Content uploaded by Boniphace Kutela
Author content
All content in this area was uploaded by Boniphace Kutela on Jun 22, 2022
Content may be subject to copyright.
Analysis of Residents' Stated Preferences of Shared Micro-mobility Devices using Regression-Text
Mining Approach
Boniphace Kutela, Ph.D.
Assistant Research Scientist
Roadway Safety Program
Texas A&M Transportation Institute
1111 RELLIS Parkway, Bryan, TX 77807
Email: b-kutela@tti.tamu.edu
Norris Novat
Graduate Research Assistant
Cleveland State University
2121 Euclid Avenue, Cleveland, OH 44115
Email: n.novatbaraba@vikes.csuohio.edu
Emmanuel Kofi Adanu, Ph.D.
Associate Research Engineer
Alabama Transportation Institute,
The University of Alabama, Cyber Hall,
Tuscaloosa, AL 35487, United States of America
Email: ekadanu@ua.edu
Emmanuel Kidando, Ph.D., P.E.
Department of Civil and Environmental Engineering
Cleveland State University
2121 Euclid Avenue, Cleveland, OH 44115
Email: e. kidando@csuohio.edu
Neema Langa
Department of Sociology
University of Nevada, Las Vegas
4505 S Maryland Pkwy, Las Vegas, NV 89154
Email: langan1@unlv.nevada.edu
2
Abstract
Prior to establishing micro-mobility schemes, operators gather residents' willingness to use them. However,
an inadequate survey setting may lead to demand over- or under-estimation. This study used survey data
from Gilbert City, Arizona, to understand the implications of the stated preferences of micro-mobility
devices. The application of multinomial logit regression and text networks revealed a great disparity
between the stated "want" and "use" of micro-mobility devices. Male residents were more likely to respond
that they wanted and would use electric scooters. Conversely, older residents were less likely to respond
that they wanted and would use either electric scooters or dockless bikes. High-income residents were more
likely to want either electric scooters or docked bike-sharing in the city, but do not plan to use them.
Additionally, residents' comments focused more on electric scooters than other schemes. The implications
of the findings to operators, engineers, and planners are discussed in the study.
Keywords: Micro-mobility; electric scooters; bike sharing; text mining
3
1. Introduction
Shared micro-mobility devices such as dockless bikes, docked bikes, and electric scooters, have recently
become the prominent mobility devices in urban areas across the globe mainly for short trips (Abduljabbar,
Liyanage, and Dia 2021). Among the micro-mobility devices, docked bikes, which require a defined
docking station for picking up or dropping off, have existed for over fifty years. On the other hand, dockless
bikes and electric scooters do not need a docking station but a defined imaginary fence where users can ride
within (Brown et al. 2020; Almannaa et al. 2020). More than one form of these devices may exist in the
same city. The presence of multiple micro-mobility options presents a wide range of choices to users
(Abduljabbar, Liyanage, and Dia 2021).
Prior to establishing a micro-mobility scheme, operators often utilize survey questionnaires to solicit
residents’ expected use of the schemes (Reck and Axhausen 2021; He et al. 2021; Tuncer and Brown 2020).
Oftentimes, the respondents are asked to state the expected frequency of use of the proposed shared micro-
mobility scheme . Other questions, such as demographics, income, etc., are typically included in the
questionnaires to understand differences and similarities in responses across the population. Generally, the
results from the survey questionnaires are used to estimate the demand and evaluate the factors associated
with ridership of the shared micro-mobility devices (Boniphace Kutela and Teng 2019; Sherriff et al. 2020).
Hulland and Houston (2021) revealed that behaviors might differ significantly. However, it is often difficult
to obtain the revealed preferences during the planning stage. Therefore, a better-crafted survey
questionnaire and advanced analysis of the responses are the keys to obtaining the best estimate of demand
and the factors associated with the expected micro-mobility device utilization.
This paper presents a hybrid, regression-text mining, approach to understanding residents' perceptions of
several micro-mobility devices. Both open-ended and closed-ended survey responses collected from Gilbert
City, Arizona, are used to explore the residents' stated preferences on three micro-mobility devices that is
electric scooters, dockless bikes, and docked bikes. It is important to note that prior to when this survey
was conducted, there had been docked bikes operating in Phoenix city since 2018 (Cusimano 2018).
Phoenix is connected to Gilbert City, and thus Gilbert City residents might have been aware of the docked
bike sharing program. The current study leveraged the survey data to: (i) quantify the implied differences
between stated "want" and plans to "use" the micro-mobility devices to aid proper survey questionnaire
setup; (ii) understand differences in the factors affecting want and plan to use of micro-mobility; and (iii)
use open-ended comments from the respondents to explore the patterns of associated factors for residents
who either welcome or object to the micro-mobility devices. The regression-text mining approach involved
three analyses: (i) descriptive statistics; (ii) multinomial regression; and (iii) text network analysis. The
following section provides in detail the literature review, data description, and methodologies, followed by
4
the study results and discussion. The last section of this paper presents conclusions and future research
directions.
2. Literature Review
A study by Lee et al., (2021) investigated the sources of potential heterogeneity in the factors that affect the
willingness to use electric scooters. They observed that the intention to use shared scooters varies among
different socio-demographic groups and level of satisfaction with alternative travel modes. Interestingly,
gender did not impact the preference for shared electric scooters. Sanders et al., (2020) observed that electric
scooters were used more for commuting than recreation, and non-white non-riders were more likely to try
electric scooters. The same study also found that electric scooters were significantly used in place of
walking and bicycling for all trip types. Further, Sanders et al., (2020) added that women were less likely
to use electric scooter due to safety concerns. Contrary to the findings by Sanders et al., (2020),
Bieliński and Ważna, (2020) observed in the Tricity, Northern Poland, that electric bicycles were
predominantly used as first and last mile travel option and for direct travel to places of interest. On the other
hand, electric scooters were used mainly for leisure rides. Eccarius and Lu, (2018) explored the reason
among potential users who were for or against electric scooter sharing. The study focused on their usage
intentions, types of trips for which they would use scooters, and their experiences with and attitudes toward
shared-use mobility. It was revealed that the reasons for or against electric scooters are rooted in different
socio-cultural currents. Indeed, their study findings reinforce the argument for an in-depth assessment of
potential user perceptions and intentions prior to deploying micro-mobility devices in a city.
Furthermore, various factors are associated with the adoption and utilization of bike-sharing (Eren and
Emre Uz 2019). For instance, bike-sharing trip demand is affected by weather conditions and temporal
factors (Talavera-Garcia, Romanillos, and Arias-Molinares 2021; Boniphace Kutela and Teng 2019), built
environment, and land-use patterns (Wang and Akar 2019), socio-demographic factors, and integration with
public transport sharing (Eren and Emre Uz 2019). Politis et al., (2020) explored user's intentions to use
dockless bike-sharing compared to private cars, buses, and walking for short and long-duration trips. They
found that users were more willing to make a switch to bike-sharing systems for short-duration trips, and
bike-sharing was more attractive mobility option for certain population groups. Also, Orvin and Fatmi,
(2021) found that built environment attributes such as bike index (defined as the measure of bike-
friendliness of an area), land use diversity index (defined as the percentage of different land uses),
affect the choice of dockless bike-sharing services. Further, Elmashhara et al., (2022) provided a
systematic review of factors influencing micro-mobility sharing systems user behavior. In that study, the
5
authors explored in depth (i) temporal, spatial, and weather-related factors, (ii) system-related factors, and
(iii) user-related factors that influence micro-mobilty sharing systems.
Hulland and Houston, (2021) argue that most researchers tend to neglect the intention-actual use gap in
their studies. The study observed that there are many important contexts (such as marketing) where the
intention– actual use gap is substantial and where behavioral outcomes are needed (Hulland and Houston
2021). In the micro-mobility context, most studies that utilize survey questionnaires do not solicit whether
the residents want the schemes and plan to use them. Estimation of the demand, for example, without
understanding whether the respondents would use the devices, tends to overestimate, or underestimate the
demand since some residents might want to see the micro-mobility devices in their city, but they are not
sure if they would use them. Similarly, evaluation of associated factors for ridership would result in
unrealistic estimates if a significant portion of respondents want but would not use the devices.
Moreover, closed-ended questions in a survey questionnaire tend to limit the power of expression of the
respondents (Çakır and Cengiz 2016; Zhou et al. 2017). In the past, limited analysis capabilities for
unstructured text data was among the reasons for not including open-ended questions in the questionnaires.
However, in recent years, improvements in data analytics using sophisticated supervised and unsupervised
machine learning algorithms have simplified the analysis of unstructured text data. Although text analysis'
current state of the art shows high powers to produce better insights from texts (Kwayu et al. 2021; Das,
Sun, and Dutta 2016; Boniphace Kutela and Teng 2021), it is argued that text analysis alone lacks the
traditional statistical significance measures. Thus, a hybrid approach that considers text mining and
traditional regression is preferred (Boggs, Wali, and Khattak 2020). Furthermore, the frequency-based text
mining approaches do not provide an opportunity to visualize the formulation of themes and the connections
of the key topics (Kutela, Das and Dadashova, 2021; Kutela et al., 2021). Text networks provide the
flexibility of visualizing the interaction of the keywords (Boniphace Kutela et al. 2021).
3. Data and Methodology
3.1. Data Source
This study utilized community survey results from the town of Gilbert, Arizona (Town of Gilbert 2021).
The community survey was conducted between January 3rd , 2019, and January 13th, 2019, and a total of
2,814 residents participated. The respondents consisted of 1,155 males and 1,626 females across all ages.
The survey queried the preferences and perceptions of using the micro-mobility devices in the city with a
particular interest in whether they would use them. The micro-mobility devices included electric scooters,
dockless, and docked bikes. Similar to other cities, along with other questions, respondents were asked
whether they wanted to see the shared micro-mobility schemes in their city. Unlike most surveys,
6
respondents were also given an open-ended question to provide their general comments. Along with
whether respondents want and will use the system, their gender, age, and zip codes were also recorded.
Since previous studies have indicated that the ridership of micro-mobility devices is influenced by age as
well as the economic status of the users (Mitra and Hess 2021; Yang et al. 2020; Meerah Powell 2019),
these variables were also included. The zip codes helped this study to establish the median household
income from the US government census website (US Government 2021), which was used as proxy for the
economic status of residents. Furthermore, the ratings of quality of education and the commuting time
extracted from an open-source website (Bestplaces, 2021), were added in the analysis.
3.2. Multinomial Logit
Survey respondents were presented with several discrete choice options accounting for their want and plan
to use the micro-mobility devices. A respondent can choose to want a device and use it, want but not use it,
not want it but will use it, and lastly, not want it and will not use it. However, ideally, a relatively small
portion of respondents are expected to choose to use a device that they do not want in the city. Therefore,
three unordered possible outcomes were created for this study, which are want the device in the city and
will use it, want it in the city but will not use it, and do not want it in the city and will not use it.
Considering that the choice options available to respondents are discrete and unordered, the multinomial
logit modeling technique was used to establish the relationship and to predict the likelihood of different
possible choices for a given set of independent variables. The multinomial logit models have been used in
many studies, including highway safety (Tay et al., 2011; Kutela et al., 2022; Chimba et al., 2012) and
transportation planning (Ashford and Benchemam 1987), among others. This model can be applied to assess
utility maximization where an individual is assumed to have preferences defined over a set of alternatives
or choice options:
The error terms in this framework are assumed to be independently and identically distributed with identical
extreme value distribution (McFadden, 1981); the cumulative distribution function is:
Based on this specification, the choice probabilities,
7
where 'i' indexes the observation or individual, is the vector of estimable parameter, is the vector of
explanatory variables that affect the likelihood of choices, and 'j' and 'm' index the choices.
Relative risk ratio (RRR) is often computed by exponentiating the multinomial logit coefficients, to
investigate how a variable affects a choice outcome in comparison to the referent group. An RRR > 1
indicates that the probability of choosing particular category relative to the comparison group increases as
the variable increases (continuous variable) or in the presence of specific group (categorical
variable). Similarly, an RRR < 1 indicates that the probability of the chosen outcome relative to the
comparison group outcome decreases as the variable increases.
3.3. Text Network Analysis
Analysis by text networks is common in the field of social networks, but it is a relatively new technique in
the field of transportation engineering. It is simply a family of techniques that dissects relationships between
units of analysis and unveils existing relationships in an unstructured data (Yoon and Park 2004).
In text networks, the nodes or clusters often comprise keywords of interest. The co-occurrences of these
keywords are represented in a network as edges, with a thicker edge describing a stronger frequency
relationship of the keywords (B. Kutela, Novat, and Langa 2021). The clustering together of the keywords
is represented as communities, which portray similar patterns (Kim and Jang 2018).
Four key steps are used in the text network analysis: i) Preparing and normalizing texts, ii) creating text
networks, iii) visualizing text networks, iv) Detecting the context of the keywords (Andrew Bail 2016).
Data preparation and normalization involves cleaning the data, converting the words to lower case letters,
and filtering non-textual data for ease and accuracy of the analysis. The structured texts are stored in a
matrix of keywords. During a network mapping, the algorithm searches a pair of keywords and their
frequency in the matrix. The pair is then mapped with its frequency portrayed by the edge size. The
algorithm then searches for the next pair; if the next pair has one of the keywords already mapped, the
algorithm then maps the second keyword and then add the frequency of both keywords as well as the edge;
if not, then a new pair is mapped (B. Kutela, Novat, and Langa 2021). The text network analysis in this
research was conducted using the R-statistical software using the quanteda and igraph packages (R Core
Team 2020; Benoit et al. 2018). After mapping, the discussion of the resulting network was done. In this
study, the discussion was based on the network topology, keywords, and co-occurrences of the keywords.
8
4. Results and Discussion
Three types of results are presented: descriptive analysis, multinomial logit, and text analysis. The
descriptive analysis results facilitate quantifying the distribution of the respondents who need and will use
the micro-mobility devices. The multinomial logit results facilitate to understand the association between
the use of micro-mobility devices and predictor variables such as age, gender, and income. On the other
hand, text network results provide the patterns in the respondents' narratives for different wants/use of
micro-mobility devices.
4.1. Descriptive Analysis Results
The findings from the descriptive analysis of the respondents' views in Figure 1 revealed that, on average,
the majority of them (48%) do not want and would not use the micro-mobility devices, while 29%
responded that they want it and would use it. About 20.5% of respondents indicated that they wanted it but
would not use it, and the rest of the respondents (2%) showed that they did not want it, but they would use
it. Regarding the question on want and intention to use e-scooters, 49% of the respondents responded that
they do not want and would not use them, while 9% want it but would not use them. About 6% of
respondents do not want e-scooters but would use them if they were available, and 36% responded that they
wanted them and would use them.
[Figure 1 near here]
The question on the “want” and intention to use dockless bikes revealed that 59% of respondents do not
want and would not use them, while 20% responded that they want it but would not use it, and 2% responded
that they do not want it but would use them. About 20% also revealed that they want dockless bikes and
would use them. Furthermore, the response on the use and want of docked bikes showed that 34% of the
respondents do not want and would not use the docked bikes, while 33% want it but would not use them.
The lowest proportion of the respondents (2%) showed that they did not want but would use docked bikes,
while 20% wanted and would use them.
The distributions of selected options across the age of respondents are shown in Figure 2. It can be observed
that generally, the preference for and intention to use micro-mobility devices decreases with an increase in
respondents' age. This observation is with exception to the question on "want docked bikeshare" which
exhibited a fairly constant trend across ages. The option for wants docked bikeshare received the highest
response, ranging from more than 60% among respondents less than 25 years to 55% for people more than
65 years. On the other hand, a small proportion of respondents who would use dockless bikeshare was
observed.
9
[Figure 2 near here]
A comparison across the "want" and the "use" of the same device shows a relatively larger gap for docked
bikeshare compared to other devices. For instance, about 62% of respondents aged up to 24 years would
want bikeshare in the city, but only 39% would use it, which makes a difference of 23 percentage points.
In contrast, the difference between respondents who want and who would use dockless bikes is 18
percentage points for the same age group. Contrary to docked and dockless bikeshare, respondents who
would use electric scooters are higher than those who want them in the city for respondents aged up to 24
years. The gap between the proportion of respondents who "want' and who "would use" micro-mobility
devices increase as the respondents' age increases. Figure 2 shows that for respondents aged 65 years and
above, the gap between those who want docked bikeshare and those who would use them is about 42
percentage points. On the other hand, the gap between respondents who want and those who would use
dockless bikes and electric scooters is about 23 and 16 percentage points respectively for the same age
group.
Generally, the findings from Figure 2 show the importance of adding a specific question regarding the
expected usage of the devices. Most survey questionnaires used in previous studies collected information
on whether residents want the devices/systems in the city as well as the associated factors (Fitt and Curl 2019;
Singh et al. 2021; Scorrano and Danielis 2021). However, this study shows a great difference between wanting and
using a system. The next section provides the associated factors for wanting and willingness to use the three
micro-mobility devices.
4.2. Multinomial Logit Results
Three variables, gender, age, income, education quality, were associated with whether respondents want
the micro-mobility devices in the city and whether they will use them. The "want" and "use" of the micro-
mobility devices were divided into three categories: "Want it and will use it", "want it but will not use it",
and "do not want it and would not use it". The "do not want it and would not use it" category was used as
the base category. Table 1 presents the multinomial logit results.
4.2.1. Electric scooters
According to the results in Table 1, male respondents are more likely to want and intend to use electric
scooters. In fact, the RRR value shows that the chance of wanting and intention to use electric scooters
increases by 46% for male respondents compared to female respondents. Additionally, the chance that a
person will want but not use electric scooter decreases by about 29% for male respondents compared to
female respondents. The possible explanation is that females consider electric sooter dangerous for their
10
safety (Yang et al. 2020). This observation is supported by previous studies (Swordfisk 2019; Asgarzadeh et al. 2017),
which revealed that males are more likely to ride electric scooters than females, although Meerah Powell,
(2019) made a contrary finding. The model estimation results further show that there is a negative
association between age and the use of e-scooters. As the respondents' age increases, the chance of wanting
and planning to use the electric scooters decreases.
[Table 1 near here]
Compared to respondents aged below 25 years, respondents aged 25-34 years are less likely to want the e-
scooters and would not use them. On the other hand, respondents aged 35 years and above are more likely
to want the electric scooters in the city but not use them. This observation shows a dichotomy between the
"want" and the intention to use electric scooters. Compared to respondents aged below 25 years,
respondents aged 65 years and above are about 82% times less likely to want and intend to use electric
scooters. This finding is similar to Lin, Yang and Suan, (2007), who argue that elderly people are not able
to handle electric scooters. Similarly, respondents aged 45-54 and 55-64 are 66% and 83% less likely to
want and would use electric scooters. Lastly, respondents who are living in high-income zip codes are less
likely to want and intend to use electric scooters. Hosseinzadeh et al., (2021) also concur with this finding
as they observed that high-income area residents own private means of travel vehicles and are less likely to
use public electric scooters. Additionally, for education quality variables with poor/average as base
category, respondents with excellent education quality are 14% more likely to want but would not use
electric scooters. While, in response to wants it and would use it question, respondents with excellent
education are 3% less likely to want but would not use electric scooters. Furthermore, the model results
show that respondents are 16% more likely to want but wouldn't use them and 17% less likely to want and
would use them if they commute for more than 30 min compared to those whose commute time is less than
30 min.
4.2.2. Dockless bike-sharing
Some contrary and similar observations to electric scooters were inferred from the multinomial logit results
with respect to dockless bike-sharing (Table 2). The RRR indicates an 8% decrease in chance for male
respondents to want and would use the scheme and a 22% decrease in chance to want it but have no intention
to use it compared to the female respondents. It can be inferred from this finding that female respondents
are less interested compared to male respondents in dockless bike-sharing devices. This finding is consistent
with those found by other researchers such as Gu, Kim and Currie, (2019) and Orvin and Fatmi, (2021). As
observed with respondents' attitude towards electric scooters, residents' want and intention to use dockless
bike-sharing are also negatively associated with respondents' age. The chances of residents wanting but not
11
intending to use the dockless bike-sharing system decrease with respondent age. Sun, (2018) also found
similar results in survey results of the use of dockless bike-sharing system, where it was found that 52% of
the users were 18-30 years and 48% were between 30 and 60 years old. Other studies, such as (Politis et al.
2020; Sherriff et al. 2020), also made similar findings. Also, the RRR value reveals that respondents in
income group above $100k are less likely to want the dockless bikes but wouldn't use them and 12% less
likely to want and use the dockless bikes. The dissimilarity with the findings from the electric scooter may
be due to the fact that dockless bike-sharing is easier and more convenient to use (Liu and Xu 2018). This
perhaps explains the higher likelihood that people in the highest income category would want and actually
use dockless bikes. Furthermore, respondents with excellent education quality are 11% more likely to want
but won't use and 9% more likely to want and use dockless bikes. Additionally, respondents with longer
than 30 minutes of commuting time are 11% more likely to want but won't use the dockless bikes, and 22%
less likely to want and use the dockless bikes. A higher percentage of use is expected since the commuters
are not required to return the bikes after a long commute distance (Guo et al. 2021; Chen, van Lierop, and
Ettema 2020).
[Table 2 near here]
4.2.3. Docked bike-sharing
The results in Table 3 show that male respondents were 27% less likely to want but would not use docked
bike-sharing but have a 16% increase in the chance that they want and would use docked-bike sharing than
female respondents. Response by the age groups showed a steady increase in RRR values with an increase
in age group on the "want it but would not use it" alternative compared to the base category (Below 25).
The majority of the older respondents want the micro-mobility device, perhaps because of its
benefits but would not use it due to safety concerns and physical constraints. Similarly, different
travel habits of older people may have contributed to this finding. However, the RRR values of the
counter-question regarding "wants it and would use it" shows a lower likelihood against the base category
(Below 25) for the remaining age groups. This again explains the fact that there are underlying factors that
limit the older population group from using the docked bike-sharing program. Shaheen and Cohen, (2019)
and Chen, van Lierop and Ettema, (2020) observed that the older population group appears to positively
accept and appraise the program due to its health promotions, but a lower proportion of this group actually
use it. In terms of income category, respondents in income class above $100k showed a decreased chance
(10%) to want docked bike-sharing but would not use it compared to respondents in the base category
(Below $100k). This can be explained by the fact that in high-income neighborhoods leaving these devices
in open space is perceived as a nuisance (Fitt and Curl 2020). However, this income group has a slightly
12
higher chance (1%) of wanting and intending to use the docked bikes. This may be attributable to the
presence of better infrastructure to accommodate docked bike-sharing in higher-income zip codes. This
finding is consistent with (JMD BruceIV, 2020), which also found that the use of the docked bike-sharing
program is influenced by social inequality; it is more accessible to wealthy neighborhoods. In addition,
respondents with excellent education quality are 17% more likely to want but wouldn't use the docked bike-
share and 21% more likely to want and would use the docked bikes. Moreover, respondents with more than
30 minutes of commute time are 11% more likely to want but wouldn't use docked bikes, while 19% less
likely to want it and use docked bikes.
[Table 3 near here]
The Multinomial logistic regression results have facilitated understanding the factors associated with
variations of "want" and "use" of different micro-mobility devices. The different demographic and
economic statuses were observed to vary in the "want" and "use" of the micro-mobility devices. However,
the results do not explicitly show why different factors are associated with the variations of "wants" and
"use" of those micro-mobility devices. Thus, the application of text mining on the open-ended answers was
necessary to understand why respondents either want or do not want and will use or not use the devices.
The next section presents the text network results.
4.3. Text Analysis Results
This section presents the text network results for respondents who want and will use all three schemes, do
not want, and will not use any of the three schemes, wants but won't use electric scooters. The development
of the text network for respondents who want but would not use electric scooters was based on the fact that
electric scooter was the main focus for respondents for all three schemes.
4.3.1. Wants and will use all three schemes
Figure 3 presents a text network of the responses related to residents who want and will use all three micro-
mobility devices. The network is centered on the heaviest node of the network - scooters. The keyword
bikes appear less frequently compared to electric scooters as indicated by the size of the nodes. The
observation implies that although the respondents wanted all three devices, most of them related their
preferences to electric scooters. This observation is also reflected in the descriptive analysis (see Figure 1),
where e-scooters had the most responses (36%) of the people who would want and use it. Other keywords
that appear to dominate the network include transportation, people, and love. The presence of the keyword
love and its co-occurrence with transportation and people can imply that not only do the people of Gilbert
want and would use, but they also love these micro-mobility devices as alternative modes of transportation.
13
Furthermore, there are several positive keywords for micro-mobility devices. The keywords such as great,
better, alternative, options, embrace, fun, like signify how the respondents were affectionate to these micro-
mobility devices. Similar studies that justify the findings of this network include Zhang and Mi, (2018),
Cohen, (2019), and Tuncer and Brown, (2020). A relatively small number of respondents are cautious with
these micro-mobility devices. This group is concerned with the safety of the devices as they are likely to
clog the sidewalks; hence the respondents feel that they are dangerous.
[Figure 3 near here]
4.3.2. Do not want and will not use any of the three schemes.
The text network in Figure 4 presents the keywords from respondents who do not want and will not use any
of the three micro-mobility devices. Similar to the respondents who want and will use all three devices, the
network is centered on keyword scooters; given its node size, electric scooters are the devices that most
respondents do not want and will not use. This is similar to the descriptive analysis results of the closed-
ended question responses in Figure 1, where most of the respondents showed that they do not want and
would not use docked bikes.
Several negative keywords are observed in the network. It can be observed that several respondents
described the devices as trashy, dangerous, nuisance, eyesore to express their negative perceptions of these
devices. Respondents also linked these micro-mobility devices with blockage of sidewalks, leave,
everywhere, thus considering them not appropriate devices within the city. Studies by James et al (2019)
and Beatrice Hügler and Birte Manke (2020) also discuss similar negative perceptions of using these micro-
mobility devices and integrate this discussion with the future of autonomous vehicles mobility.
Despite these findings on negative response, the nodes scooters and bikes are still linked to the nodes like,
need, want, which explains that a significant proportion of respondents would want and would use scooters
and bikes.
[Figure 4 near here]
4.3.3. Wants but will not use Electric Scooters.
Since most of the respondents pointed to the electric scooters as depicted by the heavily centered node
scooters in Figure 3 and Figure 4, it was of interest to understand the perceptions of people who would
want but not use electric scooters. Figure 5 presents the text network of the respondents who would want
but will not use electric scooters. The network shows three major keywords, scooters, bikes, and
transportation. The first two-scooters and bikes describe the micro-mobility devices under discussion. The
keywords connected to the three major nodes provide some patterns for this group of respondents. The
14
keyword cars is linked to all three major nodes, implying that the respondents were linking the micro-
mobility devices and cars. It is highly likely that this group of respondents own cars, and do not consider
the transition to micro-mobility. Furthermore, the connection between cars and transportation suggests that
respondents were describing their current device of transportation. A significant proportion of respondents
seems to embrace electric scooters. The respondents show that electric scooters are a good idea because
they provide more options to get around, which matches previous studies (Field and Jon 2021; Espinoza et
al. 2019).
[Figure 5 near here]
Conversely, the association between electric scooters and age is observed in the network. Several
respondents feel that they are too old to use electric scooters, which are suitable for young people. Such
observation is deduced from the keywords young, people, seniors, old, and years (Kobayashi et al. 2019).
Also, the presence of the co-occurred keywords public transportation suggests that a number of the
respondent are linking the electric scooters and public transportation (Clewlow 2019; Choron and Sakran
2019).
5. Conclusions and Future Studies
With the obvious growth in shared micro-mobility transportation, namely, dockless, docked bikes, and
electric scoters across the United States, it is important to understand user preferences to help foster a more
sustainable micro-mobility system for all. Our research explores the perception of respondents on these
schemes in terms of their need in society and if at all people expect to use them with insights from the social
demographics' characteristic from the Gilbert City respondents.
The descriptive statistics suggest that, aggregately, 48% do not want and would not use the micro-mobility
devices, while 29% responded that they want and would use them. Furthermore, 20.5% revealed that they
wanted but would not use them, while the rest of the respondents (2%) showed that they did not want but
would use them. The higher count proportion negatively portrays not wanting or using them. However,
such sentiments are influenced by the location of the facilities, safety concerns, and bikeability of the
schemes in Gilbert city. Overall, the descriptive statistics showed that there is a great difference between
wanting and using micro-mobility devices. Thus, the results point out a very important aspect in setting up
the survey questionnaires. It is important that respondents are asked proper questions that would yield a
better estimation of demand. For this case, utilizing "use" than "want" would be more beneficial.
To address the need for accurate deduction of these findings multinomial logit model that considers gender,
age, and income variables for each of the micro-mobility devices under scope was done. The RRR value
15
shows that the chance of wanting and using electric scooters is higher by 54% for male respondents
compared to female respondents. The higher preference of males towards micro-mobility devices was also
reported in the previous studies. Furthermore, a higher likelihood of wanting and using e-scooters is
observed from younger adults and male residents in lower- and middle-income areas. While younger adults
below 25 and females from middle income ($85k-$100k) areas, have shown a higher likelihood of wanting
and using dockless bike-sharing devices. In addition, male adults of age group 35-44 and income below
$85k and above $100k showed a higher likelihood for docked bike-sharing devices. The significant
disparities in preference of want and use among different micro-mobility devices indicate that each device
addresses the needs of specific groups of people with different social demographic settings. Various factors
such as land use and neighborhood safety status identified in the previous studies, can influence the choice
of a device.
As both descriptive analysis and multinomial regression do not provide detailed reasons for associations of
the want/use of micro-mobility and other predictors, the text mining approach was applied for this purpose.
Although respondents were given the open-ended space to provide their comments without any
constraints/guidelines, the focus for most respondents was on the electric scooters. Some praised them as
an alternative mode of transportation, while others had safety concerns.
Consequently, this study provides further evidence for a study involving a wide range of stakeholders,
particularly transport system users, to understand their perspectives, needs, and challenges with the
transportation system. Also, it contributes to practice by providing micro-mobility device operators with a
range of influencing factors that play a crucial role in attracting micro-mobility users, helping to guide their
strategies and leverage their activities. Furthermore, this study highlights safety as a concern by the
respondents, and such situations call for more safety considerations by transportation officials.
Similar to other preference analysis studies, this study inevitably carries limitations. The study used data
from Gilbert city, Arizona; the findings might be applicable to cities with similar demographic
characteristics to Gilbert city. Further, it is expected that future studies would include more activity-based
variables to understand the heterogeneity of residents' perceptions on these emerging technologies at a
planning level. Further, future studies can cover a wider geographical area to include responses from as
many residents as possible. In addition to that, future studies can explore the intention to use versus the
actual use gap; many people may indicate in the surveys that they intend to use, but then they do not use.
Disclosure statement
No potential conflict of interest was reported by the author(s).
16
References
Abduljabbar, Rusul L, Sohani Liyanage, and Hussein Dia. 2021. “The Role of Micro-Mobility in Shaping
Sustainable Cities: A Systematic Literature Review.” doi:10.1016/j.trd.2021.102734.
Almannaa, Mohammed Hamad, Huthaifa I. Ashqar, Mohammed Elhenawy, Mahmoud Masoud, Andry
Rakotonirainy, and Hesham Rakha. 2020. “A Comparative Analysis of E-Scooter and e-Bike Usage
Patterns: Findings from the City of Austin, TX.” International Journal of Sustainable
Transportation. Taylor and Francis Ltd. doi:10.1080/15568318.2020.1833117.
Andrew Bail, Christopher. 2016. “Combining Natural Language Processing and Network Analysis to
Examine How Advocacy Organizations Stimulate Conversation on Social Media.” Proceedings of
the National Academy of Sciences 1790–2014. doi:10.1073/pnas.1607151113.
Asgarzadeh, Morteza, Santosh Verma, Rania A. Mekary, Theodore K. Courtney, and David C. Christiani.
2017. “The Role of Intersection and Street Design on Severity of Bicycle-Motor Vehicle Crashes.”
Injury Prevention. doi:10.1136/injuryprev-2016-042045.
Ashford, Norman, and Messaoud Benchemam. 1987. “Passengers’ Choice of Airport: An Application of
the Multinomial Logit Model.”
Beatrice Hügler, and Birte Manke. 2020. “Future Mobility Insights: A Trend Receiver Approach.”
Technium Social Sciences Journal. doi:10.47577/tssj.v10i1.1318.
Benoit, Kenneth, Kohei Watanabe, Haiyan Wang, Paul Nulty, Adam Obeng, Stefan Müller, and Akitaka
Matsuo. 2018. “Quanteda: An R Package for the Quantitative Analysis of Textual Data.” Journal of
Open Source Software 3 (30). doi:10.21105/joss.00774.
Bieliński, Tomasz, and Agnieszka Ważna. 2020. “Electric Scooter Sharing and Bike Sharing User
Behaviour and Characteristics.” Sustainability (Switzerland) 12 (22). doi:10.3390/su12229640.
Boggs, Alexandra M., Behram Wali, and Asad J. Khattak. 2020. “Exploratory Analysis of Automated
Vehicle Crashes in California: A Text Analytics & Hierarchical Bayesian Heterogeneity-Based
Approach.” Accident Analysis and Prevention 135 (October 2019). Elsevier: 105354.
doi:10.1016/j.aap.2019.105354.
Brown, Anne, Nicholas J. Klein, Calvin Thigpen, and Nicholas Williams. 2020. “Impeding Access: The
Frequency and Characteristics of Improper Scooter, Bike, and Car Parking.” Transportation
Research Interdisciplinary Perspectives 4 (March). Elsevier Ltd: 100099.
doi:10.1016/j.trip.2020.100099.
Çakır, Hamide, and Özge Cengiz. 2016. “The Use of Open Ended versus Closed Ended Questions in
Turkish Classrooms.” Open Journal of Modern Linguistics 06 (02). Scientific Research Publishing,
Inc,: 60–70. doi:10.4236/ojml.2016.62006.
Chen, Zheyan, Dea van Lierop, and Dick Ettema. 2020. “Dockless Bike-Sharing Systems: What Are the
Implications?” Transport Reviews 40 (3). Routledge: 333–53. doi:10.1080/01441647.2019.1710306.
Chimba, Deo, Daniel Emaasit, and Boniphace Kutela. 2012. “Likelihood Parameterization of Bicycle
Crash Injury Severities.” Journal of Transportation Technologies 02 (03). Scientific Research
Publishing: 213–19. doi:10.4236/jtts.2012.23023.
17
Choron, Rachel L., and Joseph V. Sakran. 2019. “The Integration of Electric Scooters: Useful Technology
or Public Health Problem?” American Journal of Public Health 109 (4). NLM (Medline): 555–56.
doi:10.2105/AJPH.2019.304955.
Clewlow, Regina R. 2019. “The Micro-Mobility Revolution: The Introduction and Adoption of Electric
Scooters in the United States.”
Cohen, Kathleen. 2019. “Human Behavior and New Mobility Trends in the United States, Europe, and
China.” SSRN Electronic Journal. doi:10.2139/ssrn.3475381.
Cusimano, Lauren. 2018. “Bike Shares in Phoenix, Scottsdale, Tempe: Rental, Cost, Apps, Parking |
Phoenix New Times.” https://www.phoenixnewtimes.com/arts/bike-share-guide-phoenix-tempe-
scottsdale-10223591.
Das, Subasish, Xiaoduan Sun, and Anandi Dutta. 2016. “Text Mining and Topic Modeling of
Compendiums of Papers from Transportation Research Board Annual Meetings.” Transportation
Research Record: Journal of the Transportation Research 2552: 48–56. doi:10.3141/2552-07.
Eccarius, Timo, and Chung-Cheng Lu. 2018. “Exploring Consumer Reasoning in Usage Intention for E-
Scooter Sharing.” 運輸計劃季刊 47 (4): 271–96. doi:10.6402/TPJ.
Elmashhara, Maher Georges, Joaquim Silva, Elisabete Sá, Ana Carvalho, and Arash Rezazadeh. 2022.
“Factors Influencing User Behaviour in Micromobility Sharing Systems: A Systematic Literature
Review and Research Directions.” Travel Behaviour and Society 27 (April). Elsevier: 1–25.
doi:10.1016/J.TBS.2021.10.001.
Eren, Ezgi, and Volkan Emre Uz. 2019. “A Review on Bike-Sharing: The Factors Affecting Bike-Sharing
Demand.” Sustainable Cities and Society. doi:10.1016/j.scs.2019.101882.
Espinoza, William, Matthew Howard, Julia Lane, and Pascal Van Hentenryck. 2019. “Shared E-Scooters:
Business, Pleasure, or Transit?” ArXiv, October. arXiv.
Field, Clare, and Ihnji Jon. 2021. “E-Scooters: A New Smart Mobility Option? The Case of Brisbane,
Australia.” Planning Theory & Practice, May. Routledge, 1–29.
doi:10.1080/14649357.2021.1919746.
Fitt, Helen, and Angela Curl. 2019. “E-Scooter Use in New Zealand: Insights around Some Frequently
Asked Questions Inclusive Design for Getting Outdoors-I’DGO View Project Age and Perceptions
of Driverless Vehicle Safety View Project.” doi:10.13140/RG.2.2.13510.93761.
Fitt, Helen, and Angela Curl. 2020. “The Early Days of Shared Micromobility: A Social Practices
Approach.” Journal of Transport Geography. doi:10.1016/j.jtrangeo.2020.102779.
Gu, Tianqi, Inhi Kim, and Graham Currie. 2019. “To Be or Not to Be Dockless: Empirical Analysis of
Dockless Bikeshare Development in China.” Transportation Research Part A: Policy and Practice
119 (January). Pergamon: 122–47. doi:10.1016/J.TRA.2018.11.007.
Guo, Yuanyuan, Linchuan Yang, Yi Lu, and Rui Zhao. 2021. “Dockless Bike-Sharing as a Feeder Mode
of Metro Commute? The Role of the Feeder-Related Built Environment: Analytical Framework and
Empirical Evidence.” Sustainable Cities and Society 65 (February). Elsevier: 102594.
doi:10.1016/J.SCS.2020.102594.
18
He, Yi, Changxin Sun, Helai Huang, Liang Jiang, Ming Ma, Pei Wang, and Chaozhong Wu. 2021.
“Safety of Micro-Mobility: Riders’ Psychological Factors and Risky Behaviors of Cargo TTWs in
China.” Transportation Research Part F: Traffic Psychology and Behaviour 80 (July). Pergamon:
189–202. doi:10.1016/j.trf.2021.04.001.
Hosseinzadeh, Aryan, Majeed Algomaiah, Robert Kluger, and Zhixia Li. 2021. “E-Scooters and
Sustainability: Investigating the Relationship between the Density of E-Scooter Trips and
Characteristics of Sustainable Urban Development.” Sustainable Cities and Society 66 (March).
Elsevier: 102624. doi:10.1016/J.SCS.2020.102624.
Hulland, John, and Mark Houston. 2021. “The Importance of Behavioral Outcomes.” Journal of the
Academy of Marketing Science 49 (3). Springer: 437–40. doi:10.1007/S11747-020-00764-
W/FIGURES/1.
James, Owain, J. I. Swiderski, John Hicks, Denis Teoman, and Ralph Buehler. 2019. “Pedestrians and E-
Scooters: An Initial Look at e-Scooter Parking and Perceptions by Riders and Non-Riders.”
Sustainability (Switzerland). doi:10.3390/su11205591.
JMD BruceIV. 2020. “Local Micromobility Regulations : Implications for Equity and Data Privacy in
Four U.S. Cities.” https://repositories.lib.utexas.edu/handle/2152/83170.
Kim, Youngji, and Soong-Nang Jang. 2018. “Mapping the Knowledge Structure of Frailty in Journal
Articles by Text Network Analysis.” doi:10.1371/journal.pone.0196104.
Kobayashi, Leslie M., Elliot Williams, Carlos V. Brown, Brent J. Emigh, Vishal Bansal, Jayraan Badiee,
Kyle D. Checchi, Edward M. Castillo, and Jay Doucet. 2019. “The E-Merging e-Pidemic of e-
Scooters.” Trauma Surgery and Acute Care Open 4 (1). BMJ Publishing Group: 337.
doi:10.1136/tsaco-2019-000337.
Kutela, B., N. Novat, and N. Langa. 2021. “Exploring Geographical Distribution of Transportation
Research Themes Related to COVID-19 Using Text Network Approach.” Sustainable Cities and
Society 67. doi:10.1016/j.scs.2021.102729.
Kutela, Boniphace, Subasish Das, and Bahar Dadashova. 2021. “Mining Patterns of Autonomous Vehicle
Crashes Involving Vulnerable Road Users to Understand the Associated Factors.” Accident Analysis
& Prevention, November. Pergamon, 106473. doi:10.1016/J.AAP.2021.106473.
Kutela, Boniphace, Neema Langa, Sia Mwende, Emmanuel Kidando, Angela E. Kitali, and Prateek
Bansal. 2021. “A Text Mining Approach to Elicit Public Perception of Bike-Sharing Systems.”
Travel Behaviour and Society 24 (July). Elsevier Ltd: 113–23. doi:10.1016/j.tbs.2021.03.002.
Kutela, Boniphace, and Hualiang Teng. 2019. “The Influence of Campus Characteristics, Temporal
Factors, and Weather Events on Campuses-Related Daily Bike-Share Trips.” Journal of Transport
Geography 78: 160–69. doi:https://doi.org/10.1016/j.jtrangeo.2019.06.002.
Kutela, Boniphace, and Hualiang Teng. 2021. “The Use of Dynamic Message Signs (DMSs) on the
Freeways: An Empirical Analysis of DMSs Logs and Survey Data.” Journal of Transportation
Technologies 11 (01). Scientific Research Publishing: 90–107. doi:10.4236/jtts.2021.111006.
Kwayu, Keneth Morgan, Valerian Kwigizile, Kevin Lee, Jun-Seok Oh, and J.-S Oh. 2021. “Discovering
Latent Themes in Traffic Fatal Crash Narratives Using Text Mining Analytics and Network
Topology.” Accident Analysis and Prevention 150: 105899. doi:10.1016/j.aap.2020.105899.
19
Lee, Hyukseong, Kwangho Baek, Jin Hyuk Chung, and Jinhee Kim. 2021. “Factors Affecting
Heterogeneity in Willingness to Use E-Scooter Sharing Services.” Transportation Research Part D:
Transport and Environment 92 (March). Elsevier Ltd: 102751. doi:10.1016/j.trd.2021.102751.
Lin, Bing Ming, Mo Hua Yang, and Theresa Suan. 2007. “Major Activities of Light Electric Scooter
Development in Taiwan.” World Electric Vehicle Journal 2007, Vol. 1, Pages 155-160 1 (1).
Multidisciplinary Digital Publishing Institute: 155–60. doi:10.3390/WEVJ1010155.
Liu, Ming, and Xifen Xu. 2018. “Dockless Bike-Sharing Reallocation Based on Data Analysis: Solving
Complex Problem with Simple Method.” Proceedings - 2018 IEEE 3rd International Conference on
Data Science in Cyberspace, DSC 2018, July. Institute of Electrical and Electronics Engineers Inc.,
445–50. doi:10.1109/DSC.2018.00072.
Meerah Powell. 2019. “Data Show Women Ride Portland E-Scooters Less Than Men - OPB.”
https://www.opb.org/news/article/electric-scooters-data-women-ride-less-than-men/.
Mitra, Raktim, and Paul M. Hess. 2021. “Who Are the Potential Users of Shared E-Scooters? An
Examination of Socio-Demographic, Attitudinal and Environmental Factors.” Travel Behaviour and
Society 23 (April). Elsevier: 100–107. doi:10.1016/J.TBS.2020.12.004.
Orvin, Muntahith Mehadil, and Mahmudur Rahman Fatmi. 2021. “Why Individuals Choose Dockless
Bike Sharing Services?” Travel Behaviour and Society 22 (January). Elsevier Ltd: 199–206.
doi:10.1016/j.tbs.2020.10.001.
Politis, Ioannis, Ioannis Fyrogenis, Efthymis Papadopoulos, Anastasia Nikolaidou, and Eleni Verani.
2020. “Shifting to Shared Wheels: Factors Affecting Dockless Bike-Sharing Choice for Short and
Long Trips.” Sustainability (Switzerland). doi:10.3390/su12198205.
R Core Team. 2020. “R: A Language and Environment for Statistical Computing. R Foundation for
Statistical Computing.”
Reck, Daniel J., and Kay W. Axhausen. 2021. “Who Uses Shared Micro-Mobility Services? Empirical
Evidence from Zurich, Switzerland.” Transportation Research Part D: Transport and Environment
94 (May). Elsevier Ltd: 102803. doi:10.1016/j.trd.2021.102803.
Sanders, Rebecca L., Michael Branion-Calles, and Trisalyn A. Nelson. 2020. “To Scoot or Not to Scoot:
Findings from a Recent Survey about the Benefits and Barriers of Using E-Scooters for Riders and
Non-Riders.” Transportation Research Part A: Policy and Practice 139 (September). Elsevier Ltd:
217–27. doi:10.1016/j.tra.2020.07.009.
Scorrano, Mariangela, and Romeo Danielis. 2021. “The Characteristics of the Demand for Electric
Scooters in Italy: An Exploratory Study.” Research in Transportation Business & Management 39
(June). Elsevier: 100589. doi:10.1016/J.RTBM.2020.100589.
Shaheen, Susan A, and Adam P. Cohen. 2019. “Shared Micromoblity Policy Toolkit: Docked and
Dockless Bike and Scooter Sharing.” UC Berkeley: Transportation Sustainability Research Center.
Sherriff, Graeme, Mags Adams, Luke Blazejewski, Nick Davies, and Daiga Kamerāde. 2020. “From
Mobike to No Bike in Greater Manchester: Using the Capabilities Approach to Explore Europe’s
First Wave of Dockless Bike Share.” Journal of Transport Geography.
doi:10.1016/j.jtrangeo.2020.102744.
20
Singh, Sandeep, B. Priyadharshni, Challa Prathyusha, and S. Moses Santhakumar. 2021. “Investigating
the Characteristics and Choice of Electric Scooter Users: A Case Study of Tiruchirappalli City.”
Springer, Singapore, 161–71. doi:10.1007/978-981-16-1119-3_15.
Sun, Yiyun. 2018. “Sharing and Riding: How the Dockless Bike Sharing Scheme in China Shapes the
City.” Urban Science. doi:10.3390/urbansci2030068.
Swordfisk, K. 2019. “PSU Researcher Finds Gender Gap in E-Scooter Use | Portland State University.”
https://www.pdx.edu/news/psu-researcher-finds-gender-gap-e-scooter-use.
Talavera-Garcia, Ruben, Gustavo Romanillos, and Daniela Arias-Molinares. 2021. “Examining Spatio-
Temporal Mobility Patterns of Bike-Sharing Systems: The Case of BiciMAD (Madrid).” Journal of
Maps 17 (1). Taylor and Francis Ltd.: 7–13. doi:10.1080/17445647.2020.1866697.
Town of Gilbert, AZ. 2021. “Home • Alex • Gilbert, AZ - Open Data.” https://alex.gilbertaz.gov/.
Tuncer, Sylvaine, and Barry Brown. 2020. “E-Scooters on the Ground: Lessons for Redesigning Urban
Micro-Mobility.” In Conference on Human Factors in Computing Systems - Proceedings.
doi:10.1145/3313831.3376499.
US Government. 2021. “Census .” https://data.census.gov/cedsci/profile?g=1600000US3915000.
Wang, Kailai, and Gulsah Akar. 2019. “Gender Gap Generators for Bike Share Ridership: Evidence from
Citi Bike System in New York City.” Journal of Transport Geography 76 (December 2018).
Elsevier: 1–9. doi:10.1016/j.jtrangeo.2019.02.003.
Yang, Hong, Qingyu Ma, Zhenyu Wang, Qing Cai, Kun Xie, and Di Yang. 2020. “Safety of Micro-
Mobility: Analysis of E-Scooter Crashes by Mining News Reports.” Accident Analysis and
Prevention. doi:10.1016/j.aap.2020.105608.
Yoon, Byungun, and Yongtae Park. 2004. “A Text-Mining-Based Patent Network: Analytical Tool for
High-Technology Trend.” Journal of High Technology Management Research 15 (1). JAI: 37–50.
doi:10.1016/j.hitech.2003.09.003.
Zhang, Yongping, and Zhifu Mi. 2018. “Environmental Benefits of Bike Sharing: A Big Data-Based
Analysis.” Applied Energy. doi:10.1016/j.apenergy.2018.03.101.
Zhou, Ronggang, Xiaorui Wang, Leyuan Zhang, and Haiyan Guo. 2017. “Who Tends to Answer Open-
Ended Questions in an e-Service Survey? The Contribution of Closed-Ended Answers.” Behaviour
and Information Technology 36 (12). Taylor and Francis Ltd.: 1274–84.
doi:10.1080/0144929X.2017.1381165.
21
Tables
Table 1. Multinomial Logit Results for Electric Scooters
Wants it but wouldn't use
Wants it and would use
Estimate
OR
z-stat
P-value
Estimate
OR
z-stat
P-value
Electric Scooters
Gender
Female (ref)
Male
-0.35
0.71
-2.22
0.027
0.14
1.15
1.46
0.145
Age (years)
Below 25 (ref)
Between 25 - 34
0.26
1.30
0.51
0.610
-0.19
0.82
-0.88
0.380
Between 35 - 44
0.36
1.43
0.72
0.469
-0.50
0.60
-2.37
0.018
Between 45 - 54
0.28
1.32
0.56
0.579
-1.07
0.34
-4.86
0.000
Between 55 - 64
0.67
1.95
1.32
0.186
-1.80
0.17
-6.85
0.000
65 and above
0.87
2.39
1.68
0.093
-2.55
0.08
-7.35
0.000
Income
Below $100k (ref)
Above $100k
-0.04
0.96
-0.14
0.892
-0.17
0.85
-0.92
0.356
Education quality
Poor/average (ref)
Excellent
0.13
1.14
0.68
0.498
-0.03
0.97
-0.25
0.806
Commute time
Short (<30 min) (ref)
Long (>30 min)
0.15
1.16
0.39
0.694
-0.19
0.83
-0.78
0.436
Intercept
-2.07
0.13
-4.32
0.000
0.44
1.55
2.15
0.032
22
Table 2. Multinomial Logit Results for Dockless Bikeshare
Wants it but wouldn't use
Wants it and would use
Estimate
OR
z-stat
P-value
Estimate
OR
z-stat
P-value
Dockless bikeshare
Gender
Female (ref)
Male
-0.12
0.88
-1.15
0.251
-0.08
0.92
-0.75
0.450
Age (years)
Below 25 (ref)
Between 25 - 34
-0.50
0.61
-1.89
0.059
-0.47
0.62
-1.92
0.055
Between 35 - 44
-0.54
0.58
-2.14
0.032
-0.50
0.60
-2.11
0.035
Between 45 - 54
-0.61
0.54
-2.38
0.017
-0.93
0.39
-3.73
0.000
Between 55 - 64
-0.69
0.50
-2.43
0.015
-1.26
0.28
-4.33
0.000
65 and above
-0.56
0.57
-1.85
0.065
-1.96
0.14
-5.08
0.000
Income
Below $100k (ref)
Above $100k
-0.03
0.97
-0.16
0.871
-0.13
0.88
-0.64
0.521
Education quality
Poor/average (ref)
Excellent
0.10
1.11
0.73
0.468
0.08
1.09
0.62
0.535
Commute time
Short (<30 min) (ref)
Long (>30 min)
0.10
1.11
0.38
0.705
-0.24
0.78
-0.85
0.398
Intercept
-0.51
0.60
-2.11
0.035
-0.29
0.75
-1.28
0.202
23
Table 3. Multinomial Logit Results for Docked Bikeshare
Wants it but wouldn't use
Wants it and would use
Estimate
OR
z-stat
P-value
Estimate
OR
z-stat
P-value
Docked bikeshare
Gender
Female (ref)
Male
-0.13
0.88
-1.29
0.197
-0.27
0.76
-2.66
0.008
Age (years)
Below 25 (ref)
Between 25 - 34
0.44
1.55
1.65
0.099
0.07
1.07
0.29
0.771
Between 35 - 44
0.47
1.60
1.83
0.067
0.01
1.01
0.04
0.965
Between 45 - 54
0.29
1.33
1.10
0.273
-0.18
0.84
-0.76
0.450
Between 55 - 64
0.56
1.74
1.98
0.048
-0.23
0.79
-0.87
0.382
65 and above
0.49
1.63
1.67
0.096
-1.16
0.31
-3.54
0.000
Income
Below $100k (ref)
Above $100k
-0.11
0.90
-0.55
0.584
0.01
1.01
0.06
0.951
Education quality
Poor/average (ref)
Excellent
0.16
1.17
1.20
0.229
0.19
1.21
1.42
0.156
Commute time
Short (<30 min) (ref)
Long (>30 min)
0.10
1.11
0.39
0.699
-0.21
0.81
-0.80
0.424
Intercept
-0.40
0.67
-1.65
0.099
0.11
1.11
0.50
0.620
24
Figures
Figure 1: Proportion of selected options
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
Do not want & wouldn't
use Wants it but wouldn't
use Do not want it but
would use Wants it and would use
Percentage of responses
Selected options
On average E-scooter Dockless bikes Docked bikes
25
Figure 2: Proportion of selected options of micro-mobility by age groups
Figure 3: Text network results for "wants and will use all three schemes"
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
Up to 24 Between 25 - 34 Between 35 - 44 Between 45 - 54 Between 55 - 64 65 and above
Percentage
Age group (years)
Wants electric scooters
Wants dockless bikeshare
Wants docked bikeshare
Would use electric scooters
Would use dockless bikeshare
Would use docked bikeshare
26
Figure 4: Text network for "do not want and won't use any of the three schemes"
27
Figure 5: Text Network for "wants but won't use electric scooters"
