To add to this, more data from more regions means the estimate of average human miles without an incident is more accurate, simply because it is estimated from a larger sample, so more likely to be representative.