To which extend do material incentives matter? : a study on Backers’ motivation behind crowdfunding behaviour

(1)

To Which Extend Do

Material Incentives Matter?

A Study on Backers’ Motivation behind

Crowdfunding Behaviour

Master Thesis

Student:

Jun Zhang (10742239)

MSc. in Business Administration - Entrepreneurship and Innovation Faculty of Business and Economics of UvA

Supervisor:

First Supervisor: Dr. G.T. Vinig Second Supervisor: Dr. W. van der Aa

(2)

Statement of Originality

This document is written by Student Jun Zhang, who declares

to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document

is 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)

Table of Figures

Figure 1: Amount Raised on Crowdfunding Platform (Data resource: crowdsourcing.org) ... 7

Figure 2: Google Trend: Crowdfunding ... 8

Figure 3: Number of Crowdfunding Platforms worldwide, Source: Crowdfunding Industry Report, May 2012, P12 ... 15

Figure 4: Self-determination Continuum and Backers' Motivation, Adapted from Gagné and Deci (2005) ... 20

Figure 5: Campaign Number and Success Situation from Feb 1st to Mar 16, 2015, Data Source: Krowster.co ... 21

Figure 6: Conceptual Model (Self-sourced) ... 26

Figure 7 Sample Statistic Graph ... 39

Figure 8: Histogram of IntrinsicCF and Intrinsic PJ ... 40

Figure 9: Data Distribution of Number Backed and Amount Contributed ... 41

Figure 10: Result of Hypothesis ... 52

Figure 11: Hypothesis Test Result on Reward-based Group ... 55

Figure 12: Hypothesis Test Result on LinkedIn Group ... 56

Figure 13: Hypothesis Test Result on Facebook Group ... 56

Figure 14: Hypothesis Test Result on LinkedIn Group (Reward-based) ... 58

(5)

Acknowledgement

This is the last assignment of my Master Degree (Innovation and Entrepreneurship track) at Universiteit van Amsterdam. No words can express the feeling I have right now, as the past year has witnessed more than what I could have experienced in 10 years had I chosen to still stay in China. Getting away from a city one had lived for over 35 years and studying again after graduating 14 years earlier wasn’t an easy decision, and I wouldn’t have gone this far without the support given by people on my way. I would like to take this opportunity to extend my sincere gratitude to them.

Firstly, I would like to thank my supervisor Dr. Vinig for his continuous support and inspiring advices during the completion of this thesis. Without his encouragement and instructions, this thesis wouldn’t have achieved the results I am presenting right now.

Secondly, I would like to thank Dr. Van der Aa and other lecturers for designing this educational program, and for challenging and pushing the limits of our knowledge and skills. I also would like to thank Ms. Charlotte Yeh and the team from the students’ service, without whom, we wouldn’t have been able to better focus on our study.

Next, great appreciation will be extended to all the participants and everyone who showed willingness to support my research.

Last but not least, many thanks to my dearest family and friends: my parents for unconditionally supporting all my life decisions, my partner Chris for being there for me all the time and my son Max for understanding my absence when he needs a “playmate”.

(6)

Abstract

Backers’ motivation in CF activities has got increasingly more attention over time. Previous studies had approached the subject from economic, marketing and consumer’s perspective, few were done based on modern motivation models. In the meantime, most of motivation models were tested under management or educational settings, other situations often receive very little interest, and have yet to be fully explored. This paper combined modern motivation models and CF studies, intended to uncover a connection between intrinsic motivation, incentives and backers’ behaviour. After reviewing related literatures, a conceptual model with 10 hypothesis was proposed. The feedback from 212 people was collected, one hypothesis was accepted, three were rejected and others had different outcomes when tested in different group settings. The results shows that intrinsic motivation, incentives and family & friend relationships plays a significant role in predicting the likelihood of backers’ activity. The difference in effects are discussed in extent in later chapters.

(7)

1. Introduction

Crowdfunding (herein after referred to as “CF”), as a new method of fundraising for individuals

and organizations, has gained rapid development within the past 5 years. Based on data collected from 1,250 active CF platforms across the world, a study by crowdsourcing.org showed that the CF market experienced accelerated growth in 2014, reaching $16.2 billion, a 167% increase compared to 2013, with an estimation of more than double that figure ($34.4 billion) by end of 2015 (Crowdsourcing.org, 2015) (figure 1). It is reported that 90% of the world’s online population from more than 160 countries have been participating on CF platforms (The Crowdfunding Center, 2015). CF has indeed become the bridge that successfully connects early-stage projects with potential investors. Due to the collective wisdom and resources from contributors, as well as the collaborative creation of project initiators and supporters, in US and UK alone, an average of 100 talented projects are brought to life every day (The Crowdfunding Center, 2015).

Figure 1: Amount Raised on Crowdfunding Platform (Data resource: crowdsourcing.org)

5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 2009 2010 2011 2012 2013 2014 2015

Crowdfunding Growth

(USD: Million)

(8)

Alongside the increasing global attention this subject has received in the previous 5 years (Figure 2) and all the momentums gained by the CF world, academic studies are accumulating as well. Common areas that have been covered to date include basic fundraising models, benefits, conditions and motivations of CF as an alternative to traditional fund-raising methods, as well as explorations of characteristics and traits of its key players. Being a newly developed research theme, the understanding of the dynamics of CF activities are still limited (Hemer, 2011; Mollick, 2014), exposure and systematic studies of this area are inadequate.

Figure 2: Google Trend: Crowdfunding

Midst present studies, the “hidden secret” of successful CF campaigns tends to be a popular subject which draws attention from earlier researchers and opinion leaders. Project categories (Hemer, 2011), project quality (Burtch, Ghose, & Wattal, 2013), action of other backers (Kuppuswamy & Bayus, 2013), goal size (Mollick, 2014), exposure rate (Ward & Ramachandran, 2010; Mollick, 2014), creator network size (Giudici, Guerini, & Lamastra, 2013a; Moisseyev, 2013; Mollick, 2014) and geographical distances (Agrawal et al., 2011; Giudici, Guerini, & Lamastra, 2013b) are believe to be relative factors, yet no consistent

(9)

conclusions have been reached. For example, even though it is true that different project types (Hemer, 2011) and categories have different success rates (Mollick, 2014; An et al., 2014), there is little chance a creator will change their original campaign to fit a specific type or category, because this could lead to a huge sacrifice of the original ambition and purpose of the campaign. Using data scraping from the same CF platform, Frydrych and his colleagues (2014) argue that project characteristics as a predictor of campaign success (Mollick, 2014) is largely unsupported. Among the five examined factors in their study, only lower funding targets correlated with the expected success rate; it was found that short duration, reward level, and visual pitch had no significant correlation with success, while team composition show differences only when comparing individual creators with team creators (Frydrych et al., 2014). Another explored route is physical distance between creators and backers. While some argued that geographic location plays an influence on campaign success (Agrawal et al., 2011), a study showed that frequent backers are not necessarily bounded by distance when seeking a potential project (An, Quercia, & Crowcroft, 2014). In view of the fact that these studies can’t fully explain the success of a project, attentions are directed from creator and project to the ones that are actually conducting the funding behaviour.

Backers and their participation are frequently mentioned in most CF related articles (Glaeser et al., 2001; Hemer, 2011; Brüggen et al. 2011; Pugacheva, 2012; Belleflamme et al., 2013 & 2014; Gerber & Hui, 2013; Mollick, 2014; Zheng et al. 2014, etc.). Similar to other early research on CF subjects, studies focusing on backers’ motivation have reached no cohesive conclusions either. Some scholars believe that exchange for a certain reward is one of the reasons that a backer is willing to financially contribute to a project (Gerber, Hui, & Kuo, 2012; Gerber & Hui, 2013; Tomczaka & Brema, 2013; Ryu & Kim, 2014; Bouaiss & Maque,

(10)

2012; Mollick, 2014; Belleflamme, Lambert, & Schwienbacher, 2014; Jian & Shin, 2014) play a more important role. Interestingly, current academic articles on CF were mostly published in either financial, marketing or computer science related fields (Bouaiss & Maque, 2015), studies done by psychology or human behaviour specialists are scarce.

Seeing that “well-motivated supporters might become active investors in innovative

start-ups in the future” (Hemer, 2011), backers’ motivation will surely become the next

prevalent topic. Indeed, does material incentives play a role in backers’ funding behaviour? If so, to which extent does it affect a backers’ funding behaviours? With theoretical supports from contemporary motivation theories and based on an online survey among backers, this article tries to answer the question by analysing the relationship between backers’ motivation and supporting behaviour.

1.1 Academic Relevance

The contradictory factor of a bursting CF market and a comparatively low funded-rate leads to a wide range of exploration on influencers for campaign success. When findings on creators and campaign designs can’t fully solve this riddle, scholars and key opinion leaders start to shift their focus towards backers. Academics and researchers that have previously revealed some answers on the subject of backers’ motivation, there are few studies that specialize on understanding this issue in detail, and the majority of the completed studies take an economic or marketing perspective. Likewise, leading studies on human motivation usually support theories with samples from education or work environments. Application of motivation theories in other occasions are incomplete. To my knowledge, there has been no attempt to apply contemporary motivation theories on a CF environment thus far.

(11)

1.2 Managerial Relevance

CF, a new practice for fundraising, has been playing a progressively more important role in the world’s economy. However, the dynamics of the CF phenomena is yet to be uncovered (Hemer, 2011; Mollick, 2014). Scholars and practitioners have been seeking causes of success campaigns with no solid conclusions reached. Some have tried to explore the drivers of individuals’ motivation to financially contribute to CF projects, however, a model that reveals backers’ behaviour influenced by both motivation and incentives has never been tested.

1.3 Thesis Outline

In order to answer the research question, this paper is divided into six sections. Section 1 is a brief introduction of the study background of the research questions, with a short forecast of academic and managerial relevance. Section 2 starts with a brief introduction of Crowdfunding and its key players, based on a detailed literature review on backers’ motivation and contemporary motivation theories, hypotheses are constructed and a conceptual model for the research is presented. Section 3 provides detailed information on the methodology used in the research, including design, data collection and data analysis. Section 4 presents the results of the analysis and other findings. The last section discusses the indication of the results, contributions and limitations of this study and suggestions for future research.

2. Literature Review

On the one hand, heated popularity of CF results in the rising of related research. An unsystematic search with the keyword “crowdfunding” on Google Scholars results in about 6,800 articles (without counting patents or citations), over 60% of which were published after

(12)

2013, whilst the first 2 months of 2015 alone welcomes more than 400 articles (Google Scholar, 2015). Due to the novelty characteristic, studies are broadly extended into different areas, from finance, entrepreneurship, to computer science and social media. On the other hand, research on human motivation has a long history since the middle of the last century. Modern psychological studies mainly focus on a limited group of contemporary theories. With the purpose of this paper being to discover the interrelationship among backers’ motivation, behaviour and the influence of incentives, only related topics from both the CF field and motivation studies are covered in the following literature reviews.

2.1 Crowdfunding and Key players

Crowdfunding as an academic study subject didn’t start until recently. So far, there are

no unified definitions for this term. Some approach the definition from creators’ perspectives, underlining the efforts made by creators to draw contributions from large number of individuals (Tomczaka & Brema, 2013; Mollick, 2014); some tried to explain the characteristics of the projects involved in the process (Hemer, 2011); many interpreted the idea from backers’ perspectives, pointing out the collective efforts and contributions provided by the crowd (Harms, 2007; Pugacheva, 2012; Lambert & Schwienbacher, 2010; Hemer, 2011; Belleflamme, Lambert, & Schwienbacher, 2014; Bretschneider, Knaub, & Wieck, 2014; An, Quercia, & Crowcroft, 2014) others combined the ideas together by emphasizing the link between those who launch projects and those who contribute them (Ryu & Kim, 2014). This article believes a neutral description better suits its research purpose, therefore simplified the definition given by Harms (2007) and Hemer (2011), into a collective funding behaviour by the crowd towards a project that initiated an open call on the internet.

(13)

There are four main players in the course of crowdfunding: people who initiate the process by posting a project on the internet (either at the website of third parties or at their own websites) to attract funding; people who provide financial contributions to the posted project; the platform where this action takes place and the projects that are posted.

2.1.1 Creators

Those who make efforts to attract contributions on their project are referred to as

“creators” in the following paragraphs. Hemer (2011) used a matrix to categorize creators by

matching their background (independent and individual, company embedded projects and start-up), with their objectives (for-profit, non-profit or intermediate). Other than seeking financial support, creators also use CF platforms to demonstrate demand for new products (Gerber, Hui, & Kuo, 2012; Mollick, 2014), market new products or services (Mollick, 2014; Zheng, Li, Wu, & Xu, 2014; An, Quercia, & Crowcroft, 2014) form connections (Gerber, Hui, & Kuo, 2012; Gerber & Hui, 2013) and acquire other resources (Gerber & Hui, 2013; Mollick, 2014). 2.1.2 Backers

The counterpart of creators are the ones who contribute financial support to the projects, hereinafter referred to as “backers”. Backers are described by scholars as groups of people who share similar interest (Schwienbacher & Larralde, 2012), who are willing to help others (Glaeser & Shleifer, 2001; Harms, 2007; Gerber & Hui, 2013; An, Quercia, & Crowcroft, 2014), invest in products they believe or enjoy participation in the open innovation process (Harms, 2007), the future consumer of a products or service (Pugacheva, 2012; Belleflamme

et al., 2013 & 2014), peer to peer loan providers in case of lending-based CF practice or the

(14)

CF activities. Further review of studies on backers’ motivation is presented in the coming chapters.

2.1.3 Platform

Due to the speedy changes happen in the CF world and lack of collective information channels, it is difficult to collect information on the number of existing platforms. Nonetheless, based on the latest statistics available online, by April 2012, there had been 452 active CF platforms worldwide, with a trend of around 40-60% increase every year (Crowdsourcing.org, 2013; figure 3). Giving the differences among creators and backers, CF platforms, as a mediator between the two groups, exists in different forms. Classified by projects, there are general platforms that support a wide range of project types (such as Kickstarter.com, Indiegogo.com and Pozible.com) and platforms dedicated to a specific area such as art, sports, science (ArtistShare.com and Experiment.com, etc.), as well as a specific course such as helping people recovering from disasters (Gofundme.com, etc.). Categorized by relationship of creators and backers, there are donation-based (Gofundme.com and Youcaring.com, etc.), reward-based (Kickstarter.com, Indiegogo.com, etc.), equity-based (Crowdfunder.com, etc.) and lending-based platforms (LendingClub.com, Fundingcircle.com, etc.). Considering profitability of projects, there are Not-for-profit-based platforms and other platforms; whilst by fundraising type, there are “All or nothing” (Kickstarter.com and Fundable.com, etc.) and “Keep it All” (Indiegogo.com, RocketHub.com, etc.). Sadly, like any developing market, the rapid increase of new CF platforms worldwide is also accompanied by the disappearing of others, Spot.us and SoMolend.com, Fundweaver.com are some typical examples.

(15)

Figure 3: Number of Crowdfunding Platforms worldwide, Source: Crowdfunding Industry Report, May 2012, P12

2.1.4 Campaign

Like platforms, CF Campaigns can be categorized by project type, creator intention, creator-backer relationships, fundraising methods and outcome (material or immaterial nature). Given that the majority of projects on CF platforms intend to attract funding, creators must ensure the way they present their project is supporting their project aim. Studies report that campaigns tend to attract funding from the following mechanisms: donation-based (Hemer, 2011; Mollick, 2014), reward-based (Hemer, 2011; Mollick, 2014; Zheng et al., 2014), and exchange-based (Hemer, 2011; Gerber, et al., 2012; Belleflamme et al., 2013 & 2014; Mollick, 2014; Zheng et al., 2014). Kappel (2009), Tomczaka and Brema (2013) considered a different angle, by sorting CF campaigns into “ex post facto” ones, which refer to funding in exchange for a completed product, and “ex ante” ones, which refer to funding for a common purpose or desired result, namely to develop a working prototype, as there may not be one available at the

(16)

Thus far, academics from different fields have contributed to this theme, with the purpose of better understanding CF phenomena, by trying different exploratory approaches. Whereas early papers tried to explain what and why from a general viewpoint, there seems to be a shifting towards more focused studies. Biding the same purpose, this paper narrows its focus on backers’ motivation, funding behaviour and incentives.

2.2 Studies on Backers’ Motivation and Funding Behaviour

Scholars hold different opinions on what drives backers to participate in CF activities and why they fund some projects, and not others.

Harms (2007) is among the first academics to provide a statistical study on backer motivations. Based on his conceptual model and an experimental study, he analysed 5 perceived values and their relation with backers’ intention to support a project. The result shows that perceived positive economic value and positive functional utility are strong predictors of backers’ intention to support a project. Whilst guaranteed tangible output, self-expressiveness and enjoyment in the supporting behaviour are also positively correlated with backers’ likelihood to support a project. High financial return, involvement in the community, involvement in a project and feeling of being supportive were not found to be related to their intention. Similar conclusions were drawn by Avakyan (2013) in her master thesis. Her paper also reported positive influence of perceived functional value and social value on backers’ intentions, however, emotional value was discovered to be a stronger predictor.

Hemer (2011) suggested altruism as motivation for backing not-for-profit or intermediate category campaigns, whilst monetary rewards are expected when backing campaigns with a commercial intent. He further pointed out that personal identification, social

(17)

importance of the mission, being part of a community, enjoyment of sharing others’ success, and contribution to a new innovation are more primary motivations, as well as the need for expanding one’s network. Creators contributing other projects in exchange for funding for one’s own project was discovered as another reason for backing others’ campaigns. Tomczaka and Brema (2013) also believed that the majority of backers involved in a financial model of CF are expecting tangible or monetary rewards or the exchange of funding of projects from each other.

Later studies indicated backers may support a commercial campaign out of altruistic or donative motivation as well when they are enthusiastic about the subject (Gerber et al., 2012), they share similar interest or feel a connection to the meaning of the cause (Ryu & Kim, 2014; Zheng, Li, Wu, & Xu, 2014), or have similar expertise, mind-set or feeling of connectedness with a like-minded community (Gerber, Hui, & Kuo, 2012; Gerber & Hui, 2013). Although a higher amount of support can be expected if the campaign provides the future product along with a reward (Gerber, Hui, & Kuo, 2012). Belleflamme et al. (2014) argued that even when a physical return is provided in reward-based and exchange-based campaigns, there is no determined correlation between material incentives and backers’ willingness to pay for a future product or service to a degree greater than normal consumers. The high failure rate in general and low on-time delivery rate even in case of successfully funded campaign (Mollick, 2014) insinuate that rewards or financial return might not be the only motivation.

Whether it is as patrons, customers or investors, as backers make their decisions based on expectations of success of the campaign, scholars reason that there must be something in the process itself that is motivating (Agrawal, Catalini, & Goldfarb, 2011; Mollick, 2014). Indeed, despite the fact that belief, altruism and contributing towards a community are stated

(18)

the only two positive predictors of actual contributions. Glaeser and Shleifer (2001) even argued that providing monetary incentive might undermine the desire to support a campaign as it distracts backers from focusing on the outcome of the campaign itself.

Among all the factors proposed by academics that might affect backers’ motivation and behaviour, three stand out with comparatively consistent conclusions across the papers: intrinsic motivation (enjoyment of participating in CF activities or in backing a campaign) , incentives (either tangible or intangible rewards) and the relationship between backers and the creators (known as Family or Friend relation, hereinafter “FF”). The hypotheses of this paper will focus on these three factors.

2.3 Intrinsic Motivation and Incentives

Both intrinsic motivation and incentives are reported to influence backers’ behaviour. It is interesting to see how and to what extend each affects the behaviour, as well as whether or not they interact with one another in a CF environment. To get a better understanding of the relationship between intrinsic motivation and incentives, contemporary motivation theories are consulted.

2.3.1 Motivation Theory

Motivation, a core concept in psychology, has long been considered as the fundamental component when explaining human behaviour. There are two major schools of studies on this field. The first school of thought represented by Equity Theory (Adams, 1963) and Expectancy Theory (Vroom, 1964), focuses on psychological mechanisms that explain what motivates people internally. Pitfalls lie in their limitation on explaining the role of contextual forces (Katzell & Thompson, 1990). The second school of thought puts more emphasis on effect of

(19)

contextual influences without much regard to psychological processes behind certain actions. A typical example of such is Goal-setting Theory (Locke & Latham, 2002). Similarly, lack of ability to explain an individual’s psychological process becomes the theory’s shortcoming. Job Characteristic Model (Hackman & Oldham, 1975; 1980), as another example, did cover the psychological experiences of individuals and implies an importance of intrinsic motivation on the final outcome, unfortunately its job-related characteristics limit its applicable scope.

Based on the two types of approaches, combined with early need-driven theories (Maslow, 1943; McClelland, Clark, Roby, & Atkinson, 1949; Herzberg, Mausner, & Snyderman, 1993), Self-determination Theory (therein after “SDT”, Ryan & Deci, 2000b) plays an important role among contemporary studies at interpreting the interactive relationship between intrinsic motivation factors and extrinsic motivation factors.

2.3.2 Role of Intrinsic Motivation

According to SDT (Ryan & Deci, 2000b; Ryan & Deci, 2000a), a behavior is classified as extrinsically motivated when an individual has an intention to acquire a particular outcome, and intrinsically motivated when performed for the inherent joy or playfulness of the action itself. In order to better illustrate the different levels from “not motivately at all” to “highly motivated”, a continuum motivation chart for SDT was presented (Gagné & Deci, 2005). Based on this chart, only the pure enjoyment of backing itself is considered intrinsic motivation, all other behaviors are in the extrinsic range but may fall in different levels (figure 4).

(20)

Figure 4: Self-determination Continuum and Backers' Motivation, Adapted from Gagné and Deci (2005)

As was stated earlier, the majority of CF campaigns miss their goal by a large margin, and there is a high percentage of late or none delivery among funded projects (Mollick, 2014). A one-month (16 Feb to 16 Mar 2015) study tracking campaign success, with the help of a platform called Krowster.co, demonstrated an even lower success rate: 2.7% for Kickstarter and 1.6% for Indiegogo (figure 5). These figures surely will hold back those who, according to scholars, make decisions to contribute a campaign only with the goal of a successful outcome (Agrawal, Catalini, & Goldfarb, 2011; Mollick, 2014). Conversely, the number of backers for both platforms were still on the increase (49,343 for Indiegogo and 148,038 for Kickstarter) during the same measured period. Another interesting discovery is that one of the interviews, conducted by Gerber et al. (2012), illustrated a situation when a backer was willing to give up rewards for the course itself: “Don’t spend that money on making t-shirts, spend it on building

(21)

Figure 5: Campaign Number and Success Situation from Feb 1st to Mar 16, 2015, Data Source: Krowster.co

If supporting a campaign bears a high risk of financial loss, low chances of equal payoff, even sacrifices of one’s “pay-backs”, pure enjoyment of the action itself seems to be the only reasonable explanation, as intrinsic motivation can provide the person involved with an obviously different quality of experience (Deci, Koestner, & Ryan, 1999). Besides, most of the CF platforms being designed in a way that only a brief introduction is shown for each live campaign, backers have to click the link before they can check details on available perks, it makes more sense by assuming that backers choose a campaign out of interest or enjoyment rather than aiming for collecting rewards.

Meta-analysis of over forty years of research shows a medium to strong predicting effect of intrinsic motivation on performance, as intrinsically motivated individuals are more likely to fully endorse and participate in a task (Cerasoli & Nicklin, 2014). In accordance with intrinsic motivation theory, a paper that studied online behaviours (Mathwicka, Malhotrab, & Rigdo, 2001; Göritz, 2004; Hsu & Lu, 2007) demonstrated that intrinsically motivated respondents show higher participation efforts and performance results. In the case of CF, intrinsically motivated backers, who enjoy participation in CF activities, are expected to show

208 143 7,457 9,017 2.7% 1.6% 0.0% 1.0% 2.0% 3.0% 2,000 4,000 6,000 8,000 10,000 Kickstarter Indiegogo

Campaign Number and Funded Situation Feb 1st to Mar 16th, 2015

(22)

Hypothesis 1a: Intrinsic motivation has a positive effect on backers’ frequency of visits.

Hypothesis 1b: Intrinsic motivation has a positive effect on the number of projects backers support.

2.3.3 Incentives and their role on performance

While intrinsic motivation means a person behaves in a certain way for the inherent enjoyment of behaviour itself, extrinsic motivation refers to a behaviour conducted in order to reach a separable outcome (Deci, Koestner, & Ryan, 1999).

Incentives include all forms of rewards, benefits and recognition. It is a typical separable outcome used to motivate individual’s behaviour (Cerasoli & Nicklin, 2014). The purpose of providing incentives is to encourage desirable behaviours, yet the value of the incentive is perceived differently by different people, and can lead to an altered effect on one’s intrinsic motivation. Scholars believe that all tangible incentives designed to “control” individual’s behaviour towards a desired outcome may decrease intrinsic motivation by diminishing one’s perceived free choice, while verbal reinforcement and positive feedback enhance intrinsic motivation by communicating perceived feeling of competence (Deci, 1971; Deci, Koestner, & Ryan, 1999). The latest study of neuroscience also suggested a similar conclusion. By tracking neural correlates using functional MRI, Murayama et al. (2010)’s experiment demonstrated less voluntary engagement in the target task in the performance-based reward group than in the controlled group. Hence, intrinsic motivation should be a better predictor for behaviour when extrinsic motivation is weak or absent (Cerasoli & Nicklin, 2014). In case of CF backers, intrinsic motivation should be a better predictor for their intention to support a campaign if incentives are not present at all.

(23)

Hypothesis 2: Intrinsic motivation has a positive effect on backers’ intention to support a campaign when no incentive is provided.

Furthermore, Cerasoli & Nicklin’s (2014) pointed out that the relationship between intrinsic motivation and performance is strengthened when monetary incentive is indirectly tied to performance, whereas weakened when it is directly tied to performance. In relation to CF campaigns, the perks provided by projects can be categorized into: (a) non-monetary rewards; usually a thank-you letter or Facebook shout-out, which is included in almost all level of perks, therefore, not directly performance related; (b) rewards which offer a higher monetary value or perk if a backer contributes more money, therefore considered as a performance-related incentive. Although Cerasoli & Nicklin’s (2014)’s discovery was proven in academic and working environments, it is possible that similar results can be predicted in the setting at CF platforms.

Hypothesis 3a: When incentivized, the relationship between backers’ intrinsic motivation and backing intention is positively moderated by the presence of indirectly performance-salient incentives (such as a thank you letter).

Hypothesis 3b: When incentivized, the relationship between backers’ intrinsic motivation and backing intention is negatively moderated by the presence of directly performance-salient incentives (levelled rewards on different contributions)

On top of that, although incentives are not expected to be the main contributing motivation in backers’ decisions to support a campaign, interviews conducted by Gerber et al. (2012) indicated when a choice is provided, backers tended to contribute higher amounts of money in exchange for more desirable rewards:

(24)

“I gave them 10 dollars…10 dollars is a high definition download of the film when it comes out. So, I was like, I’m not going to give them 5 dollars, I’m going to give them 10 dollars because 5 more dollars will give me a high definition download of this film…”

“I want to see [the film] right when it’s out. So, instead of giving $10, I gave $25.”

Contemporary motivation theory confirmed that external incentives is a better predictor of the quantity of performance (Cerasoli & Nicklin, 2014). In the meantime, Ryu & Kim (2014) reported a statistically significant influence of reward-based motivation on funding amount in their paper. This paper agrees with Ryu & Kim (2014), a positive relationship between incentives and backing amount is expected. Due to the data they used came from reward-based platforms, where levelled rewards on different contribution played dominant role, I assume the incentives they referred to were performance-salient ones.

Hypothesis 4: Performance-salient incentives (levelled rewards on different contributions) have positive effects on backers’ contribution amount.

2.4 Friends and Families relation (FF)

Friends and families (FF) are a special force involved in CF, whose contributions

usually help to build up initial funding for a large number of projects (Agrawal, Catalini, & Goldfarb, 2011). Due to their personal link with the creator, their behaviour is assumed to be different than normal backers.

Firstly, FFs tend to feel stronger obligation to support the creator than normal backers. Even though they might not be interested in CF activities, they will still register at the platform where the campaign was launched. It is worth pointing out that Kickstarter’s websites shows

(25)

that 70% of its registered accounts participated in only one campaign (Kickstarter, Inc., 2015), Jian & Shin (2014) also reported a 66% of backers only contributed once and never came back, while An et al. (2014) found that there was a significant link between occasional founders and local projects. As inviting personal social networks to support their campaign is a common practice among creators, there is a possibility that a certain percentage of these one-time backers didn’t sign up as backers out of own interest, but out of obligation as FF. Since feeling of obligation thwarted one’s perceived autonomy, there should be a weaker effect of intrinsic motivation or incentive motivation on FF’s backing behaviours.

Hypothesis 5a: FF has a negative moderate effect on the relationship between backers’ intrinsic motivation and general backing activities (frequency of visits a.1 & number of project backed a.2)

Hypothesis 5b: FF has a negative moderate effect on the relationship between backers’ intrinsic motivation and intention to support a campaign.

Hypothesis 5c: FF has a negative moderate effect on the relationship between performance-salient incentive and backer’s contribution amount.

In the next section, operational construction will be developed and an analysis strategy designed to test the conceptual model (figure 6) will be presented.

(26)

Figure 6: Conceptual Model (Self-sourced)

3. Methodology

A web-based self-report survey on Qualtrics.com is used to collect information on both CF backers’ motivations and behaviours.

There are four main reasons of adopting quantitative analysis by means of survey strategy. Firstly, an explanatory study better serves the purpose of this study. By allowing the collection of large data for analysis and comparison, survey is a recommendable method for producing models of relationship and exploring explanations of reasons for such relations (Saunders & Lewis, 2012). Secondly, statistical survey strategies are particularly cost-effective for asking same questions from a large number of people (Saunders & Lewis, 2012). Results of structured collection of data can better help the testing and verifying of the hypotheses. Thirdly, main variables tested in the model, especially “intrinsic motivation” and “intentions”, are more about revealing individuals’ perceptions with minimized interference, an anonymous

(27)

self-reporting mechanism helps to achieve this target. And lastly, since in this study, CF is defined as an online activity, it is assumed that all participants have access to internet, therefore it is reasonable to assume that collecting data via internet should be relatively efficient and productive.

This methodology session introduces how the research was designed, the data was collected and what analytical strategies were adopted to prepare the data before analysis.

3.1 Research Design

3.1.1 Measurement Development

An overview of used measurements and scales are presented in table 1. Instruments containing multiple scales are developed based on existing literatures to ensure an original

Cronbach’s alpha >0.7 (Lindquist, 1981), with small changes to fit the CF situation. Instruments that involve perceived feelings or intentions are also introduced from existing research so as to maintain proper wording. For descriptive instruments, questions will be adopted from existing literature if available, otherwise developed based on questionnaire format provided by Qualtrics.com. Only one instrument is purely self-developed by this research. Detailed information is listed below:

Intrinsic motivation. Existing CF literatures use “personal identification”, “fun”,

“curiosity”, “enjoyment”, “satisfaction”, “altruism” (Hemer, 2011; Mollick, 2014; Bretschneider et al., 2014) to measure backers’ motivation, however, these factors have relatively vague boundaries as some could easily fit in the “internalization” category of extrinsic motivation (figure 4). Fortunately, studies on motivation in different environmental

(28)

constitutions provided sufficient possible solutions that could be adopted with minor adjustment.

In order not to fall into a similar likelihood of misinterpretation, after consulting several possible options, Tow-Happiness theory (Waterman, 1993; 2005; Waterman, Schwartz, & Conti, 2008) seems to be most reliable. According to the result of their study, Waterman and his colleague (2008) proved that although hedonic enjoyment and eudaimonia (personal expressiveness) had very high correlation, a noticeable difference existed. By cross testing the correlation of these two forms of happiness with variables related to intrinsic motivation, their study further indicates that “intrinsic motivation” as a term can be employed only when activities are perceived to have both forms of happiness. Therefore, this paper shall use the combined scales for both defined forms of happiness (Waterman, 1993) to measure backers’ intrinsic motivation for both participating in CF activities (herein after “IntrinsicCF”) and backing a specific project (herein after “IntrinsicPJ”). In addition, the short version (2 scales for each variable) is adopted in order not to make the questionnaire too long.

Backers’ participation in CF activities (Visit Frequency & Projects Backed). A

7-scale descriptive question is adopted directly from the question pool of Qualtrics to test the

Visit Frequency of the backer at the CF platform the respondent last participated in, ranging

from “1” as “several times a day” to “7” as “less than once a month”. The number of projects backed was asked directly in the questionnaire, requesting answers with an actual number.

Intention to support a campaign. Scales used by Harms (2007) are adopted. An

imaginary project, accompanied by five different setting conditions, was introduced to test how different incentives affect the level of backers’ intentions to support a campaign. The five conditions are: when there were no perks/rewards provided (“Condition 1”); when there were

(29)

only one kind of perk/reward provided to all contributors (“Condition 2”); when there were one kind of perk/reward provided to all contributors and you didn’t like it (“Condition 3”); when there were only one kind of perk/reward provided to all contributors which you liked, but asked for a much higher amount than your latest contribution (“Condition 4”) and when there were similar perks/rewards provided as the one you backed most recently (“Condition 5”). Condition 1 is used to test the hypothesis and condition 5 is used as a validation value.

Directly and indirectly performance-salient incentives. Perk/reward that is available

to all backers regardless of contributing amount is defined as indirectly performance-salient ones, whilst those only available for backers who contribute above a certain amount of money are considered directly performance-salient ones.

Amount contributed. Amount contributed was asked directly in the questionnaire,

requesting answers with an actual number and information of currency.

Family and Friend relations (herein after “FF relation”). As there is no suitable

existing question, a 6-scale descriptive question is designed to test the closeness of the backer to the creator, ranging from “1” as “I don’t know him/her at all” to “6” as “yes, we are direct family”.

Table 1: Scale used in Questionnaire

Variables Reference Used Items

Intrinsic Motivation participate in CF activities (“IntrinsicCF”) Waterman, 1993 (Adjusted)

In general, participating in Crowd-funding activities…

gives me my strongest sense of enjoyment (hedonic enjoyment) gives me my greatest pleasure (hedonic enjoyment)

(30)

(7-point scale ranged from “totally disagree” to “totally agree”) Intrinsic Motivation backing a campaign (“IntrinsicPJ”) Waterman, 1993 (Adjusted)

Supporting this project…

gives me my strongest sense of enjoyment (hedonic enjoyment) gives me my greatest pleasure (hedonic enjoyment)

gives me my greatest feeling of really being alive (eudaimonia)

gives me my strongest feelings that this is who I really am (eudaimonia) (7-point scale ranged from “totally disagree” to “totally agree”)

Intention to support a campaign

Harms, 2007 (Adjusted)

State the likelihood of you still contributing to this project (7-point scale ranged from “very unlikely” to “very likely”) Visit

Frequency Qualtrics.com

How often do you visit the Crowdfunding Platform? (from “Several Times a day” to “Less than once a month”) Projects

Backed

Self-developed

How many projects have you backed so far? (a specific number is expected)

Amount contributed

Self-developed

How much did you contributed to this project? (a specific number is expected, with the currency used) Family and

Friend relations (“FF relation”)

Self-developed

Do you know personally the project creator or a member of the creator's team?

(from “no, I don’t know him/her/them at all” to “yes, we are direct family members)

3.1.2 Questionnaire Design

The questionnaire includes a cover letter with author’s gratitude, a brief introduction, guarantee of anonymous, followed by a list of 18 structured questions and an open question asking for comments or feedbacks.

The questionnaire was pre-tested with 4 native speakers, 3 CF professionals and 6 international business management students. Adjustment of ordering and wording were implemented so that participants can better understand the questions. One of the feedback mentioned that the wording of “2 happiness” sounded a bit unnatural, even so, because the constructs used in the 8 questions were directly introduced from the original peer reviewed

(31)

study, changes in wording might affect the reliability and validity of the scales, I decided to keep them as is. Except for the questions related to independent variables, potential moderators and dependant variables, the questionnaire also covers control variables on biographical information of the participants, such as gender, age, education level, occupation, income, nationality. Basic information on the project type, the platform type and incentive type were also included. Furthermore, the question regarding number of projects backed will be the key criteria as a qualification filter.

No monetary reward was offered to participants, instead, acknowledgement and a summary of final result was made possible upon request. A complete list of questions included in the questionnaire can be found in Appendix 1.

3.1.3 Sample selection

Backers who are “currently active at CF platform and has contributed to at least one campaign” are the target sample of this study. For that reason, by means of convenience sampling (Saunders & Lewis, 2012), this study used the following channels as attempts to reach the targeting audience: English CF platforms, LinkedIn CF related groups and Facebook CF related groups. Moreover, with the intention of assessing the sample source more accurately, a dedicated link was generated for specific channel. Totally 1,225 emails and individual messages were sent asking for cooperation or participation.

1) CF Platform: CF platforms are the first channel considered for sample collection, as

they are the medium where CF activities take place. Considering the survey is designed in English, in order to minimize the misunderstanding of the questions caused by language boundaries, only platforms with English as a language option were selected.

(32)

(crowdsourcing.org, 2015), within which, 78 had active projects online. Emails or online requests were sent to each of the 78 platforms, asking for cooperation and support. Within the 20 answers received, 9 of them showed willingness to help (Appendix 2). A dedicated link was sent to the contact person of each platform. Ulule.com also provided the opportunity to post the same link generated for the platform at their English and Dutch forums. Sadly, none produced a single response. 2) CF individuals: Most of the 78 platforms contacted don’t provide options of contacting

backers directly. Some allow peer to peer messages only under the condition that two people are backing the same project (such as IndieGogo.com). Others allow built-in messages after one registered as a user, but an anti-scam system prevents users from reposting similar messages to others (such as Pozible.com). The remaining sites that have almost no restrictions didn’t have as many active campaigns available. Still, 239 private messages were sent to backers of most recent projects at available platforms (27 at Pozible.com, 60 at Dancefunder.com and 152 at Incited.org), 13 responses were collected.

3) LinkedIn Groups: LinkedIn was chosen as the third channel of sample collection

because it allows individual messages to be sent between members of the same group. Also, since most of people use LinkedIn for professional connections, there are noticeably less scams, therefore more reliable participants could be expected. 367 open groups were found using “crowdfunding” as a keyword in the built-in search function of a ‘general’ or free account, many of which were created for creators to promote their CF campaigns (such as “Crowdfunding Your Future”), some were not as active (no posts within the past month). Also, it is noted that people tend to join more than one group if they are interested in a certain topic, and active members of one group are more likely to maintain a similar level of activity in the other. Based on these facts,

(33)

“Crowdfunding BASICS: theory, software & tools” and “Crowd Experts – Crowdfunding Professionals” were chosen because more backers were included. 733 private messages were sent (Appendix 4), generating 163 responses.

4) Facebook Groups: The forth channel selected was Facebook. As one of the top social

media channels that plays a role in the CF community, there were 100 groups containing the keyword “crowdfunding”. Similar to LinkedIn groups, many were created with purpose of seeking support. Some were even created for supporting a specific project. After an initial selection, 23 groups were contacted, 12 of which agreed to support the survey by allowing the posting of the survey link in the group discussion section. What’s more, 1 participant offered to share the link on her personal social media account and another participant posted the link on his personal blog. The total number of responses from Facebook was 144.

3.2 Data Collection

The data collection period lasted for 15 days, from 16th_{April to 30}th_{April, 2015, all}

data was collected through Qualtrics. Response rate from links sent to platforms are the lowest (0%), whilst from LinkedIn individuals are highest (22%), a summary listed in table 2.

Table 2 Sample Channels and Data Collecting Situation

Channels Distributed Collected Respond (%)

CF Platforms 78 platforms 0 0%

Individuals from CF platforms 239 individuals 13 5.4% LinkedIn Individuals in CF Groups 733 individuals 163 22.2% Facebook CF Groups + Individuals 12 groups + 152 individuals 144 2.5%

(34)

3.3 Analytical Strategies

Among 320 responses collected, 226 participants finished the survey, giving a completion rate of 70%. Due to the fact that the purpose of the survey is to analyse backers’ motivation and behaviour, responses from participants were not included in the final analysis when the answer for “total projects backed” is left empty or entered with “0”. There were too many missing values on Intended amount (Q15 part 2, missing >60% on total) to yield any useful results, hence they were not included in final analysis. Further filtering based on this standard in Excel yields 212 qualified samples, with a qualification rate of 66%. SPSS was used to perform the statistical analysis of the collected data.

3.3.1 Missing Values

A frequency test was run to check errors and test the missing data for all the variables. No errors were found from the data. It was found that 1 participant did not respond to one of the Eudaimonia scales and another respondent did not complete 1 Hedonic happiness scale. The “Likelihood of supporting the imaginary campaign where only one kind of perk (reward) were provided to all contributors which you liked, but asked for a much higher amount than your latest contributed” had 2 missing values. Since the missing data was <10% of the total sample collected, a Hotdeck Imputation (Myers, 2011) was processed to complete the data sets of all three scales.

There were 4 other questions containing missing values: “What is your age?” (1 missing value), “How much did you contribute to this project?” (8 missing values), “Where are you employed?” (1 missing value) and “What is your annual income range” (11 missing values).

(35)

Due to the fact that these are objective answers, I decided to leave the missing value of these questions as is.

3.3.2 Recoding

A recoding of counter-indicative values was applied for Visit Frequency, to ensure a higher frequency of participation to be coded in a higher number, from “1” representing “less than once a month” to “7” representing “several times a day”; coded as “VisitFreq”. Income was also recoded to ensure that “1” represented the lowest amount while “9” represented the highest amount, and coded as “Income”.

3.3.3 Normality Test

Next, descriptive statistics were tested, skewness, kurtosis and normality checked.

IntrinsicCF, IntrinsicPJ and VisitFreq had no normally distributed items. All items were

relatively symmetrically distributed (skewness between -0.5 and 0.5) but comparatively flat (kurtosis was around -1.0). For Intentions to support a campaign, both Condition 2 (“IntentionCon2”) and 4 (“IntentionCon4”, skewness between -0.5 and 0.5) were considered normally distributed, with dots closely fit along the normality line in Q-Q plot and 0 within the bounds of the 95% confidence intervals; Condition 1 (“IntentionCon1”) and 5 (“IntentionCon5”) had skewness between -1 and 1, which suggested moderate level of skewness; in case of Condition 3, a high skewness to the right was shown (skewness > 1,). The kurtosis of IntentionCon1, 2, 3 and 5 were between -.5 and .5, while that of IntentionCon4 was between -1 and -.5. Projects Backed (“ProjectNo”) and Amount Contributed (“ContributedAmt”) both had skewness > 2, indicating the data was substantially skewed to the right. For items with skewness between 0.5 and 1, I attempted to transform the data by applying

(36)

none of the transformed data lead to better results. Although only 2 items had normal distribution data, most of the items had skewness and kurtosis between -1 and 1. Tabachnick and Fidell (2001) suggested that the effect of skewness and kurtosis can be reduced when using a large sample (200+ cases). Following this argument and based on the fact that this research had more than 200 respondents, I decided to use the non-transformed values and neglect the normality assumption.

3.3.4 Reliability and EFA

Reliability analysis was processed for both Intrinsic Motivation, with selection of “scale if item deleted”. The Cronbach’s α of IntrinsicCF is .926 (>.7) and IntrinsicPJ is .930 (>.7). “Corrected Item-Total Correlation” all above .8 (>.3) and all ∆α less than .03 (<0.1).The outcome indicates satisfying reliability.

In addition, since the 4 items of each Intrinsic Motivation originally came from 2 variables, a dimension reduction was adopted to check the factors. 1 component was extracted for IntrinsicCF (81.56%), KMO .775 (>.6) and Bartlett’s Test= 736.087 (p<.001). Same with

IntrinsicPJ, 1 component was extracted (82.67%), KMO .757 (>.6) and Bartlett’s Test=

790.330 (p<.001). This result confirmed that the scales are unidimensional. 3.3.5 Computing Scale Means

Next, means of the 4 items that were used to describe both IntrinsicCF and IntrinsicPJ were calculated and coded as “IntriCFtot” & “IntriPJtot”.

(37)

4. Results

The subsequent paragraphs further illustrated the results of data analysis. An overview of the samples’ demographic characteristic was presented, followed by descriptive statistics of data. The conceptual model was tested mainly by using correlation and regression techniques. Further analysis was implemented to test other relationships among variables, and feedback from respondents was presented.

4.1 Sample Characteristic

4.1.1 Demographic Characteristic

There were 212 qualified respondents, 8 of which were collected through individual messages sent from CF platforms, 108 from LinkedIn groups and 96 from Facebook groups. The ratio of male and female respondents were around 3:1. No published data regarding the actual gender allocation of total backers population was found, yet based on the fact that women angel-investors represented 26.1% of the angel-investor market in 2014 (Sohl, 2015), our result of 27% female participants seemed fairly reasonable. There was a wide coverage over respondents’ age, ranging from 19 to 74 (M = 40, SD = 10.85), 70% were between 28 and 49. A large percent of respondents resided in the United States (72%), followed by United Kingdom (5%), Canada (4%) and Australia (3%), 12% were from other European countries and the other 4% from South America, South Africa and the Middle East. Except for lack of respondents from Asia, this ratio was similar to the result of the latest CF industry report (Crowdsourcing.org, 2015). The majority of respondents had at least a college background and over 30% had received a master or higher degree. Almost half of the respondents had a job at private-for-profit businesses, around 19% working for government, public organization and

(38)

than 50% of the respondents had an annual salary higher than American’s national average wage index for 2013 (Social Security Office USA, 2015), 30% earned more than $90,000 per year.

There was a relatively even allocation of participants’ history of use on CF platforms (“HistoryUse”). 26% had less than 1 year of user experience, 27% between 1 to 2 years of experience, 25% between 2 to 3 years and 18% between 3 to 4 years. Only 4% of respondents had over 4 years of experience. As for visit frequency (“VisitFreq”), one third were active users who visited CF platforms at least once per day; more than half of respondents had a visit frequency ranging from several times a week to several times a month, and the remaining 14% had a VisitFreq of less than once per month. In addition, the type of CF platform (“CFPtype”), campaign backed (“PjType”) and incentive received (“ReType”) was also examined. The majority of respondents were backers of reward-based CF platform (83%), almost half of the campaigns involved in the study were for-profit ones (49%) and another 33% were artist-initiated ones, “incentive based on contribution” seemed to be the mainstream (68%) of ReType. A graphical presentation of the sample statistics is given in Figure 7.

(39)

Figure 7 Sample Statistic Graph

4.1.2 Variables: Means and Standard Deviations

(40)

while below 4 indicates negative perception. The mean value of IntriCFtot was 3.22, suggesting a relatively negative perception; a Std. Deviation of 1.57 showed that there was a moderately large variety of responses on IntriCFtot among respondents. The result of IntriPJtot was quite similar (M = 3.33, SD = 1.63). Neither of them were normally distributed, with slightly skewness to the left and relatively flat in shape. The histogram of IntriCFtot demonstrated a U-shape with one peak at the left outer end of the scale, showing a certain percentage of respondents participating in CF activities perceived fairly low intrinsic motivation. IntriPJtot, on the other hand, had two peaks, one at the left outer end and the other in the middle. This could mean that certain number of backers either had a negative perception for IntriPJtot or a neutral one. Figure 8 clarified these phenomena.

Figure 8: Histogram of IntrinsicCF and Intrinsic PJ

7-point Likert scales were also applied to Intentions to support a campaign, hence, a mean over 4 indicated a positive average intention whilst less than 4, a negative one.

IntentionCon2 (M = 4.43, SD = 1.52) & IntentionCon5 (M = 4.86, SD = 1.49) had positive

mean values while IntentionCon1 (M = 2.92, SD = 1.77) and IntentionCon3 (M = 2.51, SD = 1.70) had negative mean values. IntentionCon4 had a relatively neutral value, with M = 3.39,

(41)

SD = 1.60. This pictured a positive intention, on average, under situations that presented identical incentives without added extra conditions or an incentive similar to the previous project they backed, whereas a negative intention was seen, when either there was no incentive or an unfavourable one was provided. In the case of a favourable incentive in exchange for a higher contribution, the average intention scored neutral.

VisitFreq had relatively high average (M = 4.40, SD = 1.95) and FF relation relatively

low one (M = 2.00, SD = 1.63). The data of ProjectNo (M = 47.97, SD = 100.46) and

ContributedAmt (M = 627.70, SD = 4027.38) differed a lot (figure 9).

Further analysis on relationships among variables will be presented in the following paragraphs and possible reasons will be discussed in section 5.

Figure 9: Data Distribution of Number Backed and Amount Contributed

4.2 Test of Hypothesis

4.2.1 Correlations

The following correlation matrix includes the descriptive results and correlation coefficients among all the variables at both platform level (table 3) and project level (table 4).

(42)

There was a strong positive correlation between IntriCFtot and VisitFreq (r= .189, p< .01) but no significant relationship between IntriCFtot and ProjectNo was found, instead,

ProjectNo had a positive correlation with HistoryUse (r= .303, p< .01) and VisitFreq (r= .381,

p< .01). There was a strong negative correlation between FF and VisitFreq (r= -.242, p< .01),

FF and ProjectNo (r= -.203, p< .01), FF and Age (r= .138, p< .05) as well as FF and Income

(r= .141, p< .05). It is also noticed that DataSource had significant correlation with many variables, including Age (r= -.222, p< .01), Income (r= -.197, p< .01), IntriCFtot (r= -.274, p< .01), VisitFreq (r= .353, p< .01), IntriPJtot (r= -.277, p< .01), FF (r= -.311, p< .01), Education (r= -.188, p< .01) and Employment (r= -.148, p< .05).

Table 3: Correlation Matrix - Platform Level

Variables Mean Std. D 1 2 3 4 5 6 7 8 9 10 11 12 13 1. DataSource 2.42 0.57 -2. Gender 1.27 0.44 -.050 -3. Age 39.46 10.85 -.222** _-.100 -4. Country 161.07 54.23 .112 -.083 .037 -5. HistoryUse 3.37 1.43 .160* -.002 -.090 .027 -6. Income 5.41 2.95 -.197** _-.221** _.456** _.167* _0.078 -7. IntriCFtot 3.22 1.57 -.274** _{-.010 -.102 -.029 -.071 -.106 (.926)} 8. VisitFreq 4.40 1.95 .353** _{-.058 -.087 .032 .192** -.022 .189}** -9. ProjectNo 47.97 100.46 -.019 .079 .020 .012 .303** .095 .106 .381** -10. IntriPJtot 3.34 1.63 -.277** _{-.001 .029 -.041 -.146* -.058 .887}** _.099 _{.082 (.930)} 11. CFPtype 2.04 0.61 .009 -.124 .133 .036 -.049 .198** _{.122 .233}** _{-.015 .145}* -12. FF 2.00 1.63 -.311** _{-.035 .138}* _{.108 -.142* .141}* _{.118 -.242}** -.203** .148* .052 -13. Education 4.20 1.13 -.188** _{-.003 .289}** _{-.017 -.017 .390}** _{-.073 -.087 -.047 -.055 .125 .143}* -14. Employment 2.82 2.01 -.148* _.156* _.201** _{-.008 -.033 .009} _{.042 -.161}* _.002 _{.108 .079 .179}** _.093 **. Correlation is significant at the 0.01 level (2-tailed).

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

On project level, the matrix indicated a positive correlation between IntriPJtot and

IntentionCon1 (r= .141, p< .05), IntentionCon4 (r= .196, p< .01), CFPtype (r= .145, p< .05), ContributedAmt (r= .147, p< .05) and FF (r= .148, p< .05). PjType was correlated with Age (r=

-.156, p< .05), IntentionCon1 (r= -.206, p< .01), IntentionCon3 (r= -.241, p< .01) and

(43)

p< .05). Similar to PjType, ReType was also correlated with IntentionCon1 (r= -.297, p< .01),

IntentionCon3 (r= -.222, p< .01) and IntentionCon4 (r= -.189, p< .01), but not with Age. ContributedAmt was positively related with Age (r= .176, p< .05) and Income (r= .179, p< .05),

and was likewise with PjType (r= .147, p< .05) and CFPtype (r= .412, p< .01). There was a positive correlation between FF and IntriPJtot (r= .148, p< .05), as well as a significantly positive correlation between FF and IntentionCon1 (r= .260, p< .01) and IntentionCon3 (r= .219, p< .01). FF was also found negatively correlated with PjType (r= -.191, p< .01) and

ReType (r= -.213, p< .01). Again, DataSource had significant correlation with many variables

on project level, including IntriPJtot (r= -.277, p< .01), IntentionCon1 (r= -.386, p< .01),

IntentionCon3 (r= -.293, p< .01), IntentionCon4 (r= -.191, p< .01), PjType (r= .355, p< .01)

and ReType (r= .258, p< .01).

Table 4: Correlation Matrix - Project Level

Variables Mean Std. D 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1. DataSource 2.42 0.57 -2. Gender 1.27 0.44 -.050 -3. Age 39.46 10.85 -.222** _-.100 -4. Country 161.07 54.23 .112 -.083 .037 -5. Income 5.41 2.95 -.197**_-.221** _.456** _.167* -6. IntriPJtot 3.34 1.63 -.277** _{-.001 .029 -.041 -.058 (.930)} 7. IntentionCon1 2.92 1.77 -.386** _.141* _.185** _{-.088 .077 .141}* -8. IntentionCon2 4.43 1.52 .068 .039 .008 -.265** _-.090 _{.023 .318}** -9. IntentionCon3 2.51 1.70 -.293** _{.023 .190}** _{-.075 .093} _{.062 .672}**_.424** -10. IntentionCon4 3.39 1.60 -.191** _{.105 .041 -.193}** _{.037 .196}** _.454**_.497** _.501** -11. IntentionCon5 4.86 1.49 -.078 .056 .039 -.125 .098 .056 .091 .352** _.189** _.267** -12. PjType 2.43 0.78 .355** _-.156* _-.057 _{.046 .033 -.067 -.206}** _{-.016 -.241}** _-.176* _.022 -13. CFPtype 2.04 0.61 .009 -.124 .133 .036 .198** _.145* _{-.045 -.094 -.155}* _{-.054 -.025 .408}** -14. ReType 2.59 0.72 .258** _{-.009 -.100} _{.014 -.018 -.028 -.297}** _{-.103 -.222}**_-.189** _{.102 .263}**_.260** -15. ContributedAmt 627.69 4,027.38 -.109 -.085 .176* _{-.093 .179}* _.147* _{-.031 .104 -.023} _{.001 .075} _{.057 .412}** _-.081 -16. FF 2.00 1.63 -.311** _{-.035 .138}* _{.108 .141}* _.148* _.260** _{-.106 .219}** _{.069 .008 -.191}** _{.052 -.213}**_.113

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

To which extend do material incentives matter? : a study on Backers’ motivation behind crowdfunding behaviour