The possible effects of diversity and
gender on the performance of Dutch
fashion start-ups
Author: Lilian Frederique Smitskamp Student number: 10668381
Date of Submission: 19 February 2015
MSc. in Business Studies – Entrepreneurship & Innovation Track Institution name: University of Amsterdam
First supervisor: Y. Song Second supervisor: T. Vinig
Statement of Originality
This document is written by Lilian Frederique Smitskamp 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 to create it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Acknowledgement
I would like to thank my advisor, Yang Song for guiding and supporting me with her ideas, opinions and feedback.
Abstract
This thesis investigates the impact of the use of online social networks by male and female fashion entrepreneur on their business performance. Previous research has focused on IT-based companies or focused only on offline social networks and performance. This thesis focuses on gender, diversity or a combination of both and whether these variables affect the performance of a start-up. This research is based on the idea that an entrepreneur’s diverse online social network of weak ties and strong ties is correlated to the performance of the start-up.
The major finding in this thesis is the moderating effect of gender on diversity and their relationship to the performance of the start-up. We found that male entrepreneurs were more likely to achieve higher performance through diversity of their online social network compared to female entrepreneurs.
Table of Contents Statement of Originality 1 Abstract 2 Table of Contents 3 1. Introduction 4 2. Literature Review 6 a. Fashion Industry b. Social Networks
c. Structure of the social network d. Network Ties
e. Diversity f. Gender g. Performance h. Start-up Age
3. Methodology & hypotheses 19
a. Research objective b. Model
4. Data and descriptive statistics 23
5. Empirical Results 28
6. Conclusion 44
7. Limitations 46
8. References 48
1. Introduction
Due to the economic turmoil of today’s world, many companies are struggling to survive. As a result, unemployment in Europe is high and jobs are scarce. Hence people are looking for new opportunities to support themselves and to create new job opportunities through self-employment. One of the industries that is offering these opportunities and is still experiencing relative growth while being least affected by the crisis is the fashion industry (Scheffer & Duineveld, 2004).
It is interesting to note that even though many fashion start-ups are being created, some become incredibly successful while others remain in mediocrity. In order to assess if a start-up is successful, one would normally look at the performance indicators of these ventures. However, one of the main problems when analysing start-ups is that “return on investment and profitability, may not be an appropriate performance indicator for new business ventures, because many of these firms are still in the stage of product development (Hart, 1995).” Profitability indicators for start-ups are usually negative or low in the first few years due to the initial sunk costs. However, this may not necessarily mean that a start-up is unsuccessful.
Therefore, this thesis aims to research another important resource that could show the relationship with the performance of start-ups the social network of the entrepreneur (Lee et al, 2001). The social network has been thoroughly researched, however most of the research is focused on the link between the offline social network of the entrepreneur and the performance of the start-up ventures. Since the immense growth of the online networking phenomenon, this thesis will be focusing specifically on the online social network of the entrepreneur. Particularly testing assumptions made about offline social networks and finding if they hold for the online social network.
This thesis focuses on how entrepreneurs are interacting via their online social networks and trying to understand to what degree this has an influence on their start-up performance. Simultaneously, the difference between how male and female entrepreneurs build their online social network will be analysed. Specifically, how do both genders make use of diversity through the use of strong and weak ties during the start-up phase and how these numbers impacts the performance? The reason for this particular distinction is because research has indicated that male and female entrepreneurs build their online social network differently, resulting in different results when measuring their performance. Even though previous research has shown the connection between diversity, gender and performance; this thesis contributes by researching if this holds true in the fashion industry and also which variable
has the most impact on the performance of the start-up. Most importantly, to what extent does gender, diversity or a combination of both positively affect the performance of the start-up. These could be useful result for female entrepreneurs, since they may indicate that female start-ups are doomed to in general be less successful than their male counterparts, or if the diversity of their online social network limits their growth. Considering these three options, if indeed the results show that gender influences the overall performance of the start-up, then there is a disadvantageous gender gap for female entrepreneurs in the fashion industry. However, if diversity is the most influential variable to the performance of the start-up, female entrepreneurs could change the way they build their online social network. If the results indicate a combination of gender and diversity as most influential on the performance of start-ups, female entrepreneurs could consider changing the way they build their online social network, and should in this way work to decrease the gender gap between male and female entrepreneurs in this fashion industry in the Netherlands.
Following this introduction, the thesis is organized in six sections. Section two will review the literature regarding the main topics, including performance, social networks, structure, network ties, the phases of start-up, the fashion industry, diversity and gender. Section three will follow by introducing the hypotheses that will be tested in this thesis. Subsequently in the same section the methodology used and all the variables will be explained. In the last sections the empirical results will be presented and the conclusion and limitations provided about the research.
2. Literature Review
Many researchers have aimed to research the social networks of entrepreneurs in order to review the relationships between its diversity and a start-up’s performance. Although the literature covers a wide variety of such theories, this literature review aims to identify how diversity, gender or a combination of both affects the performance of start-ups. The focus will initially be on the overall characteristics of the fashion industry, setting the scene of the start-ups that will be analysed. This will be followed by the characteristics of the online social network of the entrepreneur; specifically, the difference between strong and weak ties which form the building blocks of the topic of diversity. The diversity of the online social network will subsequently be linked to gender characteristics; especially the difference between male and female entrepreneurs, considering their business characteristics and their online social network characteristics. Finally, the relationship between the performance of the start-ups and the diversity of the online social networks and gender differences will be introduced.
Fashion Industry
As previously mentioned, the focus of this thesis is on the fashion industry, specifically the research of the online social networks of male and female start-up fashion entrepreneurs. The choice for the fashion industry was based on the limited research and data available on entrepreneurs starting a business in the fashion industry (Grave & Salaff, 2003). Compared to other industries, the retail and fashion industry account for the largest share of female entrepreneurs (McManus, 2001). Facebook data (Appendix Image 1) showed that the fashion industry is the fourth largest user of Facebook with an estimated 1,960,361 company pages. E-commerce is ranked first; however, these 4,973,373 company pages also include fashion businesses.
The fashion industry is an unpredictable and hypercompetitive market with short product life cycles, which brings along many challenges (Saviolo & Testa, 2002). The challenges a fashion entrepreneur encounters can be differentiated between industry specific challenges and personal challenges. Industry specific challenges include possible economic recession, which causes consumers to spend less money in general or an increase or decrease of cost of goods sold. The personal challenges an entrepreneur can encounter are differences in consumer taste and timing. It could be that the fashion sense of the entrepreneur differs from the consumers demand for fashionable products. This could also correspond with timing since the fashion industry is a high-speed market: one day you are ‘in’ and the next day you
are ‘out’. Due to the different seasons and the seasonal patterns it is hard to produce products that are always ‘in’ fashion.
Another problem with start-ups in the fashion industry is the lack of financial support for the business. Fashion entrepreneurs usually have difficulties finding investors because the business ideas of fashion businesses are often considered high-risk investments while the margins are low. Combining the personal aspect with the business aspect of fashion is also hard for fashion entrepreneurs, since they usually see their own products as a piece of art, while their value of the product may differ greatly from that of the customer. The fashion entrepreneur needs to be able to grasp business opportunities and determine customer needs by coordinating resources to design, manufacture and supply fashion products or provide a service (Burke, 2011). Often at this stage, it is difficult to indicate the profit a fashion business may generate in the future. It is hard for an entrepreneur to predict the return on investment of the fashion business, which makes it less attractive to banks and investors. Financial support by investors, business angels and banks will only be granted if the investors see a large growth potential (www.entrepreneur.com). Fashion is challenging, dangerous, exciting, unfair, risky and timing and connections are the key to success (Saviolo & Testa, 2002). These important connections introduce the next topic of this thesis; the social network of the entrepreneur.
Social Networks
In the growing amount of literature on the founding and growth of entrepreneurial firms, the importance of a strong social network is acknowledged (Elfring & Hulsink, 2007). The article that lays the grounds for social networks is Granovetter’s (1983) “the strength of weak ties: A network theory revisited.” Granovetter (1983) focuses on the characteristics and importance of weak and strong ties and their effects on the social network of the entrepreneur and his/her position in the market. Many scholars have built on the articles of Granovetter; Aldrich is one of these scholars. Aldrich (1989) illustrates in his article the connection between social networks, gender and performance. Identifying that “Small- business owners are a particularly good example of the embeddedness of economic activities within a social context (Granovetter, 1985). Since, owners of small businesses are tied to many other actors, such as customers, employees, moneylenders, family members, and friends. In fact, these categories of people are not mutually exclusive, many small businesses employ family members or/and see them as important customers. In previous work Aldrich and Zimmer (1987), argued for
the importance of explicitly recognizing the social networks in which business owners operate.
In order to explicitly recognize the social network, the social network is defined as “a set of people (or organizations or other social entities) connected by a set of social relationships, such as friendship, co-working or information exchange (Garton, 1997).” When trying to analyse social networks, the approach used was the “social network approach that facilitates the study of how information flows through direct and indirect network ties, how people acquire resources, and how coalitions and cleavages operate. Social network analysts look beyond the specific attributes of individuals to consider relations and exchanges among social actors (Garton, 1997).” Typical resources that are used by a social network include “textual, graphical, animated, audio, or video-based media, for example support, knowledge, sharing information (news or data), discussing work, access to distribution channels, giving emotional support, or providing companionship (Haythornwaite, Wellman & Mantei, 1995).” The structure of a social network is never fixed; they are the social context of businesses and can be activated according to the different needs of the business (Granovetter, 1985; Burt, 1992). As they entertain, plan for, and actually set up a firm, entrepreneurs call on their family and others in their networks for different kinds of help and support (Rosenblatt, de Mik, Anderson & Johnson, 1985). However, throughout the start-up phase the connections drawn upon by the entrepreneur are likely to change and this acknowledges the importance of the fluctuating structure of the social network.
Structure of the social network
The structure of social networks has changed in the last few years due to the uprising of the Internet, allowing a shift to online interaction between contacts on social network sites (SNS). “The nature of online interaction has evolved rapidly, most notably through the surprisingly swift rise of online social networks such as Facebook, Myspace and Twitter. While older forms of online social interaction such as e-mail seemed especially suited to support existing offline interaction structures, these newer online social networks allow users not only to interact with their own contacts, but also traverse the network by discovering the contacts of their own contacts (Boyd, 2007:Parigi, 2013).”
Since more than 71% of the developed world is now online (World Bank, 2011) with two thirds of the U.S adult population using social network sites, a figure that has doubled in the last few years, it is not remarkable that individual’s social networks are also growing (Madden & Zickuhr, 2011). The average online social network user has around 350 friends on
the networking site (Statista, 2014). Many users have exponentially increased their network size and diversity by linking with old high school friends and new colleagues.
However, not only individual users are exponentially increasing their network size and diversity, so are entrepreneurs and their businesses. A study by the CBS Netherlands has found that Dutch businesses lead the way with social media use in Europe (Appendix Image2). Businesses in the Netherlands are increasingly using social media as a way to communicate with customers and partners, as well as simply to be visible. In 2013, social networks were among the most popular form of social media, with 47% of Dutch companies running an account on networks such as Facebook and LinkedIn compared to 35% in 2012 (Appendix Image 3). Dutch companies are making greater use of social media than the EU average. The most popular features social media brings to the companies include the development of their brand image and market products (76%), recruitment of staff (50%) and to conduct business relations (44%) and customer feedback (CBS, 2013). According to the study, Dutch businesses that are active on Facebook, Twitter and LinkedIn managed to increase their customer reach and improve customer loyalty, as well as successfully advertise online, compared to non-active competitors (TNS Nippo).
However, Burke (2011) mentions that by opening up the online social network for customers, suppliers, employees and distribution partners to communicate together with friends and family, the structure of the online social network is automatically changing compared to what previous offline social network could reach. Specifically, social network sites differ from the offline communication methods, since it allows a relationship with thousands of users through one single feed, which allows for an efficient maintenance of a larger social circle (Burke, 2011). “Online social networks seem especially suited for the creation of new connections that bridge social contexts and thus may have an upgrading effect on the divisions found in modern societies, divisions typically reinforced through certain patterns of offline interactions. Recent empirical research suggests that while online networks are firmly rooted in existing offline social networks, they are positively associated with various forms of bridging social capital (Ellison et al, 2007: Parigi, 2013).” When forming bridges in social capital, one must make the distinction between which ties form the best unique bridges. This forming of bridges and tie strength brings us back to the Granovetter Network ties (1983).
Network Ties
ties is an issue of debate when discussing network benefits (Elfring & Hulsink, 2007; Uzzie 1997, Lechner et al 2006, Jack 2005; Batjargal 2003). As previously mentioned Granovetter (1983) identifies that there are two types of tie strength: strong and weak ties. He describes the two different ties by describing the situation of any randomly selected individual who has a collection of close friends whom are all closely linked to each other (strong ties), and can be described as a densely knot clump of a social structure (Granovetter, 1983). However, this randomly selected individual also has accumulated a collection of acquaintances that barely know each other, or have no relations to each other. These are known as weak ties. “Each of these acquaintances is likely to have close friends in his own right and therefore to be enmeshed in a closely knit clump of social structure, but one different from the individual. The weak tie between the individual and his acquaintance, therefore, becomes not merely a trivial acquaintance tie but rather a crucial bridge between the two densely knit clumps of close friends (Granovetter, 1983).” The benefit of this crucial bridge is extremely important to recognize, because if these trivial acquaintances relationships did not exists, these social clumps would in fact not be connected at all. This means, that if certain individuals have a social network lacking weak ties, they will not be able to benefit from other social clusters around them and the resources they can provide.
The lack of access to unique social resources can cause an individual to be confined to learning only about thoughts, opinions, advice and news from their strong ties. “This deprivation will not only insulate them from the latest ideas and fashions but may put them in a disadvantaged position in the labour market (Granovetter, 1973).” Not just random individuals have a tendency to connect with similar individuals, Kim and Aldrich (2005) have also illustrated that entrepreneurs have a high tendency to favour formation of dense networks based on strong ties. Simultaneously, other research concluded that similarly to strong ties, weak ties are not only beneficial in terms of acquiring resources; they also play a role in other key entrepreneurial processes such as spotting opportunities (Elfring & Hulsink, 2007; Ardichvili et al, 2003) and gaining legitimacy (Aldrich and Fiol, 1994). The importance of these findings raises the question what is the significance of strong and weak ties and what are the essential performance benefits to having a diverse online social network. The debate is still out whether weak ties, strong ties or a combination of both leads to the highest level of performance and growth. Various scholars such as Bruderl & Preisendorfer (1998); Batjargal (2003), Jack (2005) have stressed the importance of strong ties at emergence of start-ups, while Greve & Salaff (2003), Steier & Greenwood (2000) have argued that a high number of weak ties are essential in the start-up phase.
The difficulty with making the distinction of which ties are important in an entrepreneur’s online social network is that ties can fluctuate according to importance throughout the different phases. This supports the various researches of both scholars such as Bruderl & Preisendorfer (1998), Batjargal (2003) and Jack (2005) about the importance of strong ties while also agreeing with Greve and Salaff (2003), Steier & Greenwood (2000) on their importance of weak ties.
The amount of research available on how to identify weak ties and strong ties is very small. Where there is a lot of research on the diversity of a social network and weak and strong ties in general, the actual characteristics on what makes a tie strong or weak are limited. The earliest research is from Granovetter (1995). He defines the strength of ties as the intensity and diversity of relationships. Specifically, the difference between strong and weak ties on the basis of four criteria: the frequency of contact, the emotional intensity of the relationship, the degree of intimacy, and the reciprocal commitments between the actors involved. However, this research is primarily based on the offline social network of an individual, therefore more research was needed to identify the connection between tie strength and the online social network.
Gilbert & Karahalois (2009) are one of the first that use a combination of sociology and computer science to create an API (application programming interface) that uses Facebook tie strength variables (Appendix Image 4) in order to create a measurement tool that accounts for an 87.2% reliable calculation of tie strength. They recruited 35 participants to rate the strength of a randomly selected subset of their Facebook friends (guarding against those with large networks dominating the results). On the friend’s Facebook profile five tie strength questions are asked based on a Likert scale (Appendix Image 5). Gilbert & Karahalois (2009) found seventy Facebook variables categorized into different predictable variables including intensity, intimacy, duration, reciprocal services, structural, emotional support and social distance (Appendix Image 6) to predict tie strength. This research shows that some Facebook variables can be used to predict tie strength between the survey participant and his/her network ties. Gilbert & Karahlois (2009) connection of tie strength and Facebook is an important assumption for this thesis research. Their tie strength questions will therefore also be used in this research, however due to privacy settings the use of their 70 Facebook variables is not realizable.
Burke (2011) builds on the approach used by Gilbert & Karahalois (2009). However, her research focuses more on the fluctuations of network ties considering social network sites (SNS). In her work, the most important addition to this thesis is how she analysed the
relationship of tie strength and Facebook activity. The collection of her data is based on a name generator, which allows the participant to enter names compared to Gilbert & Karahalois (2009) random selection. She conducted a survey which contained questions about participant relationships with up to eight Facebook friends.
Both Gilbert & Karahalois (2009) and Burke (2011) introduce the concept of network ties and the bridges between survey participants and their network ties. However, it is important to note that unlike the offline social network, network ties and bridges are referred to differently in the online social network respectively as nodes and edges. A node is representative of a specific tie in the network of an individual, while an edge is the bridge or connection this node has with another node in the network. Both edges and nodes have the characteristic of weight, this is characteristic is important to our research since we will be making a distinction between strong and weak ties based on edge weight. This distinction between the weights of the edges is necessary in order to calculate the diversity of the network, which introduces the next topic of diversity.
Diversity
Family members are present in all the phases of every entrepreneur’s online social network, since all entrepreneurs use their family and strong ties for information, knowledge, resources and support (Krackhardt, 1992). However, offline social network research suggests that female entrepreneurs depend more on their family and strong ties than male entrepreneurs. Research even indicates that entrepreneurship runs in the family (Rosenblatt et al, 1985) and entrepreneurs are more likely than the average population to have parents that also run small businesses (Rosenblatt et al, 1985). It is therefore not abnormal that entrepreneurs take advantage of their self-employed family members for initial feedback and input about their business idea (Rossenblatt et al, 1985; Aldrich, Reese & Dubini, 1989).
However, research indicates that entrepreneurs with a high ratio of family members to other network ties in their online social networks rely less on outsiders. Specifically female entrepreneurs with self-employed parents draw to a larger extent on family members than their male colleagues in their discussions of establishing and running a business, limiting their interaction with other network ties. This may be caused by less opportunity for females to expand their networks into male dominated business circles (Renzulli et al, 2000). If the ration of strong ties and weak ties in online social networks makes a difference in the initial phase of the start-up, we need to learn more about the network composition of female and male entrepreneurs (Greve & Salaff, 2003). This different network composition of strong and
weak ties and the weight assigned by the entrepreneur to nodes and edges can be seen as the diversity of the online social network
Greve and Salaff (2003) find that entrepreneurs build networks that systematically vary by entrepreneurial phase. The different activities the entrepreneur engage in during the different phases range from finding start-up capital, discussions with partners, and time spent networking. The start-up phase of a business can be identified in three phases: The motivation phase, the planning phase and the establishment phase. Entrepreneurs tend to talk with more people and colleagues during the planning phase than the other two phases. The diversity of the network is important throughout all three phases of the start-up, since this is the crucial phase of building a diverse and stable network. Since, once the business is running,
entrepreneurs are inclined to concentrate their network on key persons in their already present networks who are able to provide resources and commitment (Chu, 1996; Hansen, 1995; Greve & Salaff, 2003). We expect entrepreneurs to use the same amount of time maintaining their contacts as developing them.
There are two initial starting conditions when considering the mix of weak and strong ties with start-ups. First, there is the type of start-up to consider; independent start-ups or spin-offs. Both have a different optimal mix of strong and weak ties in their social network. Independent start-ups are start-ups founded by entrepreneurs who do not have any or not a lot prior knowledge of the fashion industry. Due to being considered the outsiders of the industry, these entrepreneurs focus primarily on connecting with people who give them direct access to the resources of the fashion industry (Elfring & Hulsink, 2007). These entrepreneurs do not really benefit from their strong ties, such as friends and relatives or relationships from previous work environments since these do not have the knowledge and resources in the fashion industry (Jack 2005). Therefore, in the case of independent start-ups the entrepreneur focuses on weak ties that are insiders of the fashion industry to help grow. The dominant networking activity of the entrepreneur is therefore focused on attracting weak ties by meeting new people at fashion events, conferences and participating in new types of networking activities (Elfring & Hulsink, 2007). The name spin off start-up does not necessarily mean that it came from another big company. It is “a start-up that is based on ideas and knowledge from insiders; the founders were employees in an established firm or research institute within the industry are more likely that they rely mainly on their strong ties to parent organizations to obtain information on opportunities, acquire resources and gain legitimacy” (Elfring & Hulsink, 2007).
The second initial founding condition is the type of innovation the start-up is enforcing. There are two types of innovation a start-up can pursue either incremental or radical innovation. Start-ups pursuing an incremental innovation rely more on their strong ties compared to start-ups involved in radical innovations (Elfring & Hulsink, 2003). Radical innovations are often based on new combinations of diverse knowledge domains, and a diverse network of weak ties enables entrepreneurs to search for information (Hansen, 1999). Since the research focuses on fashion start-ups, the type of innovation pursued will be
incremental; therefore, in general both male and females will include more strong ties in their online social network compared to other industries.
Besides initial founding conditions influencing the mix of strong and weak ties in a social network, post-founding entrepreneurial process also influence the optimal mix of strong and weak ties. Since some start-ups are considered off- spring start-ups, entrepreneurs do not necessarily have to convey legitimacy, as they already have the necessary strong tie
connections in the fashion industry. This reduces the amount of time spent on building a reputation in the industry. However, independent start-ups need to focus on building that reputation in the industry and showing that they have these qualifications to become a player in the fashion industry. It becomes apparent that each entrepreneurial process requires a different mix of weak and strong ties to be most beneficial for the start-up. Research has shown the importance of prior knowledge (Shane, 2000) and novel information (Fiet, 1996) which can be found through weak ties (Elfring & Hulsink, 2007). However, weak ties can evolve into strong ties if the tie proves to be helpful; while others that no longer deem
necessary are dropped. Elfring & Hulsink (2007) distinguish three tie formation processes that are important to the social network of an entrepreneur; adding ties, upgrading ties and
dropping ties. Entrepreneurs will build their social network based on these three processes, considering which type of innovation and which type of start-up they are. Entrepreneurs search and select ties to create the optimal diversity of weak and strong ties to the changing needs of the start-up over time (Elfring & Hulsink, 2007).
Gender
The gender of the entrepreneur is a highly fluctuating variable when considering start-up businesses. McManus (2001) illustrates that there are gender differences in business owner characteristics and states that the difference in gender should be identified when looking at “self-employed women and men both across representative samples of firms and within specific business niches.” The differences that are identified in the research state that
“Female-owned businesses are smaller than men's in size and gross receipts and male business owners were more likely than female business owners to report recent growth in sales, employment and profitability (Du Rietz and Henrekson 2000).” While the OECD (2012) finds that there are fewer Dutch female entrepreneurs than men, that they have lower profits than male entrepreneurs, and that employed women work less, earn less, than self-employed men and have less innovative enterprises (Appendix Image 7). It is important to find data that supports these statements, since it could potentially explain the change in performance based on employment growth and sales growth between male and female entrepreneurs. Studies of small business in Germany (Bruderl, Preisendorfer, and Ziegler 1992) and the US (Kalleberg and Leicht 1991) provide empirical support, showing that “business survival increases positively related to the human capital characteristics of the business owner at the time of founding, including education level, work experience, and industry specific experience. On average women differ from men on human capital characteristics, and these gender differences may explain part of the gender differences in self-employment rates (McManus, 2001).”
Female Entrepreneurs Male Entrepreneurs Network Size on Social
Media
Lower Higher
Educational Level Lower Higher
# Entrepreneurs in network
Lower Higher
Dependence on family & kin (Strong ties)
Higher Lower
# of weak ties in network Lower Higher
Access to Financing Lower Higher
Carrington and Troske (1995) show “that the gender composition of employees of small businesses owned by men is disproportionately male. Since there are more male business owners than female business owners, and male owned businesses tend to have more employees than female owned businesses, women have less access to the hands on training that can facilitate the transition to self-employment.” Illustrating that Aldrich and Zimmer conclude that “Social network approaches to self-employment extend the concept of entrepreneurial resources to include the sharing of specific expertise and general "know-how"
within social networks that include one or more entrepreneurs (Aldrich and Zimmer, 1987). Entrepreneurs tend to benefit from the business knowledge of other entrepreneurs in their social networks. Especially, aspiring entrepreneurs benefit from the advice of the experienced entrepreneurs, and the individuals linked by the network presumably serve as a potential source of financial capital, legitimacy and resources. Allen (2000) concluded that women are significantly less likely than men to report entrepreneurs (and former entrepreneurs) in their social networks. “These specific gender differences in entrepreneurial networks might account for the gender gap in self-employment, as well as the gender gap in earnings (McManus, 2001).”
Bruderl & Preisendorfer (1998) find that only support from strong ties, especially family support, contributes to the success of start-ups among men and women. The differences in human capital, financial capital, and social capital in the diversity of social networks of both male and female entrepreneurs may contribute to the earnings gap among the self-employed in the United States (Renzullli et al, 2000). This thesis tries to find if this also holds true for Dutch entrepreneurs, and studies the different composition of online social networks for both male and female entrepreneurs. Specifically, if indeed female entrepreneurs tend to make more use of strong ties due to the lack of access to resources, finance and distribution channels, while male entrepreneurs focus more on weak ties, whom they have met through previous employment, hands on training and other entrepreneurs. The gender differences and the effects on the online social networks of the entrepreneur will be measured through the change in performance. In the next section the relationships between gender, diversity and start-up performance will be introduced.
Performance
As Behn (2003) states: people use performance as a measurement tool since it allows managers to evaluate, control, budget, motivate, promote, celebrate, learn and improve. A performance measurement system can be defined as “the set of metrics used to quantify both the efficiency and effectiveness of actions (Neely et al, 1995)”. While Bourne et al (2000) defined it as “the use of a multi-dimensional set of performance measures. The set of measures is multi-dimensional as it includes both financial and non-financial measures. It includes both internal and external measures of performance and it often includes both measures which quantify what has been achieved as well as measures which are used to help predict the future.”
As previously mentioned, performance of start-ups is difficult to measure due to its multi-dimensionality (Cameron, 1978; Chakravarthy, 1986). One of the main problems when analysing the start-up’s performance is that “return on investment and profitability, may not be appropriate performance indicators for new business ventures, because many of these firms are still in the stage of product development (Hart, 1995).” This usually results in profitability indicators in the first few years being negative or low due to sunk costs, while this may not necessarily mean that the start-up is unsuccessful (Bosma, van Praag, Thurik, & de Wit, 2004). Reynolds and Miller (1992) examine four key events during the establishment of a business; intentionality or commitment, financing, hiring and sales. Hansen (2001) sees these steps as an indication of entrepreneurial progress and survival, giving each event equal weight, without regard to sequence. Since commitment/ intentionality and financing are very hard to measure in the case of start-ups, the research method will only take into consideration the two measures of survival; hiring and sales.
Therefore, similarly to the approach of Yang & Vinig (2012) the primary focus of this thesis is on the survival of a start-up as a measurement of performance. The performance data based on sales measured is a snapshot view of revenue information at one particular point in time. The snapshot view focuses on the information entered by the participant of the survey at that specific point in time rather than longitudinal data. This data will report the average revenue, first year revenue and last year’s revenue and will focus on their change and growth. However, one must take into consideration privacy issues with generating revenue data and the self-reported revenues of some entrepreneurs being too negatively or too positively biased (de Witt, 2004). The final measurement of success based on the hiring process is the employment growth rate of the start-up, which can be measured by the number of employees hired since the start of the venture (Baum et al, 2000; Yang & Vinig, 2012).
The decision to look at performance and the successfulness of a start-up is based on previous research on offline social networks, which has shown that entrepreneurs who are well connected, also seem to be more successful (Baum et al., 2000; Uzzi, 1997; Uzzi & Spiro, 2005). Hereby assuming a connection between the performance of a start-up and the social network build by the entrepreneur (Lee et al, 2001). Therefore, the aim of this research is to investigate if the same holds true for the online phenomenon, specifically Facebook, by comparing online social network structures and entrepreneurial performance. To be able to compare online social network structures to entrepreneurial performance the focus will be specifically on the strength of the network ties, average weighted degree and network diversity.
The purpose of the research is to see if there is a relationship between tie strength of an entrepreneurs’ online social network and its start-up performance. This may assist future male and female entrepreneurs to design their online social network in the best way to enhance performance in the start-up phase of their new venture or slightly older start-up firms experiencing disappointing performance results. Song & Vinig (2012) focus on this relationship by looking at data collected via social media like LinkedIn, Facebook and Twitter to show the structure, diversity and impact on start-up survival. When looking at the impact of online social networks on start-ups’ successfulness they choose to focus on measures such as survival, employee growth and revenue growth. Social networks affect three aspects of entrepreneurship: business founding’s, business success, and business turnover. Social ties are important for all three processes, and their importance applies to everyone. This is also important finding for this research, since they have identified that indeed there is a relationship between business success and online social networks, which gives our research more validity.
Age of the Start-up
One of the most important characteristic of start-ups is that they cannot be categorized by years of activity and be included in one of the three start-up phases according to age. Due to high diversity of different types of products and business building phases, start-ups can be active for many years however have yet to bring a product to the market. In this specific case, the Forbes (2008) theory was enforced stating that a start-up cannot really be considered a start-up when it is ten years or older. Therefore, the fashion businesses that are considered in this research are distinguished by age; less than nine years of business activity means it is a start-up and more than nine years of business activity indicates an established firm.
3. Methodology & Hypotheses
Research Objectives
Realizing a research gap in the academic journals available today, the objective of this thesis is to identify the benefits of a diverse online social network considering strong and weak ties. While questioning assumptions made by previous research indicating that male business owners are more likely to have more weak ties in their online social network compared to female entrepreneurs, holds in the retail fashion industry where female self- employment dominates in numbers compared to any other industry. Simultaneously analysing if indeed male business owners are more likely than female business owners to report recent growth in sales and employment as provided as results in other research (Du Rietz and Henrekson, 2000). This research is to analyse the relationships between gender and diversity and their impact on the performance of the start-up. Specifically, does gender, diversity or a combination of both positively affect the performance of the start-up, or are there other important variables that need to be considered.
Hypotheses
The five hypotheses that became apparent after reading the literature on the business characteristics of gender, performance, diversity and tie strength in the online social network include:
Hypothesis 1: Female owned businesses tend to be smaller than men's in network size and number of ties.
Hypothesis 2: Male business owners are more likely to report a larger number of entrepreneurs in their online social network than female business owners.
Hypothesis 3: Male business owners are more likely to have a higher degree of diversity in their online social network than their female business owners.
Hypothesis 4: Male business owners are more likely than female business owners to report recent growth in revenues and employment in the start-up phase of their new venture
Hypothesis 5: Male business owners are more likely to have a higher performance level due to a more diverse online social network.
Model
In order to analyse the relationships between gender, diversity and performance the online social networks of the entrepreneurs were analysed. Data on the online social network of the entrepreneurs was drawn on surveys and Facebook data that query fashion
entrepreneurs from the Netherlands in sub categories such as sunglasses, second hand designer clothing, shoes and baby clothing about which type of ties they use to run their business. To develop the arguments of this model, the gender assumptions of the online social network and their relationship with diversity will first be discussed, followed by the
relationship between gender and the performance measures of the start-up.
The first part of the model will be focusing on the gender assumptions and their relationship with diversity. The first three hypothesis focus on testing the gender assumptions made by previous research based on diversity indicated that female owned businesses tend to
be smaller than men's in network size and number of ties, male business owners are more likely to report a larger number of entrepreneurs in their online social network than female business owners and male business owners are more likely to have a higher degree of diversity in their online social network than their female business owners.
Hypothesis 1 looks at the online social network size of the entrepreneur and the results will indicate if there is a significant difference between the network sizes of male, female and co-sex entrepreneurs. The network size will be represented by the variable number of total likes which represents the total number of likes the entrepreneur has accumulated on their Facebook company page. The total number of likes variable is the amount of individual people whom have liked the company Facebook page, and will be considered the online business network size, since these people whom have liked the page will be receiving daily updates and information from this page. A significance test will be done between the variable online network size and the variable Gender to identify if there is a significant difference. The gender variable consists of three different options; female, male or co-sex entrepreneurs. It is important to note that a business can be owned by both female and male entrepreneurs or a co-sex couple, this distinction is essential when comparing data between male and female entrepreneurs.
The second hypothesis focuses on the number of entrepreneurs present in the online social network of the entrepreneur. Through Facebook data the variable number of
company-to-company likes was generated by calculating all the links to other fashion entrepreneurs
Facebook company pages. The company-to-company likes variable is the total amount of entrepreneurs with similar fashion businesses whom have liked the company Facebook page, and therefore are linked to information of the liked business. Since the analysis only focuses on the relationships between Dutch fashion companies, foreign founded companies are eliminated from the datasets. These company likes will be used to build our online social network with Gephi and to see how entrepreneurs are connected in the Dutch fashion industry through network diagrams and their statistics. All these links will both be analysed through Gephi, to create a network showing how all the entrepreneurs are connected, and the significance of these connections will be tested through SPSS.
The third hypothesis focuses on the diversity of the online social network and the relationship with gender. To calculate the diversity of the online social network, many steps had to be taken to calculate the average weighted degree of the edges. When reviewing the literature, not many scholars elaborated on how to make a distinction between strong and weak ties in an online social network. As previously mentioned, Gilbert & Karahalois (2009) had created a tie strength model that used Facebook data to rate tie strength. Due to patent and privacy issues the use of this tie strength model was not an option in this case, however the tie strength based questions to qualify strong and weak ties were (Appendix Image 5). Through the use Burke’s (2011) name generator, tie strength data was collected on eighteen Facebook connections from the survey participants. The questions asked in the survey were the exact same questions as Gilbert & Karahalois (2009) and no API was necessary since the name generator provided eighteen different strong, weak and random chosen ties. This ensured that an adequate number of strong and weak ties were available to make up the network model. Through the ranking of the connections of the survey participants, the average weighted
degree of the edge (strength of the tie) was calculated through Gephi, to be used as the
variable to represent the diversity of the online social network.
The second part of the model focuses on the gender assumptions and their relationship with performance. The performance will be evaluated on survival measurements, and will be linked to the entrepreneur’s online social network characteristics. This is in order to see if it is indeed true that more weak ties in the online social network leads to a higher performance level, and if indeed men do have more weak ties in their online social network. In order to collect this data Vinig and Song’s (2012) measures were used to calculate the level of performance by focusing on measures such as: survival, employee growth and revenue growth. The survey participants were asked questions concerning their performance, first
based on the employment change followed by revenue change. The employee change questions focused on questions like how many employees did the firm have when it started its
business. And how many are employed at this moment? The results of these measures were
combined to create the variable employment change, which represent one of the ways to measure performance. The employment change variable shows the amount of employment growth the start-up has received from the moment of founding up to this stage of business. Similarly, to the collection of employment change, the survey participants were asked questions about their revenues. However, measuring revenues is a little bit more complicated since this might be a sensitive subject for some participants who do not wish to disclose their exact revenues. Therefore, the questions were based on percentage growth to avoid participants having to disclose their exact revenues. The questions included; in the last year of
business how many gross revenues did you make? What were the first year revenues recorded by your business? What are the average revenues for your firm? The results allowed us to
create a variable revenue change, which showed how much the business grew from its founding day up until the business activity realized today. Since some of the start-ups have been operating for a longer period of time, the question about average revenue was added in order to see if great fluctuations of the self-reported growth and average revenues were found. The final part of the model will focus on the relationships between all three variables; gender, diversity and the performance of the start-up. The variables are respectively being represented by; the gender of the entrepreneur, average weighted degree and employee
change. The relationships between these three variables were tested with a model 1 process
regression analysis by Andrew Hayes, to test for significance difference. The following table shows all the hypotheses and the variables used in this model.
Hypothesis 1 Network Size (number of total likes) Gender
Hypothesis 2 Network of Entrepreneurs ( company-to-company likes) Gender
Hypothesis 3 Diversity (Average Weighted Degree) Gender
Hypothesis 4 Performance ( Change in Employment & Change in Revenues) Gender
Hypothesis 5 Performance (Change in Employment) Gender
4. Data and descriptive statistics Data Sources & Sample
As Saunders and Lewis (2012) mention many researchers are unable to obtain a list of the total population of the industry they are researching, this was also true for this specific case of the fashion industry. In the fashion industry, the competition is extremely high and companies rise and fall daily, therefore it is impossible to generate a sampling frame or population with all the fashion entrepreneurs in the Netherlands. Therefore, non-probability sampling, specifically quota sampling was used to generate data on these fashion entrepreneurs. This option was chosen in order to have control variables that represent certain characteristics in the population of Dutch fashion entrepreneurs whom really represent the difference between male and female entrepreneurs. Other demographics of the respondents that must be taken into consideration include educational level and self-employed relatives. The sample was generated through searches on Facebook, KvK (ZZP) registers and through fashion contacts. While the inappropriate entrepreneurs were filtered out of the sample, the remaining sample matched the selection criteria of Dutch fashion entrepreneur. The sample size of the overall Dutch Fashion industry network including both start-ups and established businesses includes the total of 1087 businesses. By using a counter if on the dataset I could eliminate all the businesses that were no longer considered a start-up by the definition and was left with 782 start-ups ranging in the founding year of 2006 – 2014. Of the 782 start-ups 15 Dutch start-ups were individually surveyed to acquire additional information about their online social network. These fifteen Dutch start-ups were selected through the willingness to provide their data through the survey and their business characteristics.
Data Description
As previously mentioned our sample group of the entire Dutch fashion industry will consist of 1087 businesses, which include 782 Dutch start-ups founded between 2006- 2014. However, when focusing specifically on the strong and weak ties reported in the online social network of an entrepreneur, 15 of the 782 Dutch start-ups will be taken into consideration due to their willingness to take part in the survey and the specific business characteristics that fit the sample searched for. Therefore, I will first discuss the characteristics of the entire fashion industry, following the specific characteristics of the Dutch start-ups and continue with the smaller survey participants’ dataset.
When analysing the data for the overall fashion industry, I found that 0.92% of the dataset (10 of the 1087 companies) had a missing gender variable. When analysing these three data sets and the information provided by online and Facebook data, we must take into consideration that some of this data may be slightly off. Since some start-ups’ have silent partners, whom are unknown to the public. This anonymity can change the gender characteristics of the data set, but since we did not have access to these numbers this is considered a limitation to the research. The solution for the missing data for the gender of the entrepreneurs, a Hotdeck was performed since the missing data was less than 10%, which replaces the missing data with a value of a similar donor. Simultaneously, Hotdeck allows us to avoid rigorous methods such as listwise deletion, pairwise deletion and mean substitution. This resulted in the dataset (Appendix Image 8) consisting of 67.2% (730) female business owners, 21.40 % (233) males, and 11.4% (124) co-sex gender business owners. The analysis of the entire fashion industry illustrated that the range of years active ranged from 1-176 years, with a mean of 10.72 years of business.
Even though these statistics are important when comparing the data of the entire fashion industry against the survey participants, another reduction must be done to create a second dataset focusing only on the start-up businesses. The businesses that are included in this specific dataset include those whom have not surpassed the nine-year limit of the definition of a start-up. Businesses whom failed to meet these requirements were removed from this specific data set; however, remain in the overall fashion industry statistics. The removal of established businesses from our second data set, change the characteristics of our data set, now consisting of 72.4% (566) female business owners, 18.8% (147) male and 8.8% (69) unisex business owners. The average number of years in business has now dropped to 4, 14 years (Appendix Image 9). The characteristics of the fifteen survey participants sample specifically focusing on the diversity of strong and weak ties consists of 53.3% female business owners and 33.3% male business owners and 13.3% co-sex business owners (Appendix Image 10). The average years of start-up activity from the sample surveyed is 3.80 years (Appendix Image 11). The average age of the fashion entrepreneurs is 33.20. (Appendix Image 12)
As mentioned in previous research literature, there are a few factors that could possibly influence the findings of this research and should be taken into consideration. These factors include the education level of the entrepreneur, entrepreneur family members, start-up location, the economic recession and the problem of the possibility of purchasing Facebook Company Likes.
The education level of the survey participants could be an influencing factor when calculating the performance of the start-up. Since entrepreneurs have varying educational backgrounds, it could be possible that higher educated entrepreneurs could have an advantage over less educated entrepreneurs. Therefore a question about education level was included in the survey. The results indicated that all the participants either had an HBO (60%) or a University Master Degree (40%) (Appendix Image 13). The educational levels of male and female educational level is also a factor to consider, since 80% of the males have a university master degree compared to only 25% of the females. The female entrepreneurs are more likely to have a HBO certificate with 75% compared to 20% of the males. It is not possible for to measure the co-sex entrepreneurs, because the information about both the male and female entrepreneurs’ educational level is not available.
The second factor that could influence the sample data is the number of participant’s who are related to self –employed business owners. As previously mentioned, research found that participants whom are related to self-employed business owners, tend to be more successful than their fellow entrepreneurs whom do not have access to these resources. In order to monitor the possible influence this has on the performance data collected in this sample, the question “Are you from a family with a history with entrepreneurship/ self-employment? Was asked.” The results showed that females were more inclined to have a background with self-employed family members (62.5%) than their male (40%) and co-sex (50%) counterparts (Appendix Image 14). These results give us an initial indication of how dependent female entrepreneurs are on their entrepreneurial family members compared to male and co-sex entrepreneurs.
The third factor that could influence the growth and performance of the start-ups is the location of the start-up. Through Facebook data all the locations of the start-ups were collected, considering these locations they were mapped on the map of the Netherlands (Appendix Image 15). Both the overall fashion industry dataset and the start-up data set were mapped to show where in the Netherlands the most start-ups and established businesses are located. This in order to check if these fashion start-ups can benefit from a locational niche. Since locational niches are very common in the fashion industry, since many businesses have a tendency to locate their business in a specific geographical area. Examples of these specific fashion industry niches that have proven to be extremely popular include places like Switzerland, Italy and Paris. When considering high quality watchmakers it is evident that all these watchmaker businesses are located in Switzerland, specifically in Geneva. Due to the knowledge, expertise and experienced labour available watchmakers tend to locate their
business in Geneva to experience economies of scale. For this same reason companies selling Italian leather goods are all located in Italy, the leather, the high skilled labour and experience is all located here. Finally, there is the most important fashion industry niche of the entire world, Paris. Paris is known to be the fashion capital of the world; the most exclusive fashion haute couture houses are located there. Every single clothing item designed is ultimately inspired by the Paris fashion week and the fashion houses unique creations. Therefore, the locations of the Dutch fashion industry start-ups are important to see if indeed, the Netherlands hosts such a locational niche where fashion businesses flourish and can make use of economies of scale through high levels of expertise, experience and knowledge. Through the mapping of the founding locations of the start-ups, the cities that are the biggest start-up incubators become visible. The results show that the Amsterdam is by far the largest incubator considering the fashion industry businesses, with a five times as many start-ups as the runner up Rotterdam. Simultaneously, when looking at the total number of likes the start-ups received on their Facebook company pages, Amsterdam also seems to inhabit the entrepreneurs whom are most popular on Facebook.
The fourth factor that could influence the data collected is the current state of the economy. Even though, the fashion industry is not as turbulent as other industries, it is still prone to the consequences of the economic recession. Since it is impossible to identify which category of the fashion industry is experiencing possible consequences from the economic recession a question was introduced in the survey. The question “Considering the recent economic recession, do you think your business performance felt any of the associated problems?” was proposed. The question aims to identify if the entrepreneur feels his or her business is experiencing any negative consequences of the economic recession (Appendix Image 16). The results can give an indication if possible disappointing performance numbers of the start-ups can be accounted for as a consequence of the economic recession. In the case of the survey participants no evidence was found that the entrepreneurs were suffering because of the economic recession. None of the co-sex entrepreneurs experienced issues, compared to 37.5% and 25% for female and male entrepreneurs whom felt the possible consequences of the recession.
The final factor that influences the dataset includes the total number of Facebook company likes the start-ups have received on Facebook. The number of likes represents the number of network ties that are connected to the entrepreneur and represents the people that will be updated on his/her feed about the activities of the start-up. Considering the immense benefits of having a large diverse online social network is for an entrepreneur, it is no surprise
that entrepreneurs lie about the actual number of total company page likes through the use of ghost likes. Ghost likes are Facebook company likes that are bought through internet companies to increase the number of total likes in order increase the popularity of the start-up online. Possible reasons for buying these ghost likes include; encouraging investors to invest and faking popularity and hoping people will go with the flow and like the page because it is popular. Therefore, the question of “in the past have you purchased likes for your company Facebook page to boost the overall number of likes?” came into play, since the survey participants might have purchased likes which would influence the overall number of likes of the start-ups (Appendix Image 17). The results of the survey participants indicate that the purchasing of company likes is also an influencing factor in this dataset, and should be taken into consideration. However, since we do not know how many ghost likes were purchased, the number of total likes will remain the same and will be considered as a limitation of the research.
5. Empirical Results
Hypothesis 1
The hypothesis focuses primarily on the network size of the online social network of female business owners compared to those of men. The reason of this specific focus is based on research performed by Baum et al (2000), Uzzi (1997); Uzzi & Spiro (2005). Who have shown in previous research that entrepreneurs, who are well connected, seem to be more successful. The assumption is that male business owners have a higher number of total likes on their Facebook company page compared to the company pages of female fashion business owners. The network size in this case is equal to the number of total likes the company Facebook page has accumulated. As previously mentioned this specific topic has some reliability issues, since self-employed business owners can purchase “ghost likes” to boost their numbers on their Facebook page. Since it is impossible to trace if companies have indeed purchased these likes, and if so how many, this possible difference should be noted. All sample sizes will be taken into consideration, and compared by gender through the use of an ANOVA test to give an indication of how representative the samples are. Since all three samples have an unequal sample size, with higher ratios of females compared to males and co-sex couples, the ANOVA one-way test is performed to remove these unequal ratios. The ANOVA one-way test was performed with as dependent variable the number of total Facebook company page likes, while the independent variable was represented by gender.
Overall Dutch fashion industry network
Through Facebook the data on the number of total likes entrepreneurs have accumulated on their company page since its activity and the creation of their Facebook page was collected. The number of total likes consists of both individual and companies likes, which like the start-up’s company page in order to receive the latest updates on sales, product information and new collections. Therefore everyone who has liked the page is updated on all the relevant information of this business on their news feed; this allows us to consider people whom have liked the page as a network tie. In order for us to identify if female or males have the largest network we will use an ANOVA test to compare the total number of likes based on gender of the overall Dutch fashion industry (Appendix Image 18). The descriptive statistics of the ANOVA test show that male entrepreneurs have a much higher mean of 10436 total likes, compared to an average mean of 3389 total likes for female entrepreneurs and a mean of 4152
for co-sex couples. The ANOVA results show that with an F value of F (2, 1084) = 9.141 and a p = 0.000, our findings are considered significant. However, the Levene test of the ANOVA descriptive statistics shows (0.00) that the data is significant for unequal variances. This significance indicates that in order for us to known if the data is significant other tests need to be performed in order to validate the data, in this case the use of the Welch and Brown
Forsythe tests to test against the unequal variances is necessary. The standard deviation statistics also support the unequal variance claims made by the Levene statistic since, the standard deviation for male entrepreneurs (42690) is four times as high as those for female (11955) and co-sex couples (10815). The results of the Welch and Brown Forsythe test of 0.040 and 0.004 support the initial findings that the data is significant and that the hypothesis proposed should be accepted according to this dataset. Indicating that male entrepreneurs on average have more total likes on their company Facebook page compared to their female and co-sex colleagues. This finding supports McManus (2001) research that there are possible gender differences in the overall fashion industry business owner characteristics, supporting the decision to look at the performance of the entrepreneurs based on gender.
Start-up business network
Similarly to the overall Dutch fashion industry dataset, a one-way ANOVA test was performed to determine the significance of the dataset of the Dutch start-ups. The start-up sample is created since we cannot assume that the overall Dutch fashion industry sample is representative for the start-ups in the Netherlands. The descriptive statistics identify that male entrepreneurs similarly to the overall Dutch fashion industry dataset are higher number of total likes per Facebook company page with a mean of 10894.41 (Appendix Image 19). The female and co-sex entrepreneurs respectively lag behind on average with means of 3586.93 and 5527.31 considering their total like numbers. Similarly to the Dutch Fashion dataset, there is a statistically significant difference between the gender means as determined by one-way ANOVA shown by F (2, 779) = 5.629, p = 0.004. However, the Levene statistic was also significant (0.000); therefore the Welch and Brown Forsythe test were performed. Unlike the overall Dutch fashion industry results, the Welch and Brown Forsythe test indicate different p levels of .126 and .058. In this case the Welch test rejects the hypothesis while the Brown – Forsythe test barely accepts the hypothesis for this specific dataset. Many researchers indicate that if the two test show differentiating results, to accept in most of the situations the results of the Welch test over the results of the Brown Forsythe test. Considering the start-up dataset the
