Robert Samuelson notes three problems
First, comparisons are made to an artificially high benchmark -- the late 1990s "tech bubble." Second, immigration distorts commonly cited statistics. Third, the census figures understate income gains by not counting fringe benefits.
I would add one more problem with these data: The income used to measure the poverty rate fails to include the support from numerous anti-poverty programs. The NY Times reports
Officials also point out that the current [poverty] measure only counts cash as income. They say a more accurate model would include government assistance like food stamps, housing subsidies and tax credits. Such aid has been devised to help support the poor, but its impact is not calculated by the current measure.Update: The CD blog points out a fifth problem: Data on household income does not adjust for declining household size over time.