Last October I posted "Data Users Beware, Part 1," in which I warned about the limitations of ACS data deriving from its small sample sizes.  According to the ACS sampling design, those limitations are overcome by aggregating samples over 3 and 5 year periods, which is intended to yield large enough samples to allow accurate estimates of population statistics at the tract level.  In my post, I stressed the importance of approaching the data with a degree of skepticism and advised users to carefully assess the accuracy of ACS data as a key part of the interpretive process.  Since then, the ACS has released their first 5-year sample, and it has become clear that such caution is amply justified.  To highlight some of the potential problems in interpreting the ACS data on even basic demographic measures, I will focus on the case of race in the Tri-County Southeast Michigan area.

The most recent issue of Families in Society: The Journal of Contemporary Social Services contains an article that uses a new approach to the study of poverty in later life, and comes to some very sobering conclusions.  In essence, although according to official poverty statistics slightly less than 10% of older adults in the U.S. currently live in poverty, a much greater number of them will experience poverty at some point.  By taking a longer view, and using the Panel Study of Income Dynamics data collected each year since 1968 by the Institute for Social Research at the University of Michigan, the article's authors found that nearly half of people between the ages of 60 and 90 will experience at least one year of living in poverty.

The Center for Retirement Research at Boston College has released a new issue brief that is of great significance to those of us interested in demographic data on older adults and, more importantly, those who help funders and policy makers understand the issues facing older adults by interpreting those data properly.  The brief is entitled, "Why Did Poverty Drop for the Elderly?" and it is authored by Alicia H. Munnell, April Wu, and Josh Hurwitz.  In it, Dr. Munnell and her colleagues analyze a curious statistical anomaly that recently arose from U.S. Census data. The anomaly is that between 2008 and 2009, when the rate of poverty increased across nearly every social strata in the United States, it actually appeared to decrease among the elderly.

Despite its widespread use in the mainstream media and the best efforts by the US Census Bureau to educate the public, few people realize that the US Decennial Census no longer provides us with much of the social, economic and demographic statistics that we consume. Over the past decade, the US Census Bureau has reverted back to using the decennial census for its constitutionally mandated purpose: enumeration for redistricting purposes. They have replaced the sample portion of the decennial census (i.e., the long form) with the American Community Survey (ACS).