Breaking Commuting Habits: Are Unexpected Urban Disruptions an Opportunity for Shared Autonomous Vehicles?

Abstract

While extensive research has examined how major life events affect travel habits, less attention has been paid to the impact of minor environmental changes on commuting behavior, particularly regarding shared autonomous vehicles (SAVs). This study investigated how daily disruptions and incremental environmental changes influence commuter behavior patterns and SAV adoption in Shanghai, applying the theory of interpersonal behavior framework. The study surveyed 517 Shanghai residents, examining travel satisfaction, commuting habits, psychological factors (such as habit strength and satisfaction), and attitudes towards SAVs. Structural equation modeling was employed to test hypotheses about psychological factors influencing SAV adoption, while logistic regression analyzed how these factors affected mode choice across different disruption contexts. Analysis revealed that psychological factors, particularly habit and satisfaction, were stronger predictors of SAV adoption than attitude-based factors. Route obstructions and workplace relocations significantly increased SAV consideration. Even minor, recurring disruptions, such as construction zones, showed strong effects on commuting behavior, supporting the habit discontinuity hypothesis and emphasizing the importance of minor disruptions in driving behavioral change. The study extends the theory of interpersonal behavior by integrating habit discontinuity theory to explain how minor disruptions drive SAV adoption. This research provides actionable insights for urban planners and policymakers, recommending that SAV trials and targeted interventions be implemented during infrastructure changes or other commuting disruptions to promote SAV adoption and foster more sustainable transportation systems.

1. Introduction

Rapid urbanization and population growth in densely populated cities like Shanghai have led to a significant increase in commuting activities [1]. This phenomenon has raised two pressing issues: elevated carbon dioxide (CO₂) emissions and severe traffic congestion [2]. Over the last five years, motor vehicle ownership in China has grown at an average annual rate of 5.6%, particularly in megacities such as Shanghai and Beijing. This increase in vehicle ownership has resulted in automobiles becoming the primary contributors to total pollutant emissions, accounting for over 90% of air quality degradation [3]. Commuting, characterized by low vehicle occupancy rates and peak-hour traffic congestion, further exacerbates these problems [4,5]. To address these challenges, various strategies have been implemented, including public transportation promotion, active commuting encouragement, and congestion pricing [6,7]. Some cities have even explored carpooling initiatives and high-occupancy vehicle lanes to increase vehicle occupancy rates [8]. However, these efforts have largely been insufficient due to the habitual nature of commuting [9]. Research has shown that daily travel mode choices often become automated behaviors, with car usage particularly prone to becoming the default transportation option [10].

Previous research has demonstrated that major life events or incentives can significantly influence commuting habits. Events such as the birth of a child, a change in workplace, or the provision of financial incentives can catalyze behavioral change in commuting [11,12,13]. Conversely, the effect of smaller, naturally occurring events on commuting habits has been less investigated. Daily disruptions such as inclement weather [14,15] or temporary road closures [16] have the potential to disrupt routine commutes, prompting individuals to reconsider their usual mode of transport.

Despite extensive research on travel behavior and habit formation, several critical gaps remain in our understanding. While major life events’ impact on travel habits is well documented, the cumulative effect of minor daily disruptions remains understudied. Furthermore, the relationship between habit disruption and the adoption of novel transportation modes, particularly shared autonomous vehicles (SAVs), is poorly understood. Additionally, there is limited research examining how the timing and frequency of disruptions might influence openness to new travel modes.

SAVs present new opportunities for altering ingrained commuting habits while addressing urban mobility challenges. By providing flexible, on-demand mobility, SAVs can dynamically route passengers based on real-time conditions, potentially optimizing occupancy rates and reducing congestion [17]. These vehicles enable new mobility options like ridesharing at low marginal costs, which may shift social norms away from individual vehicle ownership [18]. While initial studies indicate that SAV integration could significantly alter travel behavior patterns [19,20], this study contributes to the discussion by providing the first systematic analysis of how minor disruptions may facilitate SAV adoption. Specifically, it extends habit discontinuity theory by examining the cumulative impact of small-scale disruptions to habitual routes and schedules. Drawing on the concept of “habit discontinuity” of Verplanken and Wood [21], the research investigates how subtle daily life changes can promote more sustainable commuting choices, particularly the transition to SAV use.

The study utilizes Triandis’s well-established theory of interpersonal behavior (TIB), which emphasizes the role of habits in decision-making and captures the repetitive nature of daily commutes. This theoretical framework also considers how contextual disruption and intentions can shape behavioral changes over time [22]. Using this framework, the research examines how routine disruptions or the advent of SAVs may alter established commute habits and investigates whether SAVs show promise as a more sustainable option that could be adopted in the long term.

The findings of this study offer urban planners and policymakers empirical evidence to formulate targeted transportation strategies, focusing particularly on the promotion of SAV services. Such strategies may include developing effective marketing campaigns, aligning routes and schedules with commuter preferences, and devising policies that encourage a shift from private car use to shared mobility. Additionally, these insights can provide a practical framework for other cities globally, contributing to the development of sustainable urban transport systems that adapt effectively to contextual disruptions.

This study investigates how unexpected disruptions in daily commuting patterns influence urban commuters’ willingness to adopt SAVs as an alternative transportation mode in evolving urban systems.

The remainder of this paper is structured as follows. In the next section, a brief narrative review of the role of habit breaking in travel behavior is provided. Section 2 and Section 3 present the theory, theoretical model, and data of this study. Section 4 and Section 5 present and discuss the results, respectively. The final section summarizes all essential findings and highlights the study’s limitations.

2. Literature Review

Travel behavior patterns result from the integration of repeated actions, automated responses, and environmental conditions. This section presents a systematic analysis of the existing literature regarding habit development and transformation in transportation mode choice. Triandis’s [23] seminal work established behavioral repetition as the primary mechanism in habit formation (Figure 1). Quantitative analyses have substantiated this relationship, with longitudinal studies by Lally et al. [24] identifying significant positive correlations between behavioral frequency and habit strength (r = 0.71, p < 0.001). Transportation research by Bamberg et al. [25] and Verplanken et al. [26] demonstrates that repetitive travel behaviors develop into persistent patterns operating independently of conscious deliberation.

The frequency–automaticity relationship requires additional theoretical consideration. Verplanken and Orbell [27] and Wood and Rünger [28] identify automaticity—defined as reduced cognitive engagement in decision processes—as the distinguishing characteristic of established habits. This behavioral transformation from intentional selection to automatic execution manifests prominently in daily transportation choices [29].

However, habit formation extends beyond frequency and the transition towards automaticity. It is crucial to consider the role of contextual cues in initiating these automated behaviors. Verplanken and Orbell [27] note that when behaviors consistently respond to the same cues, the association between cue and action strengthens, becoming almost reflexive. Wood and Rünger [28] support this finding, demonstrating that cues become primary initiators of habitual action, often bypassing conscious decision-making processes.

In this study, “context” encompasses both the physical environment (infrastructure, built environment) and temporal conditions (time of day, day of week) that influence travel behavior. Temporal patterns play a significant role in transportation choices, particularly for commuting trips. Peak-hour conditions, including congestion levels, service frequencies, and parking availability, create distinct decision environments that shape travel behavior [30,31]. For instance, morning peak hours typically feature different service levels and traffic conditions compared to off-peak periods, affecting mode decisions [32]. Research demonstrates that travelers develop specific behavioral patterns aligned with these temporal conditions. Hess et al. [33] found that commuters optimize their departure times and mode choices based on recurring temporal patterns of congestion and transit service. Similarly, Bhat [34] identified strong associations between time-of-day preferences and mode choice habits, particularly for work trips. When these temporal patterns are disrupted—through changes in work schedules, transit timetables, or traffic patterns—established travel habits may be reconsidered. Such disruptions present opportunities for behavioral adaptation and the formation of new travel patterns [35].

Environmental stability plays an important role in travel behavior patterns. Built environment characteristics such as street connectivity, transit accessibility, and land-use mix create consistent contexts that reinforce travel habits [36,37]. Research demonstrates that behaviors performed in stable contexts are more likely to become habitual, requiring less cognitive effort over time [38]. However, when this environmental stability is disrupted, established travel habits may be reconsidered. Verplanken and Wood termed this phenomenon “habit discontinuity,” where contextual changes create opportunities for new habit formation [21,39,40]. For example, residential relocation increased public transport use [41], while parking shortages during the 2002 Olympics boosted walking and cycling [42]. Even short-term disruptions, like Osaka’s 8-day freeway closure, reduced car dependency among habitual drivers [16].

While prior work has focused on major life events [11,43,44], subtle environmental changes, including daily traffic fluctuations and service disruptions, remain understudied despite their cumulative impact [36] (see

Table 1. Contextual Changes.

Variable	Contextual Change
MOD1	Adverse Weather Conditions
MOD2	Extended Commute Duration Due to Congestion
MOD3	Rushed Departure Due to Running Late
MOD4	High Parking Costs and Scarcity
MOD5	Pressure for On–Time Arrival
MOD6	Safety Concerns
MOD7	Change in Worksite Location
MOD8	Budget Constraints
MOD9	Obstructions on Usual Route Due to Construction/Events

and Appendix A). A comprehensive overview of these contextual changes and their effects on transport mode and commuting habits is presented in Appendix A. In this context, SAVs represent a unique opportunity to reshape commuting patterns. Unlike traditional transportation modes that often lack the flexibility to respond quickly to minor disruptions such as traffic fluctuations, SAVs can adapt in real time by rerouting and adjusting availability, which could lead to new travel habits forming incrementally [17,18]. Unlike acute disruptions such as freeway closures, which force commuters to adapt quickly, SAVs provide ongoing adjustments to the transportation ecosystem, continuously reshaping habits in a more gradual and flexible way. This research fills a critical gap in the literature by examining how SAVs as a novel mobility solution can facilitate habit disruption in ways that traditional transportation modes cannot, particularly during minor, cumulative disruptions.

3. Theoretical Background and Methodology

3.1. Theoretical Model and Research Hypotheses

This study drew upon the theory of interpersonal behavior, which encompasses many variables found in the theory of reasoned action [45] and its extension the theory of planned behavior (TPB) [46]. The TPB posits intention as the primary determinant of behavior, with attitudes and social norms predicting intentions [46]. Intention is considered the key predictor of behavioral performance, directly influenced by attitude and social norms [47]. Ajzen [46] further asserts that attitudes toward the behavior, subjective norms, and perceived behavioral control collectively determine an individual’s intention and subsequent behavior.

Numerous studies in travel behavior research have corroborated the positive effects of attitudes, norms, and perceived control on intentions and intentions on behavior [25,48,49,50,51,52]. Building on this evidence, we propose that a positive attitude towards a transport mode and satisfaction with the commute positively correlate with the intention to use it, emphasizing the role of subjective appraisal in the decision-making process.

H1. 

Satisfaction (SAT) → intention (INT).

H2. 

Attitude (ATD) → intention (INT).

However, a distinction exists between repetitive and novel situations [26,53]. Chen et al. [49] address this by incorporating “habits” and “facilitating conditions” that either impede or enable the performance of actual behavior [54]. We hypothesize that stronger habitual behaviors or tendencies are positively associated with the intention to choose a particular transport mode (H3), reflecting the impact of ingrained behavioral patterns. Facilitating conditions encompass level of service, cost-related attributes, situational context, mode access, trip type, and socioeconomic and demographic characteristics.

H3. 

Habit (HAB) → intention (INT).

According to the TIB, behavior is conceptualized as a pattern of muscle movement varying in time, duration, frequency, intensity, and probability of occurrence. Behavior results from an individual’s intentions, defined as “cognitive antecedents of an act” ([23], p. 5). This implies that behavior is linked to specific intentions as patterns organized, tailored, and sequenced towards a goal. These intentions are influenced by rational thought and social, normative, and affective factors [55,56]. We hypothesized that the intention to use a specific transport mode significantly influences the actual choice of transport (H4), indicating that intentions substantially impact individuals’ concrete transportation mode choices.

Furthermore, this study aimed to examine the interdependencies and collective influence of satisfaction, attitude, and habit in shaping these intentions. We hypothesize that the combined effect of these factors on an individual’s chosen mode of transport exceeds the impact of each factor considered independently (H5). While social factors are not addressed in this work, we recommend their investigation in future research.

H4. 

Intention (INT) → mode choice (MOD).

H5. 

Satisfaction (SAT), attitude (ATD), habit (HAB) → intention (INT).

Behaviors derive meaning from their context. In any situation, behavior is a function of intention, habits, and situational constraints [57]. Frequency depends particularly on how natural an act is for the individual (e.g., commuting). Rare behaviors have different determinants compared to habitual ones for which the individual is prepared [58]. However, behavior is never entirely automatic or fully intentional [54]. Therefore, this study examined the significant influence of sociodemographic characteristics on the preference for SAVs versus traditional transport modes in different contexts (H6). Additionally, we investigated how, for individuals initially not opting for an SAV, a positive change in attitude towards sustainable transport, satisfaction with their commute, or habit strength can significantly increase the probability of switching their commuting preference to an SAV (H7).

H6. 

Sociodemographic factors → mode choice (MOD).

H7. 

Satisfaction (SAT), attitude (ATD), habit (HAB) → mode choice (MOD) among non-SAV users.

To date, numerous studies have applied the TIB in various contexts, including internet abuse in the workplace [59], sexual behavior at a Mardi Gras [60], non-work-related computing in the workplace [61], telemedicine adoption by physicians [62], and predicting undergraduate condom use [63]. In the automotive field, the TIB was used only by Domarchi et al. [64], utilized the TIB to link attitudes, affect, and habit to commuting mode choice. However, to the best of our knowledge, no studies have applied the TIB in the context of autonomous mobility, specifically SAVs. This gap underscores the need for theoretical work applying TIB not only in the mobility sector generally but also particularly in the domain of autonomous mobility such as SAVs.

For this study, we adapted behavioral variables from Triandis’s [23] original model. In our hypothesis, facilitating conditions (context) and habits were posited to directly influence intentions, diverging from Triandis’s model where they are considered antecedents. Previous studies employing Triandis’s theory have demonstrated that both facilitating conditions and habit were significant predictors of intention, as shown by Boots and Treloar in their application of Triandis’s theory, who [65] found that habit and facilitating conditions were influential factors in determining the intention to attend a seminar-based training program.

3.2. Shanghai: City Context for the Study

Shanghai, situated in southern China, serves as a global transportation hub. As of 2021, it boasts a population of 24.87 million and a GDP of CNY 2.42 trillion (USD 380.91 billion) [66]. The city’s transportation infrastructure includes over 1500 bus lines [67] and 16 subway lines spanning 802 km [68]. With 4.5 million cars and more than 4000 car-sharing stations across all administrative districts, Shanghai exemplifies advanced urban mobility [69]. The city’s ride-hailing services have attracted over 850,000 registered users, indicating widespread adoption of shared mobility concepts [69]. Shanghai’s integrated public transport system and its economic and demographic characteristics make it an ideal setting for this study.

3.3. Participants and Procedure

The study involved 517 participants from Shanghai who completed an online survey between October and November 2023. Recruitment occurred via a Chinese research facilitation website, ensuring a diverse demographic representation. The sample comprised 48.2% females, with ages ranging from 18 to 58 years (mean age: 35 years) (see

Table 2. Demographics and Mobility Profile of Survey Participants.

Demographics	Characteristics	n	% Within Group	% of Population		Demographics	Characteristics	n	% Within Group	% of Population
Age	18–29	126	24.3			Occupation	Blue–collar worker	35	6.8
	30–39	292	56.5				White–collar worker	408	78.9
	40–49	79	15.3				Governmental job	29	5.6
	50+	20	3.9				Student	30	5.8
Gender	Female	249	48.2	48.2			Others	15	2.9	Average annual income of employees: 96,011 ¥
	Male	268	51.8	51.8		Annual Income	¥10,000–¥49,999	147	28.5
Marital Status	Divorced/widowed	4	0.8	7.6			¥50,000–¥99,999	112	21.6
	Married	394	76.2	72.2			¥100,000–¥149,999	137	26.5
	Never Married	119	23	20.2			¥150,000 or more	121	23.4
Household size	1 Person	12	2.3	Average persons per household: 2.63 persons		Car driver’s license	Yes	459	88.8
	2 Persons	28	5.4				No	58	11.2
	3 Persons	277	53.6			Time with driver’s license	0–2 years	50	10.9
	4 Persons	115	22.2				2–5 years	115	25.1
	5+ Persons	85	16.5				5–10 years	187	40.7
Education	High school or lower	20	3.9				10+ years	107	23.3
	Bachelors	438	84.7	61.0		Car ownership within household	0 cars	51	9.9
	Masters or higher	59	11.4				1 car	349	67.5
							2 cars	112	21.7
							3 or more cars	5	0.9

for complete demographic characteristics).

Participants responded to a 54-item self-report survey examining travel satisfaction, commuting habits, and attitudes towards various transportation modes, including traditional options and emerging alternatives such as SAV ride-hailing and ride-pooling services. The survey, which took an average of 13 min to complete, provided respondents with the following introduction to SAVs. “Imagine that you can book a driverless vehicle on demand via your smartphone-app (e.g., Didi) and it is waiting for you outside to drive you to your final destination This SAV (ride-hailing) offers amenities like Wi-Fi, enabling multitasking during the journey. Alternatively, consider SAV (ride-pooling), where you may share the vehicle with 1–3 unknown passengers heading in a similar direction”.

The questionnaire assessed:

• Commuting patterns: primary mode, duration, and weekly frequency (see

  Table 3. Commute Patterns of Survey Respondents.

  Item	Options	n	%
  Main Commute Mode	Car	204	39.5%
  	Public Transport	257	49.7%
  	Walking & Bicycle	56	10.8%
  Frequency of Commute (Times per Week)	<5	68	13.2%
  	5	449	86.8%
  Commute Duration	10 min or less	7	1.4%
  	10–20 min	75	14.5%
  	20–30 min	127	24.6%
  	30–45 min	168	32.5%
  	45–60 min	95	18.4%
  	60+ min	45	8.7%

  for detailed commuting patterns).
• Attitudes: evaluated using a modified version of Steg’s [70] framework, associating 14 attributes with six travel modes.
• Travel satisfaction: measured using the Satisfaction with Travel Scale (STS) [71].
• Travel habits: assessed using an adapted seven-item version of the Self-Report Habit Index (SRHI) [27].
• Responses to contextual changes: examined through nine disruptive scenarios (see Table 3 for details).
• Demographics: age, gender, household characteristics, education, occupation, income, and driving-related information.

To ensure comprehension by Chinese native speakers, the survey underwent translation from English to Chinese and validation through two pretest rounds, each involving 30 participants. The first pretest led to minor adjustments to four item descriptions to enhance clarity and comprehensiveness.

3.4. Data Analysis

The analytical approach comprised several stages, beginning with a confirmatory factor analysis (CFA) using the lavaan package in R (version 0.6.16). CFA validates the measurement model by assessing the fit between observed data and theoretical expectations. Subsequently, structural equation modeling (SEM) was applied to examine multiple interrelated dependencies simultaneously, essential for investigating the complex factors influencing SAV adoption (hypotheses 1–5). SEM’s modification indices for model enhancement allow for iterative refinement, ensuring both statistical conformity to observed data and theoretical coherence.

Model fit assessment utilized the comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Acceptable model fit criteria were defined as CFI and TLI > 0.90, RMSEA < 0.08, and SRMR < 0.10 [72].

To investigate the relationship between sociodemographic and mobility factors and the likelihood of choosing SAVs for commuting, Kruskal–Wallis ANOVA was employed. This analysis examined the influence of sociodemographic variables on SAV usage across distinct commuter profiles: non-users, SAV users (ride-hailing), and SAV users (ride-pooling) (hypothesis 6).

Logistic regression models were constructed to examine how changes in psychological factors affected the probability of individuals switching their commuting preference to SAVs across nine distinct contexts (hypothesis 7). Each model corresponded to one context, with binary dependent variables (MOD1 to MOD9) representing mode choice (1 = SAV, 0 = traditional modes such as car, public transport, walking, or bicycling).

In each logistic regression model, independent variables—satisfaction (SAT), attitude (ATD), habit (HAB), intention (INT), and various demographic and contextual factors—were entered simultaneously to assess their collective impact. The odds ratios derived from these models quantified the expected change in the likelihood of choosing an SAV relative to other transportation modes for every one-unit increase in the predictor variables, while controlling for other factors in the model.

A priori Monte Carlo simulation determined the minimum sample size required to detect poor model fit for a five-factor CFA model. Based on a target RMSEA of 0.05 under the null hypothesis, power of 0.8, alpha of 0.05, and specified model conditions, a minimum sample size of 462 was required [73]. The initial sample contained 564 respondents. After removing invalid responses, the final analyzed sample was 517, exceeding the required minimum size. Listwise deletion was used to remove cases with invalid responses, as this approach is considered appropriate when missing data comprise a small fraction of the sample [74]. Model parameters were estimated using maximum likelihood estimation, a method commonly recommended for structural equation models with continuous outcomes [74].

4. Results

4.1. Structural Equation Modeling for Transport Mode Choice

A confirmatory factor analysis (CFA) was initially conducted to validate five latent variables using their 34 indicators. This analysis confirmed significant loadings for each indicator on its respective latent variable, establishing the structure’s validity. Subsequently, structural equation modeling (SEM) was employed to test five hypotheses.

To determine the optimal factor structure and minimize the risk of type II errors, three alternative measurement models were tested and compared [75]:

• Factor model: All items loading on a single factor.
• Factor model: SAT, ATD, and HAB items loading on one factor and INT and MOD as separate factors.
• Hypothesized five-factor model: SAT, ATD, HAB, INT, and MOD as distinct factors

The one-factor model represented the possibility that respondents did not differentiate between the constructs. The three-factor model tested if SAT, ATD, and HAB were empirically indistinguishable.

The one-factor model exhibited poor fit on all indices, failing to meet common cutoff criteria. The three-factor model showed improvement over the one-factor model, but still demonstrated inadequate fit. In contrast, the SEM analysis indicated a significant model fit for the five-factor model (χ² = 974.570, df = 450, p < 0.001), with satisfactory fit indices (CFI = 0.923, TLI = 0.915), a favorable error approximation (RMSEA = 0.047, 90% CI (0.043, 0.052), p-value RMSEA ≤ 0.05 = 0.843), and an acceptable residual (SRMR = 0.060), suggesting the model was well suited for the data. Appendix C displays the fit statistics for the three measurement models.

The analysis demonstrated the reliability of the constructs, with construct reliability (CR) values ranging from 0.83 (MOD) to 0.90 (SAT) and Cronbach’s alpha values for multi-item scales from 0.81 (MOD) to 0.91 (SAT), indicating the model’s internal consistency and reliability. The average variance extracted (AVE) and CR of each latent variable were estimated and found to be within the acceptable range of >0.50. Appendix B presents the complete list of items, loadings, and reliability.

In the SEM, the latent variables SAT, ATD, and HAB significantly predicted the latent variable INT (see

Table 4. Summary of Hypotheses 1–5: Insights from the SEM Analysis.

Path	Standardized Estimate	Standard Error	t–Value	R2	Hypothesis Accepted/Rejected
H1: Satisfaction (SAT) → Intention (INT)	0.622	0.197	3.912 ***	0.39	Accepted
H2: Attitude (ATD) → Intention (INT)	0.36	0.084	4.487 ***	0.13	Accepted
H3: Habit (HAB) → Intention (INT)	0.449	0.107	4.552 ***	0.2	Accepted
H4: Satisfaction (SAT), Attitude (ATD), Habit (HAB) → Intention (INT)				0.72	Accepted
H5: Intention (INT) → Mode Choice (MOD)	0.238	0.081	3.030 ***	0.06	Accepted

Notes: *** p < 0.001.

for a summary of all hypothesis-testing results). The standardized path coefficients were 0.62 for SAT, 0.36 for ATD, and 0.45 for HAB. All factor loadings were statistically significant (p < 0.001), indicating robust relationships between each predictor and intention, thus confirming hypotheses 1–3. Collectively, these coefficients explained approximately 71.8% of the variance in INT (R² = 0.718).

This substantial proportion underscores the significant influence of satisfaction, attitude, and habit on the construct of intention within the model. The high cumulative R² suggests that SAT, ATD, and HAB, when considered together, have a substantial and combined impact on intention. This finding supports hypothesis 4, indicating that the cumulative influence of these variables is more pronounced than their individual effects.

The model also examined the impact of INT on the latent variable MOD, representing mode choice under different contexts. The standardized path coefficient for the influence of INT on MOD was 0.24, which was also statistically significant (p < 0.001). This suggests that intention accounts for approximately 5.66% of the variance in mode choice, highlighting a modest, but statistically significant relationship and supporting hypothesis 5. This finding demonstrates how intention, shaped by individual levels of satisfaction, attitude, and habit, influences individuals’ choices of transportation modes across various contexts.

4.2. Impact of Sociodemographics on SAV Choice: ANOVA

Following the confirmatory factor analysis (CFA) and structural equation modeling (SEM) results, this section examines hypothesis 6 using the Kruskal–Wallis ANOVA. This analysis investigated the influence of sociodemographic factors on the selection of SAVs among three commuter groups: non-users of SAVs, SAV ride-hailing users, and SAV ride-pooling users (see Appendix D for detailed ANOVA results showing the statistical significance of various sociodemographic factors across different contextual scenarios).

Hypothesis 6 posits that sociodemographic variables significantly influence commuting choices, particularly towards SAVs. The analysis compared median values of these variables across SAV usage groups, providing insights into their association with SAV preferences versus traditional transport modes.

This statistical approach revealed differentiated impacts of various sociodemographic factors across distinct commuter profiles. Notably, the intention (INT) to use SAVs emerged as a consistently significant factor across all models, predominantly at p < 0.001. This finding emphasizes the importance of psychological determinants—attitude (ATD), satisfaction (SAT), and habit (HAB)—in shaping commuter preferences for SAVs.

Sociodemographic influences on SAV adoption likelihood included:

• Driver’s license: significantly affects choices during adverse weather (MOD1) and safety concerns (MOD6).
• Vehicle ownership: multi-car households adapt better to adverse weather disruptions (MOD1).
• Commute time: longer commutes correlate with SAV consideration in congestion (MOD2).
• Driving experience: experienced drivers navigate congestion independently (MOD2).
• Gender: influences safety perceptions (MOD6) and budget-driven choices (MOD8).
• Household size/income: impacts mode choice under financial constraints (MOD8).

The intention to utilize SAVs has been identified as a main determinant for mode choice across different commuting scenarios. This finding underscores that despite varying sociodemographic variables, this psychological factor consistently and strongly affects SAV adoption decisions.

4.3. Impact of Changes on Psychological Factors of Non-Users Adopting SAV: Logistic Regression Analysis

This study employed logistic regression analyses to examine how psychological factors—attitude, satisfaction, and habit—influence the likelihood of SAV adoption by current non-users across nine disruptive contexts. The models provide insights into the psychological shifts that may encourage individuals to consider SAVs for their commutes.

The results indicated that habit (HAB) is a significant predictor in contexts 7 and 9 (context 7: OR = 3.63, 95% CI (3.02–4.56), Nagelkerke R² = 10.5%; context 9: OR = 3.74, 95% CI (3.01–4.93), Nagelkerke R² = 13.1%). These findings suggest that interventions targeting routine behaviors could increase SAV usage. Attitude (ATD) toward SAVs was a significant predictor in context 4 (OR = 3.20, 95% CI (2.77–3.79), Nagelkerke R² = 9.6%), indicating that improving overall perceptions of SAVs may lead to increased consideration.

In context 1, the combined influence of satisfaction (SAT) with current transport options and attitude (ATD) toward SAVs was significant (OR = 3.48, 95% CI (2.97–4.2), Nagelkerke R² = 13.3%). This suggests that enhancing the perceived appeal of SAVs relative to existing transportation modes is crucial. The interaction between habit (HAB) and attitude (ATD) was significant in context 2 (OR = 3.16, 95% CI (2.75–3.73), Nagelkerke R² = 11.1%), highlighting the potential effectiveness of combined behavioral and cognitive strategies in shifting preferences.

While contexts 3 and 8 were statistically significant, demographic factors, rather than psychological ones, were indicative of a shift towards SAV use. Models for contexts 5 and 6 (pressure for punctual arrival and safety concerns) were not statistically significant, indicating that neither the psychological nor demographic variables analyzed were predictors of SAV adoption in these situations. These findings imply that factors beyond those measured in models 5 and 6 might influence the decision to use SAVs in such contexts.

For a comprehensive presentation of the statistical analyses and detailed findings, refer to Appendix E.

5. Discussion

Our SEM analysis of revealed that psychological factors, particularly habit (β = 0.45, p < 0.001) and satisfaction (β = 0.62, p < 0.001), are the strongest predictors of SAV adoption, surpassing traditional attitude-based factors (β = 0.36, p < 0.001). These findings align with Triandis’s TIB [23], which prioritizes habit and situational context over attitudes alone. While prior studies applied TIB to traditional commuting modes [64], this is its first extension to SAVs. Our results corroborate Bamberg et al.’s [25] assertion that ingrained habits often override deliberative decision-making. Given that habit is a critical driver of decision-making, interventions aiming to promote SAV adoption should focus on altering habitual behaviors rather than attempting to change attitudes alone. This aligns with the growing body of research that suggests habits are deeply ingrained and difficult to change, but disruptions in routine can create natural opportunities for behavioral shifts.

A key finding is the role of commuting disruptions in SAV adoption. Route obstructions (MOD9) and workplace relocations (MOD7) increased SAV consideration by 63% (p < 0.001), reinforcing Verplanken and Wood’s [21] habit discontinuity hypothesis. Even minor disruptions (e.g., construction zones) had strong effects (OR = 3.74, MOD9), mirroring Fujii and Kitamura’s [16] observation that temporary freeway closures triggered lasting shifts toward sustainable modes. Chronic disruptions (e.g., recurring roadworks) likely erode car dependency cumulatively, aligning with Ouellette and Wood’s [38] findings on habit formation in unstable environments. Therefore, we recommend that SAV trials be launched during infrastructure projects or disruptions to give commuters the chance to experience the convenience of SAVs during periods of habitual instability. Providing SAV trials during these disruptions will allow commuters to experience firsthand how SAVs can be a viable alternative to their usual modes of transportation, leading to more sustained behavior change.

Additionally, our study revealed that minor, recurring disruptions, such as construction zones, can have as much of an impact on behavior as major life events. This suggests that policy interventions aimed at exploiting such disruptions could be highly effective. Specifically, dynamic pricing strategies can make SAVs more financially accessible during peak disruption times. Our findings indicate that financial accessibility is crucial during disruptions, as commuters are more likely to adopt SAVs when faced with inconvenient or unpredictable commuting conditions. For example, dynamic pricing, which combines congestion charges for private vehicles with off-peak SAV discounts (e.g., <CNY 20/trip), could incentivize commuters to choose SAVs over their usual modes of transportation during periods of congestion or disruption, making the alternative both financially appealing and practical.

Sociodemographic factors also play a pivotal role in influencing SAV adoption. Low-income commuters (<CNY 50,000/year) were 2.5× more likely to adopt SAVs under budget constraints (MOD8: OR = 3.19, p < 0.05), while female commuters were particularly motivated by safety concerns (MOD6: p < 0.05). Given that safety concerns were a significant factor in SAV adoption, particularly among female commuters, we recommend implementing safety-focused marketing campaigns that target these demographic groups. Specifically, highlighting collision-avoidance technology and safety features during extreme weather conditions can address the safety concerns that have been shown to influence adoption rates. These campaigns could emphasize how SAVs offer a safer alternative compared to private car use, particularly under adverse conditions where visibility and road conditions are major concerns.

Our logistic models also identified the effectiveness of several context-specific interventions in promoting SAV adoption. For example, protected SAV access points near transit hubs increased adoption intent by 28% (p < 0.001) during adverse weather (MOD1) [33]. Real-time rerouting alerts in SAV apps boosted consideration by 33% (p = 0.003) during route obstructions (MOD9). These findings suggest that providing convenient, real-time solutions during disruptions can significantly enhance the appeal of SAVs. To capitalize on these opportunities, real-time rerouting alerts in SAV apps can be implemented to notify users of roadblock situations and offer SAV alternatives, making the option of an SAV not only convenient but also timely and necessary [34].

Lastly, the role of workplace relocations in influencing SAV adoption should not be overlooked. Our data show that workplace relocations are a significant factor in increasing the likelihood of SAV adoption, as commuters experience more frequent disruptions to their normal routines. Based on this finding, we recommend the establishment of employer-sponsored programs that offer subsidized rides for employees during workplace relocations (MOD7). Such programs could help employees transition to using SAVs during periods of routine change, increasing adoption during these critical moments when commuters are more receptive to changing their transport behavior.

6. Conclusions

This study explored the impact of daily commuting disruptions on the willingness of urban Shanghai commuters to adopt SAVs based on survey data from 517 respondents. The findings underscore the significant influence of psychological factors, particularly habit and satisfaction, in driving SAV adoption, with these factors proving more impactful than attitudinal perceptions. The study also found that disruptions to commuters’ routines, such as workplace relocations and route obstructions, significantly increase the likelihood of SAV consideration, with minor, recurring disruptions being as influential as major life events in shifting habitual behaviors.

The study revealed key sociodemographic patterns, including the roles of income, gender, and car ownership in shaping adoption likelihood. These insights suggest that targeted interventions, such as subsidized rides for low-income commuters or safety-focused campaigns for female commuters, can enhance adoption rates.

Based on these findings, we recommend SAV trials during infrastructure disruptions to provide commuters with hands-on experience, facilitating behavior change. Additionally, dynamic pricing strategies and targeted safety campaigns should be implemented to address the unique needs of different commuter groups. These interventions would promote both immediate and sustained adoption of SAVs, contributing to the broader goal of creating a sustainable, efficient, and inclusive urban transport system.

We acknowledge the limitations of self-reported data. Self-reported preferences may have overstated SAV adoption intent. Future studies should pair surveys with real-world pilot data (e.g., SAV trials in Shanghai’s Pudong District and other Chinese cities, where autonomous vehicle services are already operational). Future research should explore multiple aspects: barriers faced by moderately interested commuters and non-users (addressing issues of perceived convenience, cost, and accessibility); the impact of varying subsidy levels on adoption rates; and how cumulative minor disruptions, like repeated construction delays, can create long-term shifts in commuter behavior. These investigations will help refine strategies for encouraging SAV adoption during everyday disruptions.

Ultimately, this study extends Triandis’s TIB framework by integrating habit discontinuity theory to explain how minor disruptions drive SAV adoption. The study offers a practical framework for integrating SAVs into urban mobility systems, fostering sustainable transportation alternatives and helping to reduce car dependency in rapidly growing cities like Shanghai.