I don’t know how you’d measure driving “goodness”, but I expect the distribution would be something like exponential (there are billions of non-drivers, and only a few rally/stunt drivers). So the average is likely to be higher than the median.