Social capital, firm innovation, and family management

According to Coleman (Reference Coleman1990), Nahapiet and Ghoshal (Reference Nahapiet and Ghoshal1998), and Tsai and Ghoshal (Reference Tsai and Ghoshal1998), the concept of social capital encompasses the social interactions, network ties, trusting relations, and value systems that facilitate creativity in a group context. Social capital reflects the value of relationships and includes the interrelationship network and the assets that could be mobilised through that network (Nahapiet & Ghoshal, Reference Nahapiet and Ghoshal1998; Pearson, Carr, & Shaw, Reference Pearson, Carr and Shaw2008; Carr, Cole, Kirk-Ring, & Blettner, Reference Carr, Cole, Kirk-Ring and Blettner2011). It involves relationships among the individuals working in the organisation (internal social capital) as well as between the organisation and external parties (external social capital) (Adler & Kwon, Reference Adler and Kwon2002).

Several scholars propose that social capital facilitates the development of a distinctive knowledge base, thereby providing a foundation for the creation of firm innovation and competitive advantages (e.g., Adler & Kwon, Reference Adler and Kwon2002; Calantone, Cavusgil, & Zhao, Reference Calantone, Cavusgil and Zhao2002; Yli-Renko, Autio, & Tontti, Reference Yli-Renko, Autio and Tontti2002; Hult, Hurley, & Knight, Reference Hult, Hurley and Knight2004; Song & Thieme, Reference Song and Thieme2006). Additionally, social capital facilitates innovation by motivating cooperation and coordination among different members of the firm (Nahapiet & Ghoshal, Reference Nahapiet and Ghoshal1998; Adler & Kwon, Reference Adler and Kwon2002; Levin & Cross, Reference Levin and Cross2004; Leana & Pil, Reference Leana and Pil2006). According to Baregheh, Rowley, and Sambrook’s (Reference Baregheh, Rowley and Sambrook2009) and Craig and Moores’s (Reference Craig and Moores2006), innovation is a multi-stage process whereby organisations transform ideas into new/improved products, services, or processes in order to advance, compete, and differentiate themselves successfully in the marketplace. Hence, innovation is an essential process for the success, survival, and renewal of organisations. The exchange of ideas and knowledge and the existence of certain relationships among employees contribute to enhancing a firm’s ability to identify and develop new opportunities that could not be identified and developed otherwise (Subramaniam & Youndt, Reference Subramaniam and Youndt2005; Carrasco-Hernández & Jiménez-Jiménez, Reference Carrasco-Hernandez and Jimenez-Jimenez2013). Leana and Pil (Reference Leana and Pil2006) argue that social capital is not just the network itself or the links among the people that constitute it; rather, it is the resources created by the existence and characteristics of these links, such as information sharing and trust. Thus, intra-firm collaboration is considered important for innovation because network relationships provide channels through which knowledge can flow (e.g., Kostova & Roth, Reference Kostova and Roth2003; Subramaniam & Youndt, Reference Subramaniam and Youndt2005; Gupta, Tesluk, & Taylor, Reference Gupta, Tesluk and Taylor2007). Intra-organisational knowledge sharing also influences a firm’s capacity to innovate because it supports creativity and inspires new knowledge and ideas (Song & Thieme, Reference Song and Thieme2006; Aragon-Correa, García-Morales, & Cordon-Pozo, Reference Aragon-Correa, García-Morales and Cordon-Pozo2007).

Although social capital exists in all types of organisations, family firms – firms in which a family possesses a significant ownership stake and in whose operations multiple family members are involved (Sirmon, Arregle, Hitt, & Webb, Reference Sirmon, Arregle, Hitt and Webb2008; Chirico, Sirmon, Sciascia, & Mazzola, Reference Chirico, Sirmon, Sciascia and Mazzola2011) – share many similarities. Social structures and affective commitments are particularly salient in family firms because of the intersection of the family and business systems (e.g., Sirmon & Hitt, Reference Sirmon and Hitt2003). Family firms have their own idiosyncratic characteristics, such as family language, motivation, loyalty, and trust, which allow their members to communicate more efficiently and exchange more information with greater privacy (Tagiuri & Davis, Reference Tagiuri and Davis1996; Anderson & Reeb, Reference Anderson and Reeb2003; Botero et al., Reference Botero, Thomas, Graves and Fediuk2013; Vallejo-Martos & Puentes-Poyatos, Reference Vallejo-Martos and Puentes-Poyatos2014). Habbershon, Williams, and MacMillan (Reference Habbershon, Williams and MacMillan2003) refer to the idiosyncratic bundle of resources and capabilities possessed by family firms and resulting from familial interactions as the ‘familiness’ of the firm. Thus, family involvement is an important factor, which influences the interactions between the firm and family (Zahra, Reference Zahra2005). Family involvement may provide a firm with unique abilities to act idiosyncratically (Chrisman & Patel, Reference Chrisman and Patel2012) and, thereby, influence innovative decision making (Chrisman et al., Reference Chrisman, Chua, De Massis, Frattini and Wright2015). This idea agrees with existing studies on family firm innovation, which suggest that the development of new products, services, and processes results from creative cooperation (Nahapiet, Reference Nahapiet2009) and family involvement (e.g., Le Breton-Miller & Miller, Reference Le Breton-Miller and Miller2006; Zahra, Neubaum, & Larrañeta, Reference Zahra, Neubaum and Larrañeta2007).

Although the majority of family firms involve a family group as well as a non-family group (Mitchell, Morse, & Sharma, Reference Mitchell, Morse and Sharma2003; Miller, Le Breton-Miller, Minichilli, Corbetta, & Pittino, Reference Miller, Le Breton-Miller, Minichilli, Corbetta and Pittino2014), which form two distinct but complementary groups that coexist in family businesses (Sanchez-Famoso, Maseda, & Iturralde, Reference Sanchez-Famoso, Maseda and Iturralde2013, Reference Sanchez-Famoso, Maseda and Iturralde2014), few analytical studies in the extant family firm literature consider these two social groups separately (e.g., McCollom, Reference McCollom1992; Ram, Reference Ram2001; Arregle et al., Reference Arregle, Hitt, Sirmon and Very2007; Sanchez-Famoso, Maseda, & Iturralde, Reference Sanchez-Famoso, Maseda and Iturralde2014). With respect to the relationship between family and non-family social capital and innovation, Sanchez-Famoso, Maseda, & Iturralde (Reference Sanchez-Famoso, Maseda and Iturralde2014) show that family social capital encourages a culture of long-term goal orientation, trust, and mutual understanding, while non-family social capital also has a positive effect on family firm innovation because of the diversity and professionalism of the non-family group members. Further, Sanchez-Famoso, Maseda, & Iturralde (Reference Sanchez-Famoso, Maseda and Iturralde2014) research suggests that non-family social capital could have a greater influence on family firm innovation.

Moderating effects of family involvement in management

Recent research has focussed on family firms as heterogeneous organisations since the relationships within them are not expected to be the same (e.g., Astrachan, Klein, & Smyrnios, Reference Astrachan, Klein and Smyrnios2002; Sharma, Reference Sharma2004; Miller & Le Breton-Miller, Reference Miller and Le Breton-Miller2006; Melin & Nordqvist, Reference Melin and Nordqvist2007; Howorth, Rose, Hamilton, & Westhead, Reference Howorth, Rose, Hamilton and Westhead2010; Chrisman et al., 2015; De Massis et al., Reference De Massis, Kotlar, Chua and Chrisman2014; Nordqvist, Sharma, & Chirico, Reference Nordqvist, Sharma and Chirico2014). In other words, while sharing several characteristics, family firms are not homogeneous in all aspects. They vary significantly regarding their goals, motivations, and uniformity of relationships (Miller & Le Breton-Miller, Reference Miller and Le Breton-Miller2006; Chrisman et al., Reference Chrisman, Chua, De Massis, Frattini and Wright2015). Melin and Nordqvist (Reference Melin and Nordqvist2007) identify different types of family firms based on family involvement, business, and family differences. Similarly, De Massis et al. (Reference De Massis, Kotlar, Chua and Chrisman2014) discuss why family involvement may be used to explain family firms’ heterogeneity. Moreover, family SMEs are governed and managed through key personal relationships among family and non-family members that could affect intra-organisational and inter-organisational relationships (Arregle et al., Reference Arregle, Hitt, Sirmon and Very2007).

The TMT – defined as the chief executive officer (CEO) and the senior managers who directly report to the CEO (Boeker, Reference Boeker1997) – is widely recognised as one of the most important decision-making units in organisations (Sanchez-Marin & Baixauli-Soler, Reference Sanchez-Marin and Baixauli-Soler2015; Vandekerkhof et al., Reference Vandekerkhof, Steijvers, Hendriks and Voordeckers2015), which is responsible for innovation-related decisions in firms (Talke, Salomo, & Rost, Reference Talke, Salomo and Rost2010). Hence, the involvement of family members in the TMT may have distinct consequences for the innovation of family firms (e.g., Miller, Le Breton-Miller, Lester, & Cannella, Reference Miller, Le Breton-Miller, Lester and Cannella2007; Westhead & Howorth, Reference Westhead and Howorth2007; Howorth et al., Reference Howorth, Rose, Hamilton and Westhead2010). In Hambrick and Mason’s (Reference Hambrick and Mason1984) work, in which they analysed ‘upper echelons perspective’, they pointed out that organisational outcomes could be predicted from managerial backgrounds, which are reflected in strategic outcomes. Therefore, TMT composition could influence innovation processes and outcomes (Hambrick & Mason, Reference Hambrick and Mason1984; Hambrick, Reference Hambrick2007; Sanchez-Marin & Baixauli-Soler, Reference Sanchez-Marin and Baixauli-Soler2015), especially in family SMEs, where senior managers direct the firm’s strategy and participate directly in the implementation of that strategy (Lubatkin et al., Reference Lubatkin, Simsek, Ling and Veiga2006). However, TMT research in the context of family firms remains scarce, despite the fact that these managers are closer to the firm’s existing competencies and, therefore, are knowledgeable about when and how to exploit them (Lubatkin et al., Reference Lubatkin, Simsek, Ling and Veiga2006; Basco & Voordeckers, Reference Basco and Voordeckers2015).

Family firms that pursue an innovative strategy need to be receptive to environmental change, and the inclusion of non-family managers could widen the available pool of expertise in the TMT (Chirico & Salvato, Reference Chirico and Salvato2008; Block, Reference Block2010; Sanchez-Marin & Baixauli-Soler, Reference Sanchez-Marin and Baixauli-Soler2015; Vandekerkhof et al., Reference Vandekerkhof, Steijvers, Hendriks and Voordeckers2015). Non-family involvement in management signals an inclusive working environment where multiple perspectives are appreciated and considered (Kellermanns & Eddleston, Reference Kellermanns and Eddleston2004). Moreover, family managers and non-family managers are socialised in different ways, and this different socialisation experience is likely to have an influence on a firm’s strategy (Block, Reference Block2011). The dispersion of power across family and non-family managers could lead to the modernisation of the firm’s objectives and strategies.

The authors are aware that there are contrary arguments. For example, family involvement could orient the firm management to a long-term future, and innovation use could be related to long-term decisions. However, the extant research reveals the negative consequences arising from family involvement in management, which could be especially relevant when such involvement is very high (De Massis, Kotlar, Campopiano, & Cassia, Reference De Massis, Kotlar, Campopiano and Cassia2015), suggesting that the higher level of non-family members in TMT means that internal social capital positively influences family firm innovation (Sanchez-Marin & Baixauli-Soler, Reference Sanchez-Marin and Baixauli-Soler2015). In the context of social capital, family involvement in management could reduce the benefits of family social capital and non-family social capital for innovation in several ways. First, family members who are involved in day-to-day management tend to acquire and develop specific, in-depth knowledge (e.g., Carney, Reference Carney2005) within a narrow market niche (Cabrera-Suarez, De Saa-Perez, & Garcia-Almeida, Reference Cabrera-Suarez, De Saa-Perez and Garcia-Almeida2001; Westhead & Howorth, Reference Westhead and Howorth2006). This could lead to a desire to accommodate other team members for the ‘good’ of the team (Amason & Sapienza, Reference Amason and Sapienza1997) and ‘groupthinking’ (Arregle et al., Reference Arregle, Hitt, Sirmon and Very2007), thereby compromising the people’s ability to generate alternative ideas (Arregle et al., Reference Arregle, Hitt, Sirmon and Very2007). Consequently, excessive levels of family involvement in the TMT could result in the limited availability of diverse knowledge and multiple perspectives, which would limit innovation (Ruekert & Walker, Reference Ruekert and Walker1987; Handler, Reference Handler1992; Howorth et al., Reference Howorth, Rose, Hamilton and Westhead2010).

Second, high family involvement in management could lead to a more collectivistic orientation of senior managers, which would encourage them to consider the effect of their actions on other family members and would reduce the firm’s risk exposure in order to preserve family wealth (Cabrera-Suarez & Martin-Santana, Reference Cabrera-Suarez and Martin-Santana2013) and to maintain a familial atmosphere (Berrone, Cruz, Gomez-Mejia, & Larraza-Kintana, Reference Berrone, Cruz, Gomez-Mejia and Larraza-Kintana2010), thereby leading to resistance towards nonfamily managers. However, the generational perspective maintains that the degree of collectivistic orientation and that of risk exposure change as the firm moves through generations (Gersick, Davis, Hampton, & Lansberg, Reference Gersick, Davis, Hampton and Lansberg1997). In their analysis of Polish family firms, Kowalewski, Talavera, and Stetsyuk (Reference Kowalewski, Talavera and Stetsyuk2010) show that the social capital provided by first-generation managers is likely to benefit firms, resulting in better performance, because of the managers’ expertise with respect to the firm. It has been shown that first-generation family founders and managers have the necessary background to create a business (Aldrich & Cliff, Reference Aldrich and Cliff2003) because, in the first generation, family managers usually perceive a high level of commitment and trust (Vallejo-Martos & Puentes-Poyatos, Reference Vallejo-Martos and Puentes-Poyatos2014), which pushes them towards similar perspectives and common shares, that is, for new ways of doing thing (Kepner, Reference Kepner1991). These first-generation family founders and managers are often the driving force behind innovation. However, as the number of family members involved in management increases, different perspectives arise, and these differences can engender conflict (Gersick et al., Reference Gersick, Davis, Hampton and Lansberg1997). As Kellermanns, Eddleston, Barnett, and Pearson’s (Reference Kellermanns, Eddleston, Barnett and Pearson2008) work pointed out, such conflict persists and surfaces in most aspects of their relationship, including those of both family and non-family members. They can generate tension, irritation, suspicion, and resentment among organisational members, undermining the potential advantages of group interaction and reducing the family firm’s effectiveness by preventing cross-understanding of different individual team members (Chirico & Salvato, Reference Chirico and Salvato2016).

Third, innovation entails a greater involvement of specialised human capital, managerial talent, and expertise, which are normally not available within the family group, especially in family SMEs (Chrisman & Patel, Reference Chrisman and Patel2012). Therefore, a high degree of family management can deliver weaker collective cognition and disband family managers in the discussions, developments, and decisions affecting both the family and business (Kellermanns & Eddleston, Reference Kellermanns and Eddleston2004; Miller & Le Breton-Miller, Reference Miller and Le Breton-Miller2005; Sanchez-Marin & Baixauli-Soler, Reference Sanchez-Marin and Baixauli-Soler2015). Therefore, problems that are frequently present in family firms, such as centrality of control, conflicts of opinions, and resistance to ideas about new products, services, and processes, are likely to be mitigated when both family and non-family members are involved in management. In addition, non-family managers may contribute significantly by advising on family topics that affect the enterprise over time (Basco & Voordeckers, Reference Basco and Voordeckers2015).

We agree with Vandekerkhof et al. (Reference Vandekerkhof, Steijvers, Hendriks and Voordeckers2015) that non-family managers are important stakeholders who could add knowledge, solve family succession problems, or even mediate family conflict (Dyer, Reference Dyer1989; Klein & Kozlowski, Reference Klein and Kozlowski2000; Sonfield & Lussier, Reference Sonfield and Lussier2009; Block, Reference Block2011; Basco & Voordeckers, Reference Basco and Voordeckers2015). The inclusion of non-family members in the management team increases the social capital of firms (Portes, Reference Portes1998), and it facilitates the acquisition and promotion of information from diverse sources (Blyler & Coff, Reference Blyler and Coff2003), with a positive effect on innovation. The information base of non-family managers is expected to be different and higher than that of family managers, thereby affecting new products, services, and/or processes (Shane, Reference Shane2003). Considering these arguments, we propose the following hypotheses (see Figure 1).

Figure 1

Standardised path loading for hypothesised model

Sample and data collection

We chose Spanish firms for this study because according to the estimates provided by the Spanish Family Enterprise Institute (2009), there are around 2.9 million family enterprises (out of a total of 3.4 million enterprises) in Spain, which generate 70% of the total Spanish gross domestic product (GDP) and employ nearly 13.9 million people (representing around 75% of total private employment in Spain). The final report of the European expert group in the field of family enterprises (GEEF, 2009) reports that in most nations for which the data were compiled (the United States, Europe, Asia, and Australia), more than 60% of all firms were classified as family firms. These data show that Spain has more family enterprises than the average share.

Given the lack of a reliable business census that clearly distinguishes between family and non-family firms in Spain, we used the Iberian Balance Sheet Analysis System (SABI)Footnote ¹ database to establish the population of family businesses in this study. We used demographic parameters (Shanker & Astrachan, Reference Shanker and Astrachan1996) to identify the sample used in this study. To obtain a sample, we selected the following types of firms from the SABI database: (1) non-listed Spanish companies with more than 10 employees – because the presence of at least 10 employees could reinforce communication at work (Sorenson, Reference Sorenson2012); (2) companies that provided financial information for (at least) the last 5 years; (3) companies not affected by special situations such as insolvency, wind-up, liquidation, or zero activity (in order to obtain a sample that was representative of the population); and (4) companies in which at least 51% of the firm is owned by members of the same family (Westhead & Howorth, Reference Westhead and Howorth2006). These criteria were applied to the SABI database, along with an exhaustive review of shareholder structures (percentage of common stock) and composition (names and surnames of shareholders)Footnote ² . From the original dataset of 26,064 Spanish enterprises that fell within the set parameters, 1,122 enterprises met the specified family criteria.

It was necessary to survey firms because secondary data for Spanish non-listed family firms were not readily available, especially in relation to the factors of theoretical interest in this study (relationships within the family group, relationships within the non-family group, and innovation perceptions). The CEOs of the selected firms were sent letters in December 2012, requesting the participation of one family member as well as one non-family member (both with managerial functions) in the study. In the letter, we explained that a professional survey research firm would get in touch with them in the following 30–60 days to conduct telephonic interviews in order to complete the questionnaires. To ensure a high response rate and reliable and accurate responses, the CEOs were promised that the information about the respondents and the company would remain strictly confidential. Additionally, they were promised a summary of the study’s results. Two weeks after the first letters were sent, another set of letters was sent to remind them about the study. Finally, in January 2013, 1 week after the reminders were sent, a professional survey research firm was commissioned to conduct telephonic interviews. In each surveyed family firm, the professional survey research firm encouraged participation provided a reminder about the importance and purpose of the study, and guaranteed confidentiality. Information was collected from a family member (to collect information about the relationships among family members and innovation) and from a non-family member (to collect information about relationships among non-family members), both with managerial functions. Since people with managerial functions are directly involved in day-to day operations and have first-hand information about what is going on, they are the appropriate respondents for studies of firm-level processes such as social capital and innovation. We also looked for family firms in which more than one family member and more than one non-family member worked in order to understand the relationships among family members and among non-family members (Sorenson, Reference Sorenson2012). The professional research company assisted us in meeting this requirement by asking the family member and/or the non-family member (with managerial functions) whether there were at least two family members and/or two non-family members working in the family firm. As Brass, Galaskiewicz, Greve, and Tsai (Reference Brass, Galaskiewicz, Greve and Tsai2004: 801) point out, ‘when two individuals interact, they not only represent an interpersonal tie, but they also represent the groups of which they are members’.

After reviewing the most relevant literature on the topics relevant to this study, the questionnaire was designed in three parts. The first part requested general information about the family firm. The second part contained questions addressed to family members, and the final part included questions addressed to the non-family members of the family firm. The questionnaire consisted of closed questions, using Likert-type scales (ranging from ‘strongly disagree’ to ‘strongly agree’), and dichotomous questions. We tested the questionnaire in seven family firms from different industrial sectors, conducting face-to-face interviews, first with a family member and subsequently with a non-family member in each firm (both with managerial functions). Their comments on the survey content, item wording, terminology, and clarity were incorporated into a revised instrument. The refined instrument was piloted on eight family firms, with one family member and one non-family member with managerial functions in each firm, and final revisions were made. These revision efforts created a highly reliable instrument (Cronbach’s α ranging from 0.71 to 0.85) (see Appendix I).

Measurement of model variables

For this study, the majority of the items were measured using 5-point Likert scales ranging from one (strongly disagree) to five (strongly agree). Consistent with our research focus and similar prior approaches assessing firm-level phenomena based on individual-level responses (e.g., De Clerq, Dimov, & Thongpapanl, 2010), the survey questions were worded in order to capture the attitudes and behaviours occurring at the firm level rather than at the managerial level (Brass et al., Reference Brass, Galaskiewicz, Greve and Tsai2004; Whetten, Felin, & King, Reference Whetten, Felin and King2009). Thus, in relation to the questions on social capital and firm innovation, the respondents offered opinions about the interactions and relationships within their firm in general, instead of focussing on their own individual situations.

Control variables

We acknowledge the importance of factors that could affect a family firm’s innovation by including several specific control variables (business size, business age, business sector, life stage, and long-term orientation)Footnote ³ in our model. Although there are undoubtedly many other factors, these specific variables were chosen because of their importance. First, business size is included because prior research has demonstrated that a family firm’s size could be linked to its greater or lesser tendency for innovation (e.g., Bantel & Jackson, Reference Bantel and Jackson1989). Business size was measured using the natural log of total assets. Second, business age was measured using the natural log of a firm’s age, calculated as the difference between the year in which the survey was conducted (2013) and the year in which the firm was founded. Third, the influence of the business sector on innovation was captured using the operationalisation of the business sector as a dummy variable, which was coded 1 for manufacturing firms and 0 for service firms following the Spanish industrial classification method (National Classification of Economic Activities: CNAE)Footnote ⁴ . Fourth, the respondents provided the life stage using Adizes’s (Reference Adizes1979) framework. According to Dyer (Reference Dyer1986), it is common to find family firms with more innovative perspectives during the first and second stages of their lifecycle. Finally, because our firms are all family owned (i.e., most of the equity is owned by a family), we controlled for long-term orientation. When a family focusses on the long term, the business is more secure, and the firm is often less willing to undergo innovation. In this context, we asked the family members whether they would like the next generations to continue running the firm (yes/no) (Vallejo-Martos, Reference Vallejo-Martos2009).

Of the 769 firms that were contacted, 232 completed the survey. However, given the multiple-informant design of the survey, family firms with only one respondent were removed from the sample. Consequently, 60 family firms were eliminated from further consideration (28 family firms in which only a family member responded, and 32 family firms in which only a non-family member responded). A total of 172 questionnaires were found to be complete. The response rate was 22.37%, which is similar to that of other studies in the Spanish context (e.g., Cruz & Nordqvist, Reference Cruz and Nordqvist2012; Vallejo-Martos & Puentes-Poyatos, Reference Vallejo-Martos and Puentes-Poyatos2014). The overall sampling error was 6.53%, with a confidence level of 95%, and p=q=0.50. With regard to the respondents, in the family group, 50.50% were CEOs, and 49.50% were managers from different areas. In the non-family group, 49.50% of the respondents were financial managers, 41.70% were managers from other areas, and 8.80% were CEOs. A brief analysis of our sample shows that only 6% of the family firms have been in existence for less than 10 years (these are mainly the family firms in the sample of consolidated companies). In other words, most of the firms in our sample have passed through their early years and have had the opportunity to establish relationships among their employees.

With regard to firm size, 23% of the firms have between 10 and 50 employees, 72% have between 50 and 250 employees, and only 5% have over 250 employees. Thus, most of the sample firms are SMEsFootnote ⁵ . In terms of firm size, the sample has an average size of 93 employees and a median of 67. When we consider the generations involved in joint management, the firms are managed by only one generation in 56% of the cases, by two generations in 40% of the cases, and by three or more generations in 4% of the cases. With regard to the generation of the firms, 37% are in their first generation, 41% are in their second generation, 12% are in their third generation, 8% are in their fourth generation, and 2% are in their fifth or older generation. Finally, 48% of the firms in the sample belong to the manufacturing sector and 52% to the service sector (see Table 1). Therefore, the descriptive statistics for our sample are in line with those of prior studies on Spanish non-listed family firms (e.g., Cabrera-Suarez, Deniz-Deniz, & Martin-Santana, Reference Cabrera-Suarez, Deniz-Deniz and Martin-Santana2015; Deniz-Deniz & Cabrera-Suarez, Reference Deniz-Deniz and Cabrera-Suarez2005). Table 2 presents the descriptive statistics and correlations for all the variables analysed in this studyFootnote ⁶ .

Table 1

Sample characteristics

Table 2

Descriptive statistics and correlations

Notes. n=172. Correlations greater than І0.19І are significant at p<.01. Correlations greater than І0.14І are significant at p<.05

To assess non-response bias, we randomly selected 50 non-responding firms. Using the Kolmogorov–Smirnov two-sample test (Siegel & Castellan, Reference Siegel and Castellan1988) between responding and non-responding firms (whose data was taken from the SABI database), the non-response analysis found no statistically significant differences with respect to firm size, age, performance, annual sales volume, number of employees, or sector.

Data analysis

The proposed models were tested using structural equation modelling (SEM). Although SEM is an increasingly popular approach in business research and the related social sciences, family firm researchers have used this method sparingly (Wilson, Whitmoyer, Pieper, Astrachan, Hair, & Sarstedt, Reference Wilson, Whitmoyer, Pieper, Astrachan, Hair and Sarstedt2014). Several family firm researchers have called for the use of more sophisticated and rigorous statistical analysis techniques, including SEM (e.g., Westhead & Howorth, Reference Westhead and Howorth2006; Debicki, Matherne, Kellermanns, & Chrisman, Reference Debicki, Matherne, Kellermanns and Chrisman2009; Dyer & Dyer, Reference Dyer and Dyer2009). In fact, with regard to statistics in family firm research, SEM represents an advanced version of the general linear modelling procedures (e.g., multiple regression analysis) (Astrachan, Patel, & Wanzenried, Reference Astrachan, Patel and Wanzenried2014). Lei and Wu (Reference Lei and Wu2007) claim that SEM can be used to assess ‘whether a hypothesised model is consistent with the data collected to reflect [the] theory’ (p. 34). In particular, partial least squares structural equation modelling (PLS-SEM), which is a variance-based technique, is preferred to covariance-based techniques (e.g., AMOS, EQS) and linear modelling (e.g., multiple regression analysis) in several contexts: first, when dealing with latent constructs that cannot be observed or measured directly (Astrachan, Patel, & Wanzenried, Reference Astrachan, Patel and Wanzenried2014); second, when the moderating effects are being analysed (Chin, Marcolin, & Newsted, Reference Chin, Marcolin and Newsted2003); third, when multiple items have been used to represent each construct (Baron & Kenny, Reference Baron and Kenny1986; McClelland & Judd, Reference McClelland and Judd1993; Iacobucci, Saldanha, & Deng, Reference Iacobucci, Saldanha and Deng2007); and finally, when the data are not normally distributed, and the sample size is small (Alavifar, Karimimalayer, & Anuar, Reference Alavifar, Karimimalayer and Anuar2012; Astrachan, Patel, & Wanzenried, Reference Astrachan, Patel and Wanzenried2014). PLS-SEM should be applied for samples with fewer than 250 observations.

We used two methods to analyse the data collected in this research. On the one hand, SPSS software was used to compute the descriptive statistics (means, standard deviations) and the correlation matrix. On the other hand, partial least squares (PLS) was used in addition to SPSS to calculate the reliability and validity indexes of the variables and to test the research hypotheses through SEM. Specifically, we used SmartPLS 2.0.M3 software (Ringle, Wende, & Will, Reference Ringle, Wende and Will2005) for Windows.

Overall, the hypotheses described in this study represent the moderating or interaction effects. To test these moderated relationships, a well-known technique was applied: the product-indicator approach (Henseler & Fassott, Reference Henseler and Fassott2010). This approach involves a straightforward modelling of the moderating effects when the moderator linearly influences the strength of the moderated direct relationship (Henseler & Fassott, Reference Henseler and Fassott2010).