A Meta-Analysis of Shared Mobility Adoption: The Role of Cultural Moderators and Key Psychological Determinants

Abstract

In order to address the theoretical deficiencies present in the current literature regarding the willingness to utilize shared transportation—specifically the fragmented influencing factors and the ambiguous cultural regulatory mechanisms—this research employed a meta-analytic methodology. Adhering to the PRISMA guidelines, a systematic screening of 35 empirical studies was conducted (17 in Chinese and 18 in English), which collectively examined 21 potential predictor variables. Following the execution of heterogeneity tests and assessments of publication bias, a random-effects model was applied for the analysis. The results reveal that attitude, subjective norms, and performance expectations exhibit medium effect sizes, while perceived behavioral control, social influence, and trust show weak effects. Conversely, perceived risk is found to have a significant negative impact. Further analysis of moderating effects indicates that collectivist cultures significantly reduce the strength of the effects of subjective norms (Δβ = −0.047) and social influences (Δβ = −0.139). Additionally, the influence of subjective norms is more pronounced in cultures characterized by low uncertainty avoidance (0.371 vs. 0.265). This study offers a theoretical framework for the cross-cultural adaptation of shared transportation systems and establishes an empirical basis for the differentiated development of sustainable urban transportation policies.

1. Introduction

The accelerating pace of global urbanization and the continuous rise in motorization levels have placed unprecedented pressure on urban transportation systems, with traffic congestion and environmental pollution emerging as critical challenges. Despite the implementation of various traffic demand management (TDM) policies—such as license plate restrictions, congestion pricing, and bus priority measures—these issues remain inadequately addressed. Furthermore, structural deficiencies in traditional public transportation systems, particularly in “last-mile” connectivity and service coverage, continue to severely constrain their effectiveness in mitigating urban mobility challenges. Against this backdrop, shared mobility models—supported by big data analysis, the Internet of Things (IoT), and information technologies—have emerged as a transformative solution. Services such as bike-sharing, e-scooter sharing, car-sharing, and ride-pooling have become deeply integrated into daily life.

Academic research on shared mobility has primarily focused on three key areas: (1) travel pattern analysis, (2) determinants of usage behavior, and (3) sustainability impact assessment. Studies on user adoption intentions are predominantly grounded in technology acceptance theories, including the theory of reasoned action (TRA), which establishes the influence of attitudes and subjective norms on behavioral intentions; the theory of planned behavior (TPB), which extends the TRA by incorporating perceived behavioral control, forming a three-dimensional predictive framework; the technology acceptance model (TAM), emphasizing the roles of perceived usefulness and perceived ease of use; the unified theory of acceptance and use of technology (UTAUT), integrating performance expectancy, effort expectancy, and social influence as core determinants [1,2,3,4].

Existing research has empirically validated the impact of key variables such as attitudes and subjective norms, while also expanding to include risk perception and trust [5,6,7,8,9]. However, the findings exhibit significant contextual dependency. For instance, Spears et al. observed a weak attitude–intention linkage (β = 0.12, p > 0.05) in low-density U.S. cities due to low habitual use [10]. Conversely, Richter reported a strong effect (β = 0.38) in high-density Berlin [11]. Si et al. demonstrated that external (e.g., social norms, media influence) and individual factors (e.g., environmental concern, risk appetite, and individual innovation) shape autonomous vehicle usage intention [12]. Meanwhile, the influence of objective environmental factors has also been empirically validated in various studies [13,14,15]. For instance, Zhang et al. innovatively incorporated weather variables into their research on shared bike usage in Asian countries, demonstrating that these variables directly impacted users’ willingness to adopt shared bikes [16]. Additionally, Sendek-Matysiak utilized data from five European countries to quantify the variations in user preferences for vehicles with different drive systems [17]. Furthermore, Nuzzolo et al. employed spatiotemporal data analysis to uncover the spatial differentiation patterns of shared travel intensity in Rome [18].

Recent research findings demonstrate considerable variability, with various influencing factors present across distinct models, and the criteria for assessing effect sizes lack consistency. Consequently, it is imperative to undertake a systematic quantitative synthesis of these results. Meta-analytic techniques, which include standardized procedures such as the integration of effect sizes, evaluation of heterogeneity, and assessment of publication bias, can yield a thorough evaluation of effects and contribute to the development of a cohesive theoretical framework [19,20]. This research adhered rigorously to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines in order to perform a systematic review and meta-analysis of global shared mobility behavior studies. The analysis incorporated effect sizes utilizing a random-effects model and conducted moderation effect assessments grounded in cultural dimensions. The objective is to develop a hierarchical model of influencing factors that is both cross-cultural and cross-modal, thereby offering evidence-based insights to inform the development of sustainable transportation policies and the optimization of services.

2. Research Methods and Data Collection

2.1. Literature Search and Selection

As shown in Appendix A Figure A1, based on the PRISMA guidelines, the literature for this study was categorized into two sections: English literature and Chinese literature. In the case of Chinese literature, relevant keywords such as “shared travel” and “shared travel intention” were utilized, and the literature was sourced from databases including CNKI and Wanfang. For the English literature, an initial screening was conducted using databases such as Web of Science and ScienceDirect, employing the keywords “shared travel” and “shared travel intention”, which yielded a total of 1531 publications. This initial set of literature was subsequently refined based on specific selection criteria: (1) the literature must be empirical research; (2) it must investigate the factors influencing the willingness to share travel; (3) it must report either a correlation coefficient or a path coefficient; and (4) the sample size of the survey must be disclosed. Following this rigorous screening process, a total of 35 publications were retained, comprising 17 from Chinese literature and 18 from English literature [5,6,8,9,11,12,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49].

2.2. Document Coding

In this study, a consistent encoder was utilized to process the literature from various time periods. Upon comparison, no significant discrepancies were observed. The selected literature was encoded to include pertinent details such as author information, correlation coefficients or path coefficients, and sample sizes. In instances where a single piece of literature contained multiple datasets, it was encoded on multiple occasions.

2.3. Variable Declaration

A total of 21 influencing factors were identified in the reviewed literature, including attitude, perceived behavioral control, subjective norms, effort expectation, performance expectation, social influencing, perceived usefulness, perceived ease of use, perceived risk, and trust. Given that perceived ease of use and perceived usefulness can be categorized under effort expectation and performance expectation, these two factors were amalgamated for the purpose of analysis. Additionally, factors such as perceived cost, safety, interest, hedonic motivation, satisfaction, perceived benefit, innovation consciousness, and environmental protection consciousness were represented in no more than five research studies each; consequently, they were excluded from the current analysis. This study focused on the eight influencing factors that were most extensively documented in the literature for further examination.

2.4. Variable Descriptive Statistical Analysis

The data presented in this study includes the sample size and the correlation coefficient, which encompasses both the correlation coefficient and the path coefficient within the structural equation model. A descriptive analysis illustrating the correlation between each influencing factor and the shared mobility intention is provided in

Table 1. Descriptive statistics of each influencing factor and usage intention.

Influencing Factor	Literature Number	Path Coefficient		Significant Rate/%	Cumulative Sample Size	Average Sample Size
		Min	Max
Attitude	16	0.125	0.830	100.00	16,590	754
Perceived Behavioral Control	10	0.010	0.595	100.00	10,224	730
Subjective Norms	14	0.107	0.725	100.00	12,878	644
Effort Expectation	14	−0.300	0.604	64.29	6023	430
Performance Expectation	20	0	0.620	91.67	10,615	442
Perceived Risk	13 (China: 13)	−0.665	0	69.23	4577	352
Social Influencing	10	0	0.51	91.67	5190	433
Trust	7 (China: 6)	0	0.498	77.78	3815	424

(the comparison table outlining the influencing factors and their associated empirical literature is presented in Appendix A 

Table A1. Comparison table of influencing factors and empirical literature.

Influencing Factor	Reference Label
Attitude	[5,11,22,25,26,28,29,31,35,38,39,41,45,47,48,49]
Perceived Behavioral Control	[11,22,25,26,28,29,35,39,41,49]
Subjective Norms	[11,12,21,22,25,26,28,29,35,37,38,39,41,49]
Effort Expectation	[6,9,23,24,27,31,32,33,34,42,43,44,45,47]
Performance Expectation	[6,9,22,23,24,27,30,31,32,33,34,36,37,40,41,43,44,45,46,48]
Perceived Risk	[9,22,27,34,38,39,40,41,42,44,45,47,48]
Social Influencing	[6,9,23,30,31,34,43,44,47,48]
Trust	[8,31,36,38,41,47,48]

).

As shown in Table 1, performance expectations along with their relationship with shared mobility adoption intention constitutes the most extensively studied factor, accounting for 20 out of 35 empirical studies (57.1%). This is followed by attitude (16 studies, 45.7%) and subjective norms (14 studies, 40.0%), both examined in relation to shared mobility usage intention. In contrast, trust emerges as the least-investigated factor, with only seven studies (20.0%) reporting its path coefficients in shared transportation adoption models.

The empirical research samples concerning risk perception and trust exhibit notable regional concentration, predominantly originating from China. This phenomenon can be attributed to several factors: First, the rapid advancement of shared transportation systems in China, coupled with the high commuting demands in major urban centers, has heightened the relevance of trust- and risk-related issues. Second, the explosive growth of the shared transportation sector in China has led to initial market disarray, exemplified by incidents such as safety concerns associated with Didi Hitch and challenges related to the refunding of deposits for shared bicycles. These circumstances have drawn significant scholarly attention to the themes of risk and trust. Third, the frequent collaborations between shared transportation platforms and academic institutions facilitate researchers’ access to empirical data. Additionally, from a cultural psychology perspective, within a collectivist cultural framework, risk cognition demonstrates unique characteristics of social contagion. Following a singular security incident that is amplified by social media, a collective risk perception can rapidly emerge (as evidenced by trending topics on platforms like Weibo). However, risk compensation is often mediated through institutional trust, such as government regulation, which ultimately diminishes the behavioral inhibition effect, resulting in a relatively low perceived risk effect value.

While direct evidence regarding emerging factors such as hedonic motivation remains scarce (k < 5), interdisciplinary studies indicate that it exerts a considerable influence on gamified transportation applications. Nevertheless, the findings across these studies exhibit considerable variability, highlighting the intricate nature of research pertaining to the motivations behind shared travel behavior. This variability suggests that conventional travel research may not fully account for the significance of non-utilitarian motivations.

3. Empirical Research

3.1. Heterogeneity Test

In meta-analytic research, substantial variations across primary studies—including differences in theoretical frameworks, methodological approaches (e.g., cross-sectional vs. longitudinal designs), sample characteristics (e.g., demographic composition), experimental designs (e.g., randomized controlled trials vs. observational studies), and operational definitions—may lead to significant heterogeneity in pooled effect size estimates. During the meta-analytic process, if the studies exhibit no statistical heterogeneity, either a fixed-effects model or a random-effects model may be employed. Conversely, in the presence of heterogeneity, the random-effects model is preferred for aggregating effect sizes. The assessment of heterogeneity can be quantified using the Q-statistic and the I² statistic. Specifically, a Q-statistic p-value ≤ 0.1 and an I² value exceeding 50% suggest the presence of heterogeneity, while a p-value greater than 0.1 and an I² value ≤ 50% indicate a lack of heterogeneity. The outcomes of the heterogeneity assessment are presented in

Table 2. Results of heterogeneity test.

	Attitude	Perceived Behavioral Control	Subjective Norms	Effort Expectation	Performance Expectation	Perceived Risk	Social Influencing	Trust
I2/%	97.17	92.40	94.04	95.80	90.30	92.31	89.86	90.70
p-value	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Model	Random-effects model	Random-effects model	Random-effects model	Random-effects model	Random-effects model	Random-effects model	Random-effects model	Random-effects model

p ≤ 0.1, I² > 50%: Exists heterogeneity; p > 0.1, I² ≤ 50%: No heterogeneity.

.

As shown in Table 2, the I² values of attitude (I² = 97.17%, p < 0.001), perceived behavioral control (I² = 92.40%, p < 0.001), subjective norms (I² = 94.04%, p < 0.001), effort expectation (I² = 95.80%, p < 0.001), performance expectation (I² = 90.30%, p < 0.001), perceived risk (I² = 92.31%, p < 0.001), social influencing (I² = 89.86%, p < 0.001), trust (I² = 90.70%, p < 0.001) exceeded 50%; the random-effects model was chosen to aggregate the effect sizes.

3.2. Publication Bias Test

Publication bias is characterized by the tendency for significant research findings to be more frequently reported and published compared to insignificant or invalid results, leading to a selective distortion of research outcomes within the database. A publication bias test is a statistical approach employed to assess whether research findings are influenced by this publication tendency. Common methodologies for conducting such tests include funnel plots, the Egger test, non-parametric rank correlation (Begg) test, and Fail-Safe number. In the present study, the Egger test was utilized to quantitatively assess the presence of publication bias within the sample. A p-value > 0.05 indicates no evidence of publication bias, whereas a p-value ≤ 0.05 suggests its presence. The findings from the publication bias test are presented in

Table 3. Publication bias test results.

	Attitude	Perceived Behavioral Control	Subjective Norms	Effort Expectation	Performance Expectation	Perceived Risk	Social Influencing	Trust
p-value	0.624	0.836	0.191	0.898	0.375	0.869	0.057	0.365

.

Table 3 indicates that the p-values obtained from the Egger test for the variables of attitude, perceived behavioral control, subjective norm, effort expectation, performance expectation, perceived risk, social influencing, and trust all exceed the threshold of 0.05. This suggests that there is no publication bias present in the analysis.

3.3. Effect Size Analysis

3.3.1. Effect Size Calculation

The meta-analysis presented in this study employed SPSSAU 24.0 software for computations. To mitigate the biases arising from factors such as research contexts and sample characteristics, effect sizes are were from the sample size and the Z-value following Fisher’s z transformation. Initially, Fisher’s z transformation was applied to the correlation coefficients (or path coefficients) obtained from the relevant literature to derive the z-values, as illustrated in Equation (1). Subsequently, the weighted average z-value (z₊) is computed based on the sample sizes, as detailed in Equation (2). Finally, the effect size (r_z) is determined from the z-values, as represented in Equation (3).

$$\mathrm{z}=0.5\times \ln {\displaystyle \frac{1+r}{1-r}}$$

(1)

$$z_{+}={\displaystyle \frac{{\displaystyle \sum N_i{z}_i}}{{\displaystyle \sum N_i}}}$$

(2)

$$r_z={\displaystyle \frac{\exp (2z_{+})-1}{\exp (2z_{+})+1}}$$

(3)

where z represents the variable that adheres to the standard normal distribution, while r denotes the correlation coefficient (or path coefficient) obtained from the existing literature. The term z₊ refers to the weighted average of z, N_i indicates the sample size of the i-th study, z_i signifies the z value corresponding to the i-th study, and r_z represents the effect size.

3.3.2. Effect Value Result

This research employed a meta-analytic approach to quantitatively assess the strength of the correlation between various influencing factors and shared transportation intention. By determining the aggregated effect size, we were able to systematically compare the explanatory power of each predictor variable in relation to the outcome variable. The findings from the effect size calculations are presented in

Table 4. Effect size calculation results.

Influencing Factor	Sample Size	Combined Effect Value	Lower Bound (95 CI)	Upper Bound (95 CI)
Attitude	22	0.369	0.284	0.447
Perceived Behavioral Control	14	0.252	0.180	0.322
Subjective Norms	20	0.340	0.272	0.404
Effort Expectation	14	0.143	0.019	0.263
Performance Expectation	24	0.306	0.249	0.361
Perceived Risk	13	−0.167	−0.268	−0.062
Social Influencing	12	0.279	0.196	0.357
Trust	9	0.208	0.104	0.307

and Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8.

As shown in Table 4 and Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7 and Figure 8, based on the criteria established by Cohen, the effect sizes for attitude (0.369), subjective norms (0.340), and performance expectation (0.306) exceed the threshold of 0.3, indicating a moderate level of correlation. Conversely, the effect sizes for perceived behavioral control (0.252), effort expectation (0.143), perceived risk (−0.167), social impact (0.279), and trust (0.208) fall within the range of 0.1 to 0.3, suggesting relatively weak correlations.

Attitudes, subjective norms, and performance expectations significantly impact the willingness to utilize shared transportation. This suggests that users’ favorable assessments of shared transportation, social pressures, and expectations regarding practicality serve as the primary motivators for their engagement with such services. Conversely, the influence of perceived behavioral control, social influence, and trust appears to be limited, indicating that users’ perceptions of their own capabilities, social endorsements, and trust in the platform exert a relatively weak direct effect. Furthermore, perceived risks notably diminish the willingness to engage with shared transportation; however, the magnitude of this effect is relatively small, potentially due to users’ heightened tolerance for risk.

3.4. Sensitivity Analysis

A sensitivity analysis was conducted using the leave-one-out method, which involved removing each included study one by one and then merging the effect sizes. It was also possible to change the inclusion and exclusion criteria or remove certain types of literature before merging the effect sizes again. The results of the sensitivity analysis are shown in

Table 5. Meta-analysis sensitivity test results.

Influencing Factor	Combined Effect Value	Range of Effect Size Variation
		Lower Bound	Upper Bound
Attitude	0.369	0.336	0.379
Perceived Behavioral Control	0.252	0.224	0.269
Subjective Norms	0.340	0.314	0.351
Effort Expectation	0.143	0.102	0.173
Performance Expectation	0.306	0.291	0.318
Perceived Risk	−0.167	−0.181	−0.152
Social Influencing	0.279	0.255	0.301
Trust	0.208	0.165	0.234

.

Table 5 indicates that the variation in effect size is relatively minimal following the exclusion of individual studies, suggesting that the research findings are robust.

3.5. Moderating Effect Test

Subjective norms and social influencing factors are shaped by social and cultural contexts. The empirical research literature reviewed encompasses samples from various countries and regions, each characterized by distinct social and cultural backgrounds (as illustrated in Figure 9). This study employed the Hofstede cultural dimension index as a framework for classification [50], focusing on individualism and collectivism (IDV) culture, as well as uncertainty avoidance (UAI), as moderating variables. The objective was to investigate the variations in the impact of subjective norms and social influence factors on the willingness to utilize shared transportation across different cultural settings. Due to data completeness constraints, the analysis was limited to examining the moderating effect of uncertainty avoidance (UAI) on the relationship between subjective norms and the willingness to engage in shared transportation. The findings of this analysis are presented in

Table 6. Moderating effect test results.

Influencing Factor	Item	Sample Size	Inter-Group Heterogeneity Q-Statistic Test	Combined Effect Value	Lower Bound (95 CI)	Upper Bound (95 CI)
Subjective Norms	Collectivism (Low IDV)	5068	Q = 214.979, p = 0.000	0.321	0.206	0.427
	Individualism (High IDV)	7810	Q = 94.541, p = 0.000	0.368	0.287	0.444
Social Influencing	Collectivism (Low IDV)	2322	Q = 25.468, p = 0.000	0.217	0.134	0.298
	Individualism (High IDV)	2868	Q = 37.113, p = 0.000	0.356	0.252	0.451
Subjective Norms	Weak Uncertainty Avoidance (Low UAI)	9629	Q = 249.799, p = 0.000	0.371	0.287	0.449
	Strong Uncertainty Avoidance (High UAI)	3249	Q = 52.912, p = 0.000	0.265	0.150	0.373

.

Table 6 reveals that within a collectivist cultural framework, such as that in China, the effect sizes of subjective norms (β = 0.321) and social influence (β = 0.217) on the intention to utilize shared transportation are notably lower than those observed in an individualistic cultural context, exemplified by the United States (subjective norms: β = 0.368; social influence: β = 0.356). This finding suggests that, in collectivist cultures, the impact of social norms on individuals is not primarily conveyed through direct verbal persuasion but rather through non-verbalized norms and contextual cues. Consequently, conventional survey methods may systematically underestimate the true magnitude of social influence in these cultural contexts.

Uncertainty avoidance pertains to the degree to which individuals within a society are willing to accept future uncertainties without experiencing stress. This concept encompasses various elements, including a propensity to evade risks and a deference to authority. In contexts characterized by low uncertainty avoidance, the influence of subjective norms on the intention to utilize shared transportation is notably greater (0.371 vs. 0.265). This finding suggests that individuals exhibiting high levels of uncertainty avoidance demonstrate heightened sensitivity to potential risks and tend to prioritize risk–return evaluations over social references when making decisions.

4. Management Suggestions

Drawing upon research findings, this study proposes several management recommendations aimed at advancing the development of a sustainable transportation system within our nation and promoting the widespread adoption of the shared travel model.

For the influencing factors with large effect sizes, several key interventions are proposed:

(1)

Attitude Intervention. It is recommended that operators implement a “carbon footprint visualization system” that utilizes life cycle assessment methodologies to calculate and provide real-time feedback on the environmental benefits associated with individual trips (e.g., “This ride reduces CO₂ emissions by 0.5 kg”). This approach aims to enhance users’ behavioral attitudes by fostering a sense of environmental responsibility.

(2)

Activation of Subjective Norms. Within collectivist cultural contexts, it is crucial to emphasize the role of descriptive norms. Mobile applications can incorporate spatial analysis features to dynamically generate and display heat maps illustrating shared vehicle usage in nearby areas, thereby leveraging geographic visualization technology to amplify the social demonstration effect. Additionally, a neighborhood carpooling matching algorithm could be developed utilizing data from residential communities to facilitate carpooling for commuting purposes. In contrast, in individualistic cultural contexts, a “friend recommendation points system” could be introduced, whereby users receive incentives, such as a CNY 5 discount coupon, for each successful referral.

(3)

Performance Expectation Improvement. It is essential to enhance the human–computer interaction design of the user interface, with a focus on improving the visibility and usability of core functional modules, such as algorithms for predicting real-time traffic conditions.

For the variables exhibiting comparatively low effect sizes, the following are recommended:

(1)

Trust Development. It is essential to adopt transparent management practices, which include the disclosure of platform security measures, such as adherence to ISO/IEC 27001 Information Security Management standards, as well as the presentation of driver qualification certification processes and vehicle maintenance records. These actions are aimed at bolstering user confidence.

(2)

Risk Mitigation. The integration of security assurance features, such as share itinerary capabilities and emergency assistance buttons, within operational procedures can help alleviate users’ perceived risks. Additionally, the establishment of a collaborative response team comprising representatives from both the platform and government entities is recommended to facilitate timely and authoritative communication regarding major public sentiment, alongside the creation of a risk compensation fund.

5. Conclusions and Prospects

This research employed a meta-analytic approach to assess the overall effect sizes between various influencing factors and the willingness to engage in shared travel. Utilizing the effect size evaluation criteria established by Cohen (1988), the analysis revealed that the comprehensive effect sizes of three variables—attitude, subjective norms, and performance expectation—exhibit a moderately strong positive correlation with the willingness to utilize shared transportation. Conversely, social influence, perceived behavioral control, trust, and effort expectations demonstrate relatively weak correlations. Notably, perceived risk was found to exert a significant negative impact. Furthermore, through the analysis of moderating effects, the study identified variations in the influence of subjective norms and social factors on the willingness to use shared transportation across different cultural contexts, specifically individualistic versus collectivist cultures and levels of uncertainty avoidance. A theoretical model encompassing the interrelations of “culture-psychology-behavior” was developed, elucidating the critical psychological mechanisms that influence the willingness to adopt shared transportation. This model offers a theoretical framework for the cross-cultural adaptation of shared transportation systems and provides an empirical foundation for the tailored formulation of sustainable urban transportation policies.

This research, through a systematic synthesis of existing empirical studies, established general conclusions regarding the mechanisms influencing the willingness to utilize shared transportation, thereby providing a foundation for evidence-based policy formulation and service enhancement. Nonetheless, this study is subject to certain limitations. Firstly, constrained by the scope of the original literature, the analysis focused solely on the direct effects of eight primary variables, neglecting potential mediating mechanisms (such as the role of attitude as a mediator between subjective norms and willingness to use) and moderating variables (such as the impact of economic development levels). Secondly, the geographical distribution of the literature samples included in the analysis exhibits certain biases, which may compromise the cultural generalizability of the findings.

Future research should aim to broaden the scope of investigation by integrating a mediation–moderation effect analysis framework. This approach will facilitate the examination of the transmission pathways and boundary conditions among various influencing factors. Additionally, it is essential to enhance the diversity of the sample population by incorporating research samples from regions with distinct cultural backgrounds, thereby improving the external validity of the findings. Finally, it is essential to move beyond the homogeneous assumption of “shared mobility” by differentiating the mechanisms between micro-mobility (bike-sharing/e-scooters) and ride-pooling services at the modal level, thereby providing evidence for tailored policy-making.