The old saw is that there are lies, damned lies, and statistics, as if statistics were in some way a variety of lie. Of course, the issue is not so much that statistics are lies, as that statistics represent an attempt to simply quantify a terribly complex reality, and with simplification comes the opportunity for error -- often error confirming the biases of the person doing the analysis.
The other day I ran into a very interesting exploration of one of those statistics which is often discussed -- that "more families are in poverty" after the last three decades than was the case in the past. In 2006 Hoynes, Page and Stevens authored a paper entitled "Poverty in America: Trends and Explanations" which was published in the Journal of Economic Perspectives. One of the interesting things they do is look at the trends in poverty by family type. The findings are fascinating:
For instance, in 1967 10.7% of married couples with children were in poverty and 51.2% of single mothers with children were in poverty. By 2003 those numbers had dropped to 8.1% and 37.3% respectively. The problem is that the percentage of households consisting of a married couple with children dropped from 67.3% to 44.2%, while the number of households consisting of a single mother with children increased from 6.2% to 11.9%.
Due to more people belonging to family types more prone to poverty, the percentage of households in poverty went up (although since those households are also smaller, the actual percentage of people living in poverty went down.)