Since large numbers of people (bottom several deciles) are doing very poorly, it means that if mean indicators would be stable, then someone else must be doing very good; and even more so if average indicators up.
It could also mean that 80% of people are better than average according to their preferred metric of good driving but that everyone has a different metric.
A few rich people on the "top end" can screw up the average, but they don't do very much to the median. If the median is higher that's pretty strong evidence that they are, as a whole, outperforming.
This is almost never true unless you have a sharply bimodal distribution. Assuming a normal distribution, ~34% are right around average (within half a standard deviation of the mean). That would mean about 33% are below average.
I'm sorry I don't want to turn this into a back and forth internet debate but I feel like you're very deliberately missing the point.
Averages are an abstraction. People are real. "getting better" for those people means ... actually getting better. Its that simple. You're overthinking this and trying to apply mathematical models and abstractions, which don't get me wrong, are very valuable tools. But don't lose your anchor in material reality.
reply