Variance is a statistic indicating how widely values of given random variable are distributed. In a sense, it is a similar concept to width of range (difference between min and max) So in almost all cases, you can't say "A percent of the variance in X is due to the Var(X) for female drivers and B percent is the rest. A + B should be 100 percent" whether that variable is independent of gender.

If you want to run conventional ANOVA, then Peter Ellis's answer is the right approach. Otherwise, describe what you want to achieve in detail.