How Did You Like This Ride? An Analysis of User Preferences in

Ridesharing Assignments

oren Schleibaum

1 a

, Maike Greve

, Tim-Benjamin Lembcke

2 b

, Amos Azaria

, Jelena Fiosina

Noam Hazon

, Lutz Kolbe

, Sarit Kraus

, J

org P. M

uller

1 c

and Mark Vollrath

Department of Informatics, Clausthal University of Technology, Julius-Albert Straße 4, Clausthal-Zellerfeld, Germany

Chair of Information Management, University of G

ottingen, Platz der G

ottinger Sieben 5, G

ottingen, Germany

Department of Computer Science, Ariel University, Israel

Department of Computer Science, Bar-Ilan University, Israel

Chair of Engineering and Trafﬁc Psychology, TU Braunschweig, Germany

Keywords:

User Preferences, Ridesharing, Assignment, Shared Mobility, Platform Economy.

Abstract:

Ridesharing can signiﬁcantly reduce individual passenger transport and thus greenhouse gas emissions gen-

erated by trafﬁc. Although ridesharing offers great potential, it is not yet popular enough to be seen as an

important contribution to solving the aforementioned problems. Our hypothesis suggests that we need to

make the assignment mechanism of ridesharing systems more human-centric and comprehensible in order to

popularise ridesharing. Therefore, we investigate factors that inﬂuence the choice of users and their satisfac-

tion with the assigned ride. Most of today’s ridesharing assignment algorithms focus solely on features such as

time, distance and price. Contrarily, this paper examines additional factors that inﬂuence customer decisions

to increase their satisfaction. Therefore, we ﬁrst conduct a literature study to identify previous preferences

relevant for ridesharing from a research perspective. Subsequently, we extract the relevant preferences for an

assignment process. From these we secondly conduct a survey. Last, we analyse the obtained survey data and

order the preferences based on their importance for participants overall and among demographic subgroups.

1 INTRODUCTION

The impact of increasing greenhouse gas emissions

on our environment has been scientiﬁcally proven

(Parmesan and Yohe, 2003) and we are facing the

fastest global warming phase since the beginning of

the weather records. One of the most signiﬁcant

contributors to emissions is the individualized trans-

portation of people, mostly through personal vehi-

cles. By sharing personal vehicles with other trav-

elers (i.e. ridesharing), improved vehicle utiliza-

tion can lead to substantial fuel savings and re-

duced emissions (Jacobson and King, 2009). Schol-

ars have researched the acceptance of ridesharing for

decades; nonetheless, there are still factors that limit

a widespread adoption of ridesharing, including pric-

ing, high-dimensional assignment, trust and reputa-

tion, as well as institutional design of such services

https://orcid.org/0000-0001-7181-5336

https://orcid.org/0000-0003-3092-5277

https://orcid.org/0000-0001-7533-3852

(Furuhata et al., 2013). One of the fundamental chal-

lenges in ridesharing is to bring driver (supply) and

riders (demand) together. Therefore, a market mech-

anism is necessary to enable ridesharing services on a

larger scale. Advancements in information technolo-

gies enabled new information systems (IS) in form

of web platforms with assignment facilities for sup-

ply and demand. However, to be successful, the

chicken-and-egg problem inherent to these platforms

must be overcome, namely suitable rides offered and

demanded. Conceptually, these platforms have two

phases: First, users announce their ride offerings and

requests, and second, these offerings and requests are

assigned. Since the assignment is the core activity

of the ridesharing IS platform, it is of particular in-

terest to understand if users perceive the assignment

as satisfactory. In this paper, we understand rideshar-

ing as at least two individuals sharing a common ride

in the same vehicle. Furthermore, we consider the as-

signment process as bringing two individuals together

based on certain criteria like a route, price or the con-

Schleibaum, S., Greve, M., Lembcke, T., Azaria, A., Fiosina, J., Hazon, N., Kolbe, L., Kraus, S., Müller, J. and Vollrath, M.

How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments.

DOI: 10.5220/0009324401570168

In Proceedings of the 6th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2020), pages 157-168

ISBN: 978-989-758-419-0

157

sideration of user’s preferences for a trip. This can

also contain the allocation of vehicles.

In principle, assignments in ridesharing can be ac-

complished in two ways: First, the provider could

do the assignment according to their own discretion.

From a user perspective, such assignments happen “as

is”, in a non-transparent fashion. User needs may or

may not be reﬂected by the assignment, potentially

rendering users dissatisﬁed. Second, the provider

could do the assignment in a transparent way, al-

lowing users to understand the assignment. Further-

more, user preferences and needs may be prompted

in advance and inﬂuence the assignment to maximize

the joint satisfaction of a driver and the according

rider(s). To align such user preferences on a large

scale in an automated and ﬂexible way, artiﬁcial intel-

ligence (AI) technologies may be helpful. Nonethe-

less, to feed such AI, it is necessary to understand

which user preferences exist that inﬂuence a user’s

satisfaction with the ridesharing assignment. In cur-

rent research, such assignment preferences are only

addressed in limited amount. For example, (Bian and

Liu, 2019), (Neoh et al., 2018), (Yousaf et al., 2014)

and (Chaube et al., 2010) considers only a handful of

individual preference factors such as price and social

relations of travelers that inﬂuence personal satisfac-

tion with ridesharing experience. To the best of our

knowledge, none of the present studies have reviewed

a great number of factors to provide insight into the

satisfaction function of users within ridesharing as-

signment processes. Therefore, our study considers

several factors simultaneously, leading to our main re-

search questions:

• Which preferences inﬂuencing ridesharing users

prevail in current literature?

• Which preferences inﬂuence a users’ satisfaction

within the assignment process of ridesharing?

• Does the order of importance for these prefer-

ences differ for subsets of people who vary in age,

gender, country, etc.?

In this study, we ﬁrstly provide the research back-

ground in Section 2 and describe the methods used in

this paper in Section 3 to enable more human-centric

assignments in ridesharing. We perform a literature

study in Section 4 to extract preferences and conduct

an online questionnaire with more than 290 partici-

pants to investigate their relevance. The results of the

latter is described in Section 5 and combined with the

ﬁndings from the literature study in Section 6. Finally,

we conclude our overall contributions in Section 7.

2 RESEARCH BACKGROUND

Ridesharing Terms. In context of this study, we de-

ﬁne ridesharing as “the formal or informal sharing

of rides between drivers and passengers with sim-

ilar origin-destination pairings” (Shared and Digi-

tal Mobility Committee, 2018). Within this deﬁn-

itory framework, multiple archetypes of ridesharing

are conceivable, from employees commuting together

to ridesharing as a service solution, providing on-

demand and door-to-door ride services. Historically,

during the Second World War the ﬁrst organized

ridesharing was implemented by the U.S. government

as a regulation to save fuel (Furuhata et al., 2013).

Later, as a result of the oil crisis several rideshar-

ing methods emerged in the 1970s. Afterwards, the

popularity of ridesharing decreased due to more com-

plex travel patterns caused by demographic changes

(Ferguson, 1997). Then, with the rise of the inter-

net ridesharing services that assign riders and drivers

became apparent (Furuhata et al., 2013) and with

technological advancements like GPS-enabled smart-

phones dynamic ridesharing services such as Uber-

Pool became possible. Dynamic services let users of-

fer rides as a driver or request rides as a passenger at

any time (Nourinejad and Roorda, 2016). Nowadays,

ridesharing offers economic, environmental and so-

cial beneﬁts by reducing the number of vehicles and

travel cost (Neoh et al., 2018).

Human-centric Assignment in Ridesharing. De-

spite the increasing trafﬁc in cities and the potential of

ridesharing to reduce the pollution caused by trafﬁc,

ridesharing in Germany is particularly not very pop-

ular (Statistisches Bundesamt, 2019). Previous liter-

ature indicates that one reason for this reluctance lies

in the assignments. In order to design a shared ride in

such a way that travelers need to make minimal effort,

a system should automate the assignment to satisfy

the customer’s needs (Agatz et al., 2012). However,

this deliberation appears to be easier to implement

than it is in practice. The conﬁguration of a selection-

based assignment process is not trivial (Washbrook

et al., 2006). Nowadays, most business models only

consider the desired route and price in their assign-

ment engine. Other factors, including personal pref-

erences such as comfort or safety of the vehicle, re-

ceive none or limited attention. Nevertheless, we ar-

gue that more personalized assignments can increase

the popularity of ridesharing and, thus, its actual use.

Research has shown that riders only feel comfortable

if they are assigned to a ride with a speciﬁc group of

people, and that the group’s preferences may be moti-

vated by personal safety or social aspects (Agatz et al.,

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

158

Figure 1: Categorization of Ridesharing Factors (Neoh

et al., 2018).

2012). At nighttime, for example, a shy person may

not be willing to share a trip with a complete stranger

and may only want to share trips with friends and col-

leagues. Clearly, the more restrictions a potential user

imposes on his pool of potential ride companions, the

more difﬁcult it will be to ﬁnd successful assignments

for that user (Dailey et al., 1999).

Systematisation of Preference Factors. To date,

few studies have focused on user preferences in

ridesharing. Instead, many studies include some pref-

erence factors, but rather as a supplement to their

primary study design. To reach a uniﬁed systemati-

sation of the inﬂuence individual decisions on share

rides, (Neoh et al., 2017) has developed a categori-

sation model shown in Figure 1. On the ﬁrst level,

inﬂuencing factors are differentiated between exter-

nal and internal factors. On the second level, inter-

nal factors are separated into socio-demographic and

judgmental factors such as users’ reasons to share

rides (Neoh et al., 2017). Previous studies assume

that demographic factors have only a very small im-

pact (Vanoutrive et al., 2012) while - in combination

with other factors - they may have an inﬂuence (Cor-

reia and Viegas, 2011). Under the judgmental fac-

tors, all psychological factors like social aspects and

feeling of independence while driving the own car are

considered (Neoh et al., 2018). On the contrary, ex-

ternal factors include situational factors and interven-

tions and take place at the environmental level of the

ridesharing user (Neoh et al., 2017). Thereby, situa-

tional factors affect the location as well as all waiting

times such as waiting time for other passengers. Usu-

ally, it makes ridesharing less attractive when one or

more of these factors lead to long journeys (Tsao and

Lin, 1999). In contrast, interventional factors like me-

diating actions that are implemented by a facilitator,

e.g. a facility which encourages people to share rides

with a parking discount, yet partner assignment sys-

tems also belong to this category. Studies lean to say

penalising single occupied vehicles are more effective

than rewards for ridesharing (Neoh et al., 2018).

3 METHOD

To identify relevant preferences for assignments in

ridesharing, we ﬁrst review current literature. We de-

cided to conduct a survey afterwards because it is an

effective and popular method for gathering informa-

tion about people. Next, the process of gathering the

literature, the design of the study and used data anal-

ysis techniques are described.

3.1 Literature Study

To systematically review existing research in the area

of ridesharing preference factors, we followed a lit-

erature review process based on (Webster and Wat-

son, 2002) and (vom Brocke et al., 2009). Accord-

ingly, we ﬁrst gathered literature from IS journals and

conferences as well as general databases to include

transportation outlets by a generalized search string.

Second, we identiﬁed the mentioned factors in each

article and third, we summarized these factors with

regard to their commonness.

To ﬁnd relevant literature, a search query was

created in phase three, using the term ridesharing

and possible synonyms: ride sharing, ridesharing,

ride pooling, ridepooling, car pooling, or carpool-

ing. This was used to search the most popular IS

journals (basket of eight), the ten mostly cited trans-

portation journals according to the scientiﬁc journal

ranking (SJR) and the IS conferences. The search

was limited to literature published between 2015 and

2019. The search query had to be found in the title or

abstract of the literature. Afterwards, the left articles

are read and relevant preferences are selected.

3.2 Empirical Study

Overall, we conducted a questionnaire consisting of

68 questions, which were provided in English and

German. Two of the questions are for attention checks

to enable a high-data quality; one is an open-ended

question enabling users to provide preferences not

considered by us. Besides that, the questionnaire

consists of four parts: 1. Present and future usage

of ridesharing (6 questions); 2. Preferences of pas-

sengers (41 questions); 3. Information for an assign-

ment algorithm (10 questions); 4. Demographic data

(8 questions).

Because ﬁve-point Likert scales are typical (Sul-

livan and Artino, 2013), we used that for the second

and third part. We prefer Likert scale questions over a

conjoint analysis because the number of preferences

to investigate is relatively high. For every question

the participants had the option to provide no answer.

How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments

159

The questionnaire was provided online to enable fast

conducting around the world and to reduce cost. The

downside of enabling easy access is that we were not

able to observe preferences like trust objectively via

an observation. We used LimeSurvey (see (Schmitz

and Team, 2012)) to create and host the questionnaire.

We shared the questionnaire via email-lists of our uni-

versities and among our social networks. The data is

stored anonymously and there were no incentives to

participate. All questions and the structure is avail-

able in an online repository

3.3 Techniques for Analysis

Order Preferences by Importance. Our approach to

create an order of preferences by importance is three-

fold:

• Firstly, to get to an initial order of importance for

the preferences, we transform the answers to nu-

meric values with similar distance and sum up all

answers for each preference. The comparison of

these sums leads to an initial ordering.

• Secondly, we limit the preferences for further

analysis to later be able to verify the order statisti-

cally in the third step and focus on the relevant re-

sults. For the limitation, we apply hierarchical ag-

glomerative clustering to cluster the preferences.

We favour this technique over partitioning, like k-

means, because thereby we use a deterministic al-

gorithm and we do not have to choose a number

of clusters in the ﬁrst place. For the concrete al-

gorithm, we chose the Ward’s method.

• Thirdly, we apply a Friedman test to get a sta-

tistically veriﬁed order of preferences to the ﬁrst

and second cluster. We apply this test to compare

all preferences of leftover importances with each

other. For this procedure we orientate on (Derrac

et al., 2011), who describe the N × N compari-

son of algorithms performances. As a post-hoc

procedure we choose Shaffer’s method. Based on

the resulting p-values (p) we construct the order.

We set the signiﬁcance level α to 0.01 to cover all

common signiﬁcance levels.

Importance Order in Demographic Groups. In this

part of the analysis we split the valid samples into

subgroups based on the collected demographic data

and again create an order based on the importance

of preferences. Thereby, we are able to identify dif-

ferences between subgroups. We consider subgroups

that appear at least 21 times in the data. To statisti-

cally verify differences between subgroups, we apply

https://gitlab.tu-clausthal.de/ss16/questionnaire-

analysis-public/

a Mann-Whitney rank test (see (Mann and Whitney,

1947)). We favour this test over the Wilcoxon signed-

rank test, because the compared groups are indepen-

dent. Similar to before we set α to 0.01.

Software. For the Wilcoxon signed-rank, the Mann-

Whitney rank test and clustering of preferences we

use the implementation provided by (Jones et al.,

2001). The complete source code used for our analy-

sis in Section 5 is available online in a repository

4 PREFERENCE ANALYSIS -

LITERATURE OVERVIEW

The method described in Section 3.1 results in 63 rel-

evant articles. The detailed results can be found in

our online repository

. After having analysed the ﬁ-

nal sample, based on the 63 articles, 73 factors im-

pacting human attitude towards ridesharing were de-

termined. These factors were categorized in the cate-

gories recommended by (Neoh et al., 2017). To make

concise comparisons between categories, factors of

similar nature were merged into subcategories lead-

ing to the overview presented in Table 1.

4.1 Selection of Preferences for Our

Survey

Based on the results of the literature review, we

selected the preference factors that were assessed

though a survey. Thereby, we limited the scope so

the participants could clearly understand the setting

of the study. The literature review had resulted in a

wide scope of preferences, of which not all are plau-

sible in the context of assignment. Hence, we de-

cided to focus on factors inﬂuencing the individual

judgement and, in turn, the actual behaviour: to share

a ride or not. This leads to focusing on the prefer-

ences of the passengers and excludes preferences of

the driver. In addition, we only include factors that

are relevant for assignments when a user has already

overcome the ﬁrst barrier of using ridesharing. There-

fore, preferences such as peer pressure or living in

rural areas are excluded. Privacy is only included in-

directly, because when a person has decided to par-

ticipate in ridesharing, we assume that this person is

already willing to give up his/her privacy to a certain

degree. Going along with the categorisation of (Neoh

et al., 2017), we mainly consider judgmental factors,

as well as some situational factors. Demographic fac-

tors are surveyed separately in the last part of the sur-

vey, while interventional factors are not considered

since they refer to third-party interventions that play

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

160

Table 1: Overview of preference factors from literature.

Demographic % Judgmental % Situational % Interventional %

Gender 7.9 Economic beneﬁts 84.1 Time related 61.9 HOVL 28.6

Education 7.9 ESB 77.8 Flexibility 28.6 Parking space 20.6

Age 6.3 Convenience 61.9 Availability 15.9 Parking fee 15.9

Ethnicity 6.3 Privacy 36.5 Meeting point 9.5

Income 6.3 Safety 22.2 Finding rides 7.9

Employment type 3.2 Trust 19.0 Living location 6.3

Sexuality 1.6 Security 17.5

Peer/family pressure 1.6 Pleasure 11.1

a superior role in ridesharing, but do not affect the as-

signment process.

Based on the results of our literature review, we

derived the list of preferences shown in Table 2 to be

considered in the survey. The table describes all judg-

mental and situational factors as well as the surveyed

preferences of these. To easily identify the factor of a

preference, we introduce an abbreviation of the factor,

which will be used in later graphics. In the following

subsections, the judgmental and situational factors are

outlined in greater detail.

4.2 Judgmental Factors

This category refers to internal and judgmental factors

of ridesharing users, which include the judgement of

economic beneﬁts, environmental and social beneﬁts,

convenience, privacy and safety concerns, trust, secu-

rity and pleasure in ridesharing opportunities.

Economical Beneﬁts. The most prevalent factors in

regarded research are the ones that are economically

or environmentally and socially beneﬁcial for drivers

and passengers. Economically beneﬁcial factors like

reduced cost are referenced in 53 articles of the re-

viewed literature. The fact that ridesharing can reduce

the travel cost is the most stated factor in the analysed

literature, being mentioned 48 times. While rideshar-

ing services can operate at a lower cost compared to

traditional taxi organisations (Schweitzer and Bren-

del, 2018), private ridesharing can reduce the travel

cost by splitting it up between driver and passengers

(Wang et al., 2018). Along these lines, saving fuel

was indicated as a factor 21 times. Because this also

saves money and therefore is economically beneﬁcial

(Mourad et al., 2019), saving fuel belongs to this cat-

egory. Aside from that, saving fuel also is environ-

mental beneﬁcial (Li et al., 2017).

Environmental/Social Beneﬁts. Overall, ridesharing

does offer environmental and social beneﬁts (ESB),

which are common beneﬁts that all parties proﬁt from

like reducing the instances of drunk driving (Green-

wood and Wattal, 2017). Environmental beneﬁts like

reducing the overall energy waste or increasing sus-

tainability can also be of altruistic nature (Wang et al.,

2019). Saving CO

emissions (Li et al., 2017) and re-

ducing congestion (Mahmoudi and Zhou, 2016) can

be achieved because ridesharing increases the utiliza-

tion of a vehicle’s capacity (Lavieri and Bhat, 2019a).

This in turn saves fuel and therefore ridesharing can

play a certain role in reducing overall energy con-

sumption (Wang et al., 2019). The fact rideshar-

ing reduces trafﬁc congestion was mentioned in 30

of the analysed articles. Ridesharing can signiﬁ-

cantly reduce the number of cars on the road and

therefore limit trafﬁc congestion (Stiglic et al., 2015).

Ridesharing also reduces car ownership because it

serves as a convenient and cost efﬁcient alternative to

owning a car without the ﬁnancial and social burdens

of ownership (Liu et al., 2017).

Convenience. Factors that impact the convenience of

ridesharing are referenced in 39 of the reviewed arti-

cles. Convenience is a factor that can positively or

negatively impact human attitude towards rideshar-

ing, depending on which kind of transportation it is

compared to. Compared to driving with a private

car, ridesharing is perceived as rather inconvenient

(Xiao et al., 2016). However, ridesharing can offer

the convenience of a private car while paying a sim-

ilar amount when compared with public transporta-

tion (S

anchez et al., 2016; Nielsen et al., 2015; Wang

et al., 2019). Further factors supporting rideshar-

ing convenience are availability of different payment

methods for ridesharing-services (Hong, 2017), the

ease of use of these services (Greenwood and Wattal,

2017), avoiding transfers (Yan et al., 2019) and reduc-

ing driver stress (Mahmoudi and Zhou, 2016). Ser-

vice quality, which can also beneﬁt the convenience of

ridesharing, was only named twice in the present lit-

erature. Moreover, clear policies can reduce the con-

cerns about service surcharges (Zhang et al., 2018),

the condition of the car (Mirsadikov et al., 2016) and

options like non-smoking vehicles beneﬁt the comfort

of the ride.

How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments

161

Table 2: Overview of preference factors in survey.

Judgmental factors Preferences

Economic beneﬁts (ECB) Paid price

Environmental/social beneﬁts (ESB) Vehicle congestion and power

Convenience (CON) Payment method, short breaks during ride on longer journeys (longer

than two hours), mainly motorway usage or rural road usage, short

duration of journey, pets allowed in vehicle

Privacy Indirect

Safety (SAF) Driver’s competence, previously information about driver, calm driv-

ing or sporty driving style, track location for security, safety and con-

dition of the vehicle

Trust (TRU) Trust in other people

Security (SEC) No trip cancelling from the driver, saying no if cancel of trip, insur-

ance of passengers during the ride

Pleasure (PLE) Small number of fellow passenger (low occupied), smoking while driv-

ing (whether desired or undesired), friendliness of other people, tem-

perature in the vehicle, interpersonal climate, volume of music (in-

cluding no music), type of music and conversation topics during the

journey, similar interests of passengers, trips pass on sightseeing loca-

tions, smell in and cleanliness of the vehicle, amount of space on seat,

space in trunk, existence of air conditioning, comfort of the vehicle

Situational factors Preferences

Time related (TIR) Low delay at start and low delay by pickup of other passengers (both

less than 10 minutes), short distance

Flexibility (FLE) Drivers respondance to wishes of passengers

Availability/accessibility Not relevant for assignment

Meeting point (MEP) Small detours to be collected or dropped off

Finding rides/ high assignment rate Not relevant for assignment

Living location Not relevant for assignment

Privacy, Safety and Security. Other factors regard-

ing perceived ridesharing risks include privacy, safety

or security concerns. The perceived privacy risk is

referenced as the utmost barrier in ridesharing (Xiao

et al., 2016). It is shown that privacy sensitive in-

dividuals are less likely to have experience in using

ridesharing services (Lavieri and Bhat, 2019b). Pri-

vacy concerns mostly are about the intentional mis-

use or disclosure of private data to third parties, which

is required for using ridesharing services, like credit

card information or the user’s living location (Hong,

2017). In recent literature the loss of privacy is of-

ten seen as a tradeoff for the ﬁnancial beneﬁts that

come with ridesharing (Tian et al., 2019). Individu-

als using ridesharing are also faced with safety con-

cerns and security risks: It is indicated that travelers

are hesitant about being in a vehicle with unfamil-

iar people (Lavieri and Bhat, 2019a). The passenger

could be worried about getting kidnapped or attacked,

while drivers could be concerned with riders damag-

ing their car (Mirsadikov et al., 2016). As a resolution

a concept is proposed using meeting points to pre-

serve the users privacy and security (A

ıvodji et al.,

2016). In this approach ridesharing users do not share

their starting point or destination and therefore the

users’ patterns of mobility cannot be traced.

Trust. Such factors are referenced in 12 of the

analysed articles. For example, existing commercial

driver’s license can have a positive inﬂuence on users

attitude towards ridesharing (Hong, 2017). Besides,

driver screening, tracking systems and rating systems

give ridesharing users a feeling of safety. For riders

a rating system can show them what service quality

they can expect and it also is a safety and security

measure (Mirsadikov et al., 2016). However, a rat-

ing system can be exploited by riders and used as a

lever to manipulate drivers into providing extra ser-

vices, because drivers often will be excluded from a

ridesharing service if their ratings are too low (Mir-

sadikov et al., 2016).

Pleasure. Under the term pleasure, we summarise all

mellow factors mentioned in the literature which have

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

162

an inﬂuence on the positive/negative state of mind of

the user. The number of passengers, for example, is

an indicator associated with social inconvenience and

positive social interactions. Expected social discom-

fort or awkwardness is one of the negative perceptions

that individuals have about sharing a ride (Nielsen

et al., 2015). But ridesharing is not only seen as so-

cially unpleasant, but also as an opportunity for posi-

tive social interactions such as fun or emotional plea-

sure by making friends and learning new knowledge

(Wang et al., 2019). A variety of factors, such as the

desire for diversity, the desire to meet with strangers,

or the equipment in the car with telephone chargers

or water (Mirsadikov et al., 2016) can inﬂuence the

person’s opinion of ridesharing opportunities and the

possible enjoyment and pleasure of a ride (Lavieri and

Bhat, 2019b).

4.3 Situational Factors

The third category refers to factors which are exter-

nal and mostly location-based (Neoh et al., 2017).

The location can inﬂuence the travel distance, travel

time and the likelihood to ﬁnd ridesharing partners

(Neoh et al., 2018). Therefore, we derived factors that

are time-related, concern the ﬂexibility or availabil-

ity/accessibility, refer to the meeting point or the rate

of ﬁnding a ride or the living location.

Time Related Preferences. The reviewed literature

reveals, with time related factors being the most men-

tioned situational factors (39 times mentioned), that

users seem to be time sensitive when it comes to

ridesharing. Waiting times are perceived as incon-

venient by ridesharing users (S

anchez et al., 2016;

Stiglic et al., 2015), however waiting at a meeting

point as a group may facilitate the safety perception

of riders (Stiglic et al., 2015).

Meeting Points. The ability to choose a pick-up and

drop-off location can offer some degree of anonymity

and safety for the rider when using a ridesharing ser-

vice because it provides the option to not share per-

sonal information such as the individual’s living loca-

tion (Mirsadikov et al., 2016).

Availability/Accessibility. The distance to a meet-

ing point can also be linked to the availability of

ridesharing, which was mentioned 7 times as well as

to the individual’s living location. Existing informa-

tion technology is an underlying prerequisite and cel-

lular phone service is mandatory for most rideshar-

ing services to work (Joseph, 2018). The availability

of ridesharing also inﬂuences the assignment rate on

ridesharing services, since a higher availability im-

plies an increased amount of people using rideshar-

ing in an area. A high assignment rate is a critical

success factor for a ridesharing service because only

successfully matched users will have a positive expe-

rience and promote the service to others (Stiglic et al.,

2015).

Flexibility. 18 articles mentioned ﬂexibility as a

factor that inﬂuences people to use ridesharing.

Ridesharing services can provide increased ﬂexibility

compared to taxi services such as types of vehicles

and pricing prior to the trip (Joseph, 2018). However,

this cannot offer the same ﬂexibility as a personally

owned car (Schweitzer and Brendel, 2018). This in-

dicates that passengers of public transportation like

train or bus are more likely to substitute with rideshar-

ing than drivers who own cars (Schweitzer and Bren-

del, 2018).

5 ANALYSIS AND RESULTS OF

SURVEY

In the ﬁrst paragraph of this section, we make our pro-

cess of cleaning the data based on attention checks

transparent. After that, we list the characteristics of

the collected sample. Then, we show our analysis re-

sults of our observed overall order of preferences and

the differences in demographic subgroups. For the

second, we list results for age, gender, education and

country of residence in separate paragraphs. After-

wards, results for working status, car owners and pet

owners are summarized in one paragraph.

Clearing of the Dataset. We exclude 17 samples

from the analysis because they did not understand

the given deﬁnition of ridesharing, failed an attention

check, or answered less than 25 percent of the ques-

tions. This results in 291 valid samples. For further

analysis we also extract the preferences of the passen-

ger part from the questionnaire and replace the text-

based answers (important, rather important, neutral,

rather unimportant, unimportant) by numeric values

(1, 2, 3, 4, 5).

Sample Characteristic. The 291 participants com-

pleted the questionnaire on average in nine minutes.

The mean of the age was 29 years with a standard de-

viation of 12. 135 of the participants were female,

149 male and seven reported no gender. Most of them

come from Germany (242), 23 from Israel, two from

the Netherlands and China each; from France, Hun-

gary, Senegal, Spain and Turkey we had one partici-

pant each. Moreover, 17 people did not provide their

country of residence. The data points where collected

from August to October 2019.

Overall Importance. To initially order the pref-

erences by their importance we apply a simple ap-

proach: We sum the values of all data points for one

How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments

163

preference and compare this with the others. The

smaller its sum, the more important a preference is.

Together with the proportions of answers, this order

is shown in Figure 2. This is combined with Ward’s

method for clustering of the preferences (not the peo-

ple), which results in four clusters shown as colors

and with a dendrogram above. The fact that the clus-

ters do not disrupt this initial order of the preferences

is remarkable.

Together, both approaches already give a good

idea about the relative importance of the preferences.

Nevertheless, this result has to be interpreted with

care because for its creation important has ﬁve times

more inﬂuence than unimportant for instance.

Therefore, we further apply a Friedman test with

Shaffer’s correction method to the ﬁrst (green) and

second (black) cluster. We excluded the third (blue)

and fourth (red) cluster to be computationally able to

apply the test and focus on the relevant results. The

resulting groups of the test are included in Figure 3 in

the labels of the x-axis. The concrete p-values of the

Friedman test are shown in a heatmap available in our

online repository

The results clearly show that no trip cancelling is

in group (a) based on the Friedman test results and

therefore is the most important preference. After-

wards, we have a group of say no if cancel, safety,

driver’s competence and smell, which slightly over-

laps with group (c). In contrast to the order by

the simple approach, short duration and low delay

by pickups appear between the preferences currently

on sixth and seventh position. Overall, it is hard to

provide a clear order because the groups heavily in-

terfere. Nevertheless, the groups can be used more

clearly to provide 1 × N comparisons. This shows for

instance, that comfort is less important than all pref-

erences before friendliness and low delay at pickup.

Importance in Demographic Groups. The orders

for age and other demographic subgroups are shown

in Figure 3 and computed with the simple approach.

On the y-axis, the condition for each subset is listed

together with the number of samples matching this

condition; the x-axis lists all preferences. Each cell

of the matrix contains the calculated rank for a sub-

set/preference combination based on the simple ap-

proach. The colors represent the sum used to calcu-

late the importance order divided by the sum for all

answers of the considered preference. After naming

the subgroups, we summarize in the following the sta-

tistical differences among them.

Age. Concerning the age of participants, ﬁrstly we

create three subgroups: younger than 21, from 21 to

35 and older than 35 years. We observed that insur-

ance, calm driving, volume of music, breaks during

ride and congestion are more important for people

older than 35 compared to the middle-aged group. On

the other hand, people between 21 and 35 care more

about a friendliness and price. Compared to people

older than 35, trust and friendliness are more impor-

tant for people younger than 21. On the contrary, calm

driving, space in trunk and volume of music matter

more for the middle-aged group. Comparing people

between 21 to 35 to people younger than 21 shows

that only space in the trunk matters more. In contrast,

insurance, track location and congestion are more im-

portant for the youngest group.

Genders. Distinguishing between genders shows that

for women information about driver, trust, respon-

dance to wishes, tracking location and congestion are

more important for women.

Education. To compare certain levels of education,

we consider four subgroups: Matriculation standard,

bachelor pr master’s or degree and doctorates. Com-

pared to people holding a matriculation standard, for

people with a bachelor, information about driver and

friendliness are less important. Relative to master’s

degree holders, people with matriculation standard

care less about time (short duration, motorway us-

age), space in trunk and air conditioning. However,

sporty driving, insurance and track location is more

important for them. Compared to people with a ma-

triculation standard, for doctorates smoking, low oc-

cupied and motorway usage are more important. On

the other hand, friendliness is more important for

people holding a matriculation standard or master’s

degree. Comparing doctorates with bachelor degree

holders shows that smoking and calm driving is more

important for the former.

Country of Residence. Comparing the importance

of preferences for Germans with the small number of

Israelis, shows that say no if cancel, safety, condi-

tion and friendliness is more important for German

residents. Conversely, temperature, air condition-

ing, sightseeing, low delay at start and smoking are

more important for Israeli residents. In contrast to

all other subgroups, for Israeli residents, smoking is

most important. Moreover, say no if cancel, safety

and driver’s competence, which are among the top

ﬁve for all other subgroups, are relatively unimpor-

tant.

Leftover Subgroups. When comparing students with

employed people, we observe that no trip canceling

of a trip is more important for employed people. On

the other hand, students seem to care more about in-

surance and track location during the ride. We were

not able to verify a difference between car owners and

those who do not own a car; similarly, there was no

inﬂuence by owning a pet. For smokers, condition,

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

164

Figure 2: Showing the proportion of answers for the preferences. The preferences are marked with the factors described in

Section 4. The order is based on the simple approach.

cleanliness and power are more important; for non-

smokers smoking is more important.

6 DISCUSSION AND SUMMARY

OF FINDINGS

Overall Importance. The ﬁve most important prefer-

ences, that being safety, habits concerning cancelling

of rides and smell seem to represent the essentials for

participation in ridesharing. After that, it is compli-

cated to make a boundary for other preferences be-

cause their importance decreases approximately lin-

ear. Nevertheless, looking from the insigniﬁcant side,

the last seven can be neglected in an assignment pro-

cess. Interestingly, among these are power and sporty

driving. Comparing the ranks observed in the survey

with the attention a judgmental factor gets in the lit-

erature shows interesting differences. For instance,

the economic beneﬁt price occurs most often in the

literature but is not in the group or cluster of most im-

portant preferences. Environmental and social bene-

ﬁts (congestion and power), that show up secondly in

the literature, end up in the third and fourth most im-

portant cluster. The convenience factor group appears

in all clusters except for the ﬁrst one, and our results

show that short duration of the trip is the most im-

portant among its preferences. The factor safety, with

preferences such as driver’s competence and condi-

tion of the vehicle, appears in the ﬁrst and second

cluster, showing a relatively highly observed impor-

tance. The same applies for the factor security. The

factor pleasure occurs mostly in the third and fourth

cluster, which is similar to its received attention in

the literature. However, our survey shows that space

on seat, cleanliness and especially smell are far more

important than their occurrence in research. Consid-

ering situational factors: The time related factor with

preferences such as low delay at start are with 61.9

percent relatively important in the literature and ac-

cordingly occur in our second most important cluster.

The same goes for the factors ﬂexibility and meeting

point.

Besides being underrepresented in the literature

compared to our survey results, we believe that these

differences are based on two reasons: Firstly, some

preferences like the price of a ride are easier to ad-

just in reality than preferences like smell in the vehi-

cle. Secondly, people might care about the safety of

a vehicle, but in reality, you can assume that all vehi-

cles are safe to a certain degree. Nevertheless, our re-

sults indicate that the preferences safety of a vehicle,

driver’s competence and smell are highly underrep-

resented in the current research. On the other hand,

the preferences price, power and congestion are over-

represented. Based on our ﬁndings we therefore rec-

ommend to shift the focus for assignment processes

in ridesharing towards the underrepresented prefer-

ences.

Importance in Demographic Groups. Regarding

age we can contribute the following: Interestingly,

How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments

165

Figure 3: One row shows the observed order of preferences for a demographic subset based on the simple ordering approach.

in addition to younger people, those over the age of

34 care most about congestion. For younger people,

safety and security related factors are more important.

Regarding gender: Generally, women ﬁnd security

factors such as information about driver and environ-

mental aspects more important.

Regarding education level: People holding a ma-

triculation standard or a master’s degree care more

about friendliness. However, for people with a matric-

ulation standard safety related preferences are more

important, whereas for master’s degree holders, time

and pleasure related preferences are more important.

Regarding country of residence: We believe that the

temperature in a car and its ability to regulate the tem-

perature such are more important for Israelis due to

higher temperatures in Israel. Moreover, we believe

that smoking is most important for Israeli residents

due to religious reasons. However, this cannot be

proven, because we did not collect the corresponding

demographic data. Our results indicate that the prefer-

ences of people highly depend on their cultural back-

ground. Nevertheless, because the number of Israeli

residents is relatively small, almost all of our results

are limited to German residents. Regarding smoking:

we observe a high difference concerning the rank of

smoking between smokers and non-smokers. This in-

dicates the importance of smoking for non-smokers

because they want to avoid it. Likewise, the propor-

tions in Figure 2 show an irregularity at smoking. We

believe this is caused by two things: The answers are

bipolar distributed and the question is partly misun-

derstood.

Limitations. In the questionnaire, we asked the par-

ticipants to list preferences not considered. Three

people mentioned services like free internet connec-

tion, snack food and providing electricity to passen-

gers, which could be considered in the future for

ridesharing in general and for assigning rides to pas-

sengers. The same applies to rules regarding food

during a ride, which was also mentioned three times.

Besides not including these additional preferences, it

should be noted that this paper only considered prefer-

ences of passengers and excludes the ones of drivers.

This could be investigated in the future. The results

from the questionnaire might be wrong for some pref-

erences because a mismatch between observed (im-

plicit) and self-reported (explicit) importance of cer-

tain factors, such as trust (Papenmeier et al., 2019),

can appear. Moreover, our results indicate that the

conclusions drawn in this section are limited to Ger-

man people.

Future Research. To foster human-centric rideshar-

ing, we propose two directions for future research:

• First, user preferences could be simulated based

on the gathered data. Taxi trip data such as New

York City taxi (see (Donovan and Work, 2016))

data are already publicly available, but these do

not include preference characteristics of users.

We want to apply generative models that are able

to generate synthetic results based on provided

training data to add user preferences to existing

datasets.

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

166

• Second, these preference characteristics could

be considered to enable more human-centric

ridesharing. To do so on a larger scale, we assume

that AI algorithms would be necessary.

7 CONCLUSION

After analyzing factors and preferences that inﬂuence

ridesharing based on the current literature, we con-

ducted a survey to identify the preferences important

for users to be satisﬁed within a ridesharing assign-

ment process. Based on the literature study and the

survey, we were able to provide a comprehensive list

of preferences relevant for ridesharing and we con-

tribute an order of preferences based on relative im-

portance. In addition, we compared the importance in

demographic subgroups and collected signiﬁcant dif-

ferences among them.

In summary, comparing the observed importance

and the preferences occurrence in the literature, we

could not identify differences in situational factors.

Nevertheless, we observed high differences in judg-

mental factors that should be considered in future re-

search and applications. Based on our ﬁndings re-

garding the assignment process in ridesharing, we

recommend focussing on underrepresented prefer-

ences such as safety of a vehicle, driver’s compe-

tence and smell and to not focus on the overrepre-

sented preferences price, power and congestion too

much. Comparing different demographic subgroups,

we showed some additional ﬁndings, but overall and

similar to previous work the differences are relatively

small. However, our results indicate a high inﬂuence

of the country of residence to the relative importance

of preferences.

ACKNOWLEDGEMENTS

This work was supported in part by EC-RIDER, a

research project funded by the VolkswagenStiftung,

and Mobility Opportunities Valuable to Everyone

(MOVE), an Interreg project funded by the North

Sea Program of the European Regional Development

Fund of the European Union.

REFERENCES

Agatz, N., Erera, A., Savelsbergh, M., and Wang, X. (2012).

Optimization for dynamic ride-sharing: A review. Eu-

ropean Journal of Operational Research, 223(2):295–

303.

ıvodji, U. M., Gambs, S., Huguet, M.-J., and Killijian, M.-

O. (2016). Meeting points in ridesharing: A privacy-

preserving approach. Transportation Research Part

C: Emerging Technologies, 72:239–253.

Bian, Z. and Liu, X. (2019). Mechanism design for ﬁrst-

mile ridesharing based on personalized requirements

part I: Theoretical analysis in generalized scenar-

ios. Transportation Research Part B: Methodological,

120:147–171.

Chaube, V., Kavanaugh, A. L., and Perez-Quinones, M. A.

(2010). Leveraging Social Networks to Embed Trust

in Rideshare Programs. In 2010 43rd Hawaii Inter-

national Conference on System Sciences, pages 1–8,

Honolulu, Hawaii, USA. IEEE.

Correia, G. and Viegas, J. M. (2011). Carpooling and car-

pool clubs: Clarifying concepts and assessing value

enhancement possibilities through a Stated Preference

web survey in Lisbon, Portugal. Transportation Re-

search Part A: Policy and Practice, 45(2):81–90.

Dailey, D., Loseff, D., and Meyers, D. (1999). Seattle

smart traveler: Dynamic ridematching on the World

Wide Web. Transportation Research Part C: Emerg-

ing Technologies, 7(1):17–32.

Derrac, J., Garc

ıa, S., Molina, D., and Herrera, F. (2011). A

practical tutorial on the use of nonparametric statisti-

cal tests as a methodology for comparing evolutionary

and swarm intelligence algorithms. Swarm and Evo-

lutionary Computation, 1(1):3–18.

Donovan, B. and Work, D. (2016). New York City Taxi Trip

Data (2010-2013).

Ferguson, E. (1997). The rise and fall of the American car-

pool: 1970-1990. Transportation, 24(4):349–376.

Furuhata, M., Dessouky, M., Ord

nez, F., Brunet, M.-E.,

Wang, X., and Koenig, S. (2013). Ridesharing: The

state-of-the-art and future directions. Transportation

Research Part B: Methodological, 57:28–46.

Greenwood, B. N. and Wattal, S. (2017). Show Me the Way

to Go Home: An Empirical Investigation of Ride-

Sharing and Alcohol Related Motor Vehicle Fatalities.

MIS Quarterly, 41(1):163–187.

Hong, S. J. (2017). Assessing Economic Value of Reduc-

ing Perceived Risk in the Sharing Economy: The Case

of Ride-sharing Services. In AMCIS2017, page 10,

Boston.

Jacobson, S. H. and King, D. M. (2009). Fuel saving

and ridesharing in the US: Motivations, limitations,

and opportunities. Transportation Research Part D:

Transport and Environment, 14(1):14–21.

Jones, E., Oliphant, T., Peterson, P., et al. (2001). SciPy:

Open source scientiﬁc tools for Python.

Joseph, R. C. (2018). Ride-sharing services: The tumul-

tuous tale of the rural urban divide. In 24th Ameri-

cas Conference on Information Systems, New Orleans,

LA, USA.

Lavieri, P. S. and Bhat, C. R. (2019a). Investigating objec-

tive and subjective factors inﬂuencing the adoption,

frequency, and characteristics of ride-hailing trips.

Transportation Research Part C: Emerging Technolo-

gies, 105:100–125.

Lavieri, P. S. and Bhat, C. R. (2019b). Modeling individ-

uals’ willingness to share trips with strangers in an

autonomous vehicle future. Transportation Research

Part A: Policy and Practice, 124:242–261.

How Did You Like This Ride? An Analysis of User Preferences in Ridesharing Assignments

167

Li, Z., Hong, Y., and Zhang, Z. (2017). An Empirical Anal-

ysis of On-demand Ride-sharing and Trafﬁc Conges-

tion. In Hawaii International Conference on System

Sciences.

Liu, W., Zhang, F., and Yang, H. (2017). Modeling and

managing morning commute with both household and

individual travels. Transportation Research Part B:

Methodological, 103:227–247.

Mahmoudi, M. and Zhou, X. (2016). Finding optimal so-

lutions for vehicle routing problem with pickup and

delivery services with time windows: A dynamic pro-

gramming approach based on state–space–time net-

work representations. Transportation Research Part

B: Methodological, 89:19–42.

Mann, H. B. and Whitney, D. R. (1947). On a Test of

Whether one of Two Random Variables is Stochasti-

cally Larger than the Other. The Annals of Mathemat-

ical Statistics, 18(1):50–60.

Mirsadikov, A., Harrison, A., and Mennecke, B. (2016).

Tales From the Wheel: An IT-Fueled Ride as an

UBER Driver. In San Diego, page 10.

Mourad, A., Puchinger, J., and Chu, C. (2019). A survey of

models and algorithms for optimizing shared mobil-

ity. Transportation Research Part B: Methodological,

123:323–346.

Neoh, J. G., Chipulu, M., and Marshall, A. (2017). What

encourages people to carpool? An evaluation of fac-

tors with meta-analysis. Transportation, 44(2):423–

447.

Neoh, J. G., Chipulu, M., Marshall, A., and Tewkesbury, A.

(2018). How commuters’ motivations to drive relate

to propensity to carpool: Evidence from the United

Kingdom and the United States. Transportation Re-

search Part A: Policy and Practice, 110:128–148.

Nielsen, J. R., Hovmøller, H., Blyth, P.-L., and Sovacool,

B. K. (2015). Of “white crows” and “cash savers:”

A qualitative study of travel behavior and perceptions

of ridesharing in Denmark. Transportation Research

Part A: Policy and Practice, 78:113–123.

Nourinejad, M. and Roorda, M. J. (2016). Agent based

model for dynamic ridesharing. Transportation Re-

search Part C: Emerging Technologies, 64:117–132.

Papenmeier, A., Englebienne, G., and Seifert, C. (2019).

How model accuracy and explanation ﬁdelity inﬂu-

ence user trust. In arXiv:1907.12652 [Cs].

Parmesan, C. and Yohe, G. (2003). A globally coherent

ﬁngerprint of climate change impacts across natural

systems. Nature, 421(6918):37–42.

anchez, D., Mart

ınez, S., and Domingo-Ferrer, J. (2016).

Co-utile P2P ridesharing via decentralization and rep-

utation management. Transportation Research Part

C: Emerging Technologies, 73:147–166.

Schmitz, C. and Team, L. P. (2012). LimeSurvey: An Open

Source survey tool. LimeSurvey Project.

Schweitzer, S. and Brendel, J. (2018). The Segmented

Introduction of Transaction Fees in the German

Ridesharing Market. In PACIS 2018 Proceedings,

page 8, Japan.

Shared and Digital Mobility Committee (2018). Taxonomy

and Deﬁnitions for Terms Related to Shared Mobility

and Enabling Technologies. Technical report, SAE

International.

Statistisches Bundesamt (2019). 6 Prozent der EU-B

urger

nutzen Mitfahrzentralen.

Stiglic, M., Agatz, N., Savelsbergh, M., and Gradisar, M.

(2015). The beneﬁts of meeting points in ride-sharing

systems. Transportation Research Part B: Method-

ological, 82:36–53.

Sullivan, G. M. and Artino, A. R. (2013). Analyzing and

Interpreting Data From Likert-Type Scales. Journal

of Graduate Medical Education, 5(4):541–542.

Tian, L.-J., Sheu, J.-B., and Huang, H.-J. (2019). The morn-

ing commute problem with endogenous shared au-

tonomous vehicle penetration and parking space con-

straint. Transportation Research Part B: Methodolog-

ical, 123:258–278.

Tsao, H.-S. J. and Lin, D.-J. (1999). Spatial and Tempo-

ral Factors in Estimating the Potential of Ride-sharing

for Demand Reduction. California PATH Research

Report.

Vanoutrive, T., Van De Vijver, E., Van Malderen, L.,

Jourquin, B., Thomas, I., Verhetsel, A., and Witlox, F.

(2012). What determines carpooling to workplaces in

Belgium: Location, organisation, or promotion? Jour-

nal of Transport Geography, 22:77–86.

vom Brocke, J., Simons, A., Niehaves, B., Plattfaut, R.,

Cleven, A., and Reimer, K. (2009). Reconstructing

the giant: On the importance of rigour in documenting

the literature search process. European Conference on

Information Systems.

Wang, J.-P., Ban, X. J., and Huang, H.-J. (2019).

Dynamic ridesharing with variable-ratio charging-

compensation scheme for morning commute. Trans-

portation Research Part B: Methodological, 122:390–

415.

Wang, X., Agatz, N., and Erera, A. (2018). Stable Matching

for Dynamic Ride-Sharing Systems. Transportation

Science, 52(4):850–867.

Washbrook, K., Haider, W., and Jaccard, M. (2006). Es-

timating commuter mode choice: A discrete choice

analysis of the impact of road pricing and parking

charges. Transportation, 33(6):621–639.

Webster, J. and Watson, R. T. (2002). Analyzing the past

to prepare for the future: Writing a literature review.

MIS Quarterly, 26(2):13–23.

Xiao, L.-L., Liu, T.-L., and Huang, H.-J. (2016). On the

morning commute problem with carpooling behavior

under parking space constraint. Transportation Re-

search Part B: Methodological, 91:383–407.

Yan, X., Levine, J., and Zhao, X. (2019). Integrating rides-

ourcing services with public transit: An evaluation of

traveler responses combining revealed and stated pref-

erence data. Transportation Research Part C: Emerg-

ing Technologies, 105:683–696.

Yousaf, J., Li, J., Chen, L., Tang, J., and Dai, X.

(2014). Generalized multipath planning model for

ride-sharing systems. Frontiers of Computer Science,

8(1):100–118.

Zhang, N., Kien, S. S., and Lee, G.-w. (2018). The Insti-

tutional Legitimacy of Disruptive Start-ups in Sharing

Economy. In PACIS 2018 Proceedings, page 8, Japan.

VEHITS 2020 - 6th International Conference on Vehicle Technology and Intelligent Transport Systems

168