Assessing the Influence of Co-Sociality on a User’s Willingness to Participate on a Car Sharing Platform through Trust: A Vignette Study

(1)

Bachelor thesis 2019/2020 - University of Amsterdam

Assessing the Influence of Co-Sociality on a

User’s Willingness to Participate on a Car

Sharing Platform through Trust: A Vignette

Study

Trust in the Sharing Economy

Britt Markman 11807733 09-07-2020 Nicole Stofberg University of Amsterdam Faculty of Economics and Business

(2)

2

Statement of Originality

This document is written by Britt Markman, who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

3. Methodology... 18 3.1 Vignette Design ... 18 3.2 Pre-Test ... 19 3.3 Survey Questionnaire ... 19 3.3.1 Mediator variables ... 19 3.3.2 Dependent variables ... 19 3.3.3 Control variables ... 20 3.3.4 Manipulation checks ... 20 3.4 Data Collection ... 21 3.4.1 Sample ... 21 3.5 Data Analysis ... 21 4. Results ... 23 4.1 Distribution ... 23 4.2 Descriptive Statistics ... 23 4.2.1 Correlations ... 23 4.2.2 Control Variables ... 25 4.3 Regression Analysis ... 26

4.3.1 Dependent and independent variables ... 26

4.3.2 Control variables ... 26

4.4 Hypotheses Testing ... 27

4.4.1 Hypothesis 1 ... 28

(4)

4

5. Discussion ... 31

5.1 Summary of the Results ... 31

5.2 Discussion of the Results ... 31

5.2.1 The direct relationship between sociality and willingness to participate ... 31

5.2.2 Mediation of platform trust ... 32

5.2.3 Mediation of peer trust ... 32

5.2.4 Sequential mediation of platform trust and peer trust ... 33

5.3 Implications ... 33

5.4 Limitations and Suggestions for Future Research ... 34

6. Conclusion ... 35

7. References ... 36

8. Appendices ... 41

8.1 Appendix A: Pre-test ... 41

8.2 Appendix B: Vignettes ... 41

8.3 Appendix C: Survey questionnaire ... 45

8.4 Appendix D: Credibility and realism checks ... 52

8.5 Appendix E: Overview of the dummy variables ... 53

8.6 Appendix F: Cronbach’s alpha ... 54

8.7 Appendix G: Skewness and kurtosis ... 55

8.8 Appendix H: Kolmogorov-Smirnov and the Shapiro-Wilk tests ... 55

8.9 Appendix I: Linear regression analysis ... 55

(5)

5

List of Tables

Table 1. Overview of the hypotheses………. 17

Table 2. Correlation matrix ..……… ……...………… 24

Table 3. Coefficients and significance levels mediation effect ....………. 29

Table 4. Effect size and significance levels mediation effect .………. 30

Table 5. Questions of the pre-test………..……….……… 41

Table 6. Overview of the dummy variables...……… 53

List of Figures

Figure 1. Conceptual framework with hypotheses……… 17

(6)

6

Abstract

The biggest drawback for engaging in a sharing encounter is the lack of trust. Research has shown that by incorporating a reputation system on a peer-to-peer sharing platform, the trust in the peers who participate in sharing activity increases. In turn, a higher trust translates to a higher willingness to participate. However, to date it has not been systematically investigated if the willingness to participate is linked with if sharing products are privately or communally consumed. This study bridges this gap, by investigating the influence of co-sociality and trust in the platform or in the peer on the willingness to participate. Building on existing theory concerning peer-to-peer sharing, it asks: To what extend does high co-sociality influence the perception and trust in the sharing encounter or the sharing platform and thus the willingness to participate? In this context, co-sociality is defined as the virtual copresence of social factors on a platform in a way that the peer providers and the peer users can interact with each other. This is contrasted with sharing platforms in which users never interact in order to increase convenience and ease of use. Based on the literature, four hypotheses were formulated and answered by conducting an online vignette experiment (N=419). Respondents were randomly divided between several vignettes and asked to answer multiple questions regarding trust in the platform, trust in the peer and overall willingness to participate. The analysis of the results demonstrated that in contradiction to expectations, a high degree of co-sociality leads to a lower willingness to participate. Platform trust and peer trust were both crucial elements for increasing the willingness to participate. However, a high degree of co-sociality neither leads to a higher platform trust nor higher peer trust. These results indicate that high co-sociality does not influence the perception and trust in the sharing encounter or the sharing platform and has a negative effect on the willingness to participate. It is recommended that managers focus on increasing trust in the platform, which in turn positively influences peer trust and the willingness to participate. Managers should not focus on increasing the degree of co-sociality, since it neither leads to higher trust nor a higher willingness to participate. Perhaps managers should sacrifice sociality in order to establish convenience.

Keywords: sharing economy, car sharing, peer-to-peer sharing, co-sociality, platform trust, peer trust,

(7)

7

1. Introduction

Across a broad variety of fields, a shift from sole ownership and consumption to a shared use and consumption of possessions is observed. This is possible because of the new and innovative ways of peer-to-peer sharing, which are facilitated by online platforms. Peer-to-peer sharing is not a new phenomenon. However, due to technological developments it is more accessible, which include the advancements of online platforms, smartphones, and reputation systems on the platforms (Benjaafar, Kong, Li & Courcoubetis, 2019). This environment, where consumers grant each other temporary access to their underused personal possessions, can be defined as the sharing economy (Frenken, Meelen, Arets, & Van de Glind, 2015). The intentions to participate in a sharing encounter are mostly based on trust in the platform and trust in the peers (Frenken et al., 2015). Therefore, a lot of platforms use a reputation system, and include face-to-face interactions or direct online communication, in order to build trust. This increases the degree of co-sociality, which is the virtual copresence of social factors on a platform (Perren & Kozinets, 2018). This degree of co-sociality is expected to influence the willingness to participate. The reason for this is that the inclusion of reviews and ratings make deviant behaviour visible to other users (Möhlmann, 2016), which builds trust. Frenken et al. (2015) found that higher levels of trust lead to higher intentions to engage. Furthermore, it is expected that platform trust can be transferred to peer trust. However, this has not been researched extensively yet.

The goal of this paper is to explore the relationship between the degree of co-sociality and the willingness to participate, mediated by platform trust and peer trust, and platform trust and peer trust in serial. It is useful for managers to see what influences the willingness to participate the most, in order to know which approach is the most valuable. The degree of co-sociality can vary from low, where the platform limits or even restricts social interactions, to high, where the platform allows for reviews, ratings and face to face interactions or direct online communication. A high degree of sociality is expected to increase both platform trust and peer trust, which will increase the willingness to participate (Frenken et al., 2015). Therefore, the following research question is formulated:

To what extend does high co-sociality influence the perception and trust in the sharing encounter or the sharing platform and thus the willingness to participate?

This question contains three sub-questions:

1. What is the mediating role of platform trust on the relationship between co-sociality and willingness to participate?

2. What is the mediation role of peer trust on the relationship between co-sociality and willingness to participate?

3. What is the sequential mediating role of platform trust and peer trust on the willingness to participate?

(8)

8 These questions are answered by using a fictional car sharing platform named SplitCar, and by using seven different vignettes in which the degree of co-sociality is manipulated. The effects of the

vignettes on platform trust, peer trust, and the willingness to participate are analysed (N=419). Due to the current corona virus crisis, the vignettes control for this, and it is taken into consideration in the analysis.

The findings of this study contribute to science in different ways. The findings add to the current literature about trust and reputation, and thus the need for platform trust and peer trust on peer-to-peer sharing platforms, in order to get people to participate. Next to that, it shows sharing platforms on how to increase the willingness to participate through the inclusion of reputation systems. This can, on a larger scale, be beneficial for the environment.

The structure of this thesis is as follows: Firstly, the literature review explores the current theory and lays out the hypotheses and the conceptual model (chapter 2). Secondly, the methods are explained (chapter 3), and the results are analysed (chapter 4). Thirdly, the discussion summarizes results and discusses the findings, contributions, and limitations (chapter 5). Lastly, the conclusion will answer the research question (chapter 6).

(9)

9

2. Literature Review

The literature review contains the definition of the sharing economy (paragraph 2.1), which includes the different types of sharing platforms of which one is considered pseudo-sharing, which is explained next. After that willingness to participate in a sharing encounter and the importance of sociality on peer-to-peer platforms (paragraph 2.2) are explored. Next, the importance of trust is investigated (paragraph 2.3), and lastly, an overview of the hypotheses and the conceptual model is presented (paragraph 2.4).

2.1 The Sharing Economy

Nowadays, there are a lot of different platforms that make the act of sharing between people more accessible. Items being shared vary from garden tools, to cars, to houses. Sharing is crucial consumer behaviour, not to be confused with gifting and the transfer of goods (Belk, 2010). Sharing has been crucial for several thousands of years, as it is possibly the most elemental form of economic

distribution among humans (Price, 1975). Price (1975) defines the term sharing as “the allocation of economic goods and services without calculating returns, within an intimate social group, and patterned by the general role system of that group”. Sharing has also been defined as “the act and process of distributing what is ours to others for their use as well as the act and process of receiving something from others for our use” (Belk, 2007). The arrival of the internet and smartphones have enabled sharing beyond our own social circles. Once this occurs through the mediation of an online platform, we speak of sharing within the sharing economy (Stofberg & Bridoux, 2019). Currently, there are a lot of different forms of sharing, including peer-to-peer and business-to-consumer sharing. These are ways of sharing that do not specifically happen within an intimate social group but can happen with strangers found on different sharing platforms or with sharing companies. There are also multiple different ways to monetize peer-to-peer sharing or business-to-consumer sharing, in contrast to the definition of sharing stated by Price (1975). The sharing environment is constantly changing and the platforms that make this possible, belong to the umbrella term ‘the sharing economy’.

The sharing economy can be defined as consumers who grant each other temporary access to their underused personal possessions, also referred to as idle capacity, possibly for money (Frenken et al., 2015). Examples of goods that are shared in the sharing economy are cars (Car2Go, Snappcar) and homes (AirBnB, Couchsurfing). These goods can be defined as “shareable goods” (Benkler, 2004). These are goods that naturally give the owners idle capacity, so the owners gain the opportunity to lend their goods to others. Idle capacity is present when the personal possessions of the owner are underused, which thus can include cars, houses, tools, and so on. Exceptions would be possessions that

(10)

10 the owner always uses or must always have the ability to use them, like a mobile phone (Frenken & Schor, 2019).

2.1.1 Peer-to-peer vs. Business-to-consumer

There are debates about which companies, or which forms of sharing, are part of the sharing economy. There are two different ways to increase the utilization of durable assets, and thus two different ways of sharing on a platform, namely business-to-consumer platforms and peer-to-peer platforms. Peer-to-peer sharing, or consumer-to-consumer sharing, can be defined similarly as the term the sharing economy. Peer-to-peer sharing means that consumers grant each other temporary access to their underused personal possessions, also referred to as idle capacity (Frenken et al. 2015). Thus, it is not about transfer of the ownership of the good, but only temporary access to the good. Peer-to-peer car-sharing, like Snappcar, is the rental of cars between private individuals, facilitated by online-based platforms. This phenomenon is often called collaborate consumption (Botsman and Rogers, 2010). Collaborate consumption consists of a peer-provider, who offers their personal possessions for rent, and is the primary-user of that possession. This peer-provider is also the legal contractual partner for the peer-user. The peer-user is the person who rents the personal possessions from the peer-provider. And lastly, there must be a service enabler, who provides and manages the online platform which connects the peer-provider and peer-user, and in exchange keeps a share of the rental fee (Wilhelms, Henkel & Falk, 2017). Business-to-consumer sharing, on the other hand, is when the ownership of the car is in the hands of a company instead of a peer.

This study focusses on car sharing to test the hypotheses. Car sharing is a good example of sharing as cars sit idle 95% of the time (Rapier, 2017) making them a prime shareable good. Moreover, unlike the hospitality sector, it is still in its infancy phase, demonstrating the need to see how such car sharing platforms can grow. Different sharable cars can already be found in the streets of Amsterdam. This includes the German based car-sharing company Car2Go. Car2Go is a business-to-customer sharing platform. These Car2Go cars can be used by subscription holders for inner city journeys and can be parked anywhere for free (Suiker & van den Elshout, 2013). Car2Go differentiates itself from other car-sharing companies by not having fixed costs for the user, no need to book the car, possibility of one-way usage and real-time information about the availability and location of the cars (Firnkorn & Müller, 2011). Besides Car2Go, there is also car sharing platform Snappcar. Snappcar is a platform on which people can rent out their car to others, with the underlying thought that it can have a huge impact on the world. Snappcar is a peer-to-peer sharing platform. Sharing can lead to new contact between people, less cars that need to be produced and it saves money for both the owners of the cars and the renters of the cars (Boztas, 2017). In contrast to Car2Go, Snappcar is a traditional car-sharing system with fixed stations since the car needs to be returned to the owner.

(11)

11

2.1.2 Pseudo-sharing

Car2Go is thus a business-to-customer sharing platform. Belk (2014) stated that business-to-customer platforms engage in pseudo-sharing, which he defines as the act of a business relationship pretending to be communal sharing. Pseudo-sharing is in essence not a bad thing, since it still might be beneficial to the environment and the people engaging in it, but it is not the act of sharing. According to Belk (2014), to determine if something is being shared, the thing should be considered to be temporarily “ours” instead of “mine” or “yours”. There are four types of pseudo-sharing, including long-term renting and leasing, short-term rental, online sites ‘sharing’ your data, and online-facilitated barter economies (Belk, 2014). Long-term renting and leasing can include renting a house for a long period of time or leasing a car for a long period of time. During the time that the house or car is in your possession, it is considered “yours”. There is no sense of shared ownership with the former or future renter. That is why long-term rental is not considered sharing, but pseudo-sharing. Short-term rental also falls under pseudo-sharing, but for other reasons than described above. Examples of short-term rental include car rental, like Car2Go, tool rental, or clothing rental. Most of the business-to-customer sharing platforms, which are thus for-profit, can be described as short-term rental, and not as sharing (Belk, 2014). Online websites like Facebook or YouTube create a sense of community, and enable real sharing with friends, groups of interest or unknown people from all over the world. Despite the share button being used to connect with friends or others, it is mostly a disguise of the social network sites to attract more members and obtain data that can be sold to third-party firms. The last type of pseudo-sharing is online-facilitated barter economies. Barter economies can be placed somewhere in the middle of market economies and sharing economies. In barter economies, people engage in a mutual exchange of assets which typically does not involve money. It is only due to the fact that the exchange does not involve money, that some consider barter economies a form of sharing (Belk, 2014). Botsman and Rogers (2010) wrote in their article that barter economies are part of a collaborative lifestyle, and thus sharing. However, according to Belk (2014), barter economies are only pseudo-sharing, because it involves mutual exchange of goods and services and not mutual ownership. Because pseudo-sharing is not real sharing, it can thus be exempt from this research as it does not belong to the sharing

economy. From this point on, this paper will continue to explore peer-to-peer sharing as its main point of focus.

2.2 The Willingness to Participate in a Sharing Encounter

The willingness to participate in a sharing encounter is defined by Mittendorf, Berente and Holten (2019) as “conducting the corresponding service transaction, which in case of a service provider means offering a sharing service or accepting a service request, and which in case of a customer means requesting a sharing service”. People love the idea of sharing, but are hesitant to participate, due to so

(12)

12 called ‘stranger danger’ (Belk, 2014). On peer-to-peer platforms, there is the concern that the traded assets might be damaged due to unobservable behaviour of the peer that is being shared with, which is called the moral hazard problem (Weber, 2014), which can be resolved through trust. To increase the amount of trust, internet markets often make use of a feedback system which is based on reputation, to enable traders to openly share information about previous transactions and transaction partners

(Bolton, Greiner & Ockenfels, 2013). Möhlmann (2016) suggests that simultaneous reviews, which are mutual reviews by both the peer provider as the peer user, can be seen as the development of traditional one-sided peer-based reviews. These reviews increase the level of co-sociality on peer-to-peer sharing platforms. Sociality is defined as the common affinity of people to socialize in groups and to engage with others in shared relationships (Wittel, 2001). Co-sociality is the virtual copresence of social factors on a platform, so that the peer providers and the peer users can interact with each other (Perren & Kozinets, 2018). The degree of co-sociality can vary from low, where the platform limits or even restricts social interactions, to high, where the platform allows for reviews, ratings and face to face interactions or direct online communication. An example of a peer-to-peer sharing platform that engages in a low degree of co-sociality is the car sharing platform Turo. Turo uses keyless technology to open to the car, so the peer user does not need to meet the peer provider face-to-face. This strategy enables Turo to compete based on instant booking, convenience, and availability with traditional car rental businesses (Wilhelms, Merfeld, & Henkel, 2017). However, as stated before, traditional car rental businesses are business-to-consumer platforms. Most of the business-to-customer sharing platforms can be described as short-term rental, and not as sharing (Belk, 2014). For this reason, the convenience users experience with keyless technology does not translate to ‘true’ sharing practices.

Because reviews offer other users’ information about the former transactions and presentation of another user and make deviant behaviour visible to other users (Möhlmann, 2016), it builds trust. So, digital sharing platforms decrease the idea and the risk of ‘stranger danger’ by obtaining relevant information through reviews, ratings and thus reputations (Frenken & Schor, 2019). The inclusion of ratings, reviews and face-to-face interactions or direct online communication are associated with a high degree of sociality on the platform. It is hypothesized that platforms with a high degree of co-sociality will enjoy a higher willingness to participate than platforms who restrict co-co-sociality. Thus:

H1. P2P platforms with high co-sociality will enjoy a higher willingness to participate, than platforms who restrict co-sociality.

(13)

13

2.4 The Importance of Trust

The biggest drawback for sharing appears to be the absence of trust. In peer-to-peer sharing context, there have been reports of several large incidents including theft, rape and wilful damage, which increases the confusion and the problem of low trust in the legal measures connected with peer-to-peer sharing platforms (Möhlmann, 2016). There is an increased hazard of misbehaviour when consumers have the ability to access goods they do not own (Durgee & O’Connor, 1995). The three most obvious risk factors for sharing on a peer-to-peer platform were the risk of life loss, risk of theft, and the loss of property (Kamal & Chen, 2016). Additionally, online environments and thus platforms are risky due to the fact that both the identity of the owner and the characteristics of the rented product cannot be fully evaluated (Lee, 1998).

Trust can be defined as “the extent to which one is willing to ascribe good intentions to and have confidence in the words and actions of other people.” (Cook & Wall, 1980). This will then affect the behaviour of one towards others. Trust as described by McAllister (1995) allows human beings to take risks, because trust gives people the feeling that others will not take advantage of them. With trust, people expect to find what is predicted instead of what is feared, and thus trust revolves around the concepts of competence and responsibility. McAllister combines the ideas above and defines interpersonal trust as “the extent to which a person is confident in, and willing to act on the basis of, the words, actions, and decisions of another” (McAllister, 1995).

Möhlmann (2016) argues that digital trust on peer-to-peer sharing platforms is dissimilar to other contexts, like exchange platforms (e.g. eBay), online sales markets and non-peer-to-peer traditional retail services. In contrast to traditional ways of exchanging goods via traditional retail services, peer-to-peer platforms include three parties in each transaction, instead of two. These three parties are the peer provider, the peer receiver or peer user, and the intermediary or the platform. It is important to acknowledge the difference between trust in the platform and trust between peers who engage in sharing encounters on the platform (Möhlmann, 2016). Secondly, there is more intense contact between the peers on a peer-to-peer sharing platform than on a more traditional and anonymous online platform where goods are exchanged. The former would sometimes require the peers to engage in contact in real life (e.g. for transfer of keys), while in case of the latter the customer receives a package without ever having to meet the seller. For users of peer-to-peer platforms this results in a more socially advanced interaction due to entering the personal space of the other peer (Möhlmann, 2016). Trust is important in these face-to-face interactions, given the fact that social dimensions become more relevant. It is only logical to distinguish between peer trust and platform trust, because both the peers and the platform enable other parts of the transaction (Sundararajan 2016). The differences between peer trust and platform trust are explored below.

(14)

14

2.3.1 Platform trust

It is important for a sharing platform to ensure that users trust the platform. Given the fear of ‘stranger danger’ and the moral hazard problem, users of the platform must trust that the platform will actively try to stop unwanted behaviour (Botsman 2016). Mittendorf et al. (2019) created a model including trust in the intermediary and trust in the sharing partner, which in turn leads to the willingness to participate or not to participate in a sharing encounter. Trust in the intermediary refers to an in-between entity, that is the owner of an online sharing platform or an online marketplace, and thus further referred to as platform trust. This in-between entity only provides the platform where sharing encounters can take place, but it does not own the shared assets (Belk, 2010). People who trust the platform, are certain that the sharing platform positively supports the sharing activity in their good interest (Mittendorf et al., 2019). Mohlmann (2016) found that a higher level of platform trust takes away the perception of risk. Frenken et al. (2014) found that higher levels of trust lead to higher intentions to engage. Next to that, McKnight, Choudhury & Kacmar (2002) found that platform trust is a crucial prerequisite for the willingness to participate on said platform. Because both sharing partners, the provider and the receiver, depend on the intermediate platform for asset sharing encounters, the trust in the intermediate platform assists the user’s willingness to participate in the sharing encounter (Mittendorf et al., 2019).

The presence of reviews and ratings of the platform are a form of building a reputation for that platform. Reputation is a symbol of trustworthiness created by the statements of other users (Slee, 2013). Next to reputation, also social presence is a symbol of trustworthiness. Kim, Yoon and Zo (2015) define perceived social presence as “the extent to which a user experiences other users as being psychologically present”. This is greatly related to trust. Both reputation and social presence

contribute to a higher degree of co-sociality of the platform. Therefore, it is hypothesized that the relationship between the degree of co-sociality on a peer-to-peer sharing platform and the willingness to participate on that platform is mediated by platform trust, such that a high degree of co-sociality on the platform leads to a higher platform trust, which in turn has a positive effect on the willingness to participate on that platform. Thus:

H2. The relationship between co-sociality measures on P2P sharing platforms and willingness to participate is mediated by platform trust, such that the inclusion (absence) of specific information types is associated with higher (lower) platform trust, which has a positive (negative) effect on the willingness to participate.

(15)

15

2.3.2 Peer trust

The biggest drawback for sharing appears to be the absence of trust in other sharing members. Sharing partners include both the peer user and the peer provider who trade their resources with each other on a sharing platform (Mittendorf et al., 2019). Because of the fear of ‘stranger danger’, people are instructed to distrust strangers. This hinders the willingness to participate in a sharing encounter with them (Gebbia, 2016). The plan to engage in a sharing encounter is based on in the interpersonal trust defined by McAllister (1995), that a sharing partner is confident in the words, actions, and decisions of another sharing partner. Higher levels of trust lead to higher intentions to engage (Frenken et al., 2014). Because of trust leading to willingness to participate, sharing platforms try to strengthen the trust between the peer users. But this is made difficult by the lack of face-to-face interaction, since this makes it harder to assess the trustworthiness of the other peer user (Ishaya & Mundy, 2004). In an offline environment there are certain signals, like body language, which asses the trustworthiness of the peer. These signals are absent in an online environment (Pavlou & Gefen, 2004). In turn, sharing platforms allow for face-to-face interactions or direct online communication and reviews and ratings from other peers, as a way to increase the level of identification and decrease the social distance (Ert, Fleischer and Magen, 2016), and thus increasing the degree of co-sociality of the platform.

With platform trust as well as with peer trust, the presence of review and ratings of users, and face-to-face interactions or direct online communication on the platform, are a form of building a reputation for that user. As stated before is reputation a symbol of trustworthiness (Slee, 2013). Next to reputation, also social presence is a symbol of trustworthiness. The face-to-face interactions or direct online communication is expected to lead to a higher social presence, which is greatly related to trust (Kim et al., 2015). Therefore, it is hypothesized that the relationship between the degree of co-sociality on a peer-to-peer sharing platform and the willingness to participate on that platform is mediated by peer trust, such that a high degree of co-sociality on the platform leads to a higher peer trust, which in turn has a positive effect on the willingness to participate on that platform. Thus:

H3. The relationship between co-sociality measures on P2P sharing platforms and willingness to participate is mediated by peer trust, such that the inclusion (absence) of specific

information types is associated with higher (lower) peer trust, which has a positive (negative) effect on the willingness to participate.

The following sub paragraph continues with the exploration of the effect of platform trust in relation to peer trust on the willingness to participate.

(16)

16

2.3.3 The interaction between platform trust and peer trust

In the paragraphs before, platform trust and peer trust were seen as two separate parts of what can actually be seen as a whole. This so-called trust transfer is broadly defined by Lu, Yang, Chau, & Cao (2011) as a process whereby trust in one entity influences the feelings and opinions on another entity. In a paper about trust transfer on the World Wide Web from Stewart (2003), the transfer of trust is seen as a cognitive process. A person grounds its initial trust in one entity on the trust of another connected entity. These entities are thus the platform and the peer. This is interchangeable, so platform trust can be transferred to peer trust, and peer trust can be transferred to platform trust. A reason for the transfer of platform trust to peer trust, is that platform can try to promote peer trust by

demonstrating that the platform cares about the social relationships between peers, which is a sign to the peers using the platform that the platform will not behave undesirably or deviously towards the peers (Chen, Zhang & Xu, 2009). Thus, the degree of sociality of the platform increases the platform trust, and in turn this leads to a higher peer trust. Therefore, the following hypothesis is formulated:

H4. The relationship between co-sociality measures on P2P sharing platforms and willingness to participate is sequentially mediated by platform trust and peer trust, such that the inclusion of specific information types is associated with higher (lower) levels of peer trust and platform trust, which leads to a higher (lower) willingness to participate.

In the following paragraph, an overview of the hypotheses is provided, and the conceptual framework is displayed.

(17)

17

2.4 Hypotheses and Conceptual Framework

This study investigates the effects of the degree of sociality on a peer-to-peer platform on the willingness to participate on that platform. It is hypothesized that platforms with a high degree of sociality will enjoy a higher willingness to participate than platforms who restrict sharing. The relationship between sociality and willingness to pay is expected to be mediated by peer trust and platform trust. Furthermore, peer trust and platform trust are expected to mediate the relationship between sociality and the willingness to participate sequentially. Figure 1 displays the conceptual framework with the matching hypotheses. Table 1 gives an overview of the four hypotheses.

Figure 1.

Conceptual framework with hypotheses

Table 1.

Overview of the hypotheses

H1 P2P platforms with high co-sociality will enjoy a higher willingness to participate, than platforms who restrict co-sociality.

H2 The relationship between co-sociality measures on P2P sharing platforms and willingness to participate is mediated by platform trust, such that the inclusion (absence) of specific information types is associated with higher (lower) platform trust, which has a positive (negative) effect on the willingness to participate.

H3 The relationship between co-sociality measures on P2P sharing platforms and willingness to participate is mediated by peer trust, such that the inclusion (absence) of specific information types is associated with higher (lower) peer trust, which has a positive (negative) effect on the willingness to participate.

H4 The relationship between co-sociality measures on P2P sharing platforms and willingness to participate is sequentially mediated by platform trust and peer trust, such that the inclusion of specific information types is associated with higher (lower) levels of peer trust and platform trust, which leads to a higher (lower) willingness to participate.

(18)

18

3. Methodology

In this chapter, the design of this study is described. The survey instrument of choice of this thesis is a vignette study. This vignette study measures how the degree of co-sociality influences the willingness

to participate. A pre-test is executed prior to the research to make sure that the vignettes purpose is

clear and to check the credibility of the sketched situations of the vignettes. Next, the variables are described, including the measured independent variable, mediators, dependent variable, and control variables. The method of data collection and the sample of this study are described. Lastly, the analyzation method of the data collected from the vignette study is explained in detail.

3.1 Vignette Design

An experimental vignette methodology is used in this thesis to test the hypotheses. Atzmüller and Steiner (2010) describe a vignette as a brief and carefully assembled description of either a person, an item, or a situation, which represents a mixture of characteristics. A vignette study is superior to a normal survey design as it combines the features of a survey design with the benefits of qualitative research methods (Finch, 1987). The experimental vignette methodology entails that researchers come up with realistic scenarios, which they then present to the participants in order to assess the dependent variables and to draw out their judgements about the realistic scenarios (Aguinis & Bradley, 2014). In this study, the vignettes describe a scenario where the participant is a user of SplitCar, a fictional peer-to-peer car-sharing platform. In each vignette, the independent variable is manipulated to sketch a different realistic situation. The analyses performed in this thesis are part of a larger study. In total, 24 vignettes are created, but not all 24 are relevant for this thesis. Only seven out of the 24 vignettes are analysed. In the vignettes, the independent variable sociality is manipulated, so that in some vignettes the platform SplitCar allows for reviews and user ratings as well as face to face interactions or direct online communication, and in others it does not allow for this. Respondents of the vignette study are thus placed in the hypothetical realistic situation in which they use the peer-to-peer platform SplitCar for car rental, or to rent out their own car. In the current time, it is important to include a Covid-19 control variable, to see how the corona virus influences the outcome of this research. Therefore, the vignettes control for corona; There is a hypothetical situation where the corona virus is still a large problem, and a hypothetical situation where there is a cure and a vaccine for the corona virus and it is thus not a large problem. All seven vignettes are based on a peer-to-peer version of the car-sharing platform SplitCar. The participants of the study are randomly assigned to conditions in a between-subjects design. Afterwards they were asked behavioural intentions to participate through a traditional survey technique, explained in more detail in section 3.3.

(19)

19

3.2 Pre-Test

A pre-test is executed prior to the research to make sure that the vignettes’ purpose is clear and to check the credibility of the sketched situations of the vignettes. Firstly, all the vignettes are created. There are three different independent variables manipulated in the different vignettes. The independent variables used for the vignettes include the corona variable, the used business model of the platform, which is peer-to-peer, and the sociality variable. Since the Coronavirus variable is used as a control variable, all possible combinations of the other variables must be tested with the presence of the corona virus and with the absence of the corona virus. Each vignette was shared with five people who read the vignettes and answered thirteen questions about them (see Appendix A; table 1). After, the vignettes are improved, and all the unclear elements are altered. For the final vignettes, see Appendix B.

3.3 Survey Questionnaire

After the pre-test, the vignettes are perfected, and the survey questionnaire is created. For this questionnaire, Qualtrics survey software is used (see Appendix C). In the survey, the vignettes are randomly assigned to the participants of the study in a between-subjects design. The survey serves the purpose of gaining a better understanding of the effect of the mediator variables peer trust and

platform trust on the dependent variable willingness to participate. Control variables are included to

rule out any effects which are due to these control variables instead of the independent variables.

3.3.1 Mediator variables

The mediator variables are peer trust and platform trust. Both mediator variables are measured with a 7-point scale (1 is completely disagree; 7 is completely agree) adopted from Pavlou and Gefen (2004). For peer trust, the items include: “The customers on SplitCar are in general dependable”; “The

customers on SplitCar are in general reliable”; “The customers on SplitCar are in general honest”; and “The customers on SplitCar are in general trustworthy”. For platform trust, the items include: “As a platform provider, SplitCar can be trusted at all times”; “As a platform provider, SplitCar can be counted on to do what is right”; “As a platform provider, SplitCar has high integrity”; and “SplitCar is a competent platform provider”.

3.3.2 Dependent variables

The dependent variable is willingness to participate. Participants of the study were asked to indicate with a 7-point the extent to which they agreed with the given statement (1 is completely disagree; 7 is completely agree). The scale and statements are adopted from Lamberton and Rose (2012), Pavlou and

(20)

20 Gefen (2004), and White et al. (2012). The statements include: “Given the chance, I predict that I would use the services of SplitCar in the future”; “Given the opportunity I intent to participate in car sharing initiatives like SplitCar”; “I would exert effort to make use of the services of SplitCar”; and “It is likely that I would choose SplitCar the next time I need a car”. The last statement uses a 7-point scale to indicate the likeliness of the statement (1= very unlikely; 7= very likely).

3.3.3 Control variables

The demographic variables age, gender, education level, income and place of residence are included as control variables. Furthermore, it is important to include controls for driver’s licence possession and car possession. Additionally, there is a corona virus control variable included in the vignettes. The corona measures are loosely adopted from Goodwin et al. (2009) and are measured on a 5-point scale. The statements test for benefit, risk and impact on intent, and are described as follows: “Please

indicate the safety benefits you perceive from traveling with a car on SplitCar as compared to traveling with public transportation (1=no safety benefits at all 5; great safety benefits)”; “Please indicate to what extent you believe that renting a car from SplitCar could be risky for my health (1=not at all risky; 5= very risky)”; and “To what extent do you think that your willingness to try car sharing on SplitCar is influenced by corona? (1= greatly decreases my likelihood to engage in car sharing; 3= no impact 5= greatly increases my likelihood)”. Next to the corona measures, there are the general corona perceptions, which include the severity of corona and the effectiveness of social control. These

variables are measured on a 5-point scale and are based loosely on Rosenstock (1990). The statements include: “How severe do you think the consequences of the corona virus might be to you or your family? (1=not at all severe; 5= very severe)”; “If a member of your immediate household became ill during the corona virus, how likely do you believe it is that the person might become seriously ill? (1=not at all likely; 5=very likely)”; “To what extent do you agree with the following statement: compared to the average Dutch citizen, the risks that I myself will become seriously ill if I catch corona is…. (1= much less likely; 5 is far more likely)”; and “How affective to do you think social distancing measures are in protecting you from Corona? (1=very effective; 5=not at all effective)”.

3.3.4 Manipulation checks

The vignettes are also checked on credibility and realism. This is done with a 7-point scale, in which 1 indicates that the participant completely disagrees, and 7 indicates that the participant completely agrees. For realism, the statement “I found the situation in the above‐mentioned scenario realistic” is used. For credibility, the statement “I had no problem imagining myself in the above‐mentioned situation” is used. The mean for the variable credibility is M=4.86 and for realism is M=5.10. This indicates that the hypothetical situation was relatively realistic and credible, so that the respondents could put themselves in the sketched situation relatively easily. A one-way analysis of variance is

(21)

21 performed to make sure that there are no significant differences in realism (F(1, 417)=.204, p=.652) and credibility (F(1, 417)=.403, p=.526) between the groups (see Appendix D).

3.4 Data Collection

The data is collected through the Qualtrics survey. This survey is distributed among Dutch and international people, and it is not limited to the Netherlands. This will result in a large and diverse sample. The survey is forwarded to friends and family, and the distribution happens trough social media (Facebook, WhatsApp, LinkedIn, Instagram), as well as mouth-to-mouth and e-mail.

3.4.1 Sample

A sample of 419 respondents is recruited. The sample consists of respondents with a wide variety of ages, ranging from 18 to 92 (mean = 32.11, SD=12.917). The majority is younger than 26 years old (56.1%). More women than men and other participated in this study, respectively 56.3%, 42.5% and 1.2%. The sample is highly educated with 89.8% of the respondents who have at least an applied university degree (hbo). The sample has a lot of respondents who holds a bachelor, master, or PhD university degree (74.0%). A small number of respondents filled in lower education (mbo, high school) (8.0%) or other (2.1%) as their highest education. In contrast to the high education numbers, the income numbers are rather low. The biggest group of 51.4% has a gross income of less than €1,500 per month. The reason for this could be that many students participated in this study, who do not work (much), and thus not earn much yet. Only 11.0% of the sample earns more than € 3,500 per month, of which only 5.7% earns more than € 5,500. 14.3% of the sample did not disclose their earnings information. Moreover, most of the respondents live in an urban environment (84.4%). More than half of the respondents live in the Netherlands (51.6%). 16.2% of the respondents live in Italy, 7.2% in Indonesia, 4.5% in Lithuania, 3.3% in Germany and the other 17.2% live in other countries across the world. In total, 82.3% of the respondents is from Europe, 9.5% from Asia, and 5.7% from North America. The sample is divided between car owners, and people who do not own a car. Car owners make up for 32.2% of the sample. The other 67.8% do not own a car. Lastly, 89.0% of the sample has a driver’s license, which helps the respondents to put themselves into the sketched hypothetical situations.

3.5 Data Analysis

The data collected using the survey software is loaded into SPSS 25. SPSS is a statistical computer program used for data processing and analysing. After cleaning the dataset, 419 respondents finished the survey and are used as the sample of this study. There were a lot of respondents who did not complete the survey. This is probably due to the fact that the survey took a lot of time to complete.

(22)

22 Firstly, the dataset was cleaned. This was done by looking at the outliers for time, in which the responses which took less than 10 minutes were removed, and other outliers. After this, the categorical variables were transformed into dummy variables (see Appendix E). Further, the items measuring peer-to-peer trust, platform trust and intention to participate were combined into the final variables. The Cronbach’s alphas (α >.7) were checked and had no possibilities to improve (see Appendix F). Several tests were performed on the gathered data. First, the distribution of the mediator variables and the dependent variable were investigated. Secondly, the correlations were inspected of the independent variable, the dependent variable, the mediator variables, and the control variables. The correlation matrix shows the correlation between the variables. Moreover, the relation between high sociality and the willingness to participate is tested, and the hypotheses are tested. This is done by performing a linear regression and a mediation analysis using the PROCESS test. The PROCESS test is an SPSS extension created by Hayes et al. (2012). It is used to test for a mediated effect, and it can be helpful to measure the sequential mediation effect of both the platform trust mediator and the peer trust mediator. In the following chapter, the results of the analysis are shown.

(23)

23

4. Results

The results found from the data analysis are described in this chapter. In the first paragraph, the distribution of the mediator and dependent variables are described (paragraph 4.1). In the second paragraph (paragraph 4.2), the correlation matrix is described. The following paragraph continues with the independent variable sociality, which includes testing the direct relationship with the dependent variable willingness to participate (paragraph 4.3) as well as the mediation relationship of the mediator variables in the last paragraph (paragraph 4.4).

4.1 Distribution

Both the mediator variables peer trust and platform trust and the dependent variable willingness to participate are tested on skewness and kurtosis (Appendix G). None of the values have skewness or kurtosis problems. Further analysis using the Kolmogorov-Smirnov and the Shapiro-Wilk tests, show for all three values a significant p-value (<.05) (Appendix H). This means that the variables are not normally distributed. Nevertheless, the sample size is large (N=419). According to Pallant (2007), a violation of normality is not considered a large problem when the sample size is larger than 30 or 40.

4.2 Descriptive Statistics

4.2.1 Correlations

In table 2 the correlation matrix is shown, which contains the means, standard deviations and correlations between the main variables and the control variables. The table illustrates that there is a weak significant negative correlation between sociality and the willingness to participate (r=-.131, p<.01). This is in contrast with the theoretical framework. There is a moderate significant positive correlation between platform trust and willingness to participate (r=.427, p<.001) and between peer

trust and willingness to participate (r=.433, p<.001). This is in line with the hypothesis based on the

theory. There is also a strong significant positive correlation between platform trust and peer trust (r=.677, p<.001). However, there is no significant correlation between either sociality and platform

trust (r=.054, p=.274) and sociality and peer trust (r=.056, p=.254). This indicates that the degree of platform trust and the degree of peer trust is not expected to be influenced by the degree of sociality.

(24)

24

Table 2. Correlation matrix

(25)

25

4.2.2 Control Variables

The control variables and demographics are also included in the correlation matrix shown in table 2. These variables have some significant correlations with the independent variable, dependent variable, and the mediator variables. First of all, the owner control variable is negatively correlated to

willingness to participate (r=-.239, p<.001), as well as the car possession control (r=-.260, p<.001).

This means that car owners, who rent out their car on the platform, are less willing to participate than car users, who rent a car on the platform. Also, the control variable age is negatively correlated to

willingness to participate (r=-.183, p<.001). This means that the older participants were less willing to

participate than the younger participants. Moreover, participants that live in an urban environment have more trust in peers than participants that live in a rural environment (r=.139, p<.01), and were also more willing to participate (r=.114, p<.05). Next to that, low income correlates positively with

willingness to participate (r=.189, p<.001). This indicates that the participants with a low income are

more willing to participate that participants with a medium or high income. Furthermore, participants from the Netherlands have less trust in the platform (r=-.185, p<.001) and in the peers (r=-.133, p<.01), and are less willing to participate (r=-.121, p<.05). However, participants from Lithuania have more trust in peers (r=.112, p<.05).

Some of the control variables also correlate significantly with each other. The owner control variable correlates with age (r=.217, p<.001), with the driver’s licence variable (r=.264, p<.001), with low income (r=-.201, p<.001), with Italy (r=.104, p<.05) and with Indonesia (r=.194, p<.001). Age also has significant correlations with female (r=-.171, p<.001), with driver’s license (r=.211, p<.001), with professional education (r=.168, p<.01), with academic education (r=-.143, p<.01), with urban residency (r=-.103, p<.05), with low income (r=-.463, p<.001), with car possession (r=.377, p<.001), with the Netherlands (r=.183, p<.001), with Germany(r=-.096, p<.05), with Indonesia (r=.116, p<.05), and with Italy (r=-.264, p<.001). The control variable for corona also correlates with car possession (r=-.099, p<.05) and with Lithuania (r=-.109, p<.05). The control variable female correlates

significantly with gender other (r=-.125, p<.05), with other education (r=-.123, p<.05), with low income (r=.121, p<.05) and with Lithuania (r=.099, p<.05). The driver’s licence variable correlates significantly with low income (r=-.121, p<.05) and with car possession (r=.384, p<.001). The control variable for professional education correlates significantly with academic education (r=-.719, p<.001), with low income (r=-.155, p<.01), with medium income (r=.138, p=.005), with the Netherlands (r=.264, p<.001), with Indonesia (r=-.102, p<.05) and with Italy (r=-.183, p<.001). The control variable for academic education correlates significantly with other education (r=-.220, p<.001), with low income (r=.140, p<.01), with medium income (r=-.153, p<.01), with the Netherlands (r=-.192, p<.001) and with Italy (r=.107, p<.05). Low income also correlates significantly with medium income (r=-.437, p<.001), with car possession (r=-.305, p<.001), with the Netherlands (r=-.128, p<.001), with Indonesia (r=-.123, p<.05), with Italy (r=.218, p<.001) and with Lithuania (r=.118, p<.05). Medium

(26)

26 income correlates significantly with the Netherlands (r=.148, p<.01) and with Italy (r=-.163, p<.01). Lastly, car possession correlates with the Netherlands (r=-.138, p<.001) and with Indonesia (r=.164, p<.001).

4.3 Regression Analysis

4.3.1 Dependent and independent variables

A linear regression analysis was performed with the independent variable sociality and the dependent variable willingness to participate. The model achieved significance, which means that it successfully predicted the willingness to participate (F(1,417)=7.287, p<.01).The R2_{value is .017, which indicated} that sociality predicts willingness to participate by 1.7%. The model coefficients are significant, which indicate that by each increase of 1 in sociality, the willingness to participate decreases with .131 (t=-2.699, β=-.131, p<.01). See Appendix I for the analysis.

4.3.2 Control variables

A hierarchical multiple regression analysis was performed with the independent variable sociality, the dependent variable willingness to participate and the control variables age and corona. Block 1 included the two control variables and block 2 the sociality variable. Model 1, which includes only the control variables as predictors, achieved significance, which means that it successfully predicted the

willingness to participate (∆F(2, 416)=7.232, p<.01). The R2_{value is .034, which indicated that age} and corona predict willingness to participate by 3.4%. Model 2, which includes the control variables as predictors as well as the predictor sociality, achieved significance, which means that it successfully predicted the willingness to participate (∆F(1, 415)=6.658, p=.01). The R2_{value is .049, which} indicated that sociality, age and corona predict willingness to participate by 4.9%.

The model coefficients for age are significant, which indicate that by each increase of 1 in age, the willingness to participate decreases with .177 (t=-3.702, β=-.177, p<.001). The model coefficients for sociality are also significant, which indicate that by each increase of 1 in sociality, the willingness

to participate decreases with .128 (t=-2.580, β=-.128, p=.01). The model coefficients for corona are

(27)

27

4.4 Hypotheses Testing

H1 hypothesized that P2P platforms with high co-sociality will enjoy a higher willingness to

participate, than platforms who restrict co-sociality. Platform trust (H2) and peer trust (H3) are

expected to mediate the relationship between sociality and willingness to participate. Lastly, there is an expected sequential mediation between sociality, platform and peer trust, and the willingness to

participate (H4).

The mediation hypotheses, H2 H3 and H4, are tested using the PROCESS extension of SPSS created by Hayes (2012). Using model template 6 of the PROCESS extension, the indirect effect of

sociality on willingness to participate through platform trust is measured. Secondly, the indirect effect

of sociality on willingness to participate through peer trust is measured. Thirdly, the indirect effect of

sociality on willingness to participate through platform trust and peer trust in serial is measured.

Lastly, the direct effect between sociality and willingness to participate is measured. This is all done while controlling for owner, corona, age, gender, driver’s license, urban residency, education level, income level and country. In figure 2, the indirect and direct effects are shown.

Figure 2. Mediation model

H1 = Sociality → Willingness to participate

H2 = Sociality → Platform trust → Willingness to participate H3 = Sociality → Peer trust → Willingness to participate

H4 = Sociality → Platform trust → Peer trust → Willingness to participate

Note: Adapted from Hayes, A.F., Preacher, K.J. & Myers, T.A. 2011, ‘Mediation and the Estimation of Indirect Effects in Political Communication Research’, in E.P. Bucy and R.L. Holbert (eds.), Sourcebook for Political Communication Research: Methods, Measures, and Analytical Techniques, pp. 434-465, Routledge, New York. N= 419.

*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed). ***. Correlation is significant at the 0.001 level (2-tailed).

(28)

28

4.4.1 Hypothesis 1

It is hypothesized that P2P platforms with high co-sociality will enjoy a higher willingness to

participate, than platforms who restrict co-sociality. The direct effect of sociality on willingness to participate is c’=-.304, t(397)=-2.362, p<.05. This indicates that a lower degree of sociality leads to a

higher willingness to participate. Therefore, hypothesis 1 is not supported.

Moreover, there are two control variables that interact with the willingness to participate. Participants who enjoyed a lower education are associated with a higher willingness to participate of 1.121 units thank participants who enjoyed a professional, academic, or other education (B=1.121, p=<.05). The 95% confidence interval of this effect is entirely above zero, from .141 to 2.101 (t=2.249, p<.05). Furthermore, participants who are the owner of a car on the platform are associated with an -.388 units lower willingness to participate than participants who are the user of a car on the platform (B=-.388, p<.05). The 95% confidence interval of this effect is entirely below zero, from -.734 to -.042 (t=-2.203, p<.05).

It is hypothesized that there is an indirect effect of sociality on the willingness to participate through

platform trust. Participants who were presented a platform with high co-sociality did not enjoy a

significant increase in platform trust in comparison with respondents who were presented a platform who restricts co-sociality (a1=.092, p=.389). However, the increase in platform trust does lead to a higher willingness to participate (b1=.362, p<.001), separate of peer trust. The bootstrap confidence interval is not entirely above zero (indirect effect = .033, SE = .043, CI: -.044 to .126). The indirect effect is thus not significant, so hypothesis 2 is not supported.

However, there are three control variables that have an effect on platform trust. Firstly, the participants who enjoyed a professional education or an academic education had less trust in the platform than participants who enjoyed a lower or other education, with -.920 units and -.816 units, respectively. For professional education (B=-.920, p=<.05) the 95% confidence interval is entirely below zero, from -1.705 to -.134 (t=-2.301, p<.05). For academic education (B=-.816, p<.05) the 95% confidence interval is also entirely below zero, from -1.568 to -.064 (t=-2.133, p<.05). Furthermore, Dutch participants are associated with a lower platform trust of -.395 units (B=-.395, p<.01). The 95% confidence interval of this effect is entirely below zero, from -.678 to -.113 (t=-2.749, p<.01).

It is hypothesized that there is an indirect effect of sociality on the willingness to participate through

peer trust. Participants who were presented a platform with high co-sociality did not enjoy a

significant increase in peer trust in comparison with respondents who were presented a platform who restricts co-sociality (a2=.055, p=.563). However, the increase in peer trust does lead to a higher

(29)

29

willingness to participate (b2=.314, p<.001), separate of platform trust. The bootstrap confidence interval is not entirely above zero (indirect effect = .017, SE = .030, CI: -.048 to .074). The indirect effect is thus not significant, so hypothesis 3 is not supported.

There is however one control variable that interact with peer trust. Participants who live in an urban environment are associated with a higher peer trust of .340 units than participants who live in a rural environment (B=.340, p<.01). The 95% confidence interval is completely above zero, from .119 to .601 (t=2.938, p<.01).

It is hypothesized that there is a positive effect of sociality on the willingness to participate through

platform trust and peer trust in serial. Participants who were presented a platform with high sociality

did not experience a higher level of platform trust (a1=.092, p=.389). However, a higher level of

platform trust does lead to a higher level of peer trust (d21=.788, p<.001). This increase in peer trust

does lead to a higher willingness to participate (b2=.314, p<.001). This specific indirect effect is not significantly positive, because the bootstrap confidence interval is not entirely above zero (indirect effect = .023, SE = .028, CI: -.031 to .080). Therefore, hypothesis 4 is not supported. An overview of the outcomes is displayed in table 3 and table 4 below.

Table 3.

Coefficients and significance levels mediation effect

Consequent

Platform trust Peer trust Willingness to participate

Antecedent Coeff. SE p Coeff. SE p Coeff. SE p

Sociality (X) a1 .092 .107 .389 a2 .055 .095 .563 c’ -.304 .129 .019 Platform trust (M) --- --- --- d21 .788 .044 .000 b1 .362 .081 .000 Peer trust (M) --- --- --- --- --- --- b2 .314 .068 .000 Constant i1 5.754 .467 .000 i2 .544 .486 .267 i3 .615 .661 .356 Control variables* Low education 1.121 .499 .025 Owner -.388 .176 .028 Professional education -.920 .400 .022 Academic education -.816 .383 .034 Netherlands -.395 .144 .006 Urban residency .360 .123 .004 R2 = .071 R2 = .480 R2 = .340 F(399) = 1.593, p=.055 F(398) = 18.388, p=.000 F(397) = 9.736, p=.000

*Only control variables with significant results are presented in this table.

(30)

30

Table 4.

Effect size and significance levels mediation effect

Effect SE P LLCI ULCI

Direct effect c1’ -.304 .129 .019 -.557 -.051

Total effect c1 -.231 .147 .117 -520 .058

BOOT SE BOOT LLCI BOOT ULCI

Indirect effect a1, b1 .033 .043 -.044 .126

Indirect effect a2, b2 .017 .030 -.048 .074

(31)

31

5. Discussion

5.1 Summary of the Results

This study aimed to explore the relationship between sociality and the willingness to participate on a peer-to-peer sharing platform. Contrary to the expectations, this research did not find a positive relationship between the degree of sociality and the willingness to participate. However, it did find a negative significant relationship between the degree of sociality and the willingness to participate. Also, the two indirect paths are not significant. A higher degree of sociality does not lead to a higher platform trust or a higher peer trust. The effect for both is positive, but not significant. However, a higher platform trust does lead to a higher willingness to participate. A higher peer trust also leads to a higher willingness to participate. So even though an effect is found between platform and peer trust and the wiliness to participate, no direct relationship is found between sociality and platform trust, or sociality and peer trust, causing these indirect paths to not be statistically significant. Platform trust and peer trust are strongly related to each other. However, since there is no direct relationship found between sociality and platform trust, the sequential mediation is not significant.

5.2 Discussion of the Results

5.2.1 The direct relationship between sociality and willingness to participate

There is no positive direct relationship found. The results contradict the theory from Frenken and Schor (2019), who wrote that digital sharing platforms decrease the idea and the risk of ‘stranger danger’ by obtaining relevant information through reviews, ratings and thus reputations, which in turn leads to a higher willingness to participate. However, there is a significant negative relationship between sociality and willingness to participate.

The control variables for age and corona both correlate negatively with the willingness to participate. The results from the hierarchical multiple regression analysis only confirm that age has a significantly negative effect on the willingness to participate. The age of the participants ranged from 18 to 92, which influenced the willingness to participate in a way that when a participants age increased by 1, the willingness to participate decreases with .177 units. This can be explained by the theory of Hwang and Griffiths (2017), that the consumer segment for whom the sharing economy is most attractive are the millennials. So, the older the participants, the less attractive the sharing economy and thus the less willing to participate.

Furthermore, the sample for this study was divided between 32.2% car owners and 67.8% non-car owners. This is also proved to influence the willingness to participate. Participants who are the owner of a car on the platform are associated with an -.388 units lower willingness to participate than

(32)

32 participants who are the user of a car on the platform. This might be explained by the theory of Weber (2014), that there is a concern on peer-to-peer platforms that the traded assets might be damaged due to unobservable behaviour of the peer that it is being shared with. Non-car owners who use the platform do not share their own car, and thus do not have the fear that their traded assets might be damaged.

To conclude, there is no positive relationship found between the degree of sociality and the willingness to participate. However, a negative significant direct effect was found.

5.2.2 Mediation of platform trust

The results show that there is no indirect effect of sociality on willingness to participate trough platform trust. The effect of a high degree of sociality does not lead to higher platform trust. This contradicts with the theory of Slee (2013), who wrote that the presence of reviews and ratings of the platform was a form of building a reputation for that platform, which is a symbol of trustworthiness created by the statements of other users.

The second part of the mediation focusses on the relationship between platform trust and the willingness to participate. This relationship was significant and strong. This is in line with Frenken et al. (2014), who found that higher levels of trust lead to higher intentions to participate. Next to that, it is also in line with McKnight et al. (2002), who found that platform trust is a crucial prerequisite for the willingness to participate on said platform. People who trust the platform, are certain that the sharing platform positively supports the sharing activity in their good interest (Mittendorf et al., 2019). In conclusion, even though there is a significant effect of platform trust on the willingness to participate, there is no effect found of the degree of sociality on platform trust. Thus, platform trust is not a mediator between sociality and the willingness to participate.

5.2.3 Mediation of peer trust

The results show that there is no indirect effect of sociality on willingness to participate trough peer trust. The effect of a high degree of sociality does not lead to higher peer trust. This contradicts with the theory of Kim et al. (2015), who described that the inclusion of face to face interactions or direct online communication leads to a higher social presence, which is a symbol of trustworthiness. Next to that, it contradicts with the theory of Slee (2013), who wrote that the presence of ratings and reviews build reputation for the user, which in turn creates trust.

The second part of the mediation focusses on the relationship between peer trust and the willingness to participate. This relationship was significant and strong. This is in line with the theory of McAllister (1995) and Frenken et al (2014), who wrote that the plan to engage in a sharing encounter is based on in the interpersonal trust, that a sharing partner is confident in the words,

Assessing the Influence of Co-Sociality on a User’s Willingness to Participate on a Car Sharing Platform through Trust: A Vignette Study