Friday October 18, 2013. 03:30 pm. GUGG 205
Abstract: Compared to earlier decennial censuses, there is growing recognition and debate about the uncertainty associated with American Community Survey (ACS) estimates. However, the manifestation of that uncertainty is not widely understood. One cause for concern in the ACS is that the uncertainty generally does not follow a random pattern across attributes and space. The assumption of randomly distributed uncertainty in ACS estimates is widely held and provides the foundation for not addressing the problem explicitly. In this research we investigate 2006–2010 median household income from the ACS at the census tract scale. We first outline potential sources behind the patterns of uncertainty, and then build a multivariate model that allows us to identify the relevance and spatial variation in these sources. Finally, we present a new method that can be used to address the uncertainty challenge in empirical research.
Geography colloquium series was made possible by the generous support from The Beirne Carter Foundation.