 0.43 0.51 0.53 0.56
Higher income 0.06 0.14 0.16 0.27 0.23
Household Structure
Younger adults - no child 0.54 0.28 0.28 0.13 0.32
Retired adults - no child 0.29 0.31 0.39 0.2 0.39
Household with a child 0.17 0.4 0.33 0.68 0.29
Education
High school or less 0.44 0.53 0.25 0.36 0.18
Some college or associate’s 0.25 0.22 0.27 0.29 0.3
degrees
Bachelor’s or higher degree 0.31 0.26 0.48 0.35 0.52
1
2 Table 2 stratifies the sample into five ordinal CAI categories, ranging from no access (CAI = 0)
3 to full access (CAI = 1), and reveals a strong, stepwise relationship between car access and daily mobility
4 outcomes. Mean daily trip frequency, car usage, and distance traveled increase monotonically with CAI.
5 Individuals with no car access take fewer than one trip per day (0.97), with nearly no car trips (0.09),
6 whereas those with full access average nearly two trips per day (1.96), most of which are by car (1.82).
7 Household-level patterns mirror these findings. Average vehicle count rises from nearly zero
8 among car-deficient households to 2.33 in fully sufficient ones. Labor force participation also increases
9 with CAI: only 38% of individuals with zero access are employed, compared to 58% of those with full
10 access. Similar gradients are observed in income and education. As CAI increases, the proportion of
11 lower-income and less-educated individuals declines, while higher CAI groups contain more middle/high-
12 income earners and college graduates.
13 Racial and geographic disparities are also pronounced. White individuals constitute just under
14 60% of the CAI = 0 group but account for 86% of those with full access. Black and other minority groups
15 are overrepresented in the lower CAI tiers. Further, urban residents dominate all CAI categories but are
16 particularly concentrated among those with no access (94%), whereas rural representation increases
17 substantially at higher CAI levels, reaching nearly 25% among those with full access. Within
18 metropolitan areas, residence in large, rail-served urban centers is associated with lower CAI values,
19 suggesting that robust transit networks may partially offset the need for private vehicles. Moreover,
20 households with younger adults and no children are disproportionately represented among those with no
21 access, while households with children become increasingly prevalent in the mid-to-high CAI categories,
22 peaking at 68% in the 0.66–0.99 group.
23 In sum, these descriptive analysis results demonstrate that higher CAI values are associated not
24 only with greater vehicular mobility but also with socio-economic advantage and suburban or rural
25 residency. Conversely, lower CAI scores correspond to limited access, economic hardship, minority
26 status, and urban living. These systematic patterns support the construct validity of the Car Access Index
27 and underscore its value in identifying mobility disparities that are obscured by traditional household-
28 level vehicle metrics.
29
30 Modeling Results
31 Modeling was performed for the number of car trips and vehicle miles traveled at both the
32 individual and household levels. Individual-level modeling was performed to examine the predictive
33 accuracy of the number of car trips when CAI was used instead of traditional measures such as household
34 vehicle count, vehicle-to-driver ratio, and vehicle-to-household size ratio. Household-level analysis was
35 conducted to verify the accuracy of household-level CAI, rather than traditional household-level
36 measures. Here, household CAI was calculated by taking the average of all the individuals’ CAI. Stata 16

13
 Baldev, Mitra, and Nithila

1 was used for modeling the number of car trips and the person miles traveled. Before proceeding to the
2 modeling stage, data cleaning was performed to address extreme outliers in key relevant variables, such
3 as vehicle miles traveled, respondent age, and the number of car trips. This study employed a Negative
4 Binomial regression model for modeling the number of car trips, as the outcome variable is a non-
5 negative integer count that exhibited significant overdispersion. For vehicle-miles travelled, the study
6 utilized a Tobit model because the outcome is continuous but left-censored at zero, reflecting the presence
7 of many individuals with no recorded driving. Table 3 and Table 4 present regression estimates for two
8 key mobility outcomes: number of car trips and vehicle miles traveled (VMT) across five model
9 specifications at the individual and household level, respectively. The explanatory variables evolve from
10 traditional household car availability indicators, which include the number of household vehicles, the
11 vehicles-to-household size ratio, and the vehicles-to-driver ratio, to the Car Access Index (CAI), along
12 with the components used to derive the CAI. The significance levels indicate the reliability of
13 coefficients, while Pseudo R², AIC, and BIC values at the bottom provide indicators of model fit.
14
15 Car Trips Models at the Individual Level: Across the first three specifications, the simple household
16 vehicle count, vehicles per person, and vehicles per driver ratios, each additional unit of the respective
17 metric increases daily trip frequency, but only marginally. For instance, the base coefficient of 0.049
18 corresponds to an incident rate ratio (IRR) of exp(0.049), approximately 1.05, indicating that each
19 additional household vehicle is associated with a 5 percent increase in car trips. When the Car Access
20 Index (CAI) replaces these traditional metrics, its coefficient rises sharply to 1.284, yielding an IRR of
21 3.61, meaning that individuals with a one-unit increase in CAI undertake more than three times as many
22 car trips. Even when CAI’s underlying components, such as household size, vehicle count, and driver
23 status, are reintroduced, its coefficient remains high at 1.032, confirming that the index captures intra-
24 household access patterns that are obscured by simpler measures. Model fit improves modestly but
25 consistently, with pseudo R-squared increasing from 0.029 in the baseline to 0.032 in the fully specified
26 CAI model, and both Akaike and Bayesian information criteria decreasing modestly in magnitude in the
27 CAI models.
28
29 VMT Models at the Individual Level: The VMT models mirror many of the same patterns. A single
30 additional vehicle contributes approximately 1.2 miles to daily travel distance, while ratio-based
31 indicators account for an increase of 2.3 to 2.9 miles. The introduction of CAI dramatically raises this
32 estimate to 11.964 miles in the base model and 15.131 miles when its component variables are included,
33 indicating a suppression effect that reveals the full strength of CAI when controlling for household size,
34 vehicle count, and licensing. As with the trip models, CAI remains the dominant predictor while the
35 original vehicle metrics lose statistical significance. Additionally, model performance improves
36 consistently, with pseudo-R² rising from 0.0086 to 0.0103 and AIC decreasing by roughly 100 points,
37 indicating enhanced explanatory power for both the occurrence and magnitude of driving activity.

14
 Baldev, Mitra, and Nithila

1 Table 3: Individual Level Models for Number of Car Trips and VMT Miles
Number of Car Trips Models VMT Models
Variables Vehicles CAI + Vehicles Vehicles CAI +
Base: HH Vehicles Base: HH
per HH CAI Other per HH per CAI Other
Vehicles per Driver Vehicles
Member Vars Member Driver Vars
Number of Household Vehicles 0.05*** -0.05*** -0.06*** -0.01 1.22*** 0.28 0.09 0.22
Vehicle per Household
0.25*** 2.35***
Member
Vehicle per Driver 0.28*** 2.87***
Car Access Index 1.28*** 1.03*** 11.96*** 15.13***
Household Size -0.06*** 0.01 -0.01 -0.03** -0.68*** -0.07 -0.06 -0.11
Driving Status (Baseline: Not a Driver)
Driver 1.02*** 0.99*** 1.07*** 0.33*** 5.59*** 5.37*** 5.81*** -4.69***
Worker status (Baseline: Unemployed)
Worker 0.29*** 0.29*** 0.28*** 0.27*** 0.27*** 4.36*** 4.39*** 4.33*** 4.10*** 4.14***
Some College or Associate’s
-0.09*** -0.08*** -0.08*** -0.08*** -0.08*** 1.01** 1.04** 1.09*** 1.22*** 1.14***
Degree
High School or Less -0.28*** -0.28*** -0.26*** -0.28*** -0.25*** -1.54*** -1.49*** -1.35*** -0.93** -1.15**
Annual household income (Baseline: Higher Income)
Higher Income 0.07** 0.09*** 0.11*** 0.05* 0.08*** -0.03 0.18 0.37 0.28 0.09
Middle Income 0.11*** 0.12*** 0.13*** 0.09*** 0.11*** 1.09** 1.21*** 1.31*** 1.19*** 1.07**
Race (Baseline: Race White)
Black / African American -0.25*** -0.25*** -0.25*** -0.23*** -0.23*** -0.27 -0.24 -0.24 0.01 0.05
Other Race / Ethnicity -0.29*** -0.28*** -0.28*** -0.26*** -0.26*** -1.32* -1.29* -1.23* -0.97* -0.91
Location Characteristics (Baseline: Urban)
Rural -0.12*** -0.13*** -0.13*** -0.13*** -0.12*** 5.25*** 5.21*** 5.16*** 5.38*** 5.29***
Household composition (Baseline: HH with kids)
Younger Adults No Child -0.24*** -0.25*** -0.18*** -0.14*** -0.19*** -1.84*** -1.94*** -1.16* -0.89** -1.06*
Retired Adults No Child -0.10*** -0.11*** -0.04 -0.02 -0.07** -2.22*** -2.26*** -1.56** -1.54*** -1.68***
Model Fit
Pseudo R² 0.030 0.030 0.031 0.031 0.032 0.009 0.009 0.009 0.010 0.011
AIC 46275.51 46236.43 46188.34 46158.48 46133.89 66862.11 66851.39 66831.84 66761.96 667590.13
BIC 46388.39 46356.83 46308.74 46256.31 46254.29 66967.1 66963.39 66943.84 66852.95 66862.12
2
3 Notes: Asterisks next to coefficients denote statistical significance levels: *p < 0.05, **p < 0.01, and ***p < 0.001.
 Baldev, Mitra, and Nithila

1 Car Trips Models at the Household Level: Across the three conventional vehicle access metrics, each
2 additional unit of vehicle sufficiency is associated with progressively larger increases in household-level
3 trip frequency. An additional household vehicle corresponds to approximately a 10 percent increase in
4 daily trips (exp 0.097 ≈ 1.10), while an extra vehicle per household member increases trips by 20 percent,
5 and an extra vehicle per licensed driver raises it by 23 percent. Replacing these measures with the Car
6 Access Index produces a substantially stronger effect: a one-unit increase in CAI nearly triples household
7 trip-making (exp 1.052 ≈ 2.86). When CAI is included alongside its component variables, its coefficient
8 declines modestly to 0.981 but remains the strongest predictor in the model, indicating that CAI captures
9 behavioral and structural dimensions of vehicle sharing that are not observable through raw counts or
10 ratios. Further, Model performance improves consistently with the inclusion of CAI. Pseudo R-squared
11 increases from 0.0429 in the baseline model to 0.0483 in the fully specified version, and both Akaike and
12 Bayesian information criteria decline by approximately 180 and 135 points, respectively, despite the
13 increase in model complexity.
14
15 VMT Models at the Household Level: Patterns in the VMT models are largely consistent with those in the
16 car trip models, but differ in magnitude and the significance of some variables. The base model shows a
17 strong relationship between the number of household vehicles and VMT (4.593***), and this effect is
18 even more pronounced when using normalized measures: vehicles per household member (6.568***) and
19 vehicles per driver (5.949***). When CAI is introduced, it becomes the most powerful predictor of VMT
20 (19.883***), though the coefficient decreases to 12.593*** once other related components of CAI are
21 included. Although the pseudo R² values for all VMT models remain low, the model including CAI and
22 other relevant variables shows the best fit (lowest AIC and BIC).
23 Thus, individual- and household-level modeling support the superiority of the Car Access Index
24 over traditional vehicle sufficiency measures. CAI consistently yields larger and more significant
25 coefficients and improves model fit across both car trip and VMT outcomes. These results confirm that
26 CAI more effectively captures the complexity of within-household vehicle access and provides a more
27 precise and equitable measure of car mobility potential.
 Baldev, Mitra, and Nithila

1 Table 4: Household Level Models for Number of Car Trips and VMT Miles
Variables Number of Car Trips Models VMT Models

Base: Vehicles Vehicles CAI CAI + Base: Vehicles Vehicles CAI CAI +
HH per HH per Other HH per HH per Other
Vehicles Member Driver Vars Vehicles Member Driver Vars
Number of Household Vehicles 0.09*** 0.02 4.59*** 3.45*
Vehicle per Household Member 0.18*** 6.57***
Vehicle per Driver 0.21*** 5.95***
Car Access Index 1.05*** 0.98*** 19.88*** 12.59***
Household Size 0.01 0.05*** 0.17*** 0.09*** -0.74 0.52 1.61*** 0.07
Number of Drivers 0.27*** 0.31*** 0.22*** 1.24 3.68*** 0.84
Number of Workers 0.14*** 0.14*** 0.23*** 0.29*** 0.15*** 3.65*** 3.98*** 5.26*** 5.62*** 3.88***
Educational attainment (Baseline: Bachelor’s degree or higher)
Some College or an Associate Degree -0.12*** -0.13*** -0.12*** -0.09*** -0.10*** -2.26 -2.11 -1.94 -1.35 -2.04
High School or Less -0.41*** -0.41*** -0.43*** -0.32*** -0.35*** -4.74*** -4.50*** -4.72*** -2.95** -3.92**
Annual household income (Baseline: Higher Income)
Higher Income 0.17*** 0.17*** 0.25*** 0.24*** -0.01 -1.22 -0.69 0.60 0.63 -1.39
Middle Income 0.18*** 0.18*** 0.24*** 0.21*** 0.11*** 2.53*** 2.85*** 3.76*** 3.53*** 2.24***
Race (Baseline: Race White)
Black / African American -0.24*** -0.24*** -0.26*** -0.25*** -0.21*** -1.32 -1.37 -1.86 -1.38 -0.87
Other Race -0.36*** -0.36*** -0.37*** -0.29*** -0.31*** -4.21*** -4.37*** -4.65*** -3.53** -3.48***
Location Characteristics
Rural (Baseline: Urban) -0.12*** -0.12*** -0.09*** -0.09*** -0.12*** 5.71*** 6.21*** 6.68*** 7.13*** 5.67***
Household composition (Baseline: HH with kids)
Younger Adults No Child -0.15*** -0.18*** 0.01 -0.38*** -0.06 -0.17 -1.01 2.18 -1.06 0.82
Retired Adults No Child -0.02 -0.04 0.24*** -0.07** 0.08* -2.34 -2.79** 1.82 -1.23 -1.38
Model Fit
Pseudo R² 0.043 0.043 0.038 0.041 0.048 0.002 0.002 0.002 0.002 0.003
AIC 33249.71 33238.60 33424.60 33338.65 33063.81 60740.78 60747.98 60758.13 60747.49 60733.60
BIC 33353.69 33342.59 33521.65 33428.77 33174.72 60840.25 60847.45 60850.97 60833.69 60839.70
2
3 Notes: Asterisks next to coefficients denote statistical significance levels: *p < 0.05, **p < 0.01, and ***p < 0.001.
4
 Baldev, Mitra, and Nithila

1 CONCLUSIONS
2 This study introduces the Car Access Index (CAI), an individual-level metric designed to measure
3 access to private vehicles within households. CAI is a continuous score calculated for each person, based
4 on household vehicle availability, individual driver licensing status and restrictions, and the presence of
5 licensed household members who can provide transportation support to non-drivers. Using scenario
6 analysis, descriptive statistics, and econometric models applied to the 2022 National Household Travel
7 Survey, we show that CAI meets strong internal validation standards and explains car trip frequency and
8 vehicle miles traveled more effectively than traditional measures such as household vehicle count,
9 vehicles per person, or vehicles per driver.
10 Three key findings emerge. First, CAI reveals differences in car access that are hidden when
11 using household-level metrics. Individuals with low CAI scores make significantly fewer trips and travel
12 shorter distances than those with high scores, even when living in households with multiple vehicles.
13 Second, when CAI is used in place of traditional variables in negative binomial and Tobit models, the
14 results show stronger and more stable relationships. Measures of model performance, including Pseudo
15 R-squared, AIC, and BIC, also improve. This suggests that CAI better captures actual travel behavior.
16 Third, the descriptive analysis reveals that individuals with low CAI scores are more likely to come from
17 low-income, minority, and urban households. This reflects how limited vehicle availability, lack of a
18 driver’s license, and absence of support from household drivers combine to restrict mobility.
19 CAI is based on information commonly collected in household travel surveys, making it easy to
20 apply in travel demand modeling, planning studies, and transportation assessments. It allows agencies to
21 identify individuals with limited access to cars who may be overlooked when using household-level
22 indicators. This can help inform programs such as driver training, car sharing, or on-demand services
23 aimed at improving mobility.
24 This study has limitations. The current formulation captures only intra-household support,
25 omitting sporadic assistance from non-household members due to its irregularity and limited
26 documentation in survey data. While the CAI weighting scheme is theoretically grounded and empirically
27 validated, it remains normative and may benefit from refinement through stated preference experiments.
28 Future research should expand the framework to account for external support networks, car-sharing
29 participation, and potential access via emerging technologies such as autonomous vehicles. Spatial
30 aggregation of CAI at neighborhood or regional scales could help identify geographic disparities in
31 private vehicle access. Additionally, machine learning techniques could be employed to further validate
32 the index and improve its predictive performance.
33 In summary, the Car Access Index is a practical and easy-to-implement tool that captures
34 individual variation in vehicle access using standard variables from household travel surveys. Unlike
35 traditional household-level metrics, CAI reveals differences in car access among individuals within the
36 same household. It strengthens the explanatory power of travel behavior models and provides a more
37 accurate and behaviorally grounded alternative. By highlighting these within-household differences, CAI
38 offers researchers and practitioners a strong foundation for developing more targeted and effective
39 transportation strategies.
40
41 AUTHOR CONTRIBUTIONS
42 The authors confirm their contribution to the paper as follows: study conception and design: S.
43 Mitra, B. Giri, A Nithila; analysis and interpretation of results: B. Giri, S. Mitra, A Nithila; draft
44 manuscript preparation: B. Giri, S. Mitra; supervision: S. Mitra. All authors reviewed the results and
45 approved the final version of the manuscript.
Baldev, Mitra, and Nithila

REFERENCES
1. Kühne, K., S. K. Mitra, and J.-D. M. Saphores. Without a Ride in Car Country – A Comparison
of Carless Households in Germany and California. Transportation Research Part A: Policy and Practice,
Vol. 109, 2018, pp. 24–40. https://doi.org/10.1016/j.tra.2018.01.021.
2. Saphores, J.-D., and S. K. Mitra. Carless in California: What the Carless Can Tell Us About
Shifting Behaviors and Improving Mobility. The Equalizer: Could Ride-Hailing Extend Equitable Car
Access?, 2019, p. 19.
3. Mattioli, G. Where Sustainable Transport and Social Exclusion Meet: Households Without Cars
and Car Dependence in Great Britain. Journal of Environmental Policy & Planning, Vol. 16, 2014, pp.
379–400. https://doi.org/10.1080/1523908X.2013.858592.
4. Kalter, M.-J. O., and K. T. Geurs. Exploring the Impact of Household Interactions on Car Use for
Home-Based Tours: A Multilevel Analysis of Mode Choice Using Data from the First Two Waves of the
Netherlands Mobility Panel. European journal of transport and infrastructure research, Vol. 16, No. 4,
2016, pp. 698–712.
5. Chung, Y.-S., and C.-H. Wu. Cross-State Comparison of Income and Vehicle-Ownership Effects
on Household Mobility. Transportation Research Record, Vol. 2674, 2020, pp. 813–826.
https://doi.org/10.1177/0361198120947420.
6. U.S. Department of Transportation, Federal Highway Administration. National Household Travel
Survey 2022. Federal Highway Administration (FHWA), 2022.
7. Braitman, K. A., N. K. Chaudhary, and A. T. McCartt. Restricted Licensing among Older Drivers
in Iowa. Journal of Safety Research, Vol. 41, No. 6, 2010, pp. 481–486.
https://doi.org/10.1016/j.jsr.2010.10.001.
8. Nolan, A. A Dynamic Analysis of Household Car Ownership. Transportation research part A:
policy and practice, Vol. 44, No. 6, 2010, pp. 446–455.
9. Mitra, S. K., and J.-D. M. Saphores. How Do They Get by without Cars? An Analysis of Travel
Characteristics of Carless Households in California. Transportation, Vol. 47, No. 6, 2020, pp. 2837–
2858. https://doi.org/10.1007/s11116-019-09994-6.
10. Harper, C. D., C. T. Hendrickson, S. Mangones, and C. Samaras. Estimating Potential Increases
in Travel with Autonomous Vehicles for the Non-Driving, Elderly and People with Travel-Restrictive
Medical Conditions. Transportation Research Part C: Emerging Technologies, Vol. 72, 2016, pp. 1–9.
https://doi.org/10.1016/j.trc.2016.09.003.
11. TERAYAMA, K., and M. ODANI. Measurement and Evaluation of Mobility for Daily Travel by
Residents in Housing Developments in Suburban Area Using Capability Approach.
12. Jeon, J. S., C. Dawkins, and R. Pendall. How Vehicle Access Enables Low-Income Households
to Live in Better Neighborhoods. Housing Policy Debate, Vol. 28, No. 6, 2018, pp. 920–939.
https://doi.org/10.1080/10511482.2018.1494023.
13. Beige, S., and R. Cyganski. Long-Term and Short-Term Aspects of Car Allocation within the
Household Context. 2015.
14. Beige, S., M. Heinrichs, D. Krajzewicz, and R. Cyganski. Who Gets the Key First? Car
Allocation in Activity-Based Modelling. International Journal of Urban Sciences, Vol. 22, No. 2, 2018,
pp. 201–215. https://doi.org/10.1080/12265934.2017.1351389.
15. Scheiner, J., and C. Holz-Rau. Gender Structures in Car Availability in Car Deficient
Households. Research in Transportation Economics, Vol. 34, No. 1, 2012, pp. 16–26.
https://doi.org/10.1016/j.retrec.2011.12.006.
16. Blumenberg, E., A. Brown, and A. Schouten. Auto-Deficit Households: Determinants, Travel
Behavior, and the Gender Division of Household Car Use. California. Dept. of Transportation. Division
of Research and Innovation, 2018.
Baldev, Mitra, and Nithila

17. Qawasmeh, B., S. Qawasmeh, A. Al Tawil, and D. Qawasmeh. Estimation of Trip-Based

Generation Models and Calibration of Mode Choice Models for the American Travel Behavior.
https://doi.org/10.2174/0126671212348473241010122638.
18. Tiikkaja, H., and H. Liimatainen. Car Access and Travel Behaviour among Men and Women in
Car Deficient Households with Children. Transportation Research Interdisciplinary Perspectives, Vol.
10, 2021, p. 100367. https://doi.org/10.1016/j.trip.2021.100367.
19. Dadashzadeh, N., L. Woods, D. Ouelhadj, N. Thomopoulos, M. Kamargianni, and C. Antoniou.
Mobility as a Service Inclusion Index (MaaSINI): Evaluation of Inclusivity in MaaS Systems and Policy
Recommendations. Transport policy, Vol. 127, 2022, pp. 191–202.
20. Walker, M. Amartya Sen’s Capability Approach and Education. Educational Action Research,
Vol. 13, No. 1, 2005, pp. 103–110. https://doi.org/10.1080/09650790500200279.
21. Sechopoulos, R. K. and S. How the American Middle Class Has Changed in the Past Five
Decades. Pew Research Center, Apr 20, 2022.

20