Table 1.

Variable importance (VI) scores and regression statistics of nine independent variables tested against the dependent variable Individualism (values of collectivism-individualism for n = 66 countries) ranked according to their VI scores. In machine learning, the most important variable is always given a 100% importance score.

Independent variable	VI untransformed	VI Box-Cox transformed	Spearman-rank	Regression untransformed	Regression Box-Cox transformed
Hist Path	100.0	100.0	−0.66, < 0.0001	−0.69, 0.47, < 0.0001	−0.68, 0.47, <0.0001
GDP	74.9	73.0	0.63, < 0.0001	0.66, 0.43, < 0.0001	0.65, 0.42, <0.0001
Infect Outbreaks	52.2	55.0	0.47, 0.0002	0.50, 0.25, < 0.0001	0.41, 0.17, 0.0007
Rich Path	49.6	53.9	−0.54, < 0.0001	−0.45, 0.20, 0.0002	−0.48, 0.23, <0.0001
Zoo Outbreaks	26.1	27.3	0.40, 0.001	0.43, 0.18, 0.0004	0.37, 0.14, 0.002
Area	23.5	23.7	−0.02, 0.87	0.13, 0.02, 0.31	0.02, 0.0002, 0.90
Surveys	21.6	23.3	0.15, 0.23	0.14, 0.02, 0.26	0.13, 0.02, 0.31
Population	19.6	22.2	−0.11, 0.36	−0.09, 0.008, 0.47	−0.11, 0.01, 0.37
EID Events	14.7	15.8	−0.11, 0.36	0.34, 0.12, 0.005*	0.21, 0.04, 0.09$

For the Spearman rank correlation, we report the rho-value and p-value. For the linear regression, we report the standard coefficient, the r²-value and the p-value; ‘untransformed’ refers to the raw data being used, while ‘Box-Cox transformed’ refers to all variables having been Box-Cox transformed. Negative correlations are printed in italic letters.

^*This significant p-value is due to the three highest values for EID Events, which are from Australia, the United Kingdom and the United States of America, and which are clear outliers; without them, the values in the respective cell are: 0.15, 0.02, 0.23. ^$Therefore, even when these three outliers are included, the significant association disappears in the analysis with Box-Cox transformed variables.