The role of project owners' and potential backers' implicit social ties in crowdfunding project success

Abstract

Purpose

In this study, the authors sought to investigate how the implicit social ties of both project owners and potential backers are associated with crowdfunding project success.

Design/methodology/approach

Drawing on social ties theory and factors that affect crowdfunding success, in this research, the authors developed a model to study how project owners' and potential backers' implicit social ties are associated with crowdfunding projects' degrees of success. The proposed model was empirically tested with crowdfunding data collected from Kickstarter and social media data collected from Twitter. The authors performed the test using an ordinary least squares (OLS) regression model with fixed effects.

Findings

The authors found that project owners' implicit social ties (specifically, their social media activities, degree centrality and betweenness centrality) are significantly and positively associated with crowdfunding projects' degrees of success. Meanwhile, potential project backers' implicit social ties (their social media activities and degree centrality) are negatively associated with crowdfunding projects' degrees of success. The authors also found that project size moderates the effects of project owners' social media activities on projects' degrees of success.

Originality/value

This work contributes to the literature on crowdfunding by investigating how the implicit social ties of both potential backers and project owners on social media are associated with crowdfunding project success. This study extends the previous research on social ties' roles in explaining crowdfunding project success by including implicit social ties, while the literature explored only explicit social ties.

Citation

Gupta, J.P., Li, H., Kärkkäinen, H. and Mukkamala, R.R. (2024), "The role of project owners' and potential backers' implicit social ties in crowdfunding project success", Internet Research, Vol. 34 No. 7, pp. 1-23. https://doi.org/10.1108/INTR-07-2021-0424

1. Introduction

Over the past decade, crowdfunding has become popular, attracting the attention of academics and practitioners. Crowdfunding can be defined as a type of crowdsourcing that enables entrepreneurs of all types social, cultural, artistic, or for-profit to raise capital from a crowd so that they can pursue new ventures or causes (Hong et al., 2018; Mollick, 2014). Yet, despite the crowdfunding market's overall rise, some crowdfunding projects cannot achieve their fundraising goals successfully (Mollick, 2014; Wang et al., 2018). For example, on Kickstarter, one of the world's largest crowdfunding platforms, 40.35% of total projects have achieved their fundraising goals as of January 2023 (Kickstarter.com, 2023). Accordingly, researchers have taken a significant interest in understanding crowdfunding projects' success, increasing the number of studies that explore the various factors associated with crowdfunding project success.

Most popular crowdfunding platforms, such as Kickstarter and Indiegogo, enable users to share crowdfunding campaign pages on social media platforms (Thies et al., 2016). Social media's popularity has also attracted scholars' attention. Some studies have investigated the link between social media and crowdfunding project success. In some studies, researchers have found that social media use plays an important role in crowdfunding project success (Efrat and Gilboa, 2020; Lu et al., 2014; Mollick, 2014; Saxton and Wang, 2014). For instance, campaigns on social media could help crowdfunding projects establish large social media footprints and tap into project owners' and potential backers' social media networks to secure crowdfunding (Hong et al., 2018; Mollick, 2014).

Any crowdfunding transaction involves three types of actors: the crowdfunding project's creator (the project owner), the person who funds the project (the project backer) and the crowdfunding platform itself (Belleflamme et al., 2014; Madrazo-Lemarroy et al., 2019; Mollick, 2014; Schwienbacher, 2018). In prior research on crowdfunding project success, scholars have mainly investigated factors related to project owners or crowdfunding platforms (Agrawal et al., 2015; Belleflamme et al., 2014; Kim and Zhang, 2017; Koch and Cheng, 2016; Madrazo-Lemarroy et al., 2019; Mollick, 2014). Some recent studies have explored project backers' motives for contributing to crowdfunding projects and project backers' roles in crowdfunding success (Clauss et al., 2018; Efrat and Gilboa, 2020; Tan and Reddy, 2021). In their literature review, Cai et al. (2021) also highlighted the roles of both project owners and backers in crowdfunding project success. They found that external and internal social capital across the structural, relational and cognitive dimensions affects crowdfunding campaigns, such as the social capital of project owners and backers. However, prior studies have rarely empirically validated both project owners' and backers' different roles in crowdfunding project success.

In earlier research, scholars have also found that social ties are positively associated with crowdfunding project success (Borst et al., 2018; Madrazo-Lemarroy et al., 2019) particularly project owners' social ties (Agrawal et al., 2015; Colombo et al., 2015; Kim and Zhang, 2017; Koch and Cheng, 2016; Lu et al., 2014; Zheng et al., 2014). However, such researchers have focused on using social media data related to explicit social ties. Explicit social ties form on social media platforms when users explicitly add other individuals to their networks (Reafee et al., 2016). For example, on Facebook, users can connect with other users via the “friend” functionality. Meanwhile, on Twitter, users can specifically follow other users via the “follow user” function.

In some studies, researchers have argued that different kinds of social media ties cannot be measured adequately using factors related only to explicit social ties (e.g. total Twitter followers and total Facebook friends). Such scholars have suggested that social ties' roles in crowdfunding project success should also be examined through social media data that are not related to explicit ties, such as interactions (Borst et al., 2018; Tosatto et al., 2022). Implicit social ties are all the social connections derived from or represented by the data that a social media platform provides, excluding data related to explicit relationships. A variety of social media data such as interactions, user profiles and user metadata can be used to identify implicit social ties (Zhou et al., 2014). Some recent studies have highlighted the implicit social ties' positive roles in various contexts, such as improving social recommendations' accuracy, designing better social recommendation systems and building trust (Ahmadian et al., 2020; Li et al., 2018; Weng et al., 2021). Most studies on social ties in crowdfunding have been based on explicit social ties. The link between implicit social ties and crowdfunding project success has remained largely unexplored. Specifically, research is lacking on how project backers' and owners' implicit social ties on social media are associated with crowdfunding project success. Accordingly, in the current study, we addressed this research gap by answering the following research question: Are project owners' and potential project backers' implicit ties on social media associated with crowdfunding projects' success?

To answer this question, drawing on social ties theory (Granovetter, 1973), we examined how project owners' and potential project backers' implicit social ties are related to crowdfunding project success with around 173,000 tweets related to 2,161 crowdfunding projects on Kickstarter. We identified implicit social ties from interaction data related to crowdfunding projects on Twitter. Specifically, we proposed that three variables are associated with crowdfunding project success. These variables measure project owners' and potential project backers' implicit ties in social media networks through dyadic connections. These variables were social media activities, degree centrality and betweenness centrality. Degree centrality and betweenness centrality have been suggested as important factors to measure implicit social ties via network analysis, which can measure people's positions and connections in social networks (Borgatti and Halgin, 2011; Marsden and Campbell, 2012). Moreover, we proposed that project fundraising goals (project size) moderate the relationships between implicit social ties and crowdfunding project success.

This study contributes to the crowdfunding literature by extending the research on social ties' association with crowdfunding project success. While previous studies have focused on explicit social ties (Borst et al., 2018; Kim and Zhang, 2017), we include the implicit social ties of both project owners and potential project backers. Our investigation of the relationships between implicit social ties and crowdfunding project success provides new insights into these associations, complementing the literature's previous findings on explicit social ties (Mollick, 2014; Kim and Zhang, 2017; Borst et al., 2018). This study's results show, as we expected, that project owners' implicit social ties (their social media activities, degree centrality and betweenness centrality) are positively associated with crowdfunding projects' degrees of success. However, contrary to our expectations, potential project backers' implicit social ties (their social media activities and degree centrality) are negatively associated with crowdfunding projects' degrees of success.

The remainder of this study is structured as follows. In Section 2, we outline the previous literature related to crowdfunding and social ties' roles in crowdfunding project success. Then, we discuss our proposed research model and hypotheses in Section 3 before presenting our research method in Section 4. Finally, we discuss our findings and conclude by identifying this study's theoretical and practical implications, as well as its limitations and potential avenues for future research.

2. Theoretical background

2.1 Crowdfunding

Over the past decade, crowdfunding has become an essential means of raising capital to carry out projects that were previously impossible. According to Mollick (2014, p. 2), “crowdfunding refers to the efforts by entrepreneurial individuals and groups – cultural, social, and for-profit to fund their ventures by drawing on relatively small contributions from a relatively large number of individuals using the Internet, without standard financial intermediaries.” Crowdfunding platforms can be classified broadly as either equity-based or non-equity-based. Equity-based crowdfunding refers to online crowdfunding platforms through which backers gain equity ownership of crowdfunded projects via an online platform. On the other hand, non-equity-based crowdfunding involves reward-based platforms, donation-based platforms and lending-based platforms through which backers do not acquire any equity in projects (Vulkan et al., 2016). In this study, we focused on a non-equity-based crowdfunding platform. As we mentioned in the previous section, any crowdfunding transaction involves three different actors: crowdfunding platforms, project owners and project backers. In prior research, scholars have also investigated how factors related to each of these actors are associated with crowdfunding project success.

In a stream of the literature, researchers have focused on crowdfunding platforms or intermediaries. Some previous studies have examined how crowdfunding project success is associated with a platform's design, regulations and history (Deng et al., 2022; Josefy et al., 2017; Kaartemo, 2017). Meanwhile, other scholars have examined the associations of crowdfunding platforms' different funding models (Cumming et al., 2020; Paschen, 2017), archetypes (Kromidha and Robson, 2016; Paschen, 2017), funding strategies and designs (Coakley et al., 2022; Konhäusner et al., 2021) and regulatory environments (Hornuf and Schwienbacher, 2017; Klöhn, 2018) with crowdfunding project success. Moreover, some scholars have investigated platforms' roles in explaining crowdfunding project success from other lens, such as reducing information asymmetries (Wang et al., 2021), trust-building (Greiner and Wang, 2010) and the network effect (Thies et al., 2018).

In another stream of the literature, researchers have focused on crowdfunding project owners, mainly investigating owners' roles in explaining crowdfunding project success. Some of these scholars have examined how project owners' social networks (Agrawal et al., 2015; Moritz and Block, 2016) and geographical proximity (Saxton and Wang, 2014) are related to crowdfunding project success. Additionally, some researchers have investigated how project owners' backgrounds, project owners' emotional and cultural factors (Burtch et al., 2014; Lin and Viswanathan, 2016) and interactions between project owners and backers (Clauss et al., 2018; Efrat and Gilboa, 2020) are associated with crowdfunding project success.

Finally, in another stream of the literature, scholars have examined project backers' role in crowdfunding project success. Some of these researchers have investigated backers' motivations for supporting crowdfunding projects (Baber and Fanea-Ivanovici, 2023; Bretschneider and Leimeister, 2017; Bürger and Kleinert, 2021; Lin and Boh, 2020). Meanwhile, some scholars have studied backers' crowdfunding-related behaviors, such as herding behavior (Liu et al., 2015; Kim et al., 2020; Saxton and Wang, 2014) and funding decisions (Gleasure and Feller, 2018; Kromidha and Robson, 2016; Lin and Viswanathan, 2016). Some scholars have also examined the association between project backers and crowdfunding project success by focusing on backers' affiliations (Herd et al., 2021), trust (Rodriguez-Ricardo et al., 2019; Shneor et al., 2022), interactions with project owners (Clauss et al., 2018; Efrat et al., 2021), influence (Tan and Reddy, 2021) and social networks (Chung et al., 2021).

The prior literature has shown that crowdfunding project success is associated with various factors related to the three different actors involved in crowdfunding (Efrat and Gilboa, 2020; Deng et al., 2022; Baber and Fanea-Ivanovici, 2023; Coakley et al., 2022). However, the authors of previous studies have mainly examined factors related to platforms, project owners, or project backers (Klöhn, 2018; Clauss et al., 2018; Lin and Boh, 2020). Few researchers have yet examined crowdfunding project success through factors related to both project owners and backers, though such an approach could explain their different roles in depth. Therefore, such an investigation that incorporated both project owners and project backers was needed. This investigation will help crowdfunding project owners understand both their own and project backers' different roles in achieving their fundraising goals.

2.2 Social ties in crowdfunding

Granovetter (1973) introduced the concept of tie strength and different kinds of social ties in a seminal study titled “The Strength of Weak Ties.” He defined tie strength as “a (probably linear) combination of the amount of time, the emotional intensity, the intimacy (mutual confiding) and the reciprocal services which characterize the tie” (p. 1361). Tie strength connects micro-level interactions with macro-level patterns in two individuals' dyadic relationships (Granovetter, 1973). In other words, tie strength explains the degree of closeness between two individuals. In the literature, many different measures and proxies have been developed and used to calculate tie strength, such as communication frequency, reciprocity, mutual friends, communication recentness, interaction frequency and network topology (Aral and Walker, 2014; Fogues et al., 2018; Gilbert and Karahalios, 2009; Marsden and Campbell, 1984, 2012; Onnela et al., 2007).

Moreover, Granovetter (1973) characterized two kinds of social ties, based on the tie strength concept: strong ties and weak ties. Generally, the term strong ties refers to trusted people whose social circles tightly overlap with one's own social circle. Family members and close friends are common examples of this type. Strong ties provide emotional support, and they are stable and more reliable than weak ties (Gilbert and Karahalios, 2009; Granovetter, 1973). On the other hand, the term weak ties refers to people with whom one is merely acquainted or with whom one distantly and infrequently interacts. In many cases, weak ties provide access to novel or non-redundant information, and they can help diffuse new ideas or new knowledge (Aral and Walker, 2014; Burt, 2004; Granovetter, 1973; Levin and Cross, 2004; Shi et al., 2014).

Over the past decade, social media's rise and proliferation in individuals' daily lives have provided new ways to manage and establish social relationships or social ties (Ahn and Park, 2015). This development has provided a new data source for the development of methods that measure tie strength in the online environment and identify different kinds of ties from social media data. Various models have been developed to identify different types of ties using such data. Many of these studies' authors have extensively used explicit relationship data from social media (such as Facebook friends) to develop such methods (Fogues et al., 2018; Gilbert and Karahalios, 2009; Huang et al., 2015; Jones et al., 2013).

Researchers have also examined how social media ties are associated with crowdfunding project success (Borst et al., 2018; Kim and Zhang, 2017; Madrazo-Lemarroy et al., 2019). Digital platforms (including social media platforms and crowdfunding platforms) allow participants to reinforce both weak and strong ties in crowdfunding. On the one hand, these platforms strengthen existing social ties. On the other hand, they help develop new social ties as a campaign evolves. Thus, crowdfunding projects can leverage digital platforms to reach crowdfunding goals (Granovetter, 1973; Madrazo-Lemarroy et al., 2019). Additionally, researchers have found that project owners' ties on social media play important roles in shaping their crowdfunding projects' success (Borst et al., 2018; Kim and Zhang, 2017; Madrazo-Lemarroy et al., 2019). For instance, some scholars have examined how project owners' social ties such as the presence or number of social media contacts are positively associated with crowdfunding project success (Colombo et al., 2015; Mollick, 2014). Researchers have also found that the background of project owners and the closeness of project owners' social ties are positively associated with achieving fundraising goals and the timely project funding (Agrawal et al., 2015; English, 2014). Some scholars have examined how project owners' social ties are associated with early project backers and crowdfunding success (Ordanini et al., 2011; Shneor and Vik, 2020). Based on data collected from Facebook and Kickstarter, Jin et al. (2020) examined the temporal association between project owners' Facebook activities and crowdfunding project success. These authors found a J-curved relationship between owners' Facebook activities and project success, and they also identified a herding effect during a project's closing period.

Researchers have also explored project backers' different motives for supporting crowdfunding projects, as well as these motives' associations with crowdfunding project success (Tan and Reddy, 2021; Chung et al., 2021). However, scholars have published very little research examining how project backers' ties on social media are associated with crowdfunding project success. In a recent study, Tan and Reddy (2021) attempted to examine backers' associations with crowdfunding project success through their affiliation networks. These authors found that backers with central positions in these networks were positively associated with various crowdfunding project outcomes—including success rates, goal attainment speed and funds raised. However, Tan and Reddy focused on social ties based on backers' affiliations, rather than social media data. Meanwhile, Chung et al. (2021) investigated how backers' social media networks are associated with their backing decisions. However, these researchers did not consider the networks' associations with crowdfunding project success.

Additionally, although researchers have examined social ties' roles in crowdfunding project success, most have focused on explicit social ties (Kim and Zhang, 2017; Madrazo-Lemarroy et al., 2019). To the best of our knowledge, no scholars have attempted to examine the relationships between project owners' and backers' implicit social ties and crowdfunding projects' degrees of success.

3. Research model and hypothesis development

3.1 Research model

Based on the previous literature concerning social ties' roles in crowdfunding project success, as well as social ties theory, we developed a research model to examine the links between degrees of crowdfunding project success and the implicit social media ties of both project owners and potential project backers. Specifically, we assumed that owners' and potential backers' implicit ties on social media – represented by social media activities, degree centrality and betweenness centrality – would be associated with crowdfunding projects' degrees of success.

Degree centrality and betweenness centrality are two factors related to network structure. They can be calculated from a social network and used to identify different kinds of social ties in that network (Borgatti and Halgin, 2011; Marsden and Campbell, 2012). The authors of some network-related studies have used explicit relationship-related social media data (such as Facebook friends or Twitter followers) to directly create social networks (Mollick, 2014; Kim and Zhang, 2017). However, this approach cannot yet reflect implicit ties on social media. In the current study, we based project owners' and potential backers' degree centrality and betweenness centrality on their social media interactions concerning crowdfunding projects (Zhou et al., 2014).

Additionally, the authors of previous studies have found that crowdfunding project size (specifically, projects' fundraising goals) is associated with backers' decisions and projects' success (Jin et al., 2020; Mollick, 2014). For example, projects with higher fundraising goals might need more backers than projects with lower fundraising goals. Accordingly, in this study, we assumed that project size (large or small, based on fundraising goals) would moderate the relationships between owners' and potential backers' implicit social ties (i.e. social media activity, degree centrality and betweenness centrality) and degrees of crowdfunding success. Fundraising goals, the number of tweets about a project, and the total followers of both project owners and potential project backers have been associated with crowdfunding project success in the literature (Hong et al., 2018; Kim and Zhang, 2017; Jin et al., 2020). Therefore, we used these factors as control variables in the current study. Figure 1 presents our research model.

3.2 Hypotheses

Social media use has led to a new kind of user behavior: the ability to reiterate a friend's activity, that is, to replicate and redistribute content (e.g. text, videos, or pictures) that a friend has posted online (Geva et al., 2019). This behavior has been shown to influence other social media users' behaviors by providing an effective way to share information with other users (Fischer and Reuber, 2011; Lynn et al., 2020). Social media platforms offer various redistribution features. For example, Facebook and LinkedIn allow “sharing,” while Twitter uses “retweets.”

In previous studies, researchers have also investigated how social media activities are associated with effective personal branding strategies, social broadcasting and social commerce (Geva et al., 2019; Lee et al., 2015; Shi et al., 2014). In the crowdfunding context, some scholars have also found that project owners' social media activities are positively associated with crowdfunding project success (Liu and Ding, 2020; Thies et al., 2018). As Jin et al. (2020) explained, when project owners share a crowdfunding project on social media, their posts are publicly visible on their timelines to potential backers. Potential backers can also observe other social media users' activities in relation to a crowdfunding project, which encourages potential backers to fund the project. Potential backers can express approval for the project and share related posts publicly on social media, which may allow the project to attract more backers via information spillover effects (Tan and Reddy, 2021). Accordingly, based on findings in the literature, we hypothesized that project owners' and potential backers' social media activities are positively associated with crowdfunding projects' degrees of success.

H1a.

Crowdfunding project owners' social media activities related to their projects are positively associated with their projects' degrees of success.

H1b.

Potential backers' social media activities related to crowdfunding projects are positively associated with these projects' degrees of success.

The research on social networks has suggested that nodes' relational positions in such networks crucially shape social ties (Borgatti and Halgin, 2011). Centrality has been widely used as a metric to capture nodes' relational properties in a network, and it includes degree centrality and betweenness centrality (Chen et al., 2012; Freeman, 1978). According to graph theory, degree centrality counts a node's neighbors in a network. This measure is useful for analyzing which individuals are likely to have the most information or be able to quickly connect with a wider network (Hansen et al., 2020). On the other hand, betweenness centrality measures centrality on a graph, based on the shortest paths between nodes. In other words, it measures the number of times a node lies on the shortest path between other nodes in a network. Betweenness centrality is useful for analyzing a network's communication dynamics, and it helps identify individuals who influence a network's information flow (Hansen et al., 2020). On social media, for instance, a user with high degree centrality and betweenness centrality could access many disparate groups and influence their network's information flow. Studies have shown that users with high betweenness centrality have influential opinions and become thought leaders on social media, and they have been associated with information flow in contexts such as destination marketing (Bokunewicz and Shulman, 2017; Jin and Cheng, 2020), promoting niche products (Phang et al., 2013) and political activism (Xu et al., 2014). Researchers have also found that social media users with high degree centrality have positive relationships with some outcomes, such as the dissemination of health information (Meng et al., 2018) and brands' purchasing decisions (Britt et al., 2020).

Recent studies by Chung et al. (2021) found that the degree centrality and betweenness centrality of backers' social networks are positively related to backers' pledge decisions regarding crowdfunding projects. Tan and Reddy (2021) also found that the centrality of a backer's affiliation network is positively associated with crowdfunding project success. Therefore, in crowdfunding contexts, project owners' and potential backers' degree centrality and betweenness centrality on social media could be positively associated with a project's information flow and reach. Thus, their degree centrality and betweenness centrality would attract more potential backers and help reach fundraising goals. Accordingly, we proposed that both project owners' and potential backers' degree centrality and betweenness centrality are positively associated with crowdfunding projects' degrees of success.

H2a.

Project owners' degree centrality is positively associated with their crowdfunding projects' degrees of success.

H2b.

Potential backers' (average) degree centrality is positively associated with crowdfunding projects' degrees of success.

H3a.

Project owners' betweenness centrality is positively associated with their crowdfunding projects' degrees of success.

H3b.

Potential backers' (average) betweenness centrality is positively associated with crowdfunding projects' degrees of success.

Previous studies have also found that crowdfunding projects' fundraising goals (project size) can influence potential backers' expectations and funding decisions (Mollick, 2014; Tan and Reddy, 2021; Zheng et al., 2014). To succeed, projects with higher fundraising goals (large projects) require funding from a much larger social network than small projects. Additionally, as Tan and Reddy (2021) argued, smaller crowdfunding projects' outcomes are more certain since their smaller fundraising goals are easier to achieve than large projects' goals. Therefore, we hypothesized that project owners' and potential backers' social media activities, degree centrality and betweenness centrality are more strongly associated with degrees of success for crowdfunding projects with lower fundraising goals than for projects with higher fundraising goals.

H4a.

Project owners' activities, degree centrality and betweenness centrality on social media are more strongly associated with degrees of success for projects with lower fundraising goals than for projects with higher fundraising goals.

H4b.

Potential project backers' activities, degree centrality and betweenness centrality on social media are more strongly associated with degrees of success for projects with lower fundraising goals than for projects with higher fundraising goals.

4. Research method

4.1 Data collection

This study's data were collected from two sources: the social media platform Twitter and the all-or-nothing (AON), reward-based crowdfunding platform Kickstarter. In AON crowdfunding, project creators keep no pledged funding if their projects' total fundraising goals are not achieved (Cumming et al., 2020). We selected Twitter and Kickstarter as our data sources because Kickstarter is one of the world's largest reward-based crowdfunding platforms, while Twitter is among the largest social media platforms. Both platforms' global user bases can reach much wider audiences than any local crowdfunding or social media platform.

Twitter data were collected using the Twitter Premium application programing interface (API) and a search query containing keywords related to Kickstarter, such as “#Kickstarter,” “#kickstarter,” and “@kickstarter.” For each relevant tweet, we recorded the time stamp, author handle, any mentioned user handles, the text and any URLs included in the tweet's body or user's metadata. The collected Twitter data had been published between June 2016 and September 2018. Table 1 described this data set in detail.

Kickstarter is one of the largest online crowdfunding platforms in the world that enables organizations or teams to issue fundraising over the Internet and receive small investments from registered funders in return. The platform uses a reward-based AON model (Cumming et al., 2020). Kickstarter projects fall into 15 categories. This study's Kickstarter data were collected via webrobots.io, which uses a scraper robot to crawl all Kickstarter project data and export them in.csv and.json formats. This crawl is conducted once monthly, and the entire data set is available on the webrobots.io website. The data set used in this study contains all Kickstarter project data from April 2009 to September 2018. Collecting all the corresponding data from Twitter for all crowdfunding projects from April 2009 to September 2018 would have required vast resources and an enormous workload. Therefore, from this data set, we selected only crowdfunding projects that had been conducted between June 2016 and September 2018. Of these selected projects, we filtered out projects for which related tweets were unavailable up to six months before a project's start date since crowdfunding project owners must normally use social media campaigns to market their projects before seeking funding. After this filtering, we identified 2,161 crowdfunding projects that satisfied all our inclusion criteria. The basic campaign information in our crowdfunding project data sample included campaign organizer IDs, web page URLs, shortened versions of these URLs, campaign fundraising goals, fundraising durations, funding amounts raised, campaign start dates, campaign end dates and projects' countries of origin.

For researchers, knowing all possible relevant search terms for every crowdfunding project is practically impossible. Therefore, to avoid a potential data collection problem, we used broadly relevant search terms to collect the most relevant data from Twitter related to different crowdfunding projects during our initial data collection from Twitter. To prepare for our analysis, the data was cleaned and organized. Tweets related to the selected 2,161 Kickstarter projects were filtered, and only tweets posted during a crowdfunding project's campaign were included in our analysis. A code script based on Python was used in this data refinement process. Thus, we greatly reduced the initial data collected from Twitter for this study. Based on this Python script, in total, 172,892 tweets related to 2,161 Kickstarter projects were identified and used in this study.

4.2 Variables

In this subsection, we present the different data measures we used to test our hypotheses. Crowdfunding project success can be defined in multiple ways. For instance, it can be understood as a binary value that is, whether a project was successful or unsuccessful (Madrazo-Lemarroy et al., 2019; Mollick, 2014) or as a ratio between a project's fundraising goal and actual funding received (Shneor and Vik, 2020). In this study, we defined crowdfunding project success as the ratio of a project's actual funding to its goal. This definition has been suggested to provide more comprehensive insights into project success than a binary value (Yin et al., 2019).

We operationalized social media activities using Twitter's retweet function. We calculated degree centrality and betweenness centrality in an implicit social network using Twitter interactions between potential backers and project owners concerning a crowdfunding project. Specifically, we identified a network of potential backers and project owners based on their mentioning each other in tweets. The network's nodes were project owners and potential backers. Meanwhile, the network's edges were defined by the number of interactions between these different nodes. This kind of network construction has been used in many previous studies to analyze various phenomena (Aramo-Immonen et al., 2015, 2016). We created such networks for each crowdfunding project analyzed in this study.

We used three measures of project owners' and potential backers' implicit social ties as independent variables associated with a crowdfunding project's success. For example, a project that reached 100% of its fundraising goal would have a degree of success equal to 1. Meanwhile, a project that reached 50% of its goal would have a degree of success equal to 0.5. We used project size based on the median value of all crowdfunding projects' fundraising goals as a moderator. As control variables, we used project owners' and potential backers' total Twitter followers, fundraising goals and tweets about each crowdfunding project. Fixed effects for project categories were added to our models to capture the unobserved, constant heterogeneity within each crowdfunding project category. Table 2 describes all these variables.

4.3 Data analysis

Table 3 presents descriptive statistics for all measures used in this study except the project category and project size variables. The latter two variables were excluded because project category is a categorical variable and project size is a dummy variable. This table depicts the data measures' maximums, minimums, means and standard deviations.

A correlation matrix was calculated for the different measures used in this study's hypothesis testing. This matrix is depicted in Table 4.

Our data's skewness and kurtosis exceeded the respective recommended thresholds of 3 and 7 for normal distribution (Kline, 2015). Therefore, log transformation was performed for all independent, dependent and control variables except for project category and project size before our ordinary least squares (OLS) regression models were created. We used the following formula for our data analysis:

LogDCPS =β0 +Project category effect +β1(LogPBSM) +β2(LogPBDC)+β3(LogPBBC)+β4 (LogPOSM) +β5(LogPODC) +β6(LogPOBC)+β7(LogPBTF) +β8(LogPOTF) +β9(LogFundraisingGoal)+β10(PS)+β11(LogNoofTweets) +β12(LogPBSM * PS) +β13(LogPBDC * PS)+β14(LogPBBC * PS) +β15(LogPOSM * PS) +β16(LogPODC * PS)+β17(LogPOBC * PS)

To test our research instrument's multicollinearity, we calculated maximum variance inflation factors (VIFs). All VIFs were below the cutoff value of 10 for regression models (the maximum VIF value was 8.99). This finding indicates that multicollinearity is not a concern for this study (James et al., 2013).

4.4 Results

We estimated a series of OLS regressions to test our hypotheses (see Table 5). First, we tested the relationships between potential project backers' social media activities, degree centrality and betweenness centrality and crowdfunding projects' success (see Model 1 in Table 5). The results of this test showed that potential backers' social media activities are negatively associated with projects' success, while degree centrality and betweenness centrality are not significantly associated with projects' degrees of success. Therefore, H1b, H2b and H3b were not supported.

Model 3 tested the associations between project owners' social media activities, degree centrality and betweenness centrality and projects' crowdfunding success. This model showed that project owners' social media activities (β = 0.101; p < 0.001), degree centrality (β = 0.106; p < 0.01) and betweenness centrality (β = 0.262; p < 0.05) are significantly positively associated with crowdfunding projects' success. Hence, H1a, H2a and H3a were supported.

Model 5 included all the independent variables related to both project owners and project backers. This model's test results supported H1a and H2a but did not support H1b, H2b, H3a, or H3b.

Next, we tested the interaction effect between project size and potential backers' and project owners' social media activities, degree centrality and betweenness centrality on projects' success (see Model 2 and Model 4). The results of this test showed no significant interaction effect between project size and potential backers' social media activities, degree centrality, or betweenness centrality. Therefore, H4b was not supported (see Model 2). Moreover, the results of Model 4 showed no significant interaction effect between project size and project owners' degree centrality or betweenness centrality; however, the interaction between project size and project owners' social media activities has a significant negative effect on projects' success. This finding demonstrates that, for crowdfunding projects with lower fundraising goals, project owners' social media activities are more strongly associated with project success than they are for larger projects. Thus, H4a was partially supported.

Model 6 included the interaction between project size and both project owners and potential backers, based on Model 5. The results of Model 6 showed that project owners' social media activities and degree centrality are significantly associated with crowdfunding projects' success, which further supported H1a and H2a. Significant negative relationships were observed between potential backers' social media activities, degree centrality and project success, while no significant relationship was observed between potential backers' betweenness centrality and project success. These findings provided further evidence that does not support H1b, H2b, or H3b. The results of Model 6 show that project size significantly and negatively moderates the relationships between project owners' social media activities and project success, but project size does not significantly moderate the other relationships we examined. Thus, H4a was partially supported, while H4b was not supported. The proposed research model explains 25.7% of crowdfunding project success.

4.5 Robustness check

We also ran a robustness check on our research model. We used binary values for the dependent variable via logit regression models, defining a crowdfunding project as either a success or a failure. Table 6 depicts these robustness checks' results. Projects that achieved their fundraising goals were defined as successes, whereas projects that did not achieve their fundraising goals were defined as failures. Our fixed-effect logit regression model's results concerning the relationships between potential backers' and project owners' social media activities, degree centrality, betweenness centrality and crowdfunding project success were consistent with the results of our OLS regression model. However, we observed no significant interaction effects of the independent variables and project size on the dependent variable crowdfunding project success.

5. Discussion

In this study, we explored how project owners' and potential backers' implicit social ties are associated with crowdfunding projects' degrees of success. Our findings show that project owners' degree centrality is significantly associated with crowdfunding project success. Degree centrality is a simple count of a node's total number of linked connections in a network. This count helps determine which highly connected individuals are likely to have the most information or the ability to quickly connect with a wider network (Borgatti et al., 2013). In the Twitter context, degree centrality captures users' engagement with other users and their content. Users with high degree centrality act as conversational hubs on the platform (Hansen et al., 2020). A project owner's high degree centrality can enable crowdfunding project information to spread across a wider network, which could help attract more backers and funding.

Our results also show that project owners' betweenness centrality is significantly associated with crowdfunding project success. Betweenness centrality measures how often a given node falls along the shortest path between two other nodes (Borgatti et al., 2013), and it highlights the potential to control flows through a network – i.e. to play a gatekeeping or toll-taking role. Network nodes with high betweenness centrality can filter, color, or distort information when passing it along. At the same time, the ability to exploit a position with high betweenness centrality varies inversely with nodes' ability to create new social ties (Borgatti et al., 2013). On Twitter, a user's high betweenness centrality could indicate the ability to access users from other disparate network clusters or just the user's presence at both clusters' peripheries (Hansen et al., 2020). Hence, project owners' high betweenness centrality might enable them to control information flows and spread information across different groups. This ability could result in an overall information flow to more disparate networks, potentially increasing owners' chances of attracting more backers and receiving more funding.

This study also found that project owners' retweet counts are significantly associated with crowdfunding project success. Retweets redistribute content across an individual's social network. A higher retweet count indicates that a project owner has distributed more messages related to their crowdfunding project to many Twitter users. Thus, a high count can help attract more potential backers. This finding is consistent with previous studies' findings that crowdfunding project owners' retweets can cause messages to cascade to larger audiences; in many cases, this effect increases interest and trust in a project, thus increasing funding (Liu and Ding, 2020; Liu et al., 2021).

Contrary to our expectations, this study's results show that potential backers' retweet counts and degree centrality are significantly but negatively associated with crowdfunding project success. Our results align with the findings of Polzin et al. (2018). They observed that, when potential backers make funding decisions for crowdfunding projects, they likely rely on project owners' information, rather than information provided by other potential backers, due to such projects' potential risks and uncertainties. Additionally, when potential backers with high degree centrality act as conversational hubs by tweeting about a crowdfunding project and garnering more retweets, other Twitter users may suspect these potential backers to work as marketers for the project. This suspicion may lead to lower trust in these potential project backers and fewer contributions to the project. At the same time, potential backers with high betweenness centrality have no significant effect on crowdfunding project success. High betweenness centrality can offer access to disparate user groups. However, this access may not necessarily lead to new information about a crowdfunding project campaign since project owners are the primary source of project-related information. Therefore, other potential backers on Twitter might receive the same information about a project from different users with high betweenness centrality and pay little attention to this information. Therefore, their crowdfunding decisions would not be associated with crowdfunding project success.

We also found that project size negatively moderates the relationship between project owners' social media activities and project success. Owners' social media activities such as retweets are much more strongly associated with crowdfunding success for projects with lower fundraising goals than for projects with higher fundraising goals. The reason for this finding might be that projects with higher fundraising goals experience much more struggle to achieve these goals than projects with lower goals. Project owners' social media activities greatly help small projects achieve their lower fundraising goals, but these activities cannot help large projects in the same way. Project size's insignificant moderating effect on our other hypotheses might indicate that project information is the most important variable for potential backers' funding decisions, rather than project size.

6. Contributions and limitations

6.1 Theoretical contributions

This study contributes to the literature in the following ways. First, it enriches the literature on implicit social ties' roles in determining crowdfunding project success by assessing both project owners' and potential backers' implicit social ties, based on social media data. This study's results show that project owners' implicit social ties (social media activities, degree centrality and betweenness centrality) are positively associated with crowdfunding project success, whereas potential backers' implicit social ties are mainly negatively associated with such success. Specifically, potential backers' social media activities are negatively associated with crowdfunding project success, while potential backers' degree centrality and betweenness centrality are not significantly associated with this success. Previous studies have mainly investigated explicit social ties' associations with crowdfunding project success (Colombo et al., 2015; Mollick, 2014; Shneor and Vik, 2020; Kim and Koh, 2023; Tosatto et al., 2022). However, our findings extend the research on social ties' roles in determining crowdfunding project success by investigating the implicit social ties of both project owners and project backers.

Second, in this study, we used network-related measures (degree centrality and betweenness centrality) that were calculated based on social media interactions (i.e. tweets) about crowdfunding projects. These measures enabled us to deeply explain how implicit social ties are associated with crowdfunding project success through individuals' (i.e. project owners' and backers') network positions on social media. Both centrality measures (degree centrality and betweenness centrality) were based on nodes' positions in a network (Borgatti et al., 2013). By using these centrality measures to explain implicit social ties' roles in determining crowdfunding project success, we answered a call to understand the roles of individuals' network positions in crowdfunding research (Hong et al., 2018). Thus, in this study, we have partially addressed a research gap and provided evidence that the network positions of both project owners and potential backers importantly affect crowdfunding outcomes. While project owners' positions positively affect crowdfunding project success, potential backers' positions negatively affect such success.

Finally, to analyze crowdfunding project success, we constructed implicit social networks using social media interactions between project owners and potential backers as this study's data. Researchers have previously examined the link between social ties and crowdfunding project success using mainly explicit social media ties, such as the presence or number of contacts and likes (Colombo et al., 2015; Jin et al., 2020; Mollick, 2014). The current study contributes to the literature on project owners and potential backers by using social media interaction data to measure implicit social ties on social media to explain crowdfunding project success.

6.2 Practical implications

Our study's results have some important practical implications. First, our findings show that project owners' degree centrality and betweenness centrality are positively associated with crowdfunding project success. This association indicates that project owners should engage with other users on social media as much as possible to promote their crowdfunding projects, especially by directly interacting with other users. Additionally, project owners should strive to engage with different user groups (for example, networks based on different locations or interests). These direct interactions with other social media users will likely increase project funding.

Second, our findings show that project owners' social media activities are positively associated with projects' success. Practically, this finding indicates that project owners should strive to use social media platforms' information-sharing functionality (such as shares on Facebook and retweets on Twitter) as much as possible since their social media activity effectively attracts funding during crowdfunding campaigns. However, we found that project size has a negative moderating effect on the relationship between project owners' social media activities and project success. This finding indicates that project owners should consider their fundraising goals when using social media for a crowdfunding campaign.

Finally, our findings show that potential backers' social media activity and degree centrality are negatively associated with crowdfunding project success. Practically, these findings indicate that project owners should not rely on their social media followers or influencers to attract potential backers since followers' and influencers' excessive sharing of project information could lead to fewer contributions by other social media users. Accordingly, project owners should focus on actively sharing project information with social media users by serving as information hubs, which will help attract more potential backers.

6.3 Limitations and future research

This study faced some limitations. First, we used data from June 2016 to September 2018, which may have limited our findings' generalizability somewhat. In future studies, researchers could use data from different periods and examine whether similar results can be obtained. Second, in this study, we used data from a single crowdfunding platform and a single social media platform. This decision may have limited our findings' generalizability to different platforms. In future studies, researchers should consider replicating the current study using data from various crowdfunding and social media platforms. Third, to test our hypotheses, we used only three measures that we derived from social media data to assess implicit social ties. In future studies, researchers could use other measures of social ties derived from social media data, such as hashtags, text sentiment and topic clouds. Finally, we did not consider factors such as previous crowdfunding experience (for example, whether project owners had participated in multiple crowdfunding projects or whether backers had previously supported projects). In future studies, researchers could consider such factors while analyzing how project owners' and backers' social ties determine crowdfunding project success.

Figures

Figure 1

Research model

Table 1

Description of the study's collected and filtered Twitter data set

Variables	Collected data set	Filtered data set
Total number of tweets	4,206,408	172,892
Total number of retweets	1,856,063	26,350
Total number of users	4,406,408	255,419
Total number of unique users	1,128,397	21,491

Source(s): Authors' own creation/work

Table 2

Research model variables

Measure	Definition
Potential backers' total followers (PBTF)	The average number of all potential crowdfunding project backers' followers in a project
Project owners' total followers (POTF)	A crowdfunding project owner's number of followers
Fundraising goal	A project's actual crowdfunding project goal (in US dollars)
Number of tweets	The total number of tweets related to a crowdfunding project
Potential backers' social media activities (PBSM)	The average number of a potential project backer's retweets about a crowdfunding project
Potential backers' degree centrality (PBDC)	A potential project backer's average degree centrality (calculated from the implicit social network created for each project using project-related tweets)
Potential backers' betweenness centrality (PBBC)	A potential project backer's average betweenness centrality (calculated from the implicit social network created for each project using project-related tweets)
Project owners' social media activities (POSM)	A project owner's number of retweets about a crowdfunding project
Project owners' degree centrality (PODC)	A project owner's degree centrality (calculated from the implicit social network created for each project using project-related tweets)
Project owners' betweenness centrality (POBC)	A project owner's betweenness centrality (calculated from the implicit social network created for each project using project-related tweets)
Project size (PS)	A project's size value (small = 0; large = 1), based on the median value of all crowdfunding projects' fundraising goals (projects with goals below the median value were considered small, while the remaining projects were considered large)
Project category	The specific category label for each project on Kickstarter (Kickstarter defines 15 different project categories: art, comics, crafts, dance, design, fashion, film and video, food, games, journalism, music, photography, publishing, technology and theater)
Degree of crowdfunding project success (DCPS)	The ratio of the project funding received to a project's fundraising goal (in US dollars)

Source(s): Authors' own creation/work

Table 3

Descriptive statistics

Variable	Minimum	Maximum	Mean	Standard deviation
PBTF	0	2,786,937	15,548.650	94,917.720
POTF	0	960,041	2,613.979	24,343.563
Fundraising goal	0.779	50,000,000	43,0001.160	1,104,818
Number of tweets	1	5,049	80.006	310.613
PBSM	0	1,403	4.195	35.544
PBDC	0	4	0.394	0.622
PBBC	0	0.250	0.012	0.041
POSM	0	66.250	0.990	2.841
PODC	0	4	0.661	0.833
POBC	0	1	0.065	0.165
DCPS	0	438.140	2.049	11.155

Source(s): Authors' own creation/work

Table 4

Correlation matrix

	PBSM	PBDC	PBBC	POSM	PODC	POBC	DCPS
PBSM	1
PBDC	0.027	1
PBBC	−0.009	0.148	1
POSM	−0.028	0.013	−0.016	1
PODC	−0.017	0.717	0.07	0.093	1
POBC	−0.022	0.056	0.004	0.102	0.456	1
DCPS	−0.003	0.042	−0.006	0.011	0.070	−0.008	1

Source(s): Authors' own creation/work

Table 5

OLS regression with fixed-effect models

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
	β (se)	p	β (se)	p	β (se)	p	β (se)	p	β (se)	p	β (se)	p
Intercept	1.385 (0.074)	0.000	1.738 (0.094)	0.000	1.442 (0.074)	0.000	1.755 (0.092)	0.000	1.483 (0.075)	0.000	1.797 (0.093)	0.000
PBTF	0.016 (0.005)	0.001	0.015 (0.005)	0.002	0.001 (0.004)	0.733	0.001 (0.004)	0.811	0.0107 (0.005)	0.044	0.010 (0.005)	0.057
POTF	0.079 (0.006)	0.000	0.077 (0.007)	0.000	0.063 (0.007)	0.000	0.063 (0.007)	0.000	0.063 (0.007)	0.000	0.062 (0.007)	0.000
Fundraising goal	−0.121 (0.007)	0.000	−0.175 (0.011)	0.000	−0.126 (0.007)	0.000	−0.179 (0.011)	0.000	−0.128 (0.007)	0.000	−0.180 (0.011)	0.000
No. of tweets	−0.004 (0.010)	0.694	−0.002 (0.011)	0.822	−0.004 (0.010)	0.653	−0.006 (0.010)	0.586	−0.006 (0.011)	0.522	−0.006 (0.010)	0.605
PBSM	−0.073 (0.015)	0.000	−0.079 (0.019)	0.000					−0.042 (0.016)	0.010	−0.040 (0.019)	0.045
PBDC	0.022 (0.039)	0.572	0.062 (0.051)	0.228					−0.181 (0.066)	0.007	−0.162 (0.091)	0.006
PBBC	0.204 (0.322)	0.527	−0.246 (0.432)	0.568					0.483 (0.323)	0.135	−0.027 (0.434)	0.949
POSM					0.101 (0.021)	0.000	0.162 (0.029)	0.000	0.093 (0.021)	0.000	0.155 (0.030)	0.000
PODC					0.106 (0.033)	0.001	0.126 (0.044)	0.004	0.215 (0.058)	0.000	0.228 (0.080)	0.005
POBC					0.262 (0.105)	0.013	0.327 (0.174)	0.061	0.078 (0.121)	0.520	0.148 (0.192)	0.439
Project size (PS)			0.231 (0.043)	0.000			0.299 (0.046)	0.000			0.292 (0.049)	0.000
PBSM*PS			0.014 (0.024)	0.556							−0.003 (0.024)	0.899
PBDC*PS			−0.082 (0.069)	0.243							−0.040 (0.127)	0.752
PBBC*PS			1.041 (0.630)	0.099							0.994 (0.626)	0.113
POSM*PS							−0.116 (0.019)	0.003			−0.116 (0.040)	0.004
PODC*PS							−0.054 (0.031)	0.375			−0.037 (0.112)	0.742
POBC*PS							−0.092 (0.108)	0.669			−0.102 (0.243)	0.675
Fixed effects	Yes		Yes		Yes		Yes		Yes		Yes
Max VIF	3.60		3.75		3.99		4.50		4.88		8.99
R-squared	0.232		0.248		0.245		0.261		0.251		0.268
Adj. R2	0.225		0.239		0.237		0.252		0.242		0.257
F	30.89	0.000	28.21	0.000	33.06	0.000	30.20	0.000	29.81	0.000	25.14	0.000
N	2,161		2,161		2,161		2,161		2,161		2,161

Note(s): Crowdfunding projects were divided into size values based on the median value of all projects' fundraising goals

Source(s): Authors' own creation/work

Table 6

Fixed-effects logit regression models

	Model 1		Model 2		Model 3		Model 4		Model 5		Model 6
	β (se)	p	β (se)	p	β (se)	p	β (se)	p	β (se)	p	β (se)	p
Intercept	2.544 (0.391)	0.000	3.666 (0.517)	0.000	2.963 (0.399)	0.000	3.948 (0.519)	0.000	3.111 (0.404)	0.000	4.073 (0.529)	0.000
PBTF	0.077 (0.024)	0.002	0.071 (0.025)	0.004	0.033 (0.021)	0.109	−0.035 (0.021)	0.085	0.019 (0.026)	0.468	0.014 (0.026)	0.577
POTF	0.321 (0.034)	0.000	0.320 (0.034)	0.000	0.221 (0.035)	0.000	0.222 (0.035)	0.000	−0.232 (0.036)	0.000	0.233 (0.036)	0.000
Fundraising goal	−0.338 (0.038)	0.000	−0.505 (0.062)	0.000	−0.375 (0.039)	0.000	−0.533 (0.063)	0.000	−0.385 (0.039)	0.000	−0.542 (0.064)	0.000
No. of tweets	0.078 (0.054)	0.144	0.086 (0.054)	0.110	0.129 (0.051)	0.011	0.131 (0.051)	0.010	0.084 (0.054)	0.122	0.087 (0.054)	0.109
PBSM	−0.485 (0.075)	0.000	−0.491 (0.091)	0.000					−0.261 (0.077)	0.001	−0.238 (0.092)	0.010
PBDC	0.414 (0.194)	0.053	0.446 (0.274)	0.103					−0.622 (0.346)	0.073	−0.511 (0.514)	0.320
PBBC	0.985 (1.704)	0.563	−0.107 (2.367)	0.964					2.651 (1.707)	0.120	1.581 (2.373)	0.505
POSM					0.678 (0.126)	0.000	0.857 (0.201)	0.000	0.615 (0.126)	0.000	0.814 (0.202)	0.000
PODC					0.790 (0.167)	0.000	0.762 (0.241)	0.002	1.116 (0.305)	0.000	1.011 (0.461)	0.028
POBC					3.793 (0.810)	0.000	8.206 (2.514)	0.001	2.776 (0.877)	0.002	6.848 (2.536)	0.007
Project size (PS)			0.613 (0.209)	0.003			0.754 (0.223)	0.001			0.760 (0.234)	0.001
PBSM*PS			0.025 (0.112)	0.821							−0.032 (0.114)	0.774
PBDC*PS			0.035 (0.354)	0.920							−0.182 (0.678)	0.788
PBBC*PS			2.343 (3.343)	0.483							2.227 (3.331)	0.504
POSM*PS							−0.328 (0.251)	0.193			−0.358 (0.254)	0.159
PODC*PS							0.023 (0.309)	0.941			0.157 (0.609)	0.796
POBC*PS							−5.287 (2.642)	0.054			−4.881 (2.694)	0.070
Fixed effects (project category)	Yes		Yes		Yes		Yes		Yes		Yes
Pseudo R2	0.173		0.179		0.204		0.211		0.212		0.219
Log likelihood	−1045.42		−1037.37		−1006.31		−997.43		−996.74		−987.94
LR χ2	438.96	0.000	1038.83	0.000	517.19	0.000	534.93	0.000	536.33	0.000	553.92	0.000
N	2,161		2,161		2,161		2,161		2,161		2,161

Source(s): Authors' own creation/work