The effect of social media on crowdfunding success
between different industries
Master thesis
MSc. Business Administration – Marketing track
University of Amsterdam
by
Alexander van Unen (10099220)
under supervision of Adriana Krawczyk
2
Statement of originality
This document is written by Alexander van Unen 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 creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of
the work, not for the contents.
Abstract
Startup companies, facing difficulties to fund their businesses, are nowadays able to use social media to en-deavor financial support from the public to fund their projects (Lambert & Schwienbacher 2010). This ap-proach is known as crowdfunding. In this study we analyze the effect of social media on crowdfunding be-tween five different project categories, named project industries in this research, focusing on the amount of likes and followers. Previous studies have shown that crowdfunding and social media are associated. This current analysis builds on existing research and, for the first time, studies how Facebook, Twitter and Insta-gram effect crowdfunding success simultaneously. Data is extracted from the CrowdBerkley database re-garding 125 projects from the largest crowdfunding platform Kickstarter. As opposed to our expectations different project industries do not effect crowdfunding success. An explanation could be that we selected five project industries based on the highest campaign frequency, while there are several other project industries on Kickstarter. Based on our results we found that the amount of social media platforms that are being used are related to the amount of backers and not the funded total. This implies that if a project maker uses more social media platforms, this does not automatically lead to a higher funded total. Social media is a tool to create awareness and not the only predictor to achieve success. To make better predictions we need to look deeper into the content of the respective project structure to establish insights regarding the key drivers mak-ing a project successful.
Table of contents
1. Introduction --- P.4 2. Literature review --- P.6 2.1 Web 2.0 and social media --- P.6 2.2 Crowdfunding --- P. 8 2.3 Crowdfunding platforms and relationships --- P.10 2.4 Hypothesis and conceptual model--- P.13 3. Method --- P. 16
3.1 Database research --- P. 18 3.1.2 Sample --- P. 19 3.2 Measures --- P. 20 3.2.1 Independent and dependent variables --- P. 20 3.2.2 Mediating and moderating variables --- P. 20 4. Analysis --- P. 22 4.1 Descriptives --- P. 22 4.2 Correlations --- P. 24 4.3 Normality test--- P. 27 4.4 Logistic regression --- P. 28 4.5 Multiple mediation --- P. 30 4.6 Simple moderation--- P.34 5. Discussion --- P. 35
5.1 Effect of the amount of social media platforms on crowdfunding success --- P. 35 5.2 Effect of the amount of likes and followers and project industry on crowdfunding success --- P. 36 5.3 Managerial implications --- P. 37 5.4 Scientific implications --- P.38 5.5 Limitations and future research ---P. 38 6. Conclusion --- P. 39 References --- P. 42 Appendix --- P. 47
4
1. IntroductionSince 2011 the taxi service Uber raised 44.5 billion dollars through seed and additional funding, the shares of those who had invested at that time in Uber, are now worth 600 times as much. However, legal regulations prevented people to invest in new startup companies like Uber, unless you were very rich. Since Friday, July 19, 2015 the US rules with regard to social media, online buying and investing in new startups have changed (Farajian, Lauzon & Cui 2015). Up to now, this was only possible for the 3.5% of US richest households, including Silicon Valley venture capitalists. These federal changes open new investment dimensions and give founders of new startups a motive to bypass powerful firms. The JOBS Act, that was implemented in 2012 by the government gave companies the opportunity to raise funds and sell their shares, which encour-aged crowdfunding. The changes in the JOBS Act by SEC and the development of social media gave StartEngine the opportunity to launch a crowdfunding platform, which allows anyone to invest in new startups.
"Unless you're a rich person, you can't participate in the next Facebook or Apple or Tesla. Friday opens that up." - Ron Miller, CEO of StartEngine
Social media has had a major impact on how people communicate, create networks and communities in to-day’s world. It also removes obstacles and restrictions regarding period of time and location. Nowadays startup companies face difficulties to fund their businesses mainly in the initial period (Cosh, Cumming & Hughes 2009). These startup companies are often not entitled to lend money from banks. So they use social media to endeavor financial support from the “crowd” to fund their new startup projects (Lambert &
Schwienbacher 2010). This way of seeking financial support is also known as crowdfunding. Crowdfunding can be seen as a way of cooperative social media. Since the development of social media, crowdfunding has progressed simultaneously (Nann et al. 2010). Previous studies have shown that crowdfunding and social media are associated. Etter, Grossglauser and Thiran (2013) studied how to predict crowdfunding success with data from Kickstarter combined with social media data from Twitter. They stated that crowdfunding success can be predicted through: the amount of time in which the money is deposited (time series tors) and social media (Twitter ‘tweets’ predictors). Furthermore they concluded that the time series predic-tor, predicts the right amount of money for 85% after only 15% of the total length of the project campaign.
Social media use through the Twitter platform was less of an accurate predictor for crowdfunding success. Etter et al. (2013) suggest to combine both predictors and use other predictors in the future to predict crowd-funding success. However, Etter et al. (2013) only use ‘tweets’ from Twitter as a predictor for crowdcrowd-funding success. They forgot to look at the number of followers and likes on Twitter and other social media plat-forms. Smith (2013) demonstrated that in the last few years the most successful companies all increased their focus on Twitter, Facebook and especially Instagram. McNely (2010, 2011) studied the importance for busi-nesses to engage in social media platforms such as Facebook and Twitter. However, McNely (2012) states that the company's image can especially be enhanced by means of image-intensive social media software such as Instagram. Since march 2015 Instagram has over 300 million active accounts and since most users are younger than 35, many companies choose to reach them through this social media platform (Collins 2015). Forbes contributors have placed Instagram as the most effective marketing tool at the moment. To date there are no studies that have focussed on how the amount of likes and followers, of the three most pop-ular social media platforms, Facebook, Twitter and Instagram, effect crowdfunding success (Statista 2015). Clancy (2015) states that the usage of (multiple) social media platforms to improve business success varies per industry. He also states that not all businesses need to be very active on social media to achieve success. For example a big company like eBay only gets an average of about 650 likes per post (way less than Star-bucks), and there are no regular updates, nevertheless Ebay is still a very successful company (Petit 2014). This shows that social media usage has a different effect on performance and success between different in-dustries. Until now the literature, to understand the effect of social media likes and followers on crowdfund-ing success, falls short. In addition, there is no research that has focussed on how Facebook, Twitter and In-stagram simultaneously effect crowdfunding success between different industries. It is also interesting to see whether engaging in all three social media platforms, or in just one or two, has a different effect on crowd-funding success.
The aim of this Master thesis is to establish better insights in the effect that social media has on crowdfund-ing. Especially, this inquiry focuses on the interactions between social media functions, as likes and follow-ers, which shows how much people value the project. In this research we chose the social media platforms: Facebook, Twitter and Instagram because these platforms have worldwide the highest number of users (Sta-tista 2015). In addition, we also study how differences between project industries effect crowdfunding
suc-6
cess and how these project industries are related to social media. In this study crowdfunding success is de-fined by the following three variables: obtaining of the funded aim, the funded total, and the amount of back-ers. Backers are people who connect with project makers and invest money to give the project a financial or awareness boost (Kickstarter FAQ). This study is both societal and scientifically relevant because it investi-gates the latent effects of social media on crowdfunding outcomes and gives better insights for companies and entrepreneurs that engage in crowdfunding projects to determine positive effects and prognosticate pro-ject outcomes. The second chapter of this thesis is a literature review, which will explain the history and def-initions regarding social media and crowdfunding. In addition it also incorporates the research question and hypotheses of this study. The third chapter describes the methodology and research design. The fourth, fifth and sixth chapters respectively describe the analysis and results, the discussion and conclusion.2. Literature review
Although social media is a concept that has become more important in the last fifteen years it still has some uncertainties. One reason, among others, is that the world is changing at a very fast pace. Organizations and consumers find it hard to keep up with the ever-changing social media activities (Vij & James 2014). To un-derstand the impact of social media we firstly need to address a view questions like: What is social media and what is it not? How does social media differ from web 2.0 and how does this interact with crowdfund-ing? These questions will be clarified in this chapter.
2.1 Web 2.0 and social media
Researchers often place web 2.0, social media and user-generated-content (UGC) under the same umbrella, but although they are all dependent on each other, we also have to understand the differences between them to understand the different characteristics. Berthon et al. (2012) describe web 2.0 as the technical basis that is the necessary incentive for social media and UGC. Social media is about the content while UGC are the makers of this content. Web 2.0 makes it possible to create and distribute the content, which is known as so-cial media. Evans (2010) agrees with Berthon et al. (2012) and described soso-cial media as a way of how peo-ple can interact with each other to create content, share comments, and exchange content with other peopeo-ple.
According to Evans (2010) Web 2.0 is a platform where people can execute content online in a variety of ways like: building new content, recommendations, share information, and publishing etc. This can be seen as UGC, which is content in all kinds of forms, like: visuals, sounds, literal etc. (Blackshaw and Nazzaro 2004). People have become more cautions on what information resources, advertisements and people to be-lieve. The reason is that they are less confident and distrust the social environment where they live. People prefer sharing information with peers in communities (Brogan & Smith 2010). This is a reason, among oth-ers, why the popularity of social media has grown so much in last years. People can create online communi-ties where people with common interests can share information. When comparing web 1.0 and web 2.0 the major difference was the technological change but also concentrating on the sociological effects. Web 2.0 is about consumers, communities, social networks instead of web 1.0 which is about companies, the individual and intrusion (Berthon et al. 2012). Web 2.0 gave people the opportunity to express their own views on products and services and to share information with the people inside their community. In addition UGC, which people can exchange via social media, gave them the chance to engage in the present social environ-ment (Evans 2010). Brogan (2010) and Zarella (2010) state that social media consists of a diversity of multi-channels, like the Internet, where people and organizations can interact with each other. Naturally, social media is very convenient and ascendable. In addition, social media creates dialogue between lots of people at the same time, which encourages people to produce their own content instead of only using content from others. Kaplan and Haenlein (2010) describe social media as ‘set of Internet applications that are created on the ideological and technological basic knowledge of Web 2.0, and gives the opportunity to produce and trade UCG’. Social media can also work negatively for companies when customers share negative reactions to the outside world through social media (Kramer, Guillory & Hancock 2014). For example, Mcdonalds, where people often complain and share negative opinions on Facebook (Curry 2012). Kaplan (2012) states that social media can generate information in two ways in contrast to other media, namely: data accessibility in any case and at any time through the GPS networks. This gives marketers the change to use social media target and their (potential) customers in a new way.
Social media can consist of a variety of forms. Kaplan and Haenlein (2010) have conducted a simplistic range of six different social media forms: blogs, social network websites, virtual worlds, cooperative pro-jects, communities, fora. Kaplan and Haenlein (2010) describe blogs as a social media form that shows
in-8
formative content in a backward chronologic order, which can be present in a lot of different forms. Personal stories and blogs that are sponsored by organizations are the most desirable (Mangold 2009; Kaplan & Haen-lein 2010). The Huffington Post is for instance a very popular blog. Social network websites give people the opportunity to build individual web pages, which other people can visit, and where they can exchange infor-mation by chatting and emailing. On these individual profile web pages people can upload pictures, films, music and other content (Kaplan & Haenlein 2010). In virtual worlds people can make their own 3D world and appear as an icon or figure representing a particular person. Cooperative projects give independent users the ability to produce content together, like Wikipedia (Kaplan & Haenlein 2010). In addition social media communities is a great platform where community users with the same interests can share information. Fora are a medium where ideas and views on a particular issue can be exchanged and discussed between people (Bhadani 2012). In addition Evans (2010) states that customer engagement is a very important aspect in all forms of social media and the current business landscape. In contrast to traditionalistic media, social media is more focussed on people and companies working together, instead of creating only awareness (Evans 2010). He also states that the collaborative way of customer engagement is the most desirable, because it gives both people and companies the opportunity to create content and information in a cooperative manner. Google-Docs is an example where multiple people can modify and change content in the same document. There is a variety of ways where companies and stakeholders work collaboratively. In this research I focus on the col-laborative approach, which is known as crowdfunding.2.2 Crowdfunding
Crowdfunding is an activity where startup entrepreneurs can attract funders for specific projects. An example is Kickstarter, the biggest crowdfunding platform/webpage in the USA, where projects from different indus-tries can attract financiers (Mollick 2014). Crowdfunding is the only collaborative approach where consum-ers need to be involved financially. This gives crowdfunding a great advantage, to have a major impact on the business, by interfering in the product development of a company. Considering the barriers that new companies encounter in raising funding from external investments, banks and other trusts funds, more and more business owners are engaging in crowdfunding, which gives these business owners the ability to attract the public instantly. However, Crowdfunding is still a relatively new topic, the terminology has only been used since 2006, and there are only a limited number of studies in the marketing field. Belleflemme, Lambert
and Schwienbacher (2012) define crowdfunding as “an open call, essentially through Internet, for the provi-sion of financial resources either in the form of donation or in exchange for some form of reward and/or vot-ing rights in order to support initiatives for specific purposes”. To understand crowdfundvot-ing as an activity, a clear definition is important. The crowdfunding activity concerns three major actors: “project makers”, “backers” and “crowdfunding webpages”. The project makers are the people that need external financial support to fund their business, the backers are the people that are financially backing the projects and the crowdfunding webpage is the mediating platform that brings the projects makers and backers together. In addition projects makers are able to use the crowdfunding webpages to make content available for the gen-eral public and potential backers, regarding their projects, such as: what the project means, website URLs, pictures, Twitter, Facebook, Instagram. Besides that the project maker can also indicate what the required funded target is and the installment of the project on the webpage. The crowdfunding platform has the ability to share this information with a large number of potential backers through social media. When backers are interested in certain projects they can comment, like, follow projects via respectively the crowdfunding webpage, Facebook, Twitter and Instagram. Off course the most desirable effect is that the backers eventual-ly back the project financialeventual-ly. Crowdfunding webpages hold on the most important factors that are central in social media: descriptive project information, communities and available technology, also known as web 2.0, where backers can help through commenting, donating and financial support. However the Kickstarter Sta-tistics (2015) also show that only 41% of the total amount of projects on their webpage reaches their funded aim and even 11% does not get any financial funding.
The crowdfunding process has great links when it comes to charities, financial industries (Ordanini et al. 2011; Belleflame et al. 2012). However crowdfunding is a special and distinctive approach to fund projects compared to other financing approaches. The biggest difference is the absence of middleman, which gives ordinary people the possibility to get involved in the crowdfunding process without being an expert in a cer-tain topic or industry. Backers are able to be unselfish and generally want to help others (Schwienbacher 2010).
The role of the consumer in the marketing landscape has changed over the past few decades; before they were only a target but know they are active engagers in the whole business process (Hunt et al. 2012).
Orda-10
nini et al. (2012) state that crowdfunding, compared with other social media platforms, provides such an ac-tive consumer-engaging role. It gives consumers the opportunity and latitude to make entrepreneurial deci-sions, such as: product accessibility and development. There are several ways how crowdfunding is applied and organized in practice. Belleflamme et al. (2012) distinguish four types of funding, namely: equity, debts, donations and funds from goods that are ordered in advance. Equity and debts are the types where consumers can expect some form of compensation for their funding. Donations, on the other hand, are usually done from benevolent considerations. Donations are a very popular type of how projects, such as: upcoming artists or movie producers are funded on the Kickstarter webpage. Funds that are ordered in advance means that the backers want to get the product or service from the projects maker in exchange for their investment.2.3 Crowdfunding platforms and relationships
Kickstarter applies the “all-or-nothing” principle whereby the project maker only gets money when the de-sired project goal is achieved (Baumgardner et al. 2015). Baumgardner et al. (2015) also describe the “keep-it-all” principle, which is applied by crowdfunding website Indiegogo (Falcon 2013), whereby the project maker receives all the fundraised money but has to pay a higher fee. Lambert and Schwienbacher (2010) state that crowdfunding platforms do not only play a financial role but also give projects makers the oppor-tunity to create project awareness through social media for their projects. Social media such as Twitter, Fa-cebook and Instagram gives (potential) backers opportunity to connect and discuss information about pro-jects with each other, which can create multiple benefits for both the project makers as the backers (Lambert & Schwienbacher 2010). Project makers can see how many people like or back a project, which gives them the ability to shed a quick glimpse at which projects are interesting for the future. Backers can also comment on projects and share information with other interested people before backing the projects (Lambert and Schwienbacher 2010). Other benefits are more project awareness to attract potential backers. In addition they can attract backers by giving them samples of their products of services when they donate or fund the pro-jects and project backers have the ability to discuss information with others so that they can make a good overall funding decision (Schwienbacher 2010).
Although crowdfunding has many positive features, there are also a number of projects that do not achieve their financial goal. Financial input and approval by friends and family is a very important aspect for small
companies to reach their financial target, especially to convince other backers that the project is safe and fair (Cumming & Johan 2009; Agrawal, Catalini & Goldfarb 2011). Agrawal et al. (2011) concluded that most projects are backed financially by friends and family. The reason is that friends and family can make better funding decisions because they have a privileged position regarding the information about the projects. Burtch, Ghose & Wattal (2013) state the importance of transparency in crowdfunding. If potential backers can find content/information, such as: backers, comments, likes and followers from people who already in-vested in the project, they are more likely to invest in the project for themselves. In this case social media can provide potential backers with information to evaluate and make decisions in a more specific and honest matter (Burtch et al. 2013). Providing potential backers with transparent information on crowdfunding web-sites, such as: the funded aim, the funded total, the amount of backers, funding ratio, followers and likes, can have a major impact to achieve the financial goal (Burtch et al 2013). Frydrych et al. (2014) state that crowd-funding has developed strongly in recent years by the arrival of the Internet and especially social media. Since recent years, crowdfunding webpages as Kickstarter have integrated more and more social media func-tions such as: direct links to Facebook, Twitter, Instagram, Pinterest etc. This has brought great change with it, but the crowdfunding webpages still lack behind when it comes to user-friendly communication between business owners and backers. It can, therefore, be concluded that crowdfunding webpages have the basic functions but not much more than that. As mentioned above Kickstarter facilitates the opportunity for project owners to share their Facebook page, which can eventually lead to more project credibility and belief to back the project (Lambert & Schwienbacher 2010). Obviously project makers are not required to share personal information from Facebook, Twitter and Instagram with potential backers via the crowdfunding webpage. They can also just use the primary options, but if project makers use these secondary social media options it is interesting to research if these social media functions influence the crowdfunding business. Project makers that share social media features, assume that this has a positive contribution to achieve their financial goal quicker. In addition it is interesting to important in what way social media functions as likes, followers influ-ence the crowdfunding business differently. Mollick (2012) states that the number of Facebook ‘friends’ ef-fect crowdfunding success, but does this also apply for likes and followers on Facebook, Twitter and Instra-gram? And what role does the projects industry play?
12
A variety of researchers have already studied the relationship between social media and crowdfunding. Mol-lick (2014) did an exploratory study about crowdfunding and concluded that: most projects face postpone-ment, quality is important and geographic differences have an impact on crowdfunding performance. In an earlier research Mollick (2012) determinate that there is a relation between ‘friends’ on Facebook and crowdfunding success in the movie industry. However, as Etter et al. (2013) who only focussed on Twitter ‘tweets’, Mollick (2012) also did not focus on likes and followers on Twitter and Instagram and projects makers who don’t have Facebook. Therefore it is important to study if Facebook, Twitter and Instagram ef-fect crowdfunding success simultaneously. For example, potential backers can also back a project by onlyliking or following the project (makers) on multiple social media platforms, which shows how much they
value the project, because this creates project awareness pertaining to other potential backers. Therefore they can show that they find the project credible and honest, which can suggests that social media gives a differ-ent twist to the crowdfunding process. Kickstarter implemdiffer-ents this strategy because on every project page people can like or comment on the project. Besides that it is also very interesting to look if likes and
follow-ers are related to each other and how this difffollow-ers between Facebook, Twitter and Instragram. McNely (2012)
stated that the company's image can particularity be enhanced through image-intensive social media software such as Instagram. Collins (2015) states that Instagram is growing with a fast pace and has over 300 million active users at the moment. Besides that, Forbes contributors described Instagram as the most effective mar-keting tool at the moment. This shows how interesting it is to investigate what effect Instagram has on crowdfunding and how this social media platform relates to the other two most popular social media plat-forms Facebook and Twitter (Statista 2015). Until now there has not been any research that focuses on the effect that Instagram has on crowdfunding. This study focusses on how the amount of different social media platforms (Facebook, Twitter and Instagram) can influence crowdfunding success. We also look at how the positive relationship between the amount of different social media platforms and crowdfunding success is mediated by the amount of likes and followers on social media. Petit (2014) states that social media usage has a different effect on performance and success between different industries. Therefore, we also look at how the positive relationship between the amount of different social media platforms and crowdfunding suc-cess is moderated by project industry. The biggest crowdfunding platform Kickstarter facilitates thousands of active projects out of fifteen different categories: art, comics, crafts, dance, design, fashion, film & video, food, games, journalism, music, photography, publishing, technology and theatre (Kickstarter FAQ). This
large variety of categories makes it possible for practically everyone to start a campaign. In this research we name categories project industries because the research by Massolution (2013), which is a crowdfunding in-dustry report, also named them project industries. In this research we will focus on the following five indus-tries: music, film/video, publishing, art and games. We selected these project industries from the Crowd-Berkley database. Crowdberkley is a convenient database because the data is open to all and it dives deeper into crowdfunding projects compared to other databases like Krowdster. The industries are selected, based on the highest campaign frequency. We searched within the CrowdBerkley database which project industries (categories) had the most number of active campaigns within the time period of 3 November 2015 and 4 De-cember 2015. We found that the most campaigns were active in the following industries: music, film/video, publishing, art and games. This research aims to give a more complete perspective on how crowdfunding is influenced through social media. To get a better understanding about these relations the following research question is addressed: “How does social media influence crowdfunding success between different indus-tries?”
2.4 Hypothesis and conceptual model
In addition to the research question there are several other propositions that are important to understand the effect of social media on crowdfunding success. Based on the research of Mollick (2012) we know that there is a relation between ‘friends’ on Facebook and crowdfunding success in the movie industry. Mollick (2014) also found some other interesting findings regarding the crowdfunding process, such as: most projects face postponement, quality is important and geographic differences have an impact on crowdfunding perfor-mance. Etter et al. (2013) did not focus on Facebook ‘friends’ to predict crowdfunding success; they fo-cussed on Twitter ‘tweets’. They found that Twitter ‘tweets’ are by itself not an accurate predictor for crowd-funding success. However they stated that when you combine Twitter ‘tweets’ with other predictors simulta-neously, such as the time-series of the campaign, it is possible to predict crowdfunding success fairly accu-rate. Mitra and Gilbert (2014) did an analysis of the Kickstarter webpage and studied how project makers can use certain quality phrases on their campaign page to persuade backers to invest in their project. These back-ers are directly related with crowdfunding success, because they determine whether a project reaches it ob-taining of the funded aim and funded goal, within the time period of the campaign (Mitra & Gilbert 2014). Building on the findings of Mitra and Gilbert (2014) we will define crowdfunding success based on three
14
predictors: the obtaining of the funded aim, the funded total and the amount of backers. These are convenient predictors for crowdfunding success of projects, because the figures of these predictors are accessible to eve-ryone (Kickstarter FAQ). Based on the Kickstarter FAQ, the obtaining of the funded aim is the target aim of financial funding that project makers need to attract to develop their product or service. The funded total is the total amount of financial investments by backers when the time period of the campaign has ended. The figures of obtaining of the funded aim and the funded total are all the financial sums in a certain currency. Because there are different currencies between projects we converted all figures into euros as a ratio, to make comparisons more reliable. The amount of backers are the total amount of financial investors that financially invested in the project when the time period of the campaign has ended.Furthermore, Mitra and Gilbert (2014) concluded that the Facebook connecter on the campaign page is suc-cess factor. McNely (2012) and Collins (2015) have indicated that engaging in image-intensive social media software, such as Instagram, is important for companies to enhance their image. Instagram is a free social media application that allows people to share photos and short movies with other users and via multiple so-cial media platforms, among others, Twitter and Facebook (Salomon 2013). Salomon (2013) studied the rise of Instragram. She concluded that Instagram users find it the most rewardable and innovative platform, com-pared to other social media platforms. The reason is that users have the ability to share their picture via In-stagram with other social media platforms, such as: Facebook, Twitter and Pinterest (Salomon 2013). Anoth-er recent study by Park, Ciampaglia and FAnoth-errara (2015) detAnoth-ermined that a high amount of likes and followAnoth-ers on Instagram are great predictors for a successful campaign in the fashion industry. A reason, among others, is that fashion creators and models can interact with other people by using Instagram. Park et al. (2015) con-cluded “a high number of likes and comments, as well as frequent posting, were associated with success on the runway”. Instagram can be seen as the most rewardable and popular social media platform at the moment (Salomon 2013). It also gives users the opportunity to share pictures on multiple social media platforms. It is interesting to see if the amount of likes and followers will attract more backers in the crowdfunding process and the same positive effect as in the fashion industry Park et al. (2015). In the third chapter there is an ex-planation how we collected the total amount of likes and the total amount of followers. In this research we focus on the three social media platforms: Facebook, Twitter and Instagram because these platforms have worldwide the highest number of users (Statista 2015). With that in mind the amount of social media
plat-forms that are being used in a project campaign can differentiate between: zero, one, two or three. At the moment there are no studies that have researched how Facebook, Twitter and Instagram effect crowdfunding success simultaneously, that’s why we conducted the following four hypothesis:
H1: The amount of social media platforms that are being used are positively related with the obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success).
H2a: The amount of likes are positively related with the obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success).
H2b: The amount of followers are positively related with the obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success).
H3: The amount of social media platforms used are positively related with the amount of likes and fol-lowers, so when project makers are active on more social media platforms they will have a higher amount of likes and followers.
Clancy (2015) concluded that the usage of (multiple) social media platforms to enhance company success varies per industry and not all businesses need to be very active on social media to achieve success. As we already mentioned, Park et al. (2015) stated that when a person or company, in the fashion industry, has more likes and followers on Instagram it will be more successful. But to what extent are these social media effects also generalizable in other industries? Petit (2014) made clear that a e-commerce company like eBay is very active on the internet, but not very popular and active on social media platforms like Facebook, Twit-ter and Instagram. However eBay had a revenue of 17.90 billion dollars in 2014 (eBay 2014) which shows how successful they are without actively engaging in social media. Ordanini et al. (2011) are the only re-searchers who investigated how crowdfunding projects differ between different industries. They did a case study where they compared three different crowdfunding platforms: SellaBand (music industry), Trampoline (financial services) and Kapipal (non‐profit services). Sellaband wants to empower stakeholders in the global music community. Fans can interact and support hopeful artists, which enables new opportunities for both artists and fans via Sellaband. Trampoline is financial service that looks at multiple and new ways to attract
16
financiers by instantly communicating with potential investors. Kapipal is a non-profit service, which ena-bles the opportunity for people to fund their new social projects online (Ordanini et al. 2011). Ordanini et al. (2011) found some essential differences between these three different crowdfunding companies. Sellaband can be seen as mediator and intermediary, concerning what is offered and demanded in music industry, by linking people with artists that they might like. Kapipal can also be seen as intermediary but in a different context, because this company is not specialized in one industry, but focusses on the general private needs for example individual financial purposes. Trampoline is very different compared to Sellaband and Kapipal because it cannot be seen as a mediator or intermediary. In this case the seller arranges how his project and campaign should be funded. Trampoline advocates the new project instantly, so the crowdfunding process can start immediately. This shows crowdfunding platforms work in different manners, because they are ac-tive in different industries. However, as mentioned in the literature chapter, the big crowdfunding platforms such as Kickstarter are active in multiple industries (categories). So these platforms have a more general mo-tive, like Kapipal. However, Ordanini et al. (2011) make clear that all these three crowdfunding companies need to engage in social media to connect with project makers and potential backers. In addition they also state that crowdfunding companies can connect consumers and investors, which are the most important ac-tors, by giving people the ability to communicate with each other via multiple social media platforms.We expect that the amount of social media platforms will h In the third chapter there is a explanation how we collected the total amount of likes and the total amount of followers. ave a direct positive effect on crowd-funding success, as described in the first hypothesis. In addition, we expect that the amount of social media platforms have a positive effect on the amount of likes and followers, as stated in the second hypothesis. However we also find it interesting how the amount of likes and/or followers can influence the effect be-tween the amount of social media platforms and crowdfunding success. In this case the amount of likes and the amount of followers will both be mediating variables. Barron and Kenny (1986) describe the mediating variable as a (third) variable that influences the effect of the independent variable on the dependent variable. In the third chapter there is an explanation how we collected the total amount of likes and the total amount of followers. Furthermore, we already addressed that the way companies use social media platforms to enhance company success varies per industry (Clancy 2015). In this research we chose five different project indus-tries, based on the highest campaign frequency. Ordanini et al. (2011) showed how three different
crowd-funding platforms: SellaBand (music industry), Trampoline (financial services) and Kapipal (non‐profit ser-vices) have different roles in the crowdfunding process. As stated in the first hypothesis we expect that the amount of social media platforms have a positive effect on crowdfunding success. However we also find it interesting to see the role of the different project industries. We expect that project industry can effect the strength of the amount of social media platforms on crowdfunding success. For example, a project from the Music industry can effect the strength of the relationship between the amount of social media platforms on crowdfunding success more strongly positive or negative than a project from the Art industry. In this case project industry will be a moderating variable. Barron and Kenny (1986) describe the moderating (third) var-iable as varvar-iable that indicates the direction and the strength of the relationship among the independent and the dependent variable. Based on the literature we conducted the following two hypotheses and the concep-tual model:
H4: The positive relationship between the amount of social media platforms on the obtaining of the funded aim, funded total and the amount of backers (crowdfunding success) is mediated by the amount of likes and followers on social media, so crowdfunding success will increase if the amount of likes and followers is higher.
H5: The positive relationship between the amount of social media platforms on the obtaining of the funded aim, funded total and the amount of backers (crowdfunding success) is moderated by the project industry.
Conceptual Model H1+ H5+- H3 + H2+
Amount of social me-dia platforms: 1) Facebook 2) Twitter 3) Instagram
Crowdfunding success: a) Obtaining of the funded aim in euros
b) Funded total in euros c) Amount of backers
Amount of social media likes and followers Project industry: 1) Music 2) Film & Video 3) Publishing 4) Art 5) Games
18
3. Research method3.1 Database research
This study is based on quantitative database research. The data is extracted from the CrowdBerkley database. This database gives researchers the opportunity to find open crowdfunding data from six crowdfunding webpages, among others, Kickstarter, Indiegogo, fundrazr and Rockethub. The Crowdberkely database was very convenient for this research because the data is open to all and it dives deeper into crowdfunding pro-jects compared to other databases like Krowdster. This research aims to establish how propro-jects from five dif-ferent industries (music, film/video, publishing, art and games) can have an effect on crowdfunding success; this means a large amount of projects is needed to see the differences. Therefore it was convenient to do a quantitative research.
We have analyzed two crowdfunding webpages that facilitate social media functions: Kickstarter and Indie-gogo. We chose these crowdfunding platforms because they are the two most popular crowdfunding webpages based on forbes.com and they have the highest amount of active campaigns based on the Crowd-Berkley’s database. Although these webpages work in a similar way, there are some differences. Kickstarter operates via the “all-or-nothing” principle whereby the project maker only receives money when the desired project goal is achieved (Baumgardner et al. 2015). Baumgardner et al. (2015) also describe the “keep-it-all” principle, which is applied by the crowdfunding webpage Indiegogo (Falcon 2013), whereby the project maker receives all the fundraised money but has to pay a higher fee. In addition, these webpages also differ in the way they facilitate and mention the details about social media. Kickstarter prefers to focus and link personal profiles and data from Facebook. Kickstarter also gives project owners the opportunity to share their number of friends on their campaign page. Besides that, all other social media platforms links: Twitter, Embed, Pinterest, Tumblr and Instagram could be founded at the campaign page. Indiegogo operates differ-ently compared to Kickstarter, because the personal profiles on Facebook, Twitter, Instagram were not shared on their campaign page. Indiegogo did only share the Facebook, Twitter pages of the project itself and not of the project owners. These two crowdfunding platforms differ in the way they facilitate and con-nect with social media profiles, so it is hard to compare new projects between different crowdfunding plat-forms. That is why we decided to only focus on projects from one crowdfunding platform. We chose Kick-starter for this research, because it is the largest crowdfunding webpage worldwide with the most variety
when it comes to industries (Falcon 2013). Besides that, we chose Kickstarter over Indiegogo, because Kick-starter makes the most information accessible to the public regarding how the crowdfunding company works. Kickstarter was also preferred over Indiegogo because it facilitates a more personal interaction between the backers and the personal social media accounts of the project owners. In addition, Kickstarter was also very favorable to gather the data, because it gave the possibility to store the links of the projects and the end dates for each project. When the campaign deadline was ended we could collect the final data of the social media likes and followers, the obtaining of the funded aim, the funded total and the amount of backers.
3.1.2. Sample
The data was collected between 3 November 2015 and 4 December 2015. In order to provide a representa-tive view of the projects on Kickstarter, in this time period, we randomly selected 125 projects from five dif-ferent project industries. To select the projects for this research we first examined which project industries had the highest frequency of campaigns. From the CrowdBerkley database we selected, based on the highest campaign frequency, the following five project industries: music, film/video, publishing, art and games. The projects were selected out of the CrowdBerkley's database by probability and stratified sampling. Probablity sampling gives each of the projects for each industry the same chance to be selected (Explorable 2009). Stratified sampling, which is a type of probability sampling, was convenient because we needed to divide our sample of 125 projects into five subgroups with each 25 projects. These 25 projects were randomly chosen, within our time frame of data collection, from the total amount of projects for each industry (Explorable 2009). Per project industry we selected 25 projects, once every four, within the specified time period to get a representative view between the projects in the five industries. We defined the five different industries, so there was no apparent overlap between projects industries, which is a requirement for stratified sampling. At the Kickstarter webpage we searched for the selected projects and gathered the necessary data. Since the most project campaigns were not ended at the time we have saved the links of the webpages. After the time period of the 125 projects had expired, the final data regarding the social media likes and followers, the ob-taining of the funded aim, the funded total and the amount of backers were collected (Appendix A).
20
3.2 MeasuresIn this research we aim to establish the relation between the independent social media variable and the de-pendent crowdfunding variables to predict crowdfunding success. In addition, we want to examine if the amount of likes and followers on social media can have a possible mediating effect and how project industry can have a possible moderating effect on crowdfunding success.
3.2.1. Independent and dependent variables
The predictor (independent) variable is the amount of social media platforms that are used per project cam-paign. In this research we chose the social media platforms: Facebook, Twitter and Instagram because these platforms have worldwide the highest number of users (Statista 2015). The amount of social media platforms can be: zero, one, two or three when all platforms are directly or indirectly shared on the Kickstarter cam-paign page. Project makers can directly share links to their Facebook, Twitter and/or Instagram accounts on their campaign pages or indirectly when they, for example, only share their main project webpage with social media linkages. The amount of social media platforms is the independent variable in all five hypotheses and is measured at ratio level.
The dependent variables the obtaining of the funded aim, the funded total and the amount of backers, used in the first, third, fourth and fifth hypothesis, explain crowdfunding success. These dependent variables are all gathered as a ratio after the campaign had ended. In addition, the obtaining of the funded aim and the funded total of some projects were not in the same currency. This makes it difficult to make comparisons between different projects and therefore all amounts in currency terms are converted into euros. The total amount of backers is the final number of backers, who had invested in the project, when the duration of the campaign had expired. We analyzed our data with SPSS 23. We chose SPSS because we have worked with this soft-ware at the University of Amsterdam. The first and third hypothesis were tested through correlational analy-sis using. While both second hypotheanaly-sis are tested through logistic regression analyanaly-sis
3.2.2 Mediating and moderating variables
To analyze mediating and moderating effects between variables Process is a convenient modelling tool that is available for SPSS for free (Hayes 2012). Mediating variables are related to questions about ‘how’ a
rela-tionship is established, while moderating variables described when relarela-tionships between variables occurs (Baron & Kenny 1986). Mediation analysis wants uncover the extent how a causal variable (X) influences the outcome (Y) through the mediator variable(s) (Anderson & Bushman 2002). By executing a Moderating analysis researchers want to uncover if the size of sign of the effect between a causal variable (X) on the out-come (Y) interacts with another variable, which is also known as the mediator (Markey & Markey 2010).
The mediating variables in this research are the, amount of likes and followers, and are expected to be related with each other based on previous research by Nicholls (2012). However they are two different mediating variables. Facebook and Instagram give people the opportunity to both like and follow project pages and pic-tures related to the project. By using Twitter project makers can attract followers and upload picpic-tures/text posts, which can be “retweeted” by others. In this research we considered “retweets” as likes because previ-ous research by Nicholls (2012) showed that when people “retweet” a certain post or picture they conform that they like the post.
The social media accounts are analyzed according to two criteria: the personal social media accounts of the project owner(s) and the social media accounts of the product or service for which they wish to receive fund-ing by backers. The total amount of followers from the three social media accounts are all added together, in order to give a complete view of the total amount of followers regarding each project. The same has been done for the total amount of likes. We have added the total amount of likes on Facebook and Instagram with the number of “retweets” for each crowdfunding project. Off course a prerequisite is that all the likes and “retweets” need to be related to the crowdfunding project and the linkages to the social media accounts need to be delivered directly or indirectly through the Kickstarter campaign page.
The moderating variable, project industry, is based on five industries with the highest campaign frequency on Kickstarter: music, film/video, publishing, art and games. As Clancy (2015) mentioned the usage of (mul-tiple) social media platforms to enhance company success varies per industry and not all businesses need to be very active on social media to achieve success. We find it interesting to test whether this also applies for crowdfunding projects. We expect that moderator project industry may have a positive or negative effect on crowdfunding success as mentioned in the fourth and fifth hypothesis. The fourth and fifth variables are
test-22
ed with the Process software by Hayes (2012). We chose for Process because we have worked with this software at the University of Amsterdam and it is compatible with SPSS.4. Analysis & Results 4.1 Descriptives
After all the data was collected we analyzed the data by using SPSS 23.0. In this research we used data of the independent ratio (numeric) variable: amount of social media platforms, three dependent ratio (numeric) var-iables which measure crowdfunding success: obtaining of the funded aim (in euros), funded total (in euros), number of the amount of backers and two mediating ratio variables: amount of likes, amount of followers. In addition there is one moderating categorical variable: project industry. We also added another dependent (dichotomous) categorical control variable, successful, which states if the funding of a project was successful or not, when the duration of the campaign had expired. This variable is strongly related to crowdfunding success of a project. A project is successful when the funded total is equivalent or greater than the obtaining of the funded aim. Whether a project was successful, the control variable is coded as ‘0’ for No and ‘1’ for Yes. The independent variable amount of social media platforms are coded as ’0’ when none of the platforms are used, ‘1’ where one of the platforms is used, ‘2’ where two of the platforms are used or ‘3’ where al three platforms are used. The variable total amount of likes is the sum of all the Facebook, Twitter (‘retweets’) and Instagram likes a project obtained. The same goes for the variable total amount of followers. In order to per-form the analysis we have assigned numbers for the categorical moderating variable project industry. Project industry was recoded into ‘1’ for Music, ‘2’ for Film & Video, ‘3’ for Publishing, ‘4’ for Art and ‘5’ for Games. We tested the descriptive statistics (Table 1 and 2), normality (Table 3 and 4) for each variable. In addition we executed a correlational analysis for the main variables (Table 5). In addition we made sure that there were no missing values in the data before executing any analysis. A logisctic regression analysis was used to test if we could predict crowdfunding success through both the amount of likes and the amount fol-lowers. To test the mediating effect of the amount of likes and the amount of followers and the moderating effect of project industry we used an SPSS macro of Hayes (2012).
Table 1 Descriptive statistics categorical variables
Variable Levels N %
Project Industry Music 25 20 Film & Video 25 20 Publishing 25 20 Art 25 20 Games 25 20 Successful No 69 55.2 Yes 56 44.8
Table 2 Descriptive statistics numeric variables
N Min. Max. Mean Std. Dev.
Obtaining of the funded aim in euros Funded total in eu-ros Amount of backers Amount of social media platforms Likes Followers 125 125 125 125 125 125 92,37 0 0 0 0 0 70121.21 838739 3918 3 240000 22504 9653.03 12676.33 127.52 1.55 5693.45 1751.46 14302.709 76405.817 413.004 1.074 24505.460 3884.534
Out of the 125 crowdfunding projects we have analyzed, 44.8% of the projects have been funded successful-ly (Table 1). In this research two categorical variables and six numeric variables have been used. The de-scriptive statistics of the categorical variables can de found in Table 1 and the dede-scriptive statistics of the numeric variables can be found in Table 2. Based on Table 2 we see that the minimum obtaining of the fund-ed aim is 93.7 euros in the Art industry, while the maximum obtaining of the fundfund-ed aim is 70121.21 euros in the Film & Video industry. This means that there is a big difference regarding the obtaining of the funded aim between different projects and industries. In addition there is also a big difference regarding the funded total. The highest funded total is 838739.00 euros in the Game industry while four projects in the Music in-dustry received 0 euros. Besides that the mean for the mediating variables shows that there were more likes (5693.45) compared to followers (1751.46).
24
4.2 CorrelationsTo study the relationship between the main variables of interest a correlation analysis is executed. The corre-lations between the main variables are demonstrated by the Pearson r in the correlation matrix (Table 3). A distinction is made between the correlational significance of five percent (*) and the correlational signifi-cance of one percent (**).
Based on Table 3 it can concluded that there is a tendency to a significant positive correlation between the obtaining of the funded aim and the funded total r = .367, p < .01 (p = .000). This means than if the obtaining of the funded aim of a project increases also the funded total has a tendency to increase. In addition there is also a tendency to a significant positive correlation between the obtaining of the funded aim and amount of backers r = .427, p < .01 (p = .000). This means than if the obtaining of the funded aim of a project increases also the amount of backers has a tendency to increase. Furthermore there is a significant high positive corre-lation between the funded total and the amount of backers r = .915, p < .01 (p = .000). This means that if the funded total of a project increases also the amount of backers will increase. Based on these results it can be concluded that the variables, obtaining of the funded aim, funded total and the amount of backers, which measure crowdfunding success are positively correlated with each other.
Table 3 Correlation matrix: Mean, standard deviation and Correlations of Study Variables
M SD 1. 2. 3. 4. 5. 6.
1. Obtained aim Pearson’s r 2. Funded total Pearson’s r 3. Backers Pearson’s r 4. Amount Pearson’s r media platforms 5. Likes Pearson’s r 6. Followers Pearson’s r 9653.04 12676.32 127.52 1.55 5693.45 1751.46 14302.71 76405.82 413 1.07 24505.46 3884.53 1 0.37** 0.43** 0.17 0.06 0.14 0.37** 1 0.92** 0.09 0.03 0.15 0.43** 0.92** 1 0.2* 0.04 0.28** 0.17 0.09 0.2* 1 0.27** 0.42** 0.06 0.03 0.04 0.27** 1 0.53** 0.14 0.15 0.28** 0.42** 0.53** 1
*. Correlation is significant at the 0.05 level (2-tailed). **. Correlation is significant at the 0.01 level (2-tailed).
Based on the results of the correlation analyses and Table 3 we can conclude that the first hypothesis, The amount of social media platforms that are being used are positively related with obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success), can partially be accepted. Based on Table 3 it can be concluded that there is a tendency of a significant positive correlation between the amount of social media platforms and the amount of backers r = .202, p < .05 (p = .024). Which means that when the amount of social media platforms of a project increases, also amount of backers have a tendency to increase. However there was no correlation between the amount of social media platforms and obtaining of the funded aim p >.05 (p = .059) and the funded total p > .05 (p = .306).
As expected there is significant high positive correlation between the amount of likes and the amount of fol-lowers r = .525, p < .01 (p = .000). This means that if the amount of likes of a project increases also the amount of followers will increase. The second hypothesis, the amount of likes and amount followers are, both separate and together, positively related with obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success) can also partially be accepted. We only found a tendency of a positive correlation between the amount of followers and the amount of backers r = .277, p < .01 (p = .002). This means that when the amount of followers increases the amount of backers in-creases. There is no correlation between the amount of followers and the obtaining of the funded aim p >.05 (p = .120) or the funded total p >.05 (p = .103). Furthermore, there is no correlation between the mediating variable amount of likes and dependent variables obtaining of the funded aim p >.05 (p = .519), funded total p >.05 (p = .747) and amount of backers p >.05 (p = .624). There is however a tendency of a positive correla-tion between the amount of likes and the amount of social media platforms r = .265, p < .01 (p = .003). The third hypothesis, the amount of social media platforms used are positively related with the amount of likes and followers, so when project makers are active on more social media platforms they will have a higher total of likes and followers, can be excepted. We found a significant tendency of a positive correla-tion between the amount of social media platforms and the amount of likes r = .265, p < .01 (p = .003) and the amount of followers r = .415, p < .01 (p = .000). So when the amount of social media platforms increases also the amount of likes and amount followers will increase.
26
Table 4 Distribution valuesSkewness Kurtosis
Statistic Std. Error Statistic Std. Error
Obtaining of the funded aim 2.341 .217 5.328 .430
Funded total 10.410 .217 112.665 .430
Amount of backers 6.965 .217 58.752 .430
Amount of social media platforms -.018 .217 -1.251 .430
Amount of likes 7.873 .217 70.372 .430
Amount of followers 3.689 .217 15.112 .430
Project industry .000 .217 -1.304 .430
Successful .212 .217 -1.987 .430
Table 5 Normality test
Kolmogorov-Smirnova Shapiro-Wilk
Statistic df Sig. Statistic df Sig.
Obtaining funded the aim Funded total
Amount of backers
Amount of social media platforms Amount of likes Amount of followers Project industry Successful .264 .434 .379 .192 .408 .326 .159 .367 125 125 125 125 125 125 125 125 .000 .000 .000 .000 .000 .000 .000 .000 .658 .614 .309 .866 .228 .493 .888 .632 125 125 125 125 125 125 125 125 .000 .000 .000 .000 .000 .000 .000 .000 a. Lilliefors Significance Correction
4.3 Normality test
Normality tests were executed to check if the data was normally distributed for all the dependent and inde-pendent variables. Based on the distribution values of Skewness and Kurtosis (see Table 4) we can conclude the three variables that measure crowdfundings: obtaining of the funded aim, funded total and the amount of backers are all positively skewed (2.341; 10410; 6965) which means the distribution has too much low val-ues. In addition the Kurtosis for these three variables were also positive (5.328; 112.665; 58.752), which en-tails that the distribution is heavy tailed. Both the values for the Skewness and the Kurtosis are not even close to zero, which means that the variables are probably not normally distributed. With that in mind we also executed a normally test. The Kolmogorov-Smirnova test is as expected significant p < .05 (p = .000) and the Shapiro-Wilk test as well p < .05 (p = .000) (see results in Table 5), which entails that the three de-pendent variables that measure crowdfunding success are not normally distributed. We also did normality tests for the variables amount of likes and amount of followers. However these variables were also positively Skewed (7.873; 3.689) and had positive values for the Kurtosis (70.372; 15.112), which means the distribu-tion has also too much low values. Based on the results of Shapiro-Wilk test both variables were also signifi-cant p < .05 (p = .000), which means the amount of likes and followers were not normally distributed. For all these data we expect that the data is not normally distributed through extreme values with the ratios.
Since the three variables that measure crowdfunding success were not normally distributed we use the binary dichtomous categorical variable successful (‘0’ No and ‘1’ Yes) to measure crowdfunding success. The out-come variable for crowdfunding success is now categorical instead of numeric. Therefore we have changed, Hypothesis 2a: The amount of likes are positively related with the obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success) and hypothesis 2b: The amount of followers are positively related with the obtaining of the funded aim, the funded total and the amount of backers that invest in the project financially (crowdfunding success), to measure crowdfundig success with the variable succesfull, instead of obtaining of the funded aim, the funded total and the amount of backers because these variables were not normally distributed. With that in mind we used a logistic re-gression analysis to test the second hypothesis (a and b), because the assumption of normality is not neces-sary for the independent variables: amount of likes and amount of followers to execute this analysis.
28
4.4 Logistic RegressionBased on the results of the correlational analysis we concluded that the amount of social media platforms only has a positive effect on the amount of backers, which measures crowdfunding success. However, the crowdfunding success variables were not normally distributed, that is why we used the dichotomous vari-able successful to measure crowdfunding success. To test the second hypothesis and get a better under-standing how crowdfunding success can be predicted through the amount of likes (2a) and the amount followers (2b) separately, a hierarchal logistic regression analysis is conducted (regression model in Table 6). Hypothesis 2a is: The amount of likes are positively related with crowdfunding success. To test this hypothesis we have tested the effect of the amount of likes on crowdfunding success (successful). Using the hierarchical regression method we first entered the control variable project industry in the first model and after that the important predictor variables: amount of likes (2a) and amount of followers (2b) in the second model.
Table 6 Regression
Model 1 Model 2
B Wald Sig. B Wald Sig
(Constant) Project Industry Amount of likes Amount of followers .131 -.114 .096 .790 .756 .374 -.114 -.104 .000 .000 .066 .630 1.090 3.898 .797 .427 .296 .048
Omnibus Test of Model Coefficients X² .0794 df 1 Sig. .373 X² 6.238 df 2 Sig. .044 Nagelkerke R square .008 .073
Table 7 Classification table Observed Predicted Successful? Percentage correct Step 1 Successful ? No Yes No Yes 59 10 85.5 38 18 32.1 Overall percentage 61.6
a. The cut value is ,500
The results of the logistic regression are presented in table 7. Based on the Omnibus Tests of Model coef-ficients and the X² we were able to see to what extent the data fitted the logistic regression test. Based on the results of the first model in Table 6 we can conclude that there are no significant results P > .05 (P = .373). The second model, with a X² = 6.238, on the other hand is significant P > .05 (P = .044), which means that the second model, with the variables the amount of likes and the amount of followers has a better fit with the data than the first model. Table 7 shows us that the second model predicts overall 61.6% correctly. The second model also predicts project that are not successful more precise (85.5%) than projects that are successful (32.1%). To test significance we used the Wald statistic. As table 6 shows on-ly the amount of followers had a significant effect p < .05 (p = .048), which means the amount of follow-ers is a predictor for being successful (crowdfunding success). The Nagelkerke R square of the second model is .073, which entails that 7.3% of the variance of being successful (crowdfunding success) is ex-plained by the amount of followers. Based on these results we can conclude that there is a relation be-tween the amount of followers and crowdfunding success. However there is no relation bebe-tween the amount of likes or project industry and crowdfunding success. This means that Hypothesis 2a can be re-jected, while hypothesis 2b can be expected.
30
4.5 Multiple MediationTo test the fourth hypothesis, The positive relationship between the amount of social media platforms on the obtaining of the funded aim, funded total and the amount of backers (crowdfunding success) is mediated by the amount of likes and followers on social media (so crowdfunding success will increase if the amount of likes and followers is higher), we we used a SPSS macro by Hayes (2012) and did a multiple indirect effect: serial mediation analysis. In this research crowdfunding success is based on three variables: obtaining of the funded aim, funded total and the amount of backers. To test the mediating effect of the amount of likes and amount followers on crowdfunding success we tested the possible mediating effect for each of the crowd-funding success variables.
The first crowdfunding success variable, to test mediation, was the obtaining of the funded aim. The direct effect of the amount of social media platforms is not significant. c’ = 1819.269, t (121)= 1.389 , p > 0.01. Based on the results from the regression analyses, shown in Figure 1, the amount of social media platforms is not directly related with the obtaining of the funded aim. Additionally, there are three indirect effects. The first indirect effect is the specific indirect effect of the amount of social media platforms on the amount of likes. This effect shows that when a project is engaged in to a higher amount of social media platforms, there will be a significant increase in the amount of likes (a1 = 6055.19, p < 0.01), which is however not signifi-cantly associated with the obtaining of the funded aim (b1 = -.02, p =0.78), independently of the amount of followers. The indirect effect can be interpreted as not significantly positive because the bootstrap confi-dence interval is also below zero (indirect effect = -105.984, SE = 633,79, CI: -1319.887 to 1263.789). The second indirect effect is the effect of the amount of social media platforms on the obtaining of the funded aim, through the amount of likes and followers in serial. When a project is engaged in a higher amount of social media platforms the amount of likes increased significantly, which is also related to a higher amount of followers (a3 = 0.07, p < 0.01), this increase in the amount of followers can however not be translated in to a higher obtaining of the funded aim (b2 = 0.36, p = 0.38). This specific indirect effect is not significant (indirect effect=156.064, SE = 313.43, CI: -282.043 to 1087.299). The third indirect effect shows the specif-ic effect of the amount of social media platforms on the obtaining of the funded aim, through the amount of followers. When a project is engaged in a higher amount of social media platforms the amount of followers will significantly increase (a2 = 1072.06, p <
