Page 85 - ISC PROCEEDINGS 21.4

Table 2. Descriptive statistics results

Source: Author
* Correlation analysis: The results show that the GE variable is correlated with
indicators representing research and innovation, business environment, human capital,
and economic growth.
Table 3. Correlation analysis results
SD DE CKH BPM CCN FDI SDN LD TT
SD 1.0000
DE 0.7941 1.0000
0.0061
CKH -0.5482 -0.4414 1.0000

0.1009 0.2016
BPM 0.9010 0.5726 -0.2450 1.0000
0.0004 0.0836 0.4950
CCN -0.5191 -0.0429 0.2392 -0.6722 1.0000
0.1242 0.9063 0.5056 0.0332
FDI 0.9144 0.7722 -0.8193 0.7029 -0.4663 1.0000
0.0002 0.0089 0.0037 0.0234 0.1743
SDN 0.9723 0.8579 -0.4672 0.8759 -0.4250 0.8629 1.0000
0.0000 0.0015 0.1734 0.0009 0.2209 0.0013
LD -0.9906 -0.7536 0.5540 -0.8911 0.5447 -0.9098 -0.9485 1.0000
0.0000 0.0118 0.0966 0.0005 0.1035 0.0003 0.0000
TT 0.9516 0.6897 -0.4111 0.9747 -0.6329 0.8242 0.9318 -0.9329 1.0000

0.0000 0.0273 0.2380 0.0000 0.0495 0.0034 0.0001 0.0001
Source: Author
* The initial regression model with the dependent variable SD and the independent
variables DE, CKH, BPM, CCN, SDN, LD, FDI, and TT showed R² = 0.9961 and adjusted R² =
0.9645. Multicollinearity testing using VIF revealed a very high degree of multicollinearity,
with BPM (VIF = 343.94), TT (VIF = 323.19), and FDI (VIF = 320.67) exceeding the warning
threshold. Due to the high degree of autocorrelation, the three variables BPM, TT, and FDI
were removed to reduce multicollinearity and improve the stability of the estimate.
* After removing variables with high autocorrelation levels, the regression model
with dependent variable SD and independent variables DE, LD, CKH, SDN, CCN yielded R²
= 0.9928 and adjusted R² = 0.9839, overall F statistic F(5,4) = 110.80, p = 0.0002, indicating
that the model is statistically significant. Multicollinearity testing using VIF showed that
the level of multicollinearity had decreased significantly compared to the original model,
although relatively high VIFs still existed in SDN (VIF = 22.28) and LD (VIF = 16.40), while
the remaining variables had low VIFs (DE = 7.61, CCN = 2.68, CKH = 1.64, Mean VIF =

84