The internal consistency and reliability of the main constructs was analysed using Cronbach’s alpha. Cronbach’s alpha is the most common measure of internal consistency. According to Gliem, J.A., & Gliem, R.R (
2003) while using Likert-type scales it is essential to calculate and report Cronbach’s alpha coefficient for internal consistency for any scales or subscales one may be using. According to them, “The analysis of the data then must use these summated scales or subscales and not individual items.”
11 A full list of all constructs and corresponding Cronbach’s alpha are shown on Table
1. To test the reliability of the measures, SPSS 16 was used.
Table 1
Summary of predictor Measures
SMP | 8 | 1–5 Likert scale | 0.91 |
Network Density | 6 | 1–5 Likert scale | 0.61 |
Network Centrality | 6 | 1–5 Likert scale | 0.65 |
Competitive network | 5 | 1–5 Likert scale | 0.77 |
Supportive network | 4 | 1–5 Likert scale | 0.76 |
Hierarchical regression analysis
To test the hypotheses given in section 4, this study used hierarchical regression analysis method. It is used as the main statistical procedure for examining the relationship between entrepreneurial networks and firm performance. Hierarchical regression is used when the data is organized into a tree-like structure which is shown in the research model Fig.
1. The data are stored as records which are connected to one another through links. In this study, the predictor variable and control variable is a link with each other. Therefore, sequential regression with performance as the dependent variable was tested. In the Model 1 entrepreneurs, demographic attributes such as age, gender, experience, and education were entered. In Model 2 structure of networks was entered and In Model 3 types of networks was added Accordingly, one reason to use Hierarchical regression instead of regular regressions because this method allows the researcher to separate within-group effect from between-group effect; Whereas Level 1 regression blends them together into a single coefficient (Veronika; Huta
2014). Before estimating the models, diagnostic test has been performed to detect the presence of multicollinearity. Value Inflation Factors (VIFs)
12 was calculated to examine multicollinearity among predictors. The check for multicollinearity revealed that the VIFs values were within the acceptable limit. To test the impact of entrepreneurial networks and firm performance, three regression models were formulated. Dependent Variable “Y” as indicator of subjective firm performance
(SFP). Independent variables are X
n as dimensions of entrepreneurial networks and firm related control variables (X
1= Gender, X
2 Age, X
3 Experience, X
4 Education, X
5 Network size, X
6 Density, X
7 Centrality, X
8 Competitive network, X
9 Supportive network). Therefore, the econometric models that are framed in order to capture the phenomenon are given below.
Model 1 = SFP = β1+ β2 Gender, +β3 Age + β4 Experience + β5 Education + u
Model 2 = SFP = β1+ β2 Gender, +β3 Age + β4 Experience + β5 Education + β6 Network size + β7Density + β8 Centrality+ u
Model 3 = SFP = β1+ β2 Gender, +β3 Age + β4 Experience + β5 Education + β6 Network size + β7Density + β8Centrality + β9 + Competitive network + β10 Supportive network +u
Three sequential regression models were estimated. In model 1, the effects of the demographic control variables were estimated. In model 2, the control variables and the main effects of network variables were estimated. In model 3, all variables were estimated. The results are discussed in the Table
3.
Table 3
Results of Hierarchical Regression models
Gender | .167 .333 1.188 | .137 .322 1.305 | .041 .764 1.470 |
Age | .615*** .004 1.820 | .575*** .002 2.067 | .640*** .001 2.129 |
Experience | −.210 .224 1.179 | −.283**.045 1.292 | −.247*.063 1.315 |
Education | −.264 .228 1.899 | 414**.020 1.997 | 457*** .008 2.085 |
Network size | | −.316***.003 1.579 | −.538 *** .004 2.404 |
Density | .393 **.022 1.867 | .455***.007 1.945 |
Centrality | .409***.008 1.463 | .388**.030 1.659 |
Competitive network | | .122 .441 1.985 |
Supportive network | .252 .150 2.353 |
F | 2.815** | 6.306*** | 6.275*** |
Multiple R | .522 | .788 | .833 |
R square | .273 | .620 | .693 |
Adjusted R square | .176 | .522 | .583 |
R square change | .273 | .348 | .073 |
Table
3 reveals that, in model 1, entrepreneur’s demographic control variables such as gender, age, experience and educational level were inserted. The multiple R
13 shows a substantial correlation between the predictor variables and the dependent variable
SFP (
R = .522);
p < .05. Here, R-squared value is .27 indicating that in the current stage the explanatory variables can explain only 27% of the variance in dependent variable. Regression results show that
Age found to be highly significant at 1% level of significance. The coefficient of
Age being positive, it can be concluded that given the other things,
SFP would increase with increase in age of the entrepreneurs. Regression model 2 and 3 shows entrepreneur’s
Experience is found to be statistically significant and its coefficient is negative (−.283**, −.247*). These results were not line with the traditional literature. According to Peake and Marshall (
2009) the reason for this is that most firms struggle during their initial earlier years of operation, which may not allow the experience of the entrepreneur to make a positive impact on the firm performance from an empirical stand point. Traditional experience in terms of age, maturity and life experience would be expected to positively impact the firm performance, but modern studies say that management experience is necessary to increase the performance of the firm. For example, if an entrepreneur who has experience in the technology industry but decides to launch a restaurant, than the start up experience he/she gained in the technology industry may not be helpful or useful at all in the restaurant performance. As shown in regression model 2 and 3 entrepreneurs
Education is positive and significant (414**, 457***).The coefficient of being positive, it can be concluded that given the other things,
SFP would increase with increase in education of the entrepreneurs.
In Model 2, along with control variables, network size, density and centrality were added. While controlling for the demographic variables, the results showed a significant improvement of overall multivariate relationship (R = 0.788; p < 0.001). The linear combinations of age, entrepreneurial experience, educational level, network size, density and centrality accounted for R-squared .62 which indicated that around 62% of the total variance in the dependent measure. After controlling the effect of demographic variables, a significant (R square change = 0.348; p < 0.001) degree of variance was explained by addition of network size, density and centrality.
In Model 3, along with Model 1 and Model 2 variables, and by adding competitive networks and supportive network, the result that was obtained, showed a significant improvement in the overall multivariate relationship (R = 0.833; p < 0.001) and the coefficient of determination R-squared has also improved. It is .69 which indicated that around 69% of variance in firm performance is explained by the independent variables. A significant R square change of .073 p < 0.001 was observed. The results are discussed in relation to the individual hypotheses.
Hypotheses1A proposed that network size has an impact on subjective firm performance. As shown in regression model 2 and 3 of Table
3, do not support hypothesis 1A. But the regression results (Table
3) show a negative (−.316***, −.538 ***) and highly significant relation. This happens because the total number of contacts of an entrepreneur increases but that also reduces efficiency in performance. There may be problems related to co-ordination, communication and decision making when the number of relationship increases which also ends up hindering performance.
Hypothesis 1B suggested that density of the network has an impact on subjective firm performance. As shown in model 2 and 3 of Table
3, density of network has a significant and positive impact on firm performance (.393**, .455***). Therefore, the results support the hypotheses1B.
Hypothesis 1C stated that network centrality has an impact on subjective firm performance. As shown in model 2 and 3 of Table
3, centrality of network has a significant and positive impact on firm performance (.409***, .388**). Therefore, the results support the hypotheses1C.
Hypothesis 2A and 2B argued that Competitive network and Supportive network has an impact on subjective firm performance. As shown in the regression model 3 of Table
3, the results does not support H2A and H2B that supportive and competition might not increase the performance.