Shared micro-mobility and transport equity: A case study of three European countries

Shared micro-mobility services (e


Introduction
Due to the technology innovation of GPS (global positioning system) location tracking and digitalized payment systems, many cities around the world have recently witnessed an increasing development of shared micro-mobility services (e.g., shared bikes, e-bikes, e-scooters, and emopeds).These new mobility services are intended to facilitate a more efficient and sustainable urban transport system by introducing new transport options and by becoming available as a 'first' or 'last mile' solution in connection with public transport (Hirsch et al., 2019;Ricci, 2015).Accordingly, many studies have explored the usage of shared micro-mobility services (Fishman, 2016;Sun et al., 2021), their impacts on people's travel mode choice (Nikiforiadis et al., 2021;Sanders et al., 2022), and the result on activity generation (Bai et al., 2021).At the same time, what also received great attention is the impact of shared micro-mobility on transport equity, which refers to "the fair and appropriate distribution of transport benefits and costs among individuals or population groups (Pereira et al., 2017)."As suggested by Lucas et al. (2016), some individuals suffer from transport poverty because of the following issues: transport (in)affordability (i.e., excessive amount of time or money spend on travelling), mobility poverty (i.e., lack of available transport options), accessibility poverty (i.e., low access to destinations for daily activity needs), and exposure to externalities (e.g., unsafe/unhealthy travel condition).While some researchers have suggested that shared micro-mobility services could facilitate transport equity by increasing transport disadvantaged groups' mobility options (Dill & McNeil, 2021), others have suggested that their development may aggravate transport inequality because of the unequal distribution of these services as well as barriers associated with using these services related to physical ability and tech savviness (e.g., Groth, 2019).
The impact of shared micro-mobility on transport equity depends on how it influences the mobility levels of different population groups.In other words, to increase transport equity in a region, shared micromobility services must benefit disadvantaged groups and alleviate their transport poverty equally (or more greatly) compared to groups that do not experience transport poverty.Previous studies tend to investigate the role of shared micro-mobility in transport equity from the perspective of people's unequal use of (or access to) these services only (Böcker et al., 2020;Chen et al., 2020a;Dill & McNeil, 2021;Laa & Leth, 2020;Mooney et al., 2019;Reck & Axhausen, 2021;Ricci, 2015).However, whether the adoption of these services can help reduce the users' transport poverty and how such benefit varies across population groups were largely neglected.This results in an incomplete understanding of the equity implications of shared micro-mobility.
For this reason, we argue that research on shared micro-mobility and transport equity needs to consider two aspects simultaneously: the equity of shared micro-mobility use (i.e., by whom it is used) and the impact of its use on transport poverty.However, so far limited research has paid attention to the latter aspect (Abouelela et al., 2024;Chen et al., 2022;Qian & Niemeier, 2019).Moreover, more studies are needed to compare the equity impacts of different kinds of shared micro-mobility services and their potential context difference.Shared electric microvehicles (e.g., shared e-bikes/e-mopeds) may benefit the users' mobility and accessibility more greatly than conventional bikeshare because of the higher levels of travel speed and less physical effort needed in use.Dock-based shared micro-mobility services may play a limited role in improving accessibility because they require additional travel time a user should walk to and from the station.Relatively, dockless services can provide more flexible transport options for daily travel by freeing the users' route and destination choices from around the docked stations.However, the achievement of such accessibility benefit requires a sufficient supply of vehicles around the city to guarantee that it's easy for the users to access them at any time.Also, the high dependency on smart phone applications for dockless services might exaggerate the transport inequality between groups with digital barriers and the rest population.Besides, the role that shared micro-mobility plays in daily travel may largely depend on the transport infrastructure, spatial distribution of facilities, and mobility culture, thus varies across different spatial contexts.
To fill these research gaps, this paper aims to contribute to an improved understanding on the role of shared micro-mobility in transport equity by investigating the equity of shared micro-mobility use and on the other hand, its impact on individual's perceived transport poverty, measured by the perceived mobility level, accessibility level, travel costs (monetary, time and emotional) and exposure to externalities (Lucas et al., 2016).A comparative study is conducted using an online questionnaire survey data from cities in three European countries (Netherlands, Sweden, and United Kingdom).In addition, our study involves all the different kinds of shared micro-mobility services (including shared bike, shared e-bike, shared e-scooter, shared e-moped, and shared e-cargo bike) implemented in each city simultaneously to clarify their potential different roles in alleviating transport inequity.

Research on shared micro-mobilities
The rapid development of shared micro-mobility services around the world has gained extensive research attention in the last decade.Relevant studies started from understand the effect of docked bikeshare systems (Fishman, 2016) and, more recently, research on dockless shared e-bikes and e-scooters emerged.Research on the use of shared emopeds remains scarce.Numerous studies have investigated the determinants of shared micro-mobility use, which could be roughly divided into three categories: First, personal attributes, including sociodemographic characteristics like age (Reck & Axhausen, 2021), gender (Reilly et al., 2020), income level (Frias-Martinez et al., 2021), health conditions (Barbour et al., 2019), and car ownership (Fishman, 2016), as well as psychological factors such as travel mode preferences (Chen et al., 2020b), perceived convenience of using shared micro-mobility (Guo & Zhang, 2021), environmental and health consciousness (Kim et al., 2017;Mitra & Hess, 2021), and subjective norms (Si et al., 2020).Second, environmental features.Shared micro-mobilities were found to be used more frequently in areas with high population density, better access to transit, and well-equipped cycling infrastructures (Bai & Jiao, 2020;Shen et al., 2018;Sun et al., 2021;Zhang et al., 2017).Meanwhile, the natural environment, such as air quality (Campbell et al., 2016) and weather conditions (Caulfield et al., 2017;Hosseinzadeh et al., 2021), also influences the decision to use shared micro-mobility services.Third, several studies have also focused on level of service, such as the access to shared micro-mobility services (Reck & Axhausen, 2021;Sanders et al., 2020), available fleet size (Shen et al., 2018), the lengthy of sign-up process (Fishman et al., 2012), and the price of usage (Reck et al., 2021) of shared micro-mobility services.
Many other scholars analyzed the users' travel patterns with shared micro-mobilities.In general, both shared bikes and e-bikes were observed to be used mostly during morning and evening peak hours on weekdays, and during the day on weekends (Fishman et al., 2015;McKenzie, 2019;Nickkar et al., 2019;Shen et al., 2018).This suggests that shared bikes/e-bikes are used mostly for commuting trips on weekdays and for other purposes (e.g., recreation) on weekends (Fishman et al., 2015;Sun et al., 2021).Findings on the temporal distribution of shared e-scooter trips are inconsistent.Specifically, while some studies reported evidence on commuting peaks (e.g., McKenzie, 2019), others found only afternoon peaks (Bai & Jiao, 2020;Reck et al., 2021).In accordance with the latter finding, most studies concluded that shared e-scooters are predominantly used for shopping or recreation, rather than commuting purposes (Bai et al., 2021;McKenzie, 2019).In terms of travel duration, it was found that most trips by shared bikes are under 30 min and within 2 km (Chen et al., 2020a).Trips by shared ebikes are relatively longer, with the mean trip distance between 2 and 4.5 km (Campbell et al., 2016;McKenzie, 2019), while shared e-scooter trips tend to be shorter, with a median distance of 1.3 km in Louisville, USA as an example (Reck et al., 2021).
The transport and environmental impacts of shared micro-mobility services have also been investigated.The majority of research found that shared bikes (Barbour et al., 2019;Chen et al., 2020a;Fishman, 2016) and e-bikes (de Haas et al., 2022;Fukushige et al., 2021) substituted trips which were previously made by walking, cycling, and public transport, rather than private cars.Similar findings were also found for shared e-scooters (Abouelela et al., 2024;Nikiforiadis et al., 2021;Sanders et al., 2022).These findings question the positive impact of shared micro-mobility on the environment.For example, by simulating the modal shifts before and after Amsterdam is fully covered by shared mobility (shared e-bikes, e-mopeds, and e-cars) hubs, Xanthopoulos et al. (2024) concluded that only 32 % of the "shared trips" replace previous car trips, leading to an only 1.27 % reduction of CO2 emissions.Nevertheless, the study of Fishman et al. (2014) concluded that bikeshare did reduce car use in most of the cities in the United States, the United Kingdom (except London), and Australia.Studies have also found that bikeshare could help raising the awareness, acceptance, and use amount of conventional cycling as a mode in daily travels (Fuller et al., 2013;Goodman et al., 2014;Murphy & Usher, 2015).Besides, bikeshare is suggested to be able to reduce the user's total travel time by providing alternatives for short car and public transit trips, especially in congested streets (Bullock et al., 2017;Faghih-Imani et al., 2017;Woodcock et al., 2014).
X. Guan et al.

Shared micro-mobility and transport equity
The impact of shared micro-mobility development on transport equity also received research attention.An equitable transport system needs to allocate transport resources to groups evenly (i.e., horizontal equity) on the one hand and favor disadvantaged groups in particular to improve their current travel condition (i.e., vertical equity) on the other hand (Guzman et al., 2017).Previous studies have suggested that individuals who are female, low-income, less-educated, migrants, disabled, and non-auto travelers are usually vulnerable groups of transport poverty (Gössling, 2016;Guzman et al., 2017;Lucas et al., 2019;Verlinghieri & Schwanen, 2020).In general, studies have widely found that transport poverty hinders people's access to life opportunities (Wang et al., 2021), social capital (Ma et al., 2018;Schwanen et al., 2015), social inclusion (Currie & Delbosc, 2010;Lucas, 2012), subjective wellbeing (Churchill & Smyth, 2019;Delbosc & Currie, 2011), as well as physical/mental health (Churchill et al., 2023;Ma et al., 2018).
Shared micro-mobility services are claimed to facilitate transport equity by offering more widely available travel modes for transport disadvantaged populations (Dill & McNeil, 2021).However, achieving transport equity requires that disadvantaged groups are able to access and use shared micro-mobility services equally compared to more advantaged groups.Moreover, shared micro-mobility needs to be able to significantly improve the users' overall mobility and accessibility levels, thus help reduce the transport poverty among disadvantaged groups.

Spatially/socially unequal access to and usage of shared micromobility
The spatial/social equity of shared micro-mobility access and usage has received great research attention.Scholars have pointed out that shared micro-mobility development may reproduce transport poverty due to the unequal distribution of services (Groth, 2019).Some providers may implement shared micro-vehicles in denser and wealthier areas for profit reasons, resulting in the relatively lower opportunity to benefit from shared micro-mobility for residents in marginalized areas (Bai & Jiao, 2021;Mooney et al., 2019;Ricci, 2015).Moreover, even within the service areas, some groups are less represented in adopting shared micro-mobility due to the lack of smartphones (or knowledge on using the app) and online-payment methods (i.e., digital divide), or the physical ability to ride the shared vehicles (i.e., physical divide, such as for those elderly, female, and disabled), or the affordability to use shared micro-mobility services regularly (i.e., economic divide) (Bateman et al., 2021;Chen et al., 2020a;Nikiforiadis et al., 2021;Uteng et al., 2019).Besides, the privacy concern on using digital surveillance systems and safety concerns on riding may also prevent some individuals from adopting shared micro-mobility, regardless of the access to the services (Groth, 2019;Sanders et al., 2020).
In general, studies have revealed a lower access to and usage rate of shared micro-mobility in marginalized areas and traditionally transport disadvantaged groups by different approaches (see review Dill & McNeil, 2021).Some studies examined the spatial distribution of available shared micro-vehicles (Bai & Jiao, 2021;Mooney et al., 2019 ;Qian et al., 2020) and/or their services areas (Brown, 2021;Hosford & Winters, 2018;Qian et al., 2020), and then linked them with the neighborhood sociodemographic features.Similarly, a few papers also mapped the spatial pattern of shared micro-mobility ridership (e.g., McKenzie, 2019McKenzie, , 2020;;Nickkar et al., 2019;Qian et al., 2020).Some others analyzed the socio-demographic profile of shared micro-mobility users (Bieliński & Ważna, 2020;Laa & Leth, 2020;Reck & Axhausen, 2021) and compared that with the general population within the service areas (Gavin et al., 2016).Besides, as mentioned above, many studies have regressed personal and neighborhood attributes on shared micromobility use to examine how it is adopted differently among spatial areas and groups (e.g., Barbour et al., 2019;Chen et al., 2020b;Guo & Zhang, 2021;Reck & Axhausen, 2021).
Specifically, shared bikes, e-bikes, and e-scooters were all found to be less accessible in low-density areas (Brown, 2021;McKenzie, 2020) and for lower-income and minority populations (Bai & Jiao, 2021;Brown, 2021).In particular, dockless shared micro-mobility services are more equitable than the docked ones in terms of spatial access (Mooney et al., 2019;Qian et al., 2020).In terms of usage, minority groups, lowereducated people, females, and older adults are generally less likely to be members and/or habitual users of bikeshare compared with their counterparts (Bieliński & Ważna, 2020;Dill & McNeil, 2021).Similar social disparities have been found for shared e-scooter adoptions, especially for gender and age (Mouratidis, 2022), though shared escooter users are relatively more representative of the general population than shared bike users (Reck & Axhausen, 2021).Findings on the association between income and shared micro-mobility use are mixed.For example, while many studies indicated that low-income people use shared bikes less (Chen et al., 2020b;Reck & Axhausen, 2021), some others found a negative relationship between income and shared bike use (e.g., Barbour et al., 2019).Besides, a few studies found that car owners have lower access to (Brown, 2021) and usage rate of (Guo & Zhang, 2021) shared micro-mobility compared with those without cars, which means that the unequal distribution of services may limit the role of shared micro-mobility in substituting cars, and thus their environmental and sustainable benefits.However, some unfavored groups in transport like the disabled people have less been discussed in the equity research of shared micro-mobility (Dill & McNeil, 2021).

How shared micro-mobility helps reduce transport poverty?
Despite the evidence on the equity of shared micro-mobility access and use, to what extent it can help reduce transport poverty is still less understood.Some scholars suggested that shared micro-mobility services can improve people's mobility and accessibility levels as an additional flexible travel option, which is convenience for short trips and trip chaining (Bullock et al., 2017;Chen et al., 2020a).Meanwhile, they could also improve people's access to public transport as a first/last mile solution to access transit stations (Bieliński & Ważna, 2020;Chen et al., 2020b;Hirsch et al., 2019).However, as found in many studies, the speed and capacity limitations of shared micro-mobility constrain its capability to substitute private cars and cover long-distance trips (Barbour et al., 2019;Fishman, 2016;Nikiforiadis et al., 2021).The extra monetary cost and sensitivity to weather in adopting shared micromobility may also limit its regular use in daily travels (Hosseinzadeh et al., 2021;Nikiforiadis et al., 2021).Besides, shared micro-mobility users are more exposed to road accidents and vehicle emission pollution than auto travelers (Ricci, 2015).All these issues may weaken the role of shared micro-mobility in help alleviating transport poverty.
Limited research has investigated the impact of shared micromobility on transport poverty, and most of them focused on accessibility improvement based on simulation approaches (e.g., Abouelela et al., 2024).Tsigdinos et al. (2024) simulated and found that the greatest change in workplace accessibility would occur in areas with lowest accessibility after the implementation of a metropolitan cycling network in Athens, suggesting that bicycle-oriented multimodal transport system can help increase accessibility and transport equity.However, this study did not examine the influence of bikeshare services on accessibility directly.Qian and Niemeier (2019) simulated the accessibility benefits from developing bikeshare services by comparing the accessibility to jobs and services at the census block level under two scenarios: one by walking using the current pedestrian system, and the other by (assumed) perfectly equipped bikeshare services.The findings revealed the potential of bikeshare services in improving the accessibility of disadvantaged communities in Chicago and Philadelphia.The research of Chen et al. (2022) provided empirical evidence on this conclusion.They collected the bikeshare users' self-reported accessibility improvement that bikeshare use has brought in Beijing, China, and most users agreed that bikeshare use made accessing activities easier.
Even fewer studies have examined whether shared micro-mobility use influences transport poverty equally for different population X. Guan et al. groups.Chen et al. (2022) made an initial attempt by examining the influences of the bikeshare users' personal attributes on their selfreported accessibility improvement.They found that male users tend to enjoy more benefits in accessibility from using bikeshare than female users.In another simulation-based study in Louisville, USA, Abouelela et al. (2024) found that disadvantaged groups do not receive more accessibility gains from replacing other modes by shared e-scooters than the rest of population.
However, how bikeshare influences other dimensions of transport poverty, such as mobility level (i.e., availability of transport options), travel cost (e.g., monetary, time, and emotional cost), and exposure to externalities (e.g., unhealth or unsafe travel conditions), is still unclear.Moreover, the influence of shared electric micro-vehicles (i.e., shared ebikes/e-scooters/e-mopeds) on transport poverty, which may differ from that of bikeshare because of higher travel speed (Sanders et al., 2020), also remains less explored so far.Therefore, more studies on how shared micro-mobility services affect different population groups' transport poverty are needed to better understand their equity implications.

Conceptual framework
Fig. 1 illustrates the conceptual model of this research.We are interested in (1) how the use of shared micro-mobility is linked to personal (e.g., age, gender, car ownership, etc.) and environment attributes (e.g., urban vs. suburban neighborhoods), and (2) how its use contributes to reduced transport poverty for different population groups (e.g., high-income vs. low-income groups).In line with previous studies, we assumed that shared micro-mobility use is determined by personal attributes including socio-demographics and the access to different transport modes (e.g., private cars and bicycles), as well as environmental attributes like built environment factors (Fishman, 2016;Reck & Axhausen, 2021;Shen et al., 2018).All these variables, together with the shared micro-mobility use, also directly influence individuals' perception of their transport poverty levels (Delbosc & Currie, 2011;Guzman et al., 2017).In addition, since previous studies have suggested that some disadvantaged groups may not receive comparable accessibility benefits from shared micro-mobility services (Abouelela et al., 2024;Chen et al., 2022), we also examined whether individual sociodemographics moderate the impact of shared micro-mobility use on transport poverty.
Other links among the factors of shared micro-mobility use are also possible.For example, the residential location choice could be influenced by individual socio-demographics (Guan et al., 2020;Mokhtarian & Cao, 2008).Besides, as suggested by previous studies (Guan & Wang, 2020;Van Acker & Witlox, 2010), access to private cars and other transport modes can also be affected by socio-demographics as well as the built environment.However, since this paper focuses on the social and spatial equity of shared micro-mobility use and its impact on transport poverty, we will not examine the associations between these variables and their indirect impacts on shared micro-mobility use in following modelling process.Besides, natural environment (e.g., the weather condition) is not included because shared micro-mobility use is measured weekly rather than daily in our survey.The modelling results will provide a more comprehensive picture on the equity implications of different shared micro-mobility services.

Data
Data used in this research comes from an online questionnaire survey conducted from June to September 2022 in three European cities, including Utrecht (Netherlands), Malmö (Sweden), and Great Manchester (UK).The three cities differ in terms of transport infrastructure and shared micro-mobility services implemented.
Utrecht is the fourth most populous municipality in the Netherlands with about 360,000 inhabitants living in an area of 99.21 km 2 .Like other Dutch cities, Utrecht is a cycle-oriented city with well-developed bicycle paths and nearly 50 % of trips within the city were made by bicycle in 2021.In 2022, public transport rental bicycles (OV-fiets), shared e-bikes & e-mopeds (Tier), and shared e-cargo bikes (Cargoroo) were available in Utrecht.
Malmö is the third-largest city in Sweden with an areas of 76.81 km and about 350,000 inhabitants.The city has a bike path network of km.The recent travel survey conducted in 2018 indicates that cycling accounts for around 25 % of the trips within the city.In 2022, shared escooters (Lime, Tier, and VOI) and both docked (Malmöbybike) and dockless bikeshare (Donkey Republic) were provided in the city of Malmö.
Greater Manchester is a city region in North West England, covering an area of 1276 km 2 , home to over 2.8 million people.Trips in Greater Manchester are dominated by cars and public transport, while cycling level is very low (around 5 %).In 2022, shared bikes/e-bikes (the Bee Network Cycle Hire) and shared e-scooters (Lime) were available in the city of Manchester and some other regions of Great Manchester (e.g., Salford and town of Trafford).
In all the three cities respondents were recruited from residents in the areas where shared micro-mobility services were available via local survey companies.The primary aim of the survey is to understand how shared micro-mobility is used in combination with existing transport modes and what implications this has on sustainability, equity, and inclusion.To achieve this goal, the questionnaire (available in both English and the local language) included information on the awareness and use frequency of different shared micro-mobility services, individual socio-demographics, access to different kinds of transport modes, and a series of statements on perceived transport poverty.For users of any kind of shared micro-mobility, we also asked their perceived changes in travel mode choice, accessibility, number of activities, and monthly travel expense that the adoption of shared micro-mobility has brought.The first part of postal code (i.e., first 3 digits in Malmö, 4 digits in Utrecht, and 2-4 digits in Manchester) at the residential location was collected from each respondent.In total, we received 2110 completed surveys from three cities (please refer to An et al., 2023 for more detailed information of the survey).After data cleaning, the final sample for this study includes 1518 respondents, with 278 from Utrecht, 442 from Great Manchester, and 798 from Malmö.

Variables and measurements
Table A1 in appendix lists the original questions in the survey for all the variables we used in this study.For shared micro-mobility use, respondents were asked to report the frequency of using each kind of shared micro-mobility services in their city in the 12 months prior to participating in the survey.Responses were recorded using a 7-point scale: (1) (Almost) never; (2) 1-5 days a year; (3) 6-11 days a year; (4) 1-3 days a month; (5) 1-3 days a week; (6) 4-6 days a week; (7) Daily.For each kind of shared micro-mobility services, we classified the respondents who had used the mode for at least 1-5 days a year as being "users", and those who had used the mode for at least 1-3 days a week as "frequent users".
Variables on transport poverty were derived from 15 statements on perceived disadvantage in daily travel in 5-point scale.We used principal component analysis with oblique rotation to identify the latent constructs in SPSS version 28.0.Oblique rather than orthogonal rotation was adopted because the latent factors on perceived transport poverty might be correlated with each other (Field, 2005;Lucas et al., 2016).The criterion "Eigenvalue > 1" was used to determine the number of factors.As Table 1 shows, the factor analysis resulted in five latent constructs of perceptions on transport poverty: perceived mobility (the options available for daily travel), digital barrier, perceived accessibility (access to the workplace and other facilities), travel cost (monetary, time and emotional costs in daily travel), and healthy travel (safe and healthy travel conditions).The Cronbach's alpha is larger than 0.7 for all the factors, indicating a high internal consistency for every factor.The scores of these factors (except digital barrier) will be used as the dependent variables in following transport poverty models.Respondents were considered to be suffering from transport poverty if the factor scores of digital barrier and travel cost are high, or if scores on other factors are low.Otherwise, we assumed that the respondents are likely to experience "transport adequacy", i.e., a positive subjective assessment of the quality and sufficiency of one's transport options and associated life outcomes (Ettema & Geigenmüller, 2023;Fu et al., 2024).For shared micro-mobility users (shared bike users as an example), we also collected the self-reported changes in destination accessibility and travel expense after adopting shared micro-mobility by the following questions: "Has the use of shared bikes impacted your travel in other ways: 1) the convenience to access daily travel destinations; 2) monthly travel expense" (responses in 5-Likart scale, from much decreased to much increased).
To investigate the socially inequality in shared micro-mobility use and its benefits on transport adequacy, we include some commonly used individual and household socio-demographics in our models, including age, gender, employment status, education level, the presence of children, and household income.In particular, we also asked the respondent to indicate if he/she has "a condition that substantially limits basic physical activities such as walking, climbing stairs, or carrying", and added this variable into the models to capture the equity impact of shared micro-mobility for individuals and groups with different physical abilities.In addition, having a driver's license (for a car) is also included in the models, given that it enables the adoption of shared e-scooters and e-mopeds in our study areas.
Measures on the access to different transportation modes include the ownership of public transport seasonal tickets and the access to private cars, bikes, e-bikes as well as other private micro-vehicles (i.e., motorcycle, e-scooter, e-moped, or e-skateboard).For private bikes and ebikes, we distinguished "ownership" and "regular use from others" when creating the access variable, considering that people who regularly borrow bikes/e-bikes from others may be more likely to take shared bike/e-bike services as an alternative choice than those with their own bike/e-bikes.Unfortunately, similar information on private e-scooters and e-mopeds were not collected in the survey.
The residential built environment is measured at the spatial level that defined by the first part of the postal code in each city.Two variables are used to capture the urbanity level of the home location: population density and the distance from home (the central point of the residential neighborhood) to the city center.We also collected the locations of train, bus, and tram stations as well as cycling routes in the three cities from Open Street Map and created three variables on transport infrastructure by spatial joining their locations with the postal code boundary in ArcGIS Pro: 1) the distance from home to the nearest train station; 2) density of local public transport (bus and tram) stations; 3) density of cycle routes (length per kilometer square).
Table 2 lists the sample profiles in the three cities.In general, our sample is somewhat overrepresented by younger, high-educated, and employed adults, probably because the survey was conducted online.Gender and income level are more balanced in the sample.On average, around 10 % of respondents reported a (more or less) disability that limits physical activities.The sample characteristics differ across the three cities.While car ownership levels were similar among the three cities, Utrecht sample had a higher ownership rate of public transport season tickets, and the sample in Manchester had a lower bicycle ownership level than the samples from other cities.The built environment features also differ across cities. Population density was higher in Utrecht than in Manchester and Malmö.Besides, Manchester had a higher transit density but a lower cycle route density than the other two cities (Fig. A1 in the appendix shows the distribution of rail stations and cycling routes in the three cities).In terms of shared micro-mobility use, as shown in Fig. 2, on average the respondents in Manchester had the highest usage rate of shared micro-mobility, followed by those in Utrecht, while the respondents in Malmö adopted shared micro-mobility least frequently.

Model design
To fully understand the equity impact of shared micro-mobility services, we ran two groups of models in each city following Fig. 1, and selected the appropriate regression method based on the type of the dependent variable: 1) a binary logistic regression model on the use (or not) of each kind of shared micro-mobility services (dichotomous variable), with individual socio-demographics, access to transport modes, and the built environment as independent variables.This resulted in 9 models in total: 4 in Utrecht, 3 in Manchester, and 2 in Malmö.2) two ordinary least squares (OLS) regression models on each of the perceived transport adequacy factor (continuous variable): one with shared micromobility user (or not) added as a new predictor, and the other with frequent shared micro-mobility user (used at least once a week vs. otherwise) added.This resulted in 2 × 4 × 3 = 24 models in total.The interaction terms between "low income" and shared micro-mobility use were also included to capture how population groups with different income levels received transport benefits from adopting shared micromobility differently. 1 Since the two kinds of transport adequacy models differ only in the shared micro-mobility use variable, the estimated coefficients of other independent variables were quite similar.For this reason, their coefficients were only reported for the first model in the following subsections.Besides, we conducted descriptive analyses on the users' self-reported changes in perceived accessibility and monthly travel expense after adopting shared micro-mobility.The results contribute to a better understanding on the casual relationships between shared micro-mobility use and perceived transport adequacy.
Though the factor "digital barrier" reflects the respondent's transport poverty, we included it into the first group of models as a predictor (rather than as a dependent variable in the second group of models) considering that it may affect shared micro-mobility use (negatively) but not the other way around.All the regression models were estimated using the samples in different cities separately in SPSS version 28.0.Before running the regression models, we checked the scatterplot for each pair of the continuous predictors and dependent variables and removed obvious outliers that may bias the estimation.Multicollinearity was checked via the indicator "Variance Inflation Factor (VIF)", which is lower than 5 for all the independent variables in all the models, suggesting no serious multicollinearity in the data (Mason et al., 2003).For all the OLS models, we further tested the heteroskedasticity by plotting residuals against the predicted value as well as White's/Breusch-Pagan test, finding no trends of serious heteroskedasticity that may bias the model estimation.The normality of residuals was checked by the standardized normal probability (P -P) plot.The results showed no serious deviation from normality for all the OLS models.Besides, we also checked the linearity between the dependent variables and continuous predictors by two approaches: 1) to plot the standardized residuals against each of the continuous predictors; 2) augmented componentplus-residual plot (i.e., augmented partial residual plot).Both plots suggested that non-linearity is not a critical concern in our models.

Social/spatial equity regarding shared micro-mobility use
Table 3 shows the regression modelling results on shared micromobility use.In general, we found that younger and male individuals were more likely to adopt shared micro-mobility services than others in all the three cities, which is in line with previous studies (Dill & McNeil, 2021), while driving license, the presence of children, and employment status have no or limited impacts on shared micro-mobility use.The effects of education level and household income tend to be mixed.Higheducated and high-income people used shared bikes more but shared escooters less compared to their counterparts.Household income level is also positively associated with the use of shared e-bikes, e-mopeds, and e-cargo bikes in Utrecht.Surprisingly, we found that people with physical disabilities used shared e-mopeds more than others (in Utrecht), but this is not the case for other shared micro-mobility services.This finding could be explained by the fact that e-moped, as a "sitting" e-scooter, requires much less physical efforts than (e-)bikes or conventional e-scooters.Digital barrier does not show a significant impact on the use of most shared micro-mobility services (except shared e-cargo bikes).This is probably because the respondents in this research, who completed our questionnaire survey online, were generally familiar with the use of smart phones and transport apps.
Access to other transportation modes influences shared micromobility use substantially.In all the three cities, people with public transport season tickets used shared bikes significantly more than others, which is reasonable as they may take shared bikes as the first/last mile solution to access public transport stations (Chen et al., 2020b;Hirsch et al., 2019).Interestingly, we found that shared micro-mobility users also had better access to private micro-vehicles, such as bikes and e-bikes in all the three cities.This reveals that the absence of private micro-vehicles is not an important reason for using shared micromobility.Instead, the familiarity with these micro-mobilities may facilitate people to approach shared micro-mobility more easily.Besides, access to private cars also shows a positive impact on the use of shared micro-mobility in Manchester and Malmö.
The residential built environment features also affect shared micromobility use.In accord with existing studies (Brown, 2021;McKenzie, 2020), we found that shared micro-mobility services were adopted more in high-density and central areas in Malmö.Meanwhile, shared bikes and e-cargo bikes were used more in denser areas in Utrecht, where the supply of these services was relatively sufficient.Notably, the distance to the city center shows a negative impact on shared e-bike use in Utrecht, probably because shared e-bikes were not allowed to park in the central areas of Utrecht at the time of survey.In Manchester, shared e-bikes/escooters were used more in areas with more transit stations, again, likely as a mode to access public transport.The impacts of the distance to train station and cycle route density on shared micro-mobility use are relatively limited.

Influences of shared micro-mobility use on transport adequacy
Table 4 presents the results of regression models on perceived transport adequacy factors.In line with previous research (e.g., Gössling, 2016;Lucas et al., 2019), household income level is positively related with perceived transport adequacy in Malmö.Interestingly, we found that elderly people were less likely to consider themselves as being disadvantaged in daily travel in all the three European cities.
1 Interaction terms regarding other individual socio-economics, such as gender, elderly (age > 50), and access to car were also examined initially, but were excluded from the final models due to high multicollinearity.
X. Guan et al.Employed individuals reported lower perceived mobility levels and higher travel cost in Malmö, which might be attributed to their commuting burdens.Unsurprisingly, physical disabled people reported a significantly lower transport adequacy level than their counterparts in both Utrecht and Malmö.All other socio-demographic variables show very limited impacts in perceived transport adequacy models.Access to private cars shows a positive impact on perceived mobility level in Utrecht, but a negative impact in Manchester.This result indicates that car owners in Manchester were more likely to be captive car users who had little alternative transport options.Context difference was also found for the access to private bikes, which contributes to perceived transport adequacy only in Utrecht and Malmö, possibly because of the better cycling infrastructure in these two cities.Access to other private micro-vehicles and the ownership of public transport tickets do not benefit perceived transport adequacy in all the cities.
Regarding the built environment, as expected, we found that residents in neighborhoods with high population density were more favored in transportation than their counterparts in Utrecht.Meanwhile, the proximity to the city center shows a positive association with perceived mobility and accessibility levels in Malmö.What should be noted is that the distance to the city center, as a city-scale location variable, may be associated with or even determine other local built environment variables, such as street design and accessibility to transit stations and jobs (Guan & Wang, 2019;Naess, 2022), and therefore affect people's The association between shared micro-mobility use and perceived transport adequacy is mixed.Not surprisingly, Table 4 shows that a high use frequency strengthens the role of shared micro-mobility use in transport adequacy in general."Frequent use" of shared micro-mobility exerts more significant effects on transport adequacy factors (Model II) than "use or not" (Model I).Such influences differ across different types of shared micro-mobility services and cities.Frequent use of shared bike is positively associated with perceived mobility and healthy travel

Table 5
Self-reported changes of perceived accessibility and travel cost after using shared micro-mobility.condition for low-income people, but only in the two UK and Sweden cities.In contrast, it has a tendency to increase the users' perceived mobility more greatly for high-income (than low-income) groups in Utrecht, the Netherlands (though the p-values of corresponding coefficients are slightly over 0.1).This result may be partially attributed to the fact that shared bike services in Utrecht (OV-fiets) were mostly connected with train stations as public transport rental bicycles, therefore can better serve the high-income groups who travelled by train more often.Similarly, frequent shared e-scooter use benefits the users' perceived mobility only for the low-income groups in UK/Sweden cities.However, the opposite is true for frequent shared e-bike use in UK.These results suggest that shared e-scooters could help low-income people to get rid of transport poverty more effectively than shared e-bikes.Besides, frequent shared e-moped users also reported a higher mobility level than others in Utrecht, regardless of their household income level.Shared e-cargo bike use has limited influence on perceived transport adequacy factors.

Variables
Findings regarding accessibility and travel cost are somewhat unexpected.Frequent shared e-bike use is negatively associated with perceived accessibility in Utrecht, especially for high-income groups.Similarly, for high-income people in Sweden, frequent shared bike users tend to had a lower perceived accessibility level than others.Meanwhile, frequent shared bike use is positively related with perceived travel cost for both groups in Utrecht, and such positive association was also found between shared e-bike use and travel cost in UK.These results may suggest that the shared micro-mobility use does not benefit accessibility and travel affordability.However, the reverse influences may play a role here.Respondents who perceived themselves as bearing high travel cost or low accessibility may be more likely to adopt shared micro-mobility as an alternative transport mode in daily life.
To better understand the causal influences between shared micromobility use and perceived accessibility and travel cost, Table 5 summarized the users' self-reported changes in perceived accessibility and monthly travel expense after adopting each kind of shared micromobility.In line with previous studies (Chen et al., 2022;Qian & Niemeier, 2019), the majority of respondents (in all the three cities) reported a similar or increased accessibility, suggesting that the use of shared micro-mobility did help improve accessibility to some extent, and this applied for both high-income and low-income groups.In this case, the reverse influence should be the main explanation of the negative association between shared micro-mobility use and perceived accessibility in the Netherlands and Sweden cites.This means that people with lower levels of perceived accessibility in these areas used shared bikes/e-bikes more and still experienced a relatively lower accessibility afterwards.This is also the case for shared e-bike use and travel cost in Manchester, UK.However, we found that there are more shared bike users in Utrecht who reported an increase than a decrease in travel expense.This result reveals that shared bike use did bring extra travel cost in Utrecht, possibly due to the increased expense of additional train use resulted from adopting public transport rental bicycles (OV-fiets).Nevertheless, cautions should be paid when interpretating these results since the self-reported changes are inevitably influenced by recall bias to some extent.

Discussion
Our findings suggest that current shared micro-mobility services in European cities are still facing challenges in promoting transport equity.First, some transport disadvantaged groups adopted shared micromobility less than their counterparts.In line with previous research (Dill & McNeil, 2021), it was found that in our sample the elderly, females, and those without private cars and micro-vehicles were less likely to use shared micro-mobility.Residents in low-density and marginalized areas also had a lower chance to benefit from shared micro-mobility services (Brown, 2021;McKenzie, 2020).Second, we found that shared e-bikes did not benefit the transport inequality between income groups in the car-oriented city (i.e., Manchester), since its frequent use increased the perceived mobility level only for high-income people there (Table 4).Third, shared micro-mobility played a limited role in improving perceived accessibility and travel affordability.In line with the study of Chen et al. (2020a), the respondents from Utrecht and Malmö in our survey reported that shared micro-mobility use helped improving the convenience to access daily travel destinations slightly (Table 5).Nevertheless, shared e-bike (or bike) users still had a significantly lower perceived accessibility level than non-users (Table 4).Moreover, shared micro-mobility use may also generate additional travel expense in these areas (Table 5), thus aggravate the economic burden of daily travel for some people.
Nonetheless, shared e-scooters/e-mopeds showed a greater potential in facilitating transport equity than shared bikes/e-bikes.We found that low-income people received comparable or more mobility benefits from adopting shared e-scooters and e-mopeds than high-income people, probably because these shared electric vehicles could be used as an alternative of private cars by low-income groups.In addition, shared emopeds can provide an additional mobility option for individuals with physical disabilities and therefore have the potential to reduce transport inequality between them and the others (Table 4).However, safety could be a concern for the use of these high-speed shared electric vehicles.Shared e-mopeds have been banned in Utrecht at the year 2023 because the municipality found that their usage may generate safety risks and didn't fit the aimed image of a "bicycle city".Meanwhile, there are speed-restriction zones for shared e-scooters in many cities for safety issues.Future policy makers need to balance the travel speed restrictions and safety considerations of shared micro-mobility to maximum their benefits on transport equity.
Context difference exists in terms of the impact of shared micromobility on transport equity.While shared bikes were implemented in all the three cities, they showed the greatest potential in improving transport equity between different income groups in the car-oriented city (i.e., Manchester) and lowest potential in the cycle-dominant city (i.e., Utrecht).On the one hand, shared bikes were more used by highincome (than low-income) groups in Utrecht, while its users were more balanced in terms of income levels in Manchester.On the other hand, frequent shared bike use contributed to perceived mobility and accessibility levels more greatly for low-income (than high-income) groups in Manchester, while the opposite trend was found in Utrecht.These results suggest that conventional bikeshare does not add too much in daily travel in cycling-oriented cities, especially for the low-income people, therefore can hardly alleviate transport inequality there.
The research findings generate some insights for future development of shared micro-mobility services.First, to promote the social/spatial equity in shared micro-mobility use, it is necessary to further increase people's awareness and acceptance of shared micro-mobility (e.g., via social media).This is suggested by the finding that non-owners of private bike/e-bikes were less represented in shared micro-mobility users in our sample, probably because they had relatively lower familiarity and interest of micro-vehicles from the beginning.Meanwhile, how to attract and enable females and the elderly to adopt shared micromobility also requires attention.Compact development and improved cycling routes could be possible strategies to encourage women to use shared micro-mobility, since studies have found that the presence of cycling infrastructure, good traffic safety, and shorter travel encourage women to a larger extent than men to adopt bikeshare (Dill & McNeil, 2021).Besides, providing shared e-cargo bikes may also help reduce the X.Guan et al. gender disparity in shared micro-mobility use.This is supported by our finding that shared e-cargo bike is the only mode without significant gender difference in use (Table 3), possibly because women could make use of it for childcare conveniently.For elder people, it may be unrealistic to expect that they use shared micro-mobility (especially shared escooters) as much as younger groups do due to on average their weaker physical strength and probably greater safety concerns.Shared (e-)tricycles, which are typically easier to ride and safer for the users, might be an option to facilitate older people to benefit from shared micromobility more equally.
Second, actions are needed to maximize the contribution of shared micro-mobility services to accessibility.Shared micro-mobility use does increase accessibility, but only to a limited extent at present in the sampled cities.More supportive infrastructure for shared micro-mobility use (e.g., parking areas) may help on this issue.Third, affordability should be highly considered in future development of shared micromobility since now its use does not reduce travel cost and even generates extra travel expense in some areas.Lowering the price can help improve its mobility and equity benefits by enabling its frequent usage, especially for low-income people.Besides, a reasonable distribution of shared micro-mobility services is also useful to optimize their equity benefits.Our study suggests the implementation of bikeshare services (ideally combined with public transport) in car-dominant areas, and shared electric micro-vehicles in cycling-oriented areas.Both of them would favor low-income people's mobility more greatly than that of high-income people, thus reduce the disparity between them in terms of transport adequacy (though the latter may result in an increased travel expense for some users).

Conclusions
Our research provides a more comprehensive understanding on the transport equity impact of shared micro-mobility use by investigating its influential factors and its association with perceived transport adequacy (or poverty) between different income groups.Using online questionnaire survey data in three European cities, our research indicates that current shared micro-mobility services play a limited role in promoting transport equity.On one hand, shared micro-mobility services are unequally used among population groups.They are less adopted by transport disadvantaged groups like the elderly, females, non-owners of private cars or micro-vehicles, and suburban residents.On the other hand, shared micro-mobility services have limited potential in reducing the users' transport poverty.It is found that the users tend to have relatively lower perceived accessibility and travel affordability than those non-users.Although shared e-moped/e-scooter use can benefit transport equity by favoring low-income people's mobility equally or more greatly compared with their high-income counterparts, this is not the case for shared bike/e-bike use.In general, the research findings reveal the limited impacts of current shared micro-mobility services on improving transport equity, and generate implications for future development and policy making of shared micro-mobility services.
This study has some limitations and could be extended from several directions.First, the built environment variables were measured at aggregated spatial scales.The estimated impact of built environment variables may suffer from ecology fallacy.Studies with built environment variables at more finite geographical levels are needed to verify the built environment's impact on shared micro-mobility use and transport adequacy.Second, to better understand the impact of shared micromobility services on transport equity, future studies should include the access to these services in transport adequacy models once relevant data is available.Since most shared micro-mobility services in this research are free-floating, their mobility and accessibility benefits are likely dependent on the overall supply of these services around the city.The use of shared micro-mobility should be viewed as a result of both the supply and usage demand of these services.Some individuals may consider that the access to shared micro-mobility services is important for improving accessibility, but use them only occasionally because of their low travel demands.Third, the cross-sectional nature of this study prevented us from drawing a solid conclusion on the causal relationship between shared micro-mobility use and perceived transport adequacy.While we provided some retrospective evidence on perceived accessibility change in response to shared micro-mobility use, studies based on true panel data are required to further validate their causality.Besides, future studies should also explore the underlying reasons behind the unequal use of shared micro-mobility, especially barriers of non-users, which are the base to improve the equity performance of future shared micro-mobility services.

Fig. 2 .
Fig. 2. Usage patterns of shared micro-mobility services by city.

Utrecht
Frequent; L-I: low-income.All coefficients are standardized.

Table 1
Factor analysis result on perceived transport adequacy (poverty).Extraction method: Principal Component Analysis.Rotation method: Oblimin with Kaiser Normalization.Total variance explained: 72.57%.Loadings lower than 0.3 are suppressed.
X.Guan et al.

Table 2
Sample profile.
a Please refer to Table5for the descriptive statistics of self-reported changes in perceived accessibility and monthly travel expense.bThedifference in variables among cities was examined using Chi-square test or one-way ANOVA.***p< 0.01; ** p < 0.05; * p < 0.1.X.Guan et al.

Table 3
Binary logistic regression analysis results on shared micro-mobility use.

Table 4
OLS regression analysis results on perceived transport adequacy (poverty).