4. Analysis and results
In this study, we applied a hierarchical regression model to test our hypotheses. In the extended models which included the moderating variable, the independent variable highly correlated with the interaction term. Therefore, these variables were meancentred (Aiken & West, 1991; Cohen, Cohen, West, & Aiken, 2003). We computed variance inflation factors (VIF) to assess whether there was a multicollinearity problem in our sample. Since all values were lower than 4.95, there was no problem of multicollinearity in our analyses (Kennedy, 2008). To test for heteroscedasticity, we performed a White’s general test and a CookWeisberg test. Both indicated there were no problems concerning heteroscedasticity. Table 2 provides means, standard deviations, and correlations of variables. On average, firms have 63 employees. Furthermore, the independent director ratio is relatively low (10%) and the average board size is 4.50 members, which is comparable to other studies conducted in the context of family firm boards (Basco & Voordeckers, 2015; Collin & Ahlberg, 2012; Zattoni et al., 2015). The average faultline strength included in our sample is 0.05 and the maximum faultline strength measured is 0.22. According to Shaw (2004), these numbers indicate that the average faultline strength is rather weak, but that rather strong faultlines are also present in the upper quartile, showing a wide variance of faultline strengths in our sample.3 Furthermore, our average faultline strength is comparable with other studies that calculate faultline strength using Shaw’s method, as these average values range from 0.03 to 0.14 (e.g., Choi & Sy, 2010; Meyer & Glenz, 2013).
