Narxoz Student Research-2021
Figure 3: Scatterplot matrix for the re
fi
ned linear regression model (RStudio)
As we can see from the above plot, all the variable pairs have
relatively high correlations, except for the migration
(migr1
) variable.
Considering the relatively high value of the adjusted
R
2
= 0.985
(see
Fig. 1) and presence of six highly correlated pairs (Fig. 2), we suspect
the multicollinearity. In order to see, which variables specifi cally cause
the multicollinearity, let’s run the individual multicollinearity test in
RStudio:
Figure 4: Individual multicollinearity diagnostics via mctest (RStudio)
According to variance infl ation factor (VIF) and tolerance (TOL)
tests, the only variable that is most likely not to cause multicollinearity
is migration. The rest variables, including foreign investment, salary,
287
Достарыңызбен бөлісу: | 
