I posted earlier that the Detroit Free Press screwed up the figures, and posted a link to a table of corrected figures. Here is
a link to a conservative economist (Jane Galt, a/k/a Economist writer Megan McArdle) (I hesitate to let Spanky know that I'm a regular reading and sometime commenter at her site) explaining at some length what was wrong with the first numbers.
I note that neither she nor anyone else whom I can find think there is some problem with the concept of looking at median income, and that her guess of what median income has done in the last three years (down 3%) is very, very close to the corrected numbers in the table I posted.