> I suspect that the distribution of incomes in the US remains fairly similar to what it was 40 years ago. Thus, if the median has remained the same, then the a

> I suspect that the distribution of incomes in the US remains fairly similar to what it was 40 years ago. Thus, if the median has remained the same, then the average probably also has. If the distribution of income had gone unchanged, the spread of wealth would also be unchanged. But it has changed, see http://inequality.org/wealth-inequality/ http://inequality.org/wealth-inequality/ or the source of your choice. If the distribution of income had gone unchanged, the percentage of the population receiving minimum wage would have gone unchanged. But it has changed, see http://www.weeklystandard.com/bls-percentage-of-hourly-workers-earning-minimum-wage-or-less-in-2013-falls-to-4.3/article/786466 http://www.weeklystandard.com/bls-percentage-of-hourly-worke... or the source of your choice. I could go on... but I think that's probably sufficient. While I have you here, quick question: why did you let yourself get away with saying "I suspect" rather than just looking it up? It takes all of 10 seconds to do and once you get in the habit you'll be right way more often. > I suspect that the distribution of incomes in the US remains fairly similar to what it was 40 years ago. Thus, if the median has remained the same, then the average probably also has. The metric in question isn't the distribution's median. It's the percentage of children earning less than their parents. median: sup k : { Count(xi >= k) / Length(xi) <= 0.5 } earns more: Count(xi >= pi)/Length(xi) These are very different values. So while I'll need to think more about the idea that medians say something about distribution similarities, thankfully those thoughts don't pertain here. For another example, of "earns more" not behaving the way you're expecting: if you set the earnings of everyone's parents to their wage - $0.01, we'd have a 0% "earns more" but extremely similar distributions.