Coladirienzi is correct. You can measure median income for any group. I don't know that I believe his assertion. And anytime you look at a period including a recession, chances are there is a drop in median income. I'd have to look at the data. Perhaps coladirienzi has a link?
I was wondering what happens when someone moves into the top quintile from below. He "falls off the table," so to speak. If you have more people falling off the table than moving from the bottom half of the (adjusted) median into the top, then it might explain why the new (adjusted) median is falling. The fact that the time frame encompasses a period of weak(er) economic growth makes me wonder if the statistic is so flawed as to be almost worthless.