CHAPTER IV
EMPIRICAL RESULTS Chapter IV Empirical Results
• Descriptive statistics analysis
• Inferential statistics
• Supplement analysis
- 37 -
, 4. Empirical Results
4.1 Descriptive statistics analysis
As the prerequisites of any statistical analysis, the Pearson Correlation test was
performed (Table 3). Considering the rule of thumb and correlation range between -1 and 1, an
obtained result indicates that variables are neutrally and/or moderately correlated in the dataset.
The correlation matrix and scatterplot depict a linear and positive, moderate correlation
coefficient (ρ) of DV, IV, and control variables that show the value of .553** for subsidiary
DC and exploitation (Appendix III-IV). Similarly, the correlation between DC and exploration
is .477 (p = 0.01), implying that its exploration automatically increases when the subsidiary’s
DCs increase (Table 3). These relationships are neutral/moderate and positive. Yet the
correlation matrix also displays two highly correlated coefficients for exploitation with
exploration (.758**) and IP (.937**). However, these are dependent variables so a high
coefficient can be safely ignored (Keller & Gaciu, 2019). Besides that, it is crucial to highlight
that the IP reflects the mean of exploration and exploitation, resulting in a high correlation. The
correlation coefficients of the independent and other control variables in this study are below
0.7, meaning that the information in those variables is not largely redundant. It will not make
the regression analysis unstable (Keller & Gaciu, 2019; Research Project, 2021). As a result, it
can be stated that the collinearity issue does not exist in this study. In addition, the variance
inflation factor (VIF) for all models lies below 1.5, confirming that all predictors are orthogonal
(Appendix V). The correlation for exploitation is ρ = .553 (p = 0.01), meaning that the increase
in the subsidiary’s DCs also increases its exploitation (Appendix III-IV, Table 3). Also, it can
be noticed that IVR (ρ=.114) positively and significantly correlates with exploitation,
exploration (ρ= 0.99), and IP (ρ =.112). A similar correlation occurs with the GE, having ρ
=.099, ρ =.096, and r =.104 for exploration, exploitation, and IP, respectively.
- 38 -
EMPIRICAL RESULTS Chapter IV Empirical Results
• Descriptive statistics analysis
• Inferential statistics
• Supplement analysis
- 37 -
, 4. Empirical Results
4.1 Descriptive statistics analysis
As the prerequisites of any statistical analysis, the Pearson Correlation test was
performed (Table 3). Considering the rule of thumb and correlation range between -1 and 1, an
obtained result indicates that variables are neutrally and/or moderately correlated in the dataset.
The correlation matrix and scatterplot depict a linear and positive, moderate correlation
coefficient (ρ) of DV, IV, and control variables that show the value of .553** for subsidiary
DC and exploitation (Appendix III-IV). Similarly, the correlation between DC and exploration
is .477 (p = 0.01), implying that its exploration automatically increases when the subsidiary’s
DCs increase (Table 3). These relationships are neutral/moderate and positive. Yet the
correlation matrix also displays two highly correlated coefficients for exploitation with
exploration (.758**) and IP (.937**). However, these are dependent variables so a high
coefficient can be safely ignored (Keller & Gaciu, 2019). Besides that, it is crucial to highlight
that the IP reflects the mean of exploration and exploitation, resulting in a high correlation. The
correlation coefficients of the independent and other control variables in this study are below
0.7, meaning that the information in those variables is not largely redundant. It will not make
the regression analysis unstable (Keller & Gaciu, 2019; Research Project, 2021). As a result, it
can be stated that the collinearity issue does not exist in this study. In addition, the variance
inflation factor (VIF) for all models lies below 1.5, confirming that all predictors are orthogonal
(Appendix V). The correlation for exploitation is ρ = .553 (p = 0.01), meaning that the increase
in the subsidiary’s DCs also increases its exploitation (Appendix III-IV, Table 3). Also, it can
be noticed that IVR (ρ=.114) positively and significantly correlates with exploitation,
exploration (ρ= 0.99), and IP (ρ =.112). A similar correlation occurs with the GE, having ρ
=.099, ρ =.096, and r =.104 for exploration, exploitation, and IP, respectively.
- 38 -
