The difference between the graphs are that the one linked by jusmus88 is illustrating the distribution of incomes of "families", whereas all three in the article you linked are for the distribution of incomes of "households". These are two different ways of analysing the issue that give different answers, and the differences are discussed in the article.

edit: Just to be clear, I'm not making any positive or negative statements about the article, I just wanted to help you work out what it was you had missed, and I think what I described is it.

Thanks for pointing that out! I had saved both images to my phone and flipped back and forth between them to try to understand, and all I could see changing was the data! But you are absolutely correct it's "Families" in one and "Households" in the other. Very interesting distinction...

How does that graph read differently and the captions are from what sources exactly?

I ignored the captions. The data show the bottom 80% experiencing significant wage deflation the last decade in real dollar terms. Hence, falling out of the middle class.

The title accurately of your link was "Census data on income distribution reveal evidence of rising income levels for a rising share of American households", not the opposite.

We just had the biggest recession since the Great Depression. Taking such a short window will of course show wage deflation. You cannot remove the effects from bubble income and expect to keep bubble income.

The overall trend for most slightly bigger windows is wage increase. If a certain wage level meant middle class in the 1960s, or 1970s, or 1980s, or 1990s, then we have a larger ratio of people in that class, not less.

Of course if you define middle class as also increasing in income, then you can play all sorts of tricks. But the fact remain more people than perhaps any but the last decade are richer than ever. All your graphs point to exactly this.

Remember - bubble popped. It takes a while for those effects to smooth out.