ILR School

University of Colorado at Boulder/University of Tennessee Node

Permanent URI for this collection


Recent Submissions

Now showing 1 - 2 of 2
  • Item
    Uncertain Uncertainty: Spatial Variation in the Quality of American Community Survey Estimates
    Folch, David C.; Arribas-Bel, Daniel; Koschinsky, Julia; Spielman, Seth E. (2014-11-07)
    The U.S. Census Bureau's American Community Survey (ACS) is the foundation of social science research, much federal resource allocation and the development of public policy and private sector decisions. However, the high uncertainty associated with some of the ACS's most frequently used estimates can jeopardize the accuracy of inferences based on these data. While there is high level understanding in the research community that problems exist in the data, the sources and implications of these problems have been largely overlooked. Using 2006-2010 ACS median household income at the census tract scale as the test case (where a third of small-area estimates have higher than recommend errors), we explore the patterns in the uncertainty of ACS data. We consider various potential sources of uncertainty in the data, ranging from response level to geographic location to characteristics of the place. We find that there exist systematic patterns in the uncertainty in both the spatial and attribute dimensions. Using a regression framework, we identify the factors that are most frequently correlated with the error at national, regional and metropolitan area scales, and find these correlates are not consistent across the various locations tested. The implication is that data quality varies in different places, making cross-sectional analysis both within and across regions less reliable. We also present general advice for data users and potential solutions to the challenges identified.
  • Item
    Reducing Uncertainty in the American Community Survey through Data-Driven Regionalization
    Spielman, Seth; Folch, David (2014-11-07)
    The American Community Survey (ACS) is the largest US survey of households and is the principal source for neighborhood scale information about the US population and economy. The ACS is used to allocate billions in federal spending and is a critical input to social scientific research in the US. However, estimates from the ACS can be highly unreliable. For example, in over 72% of census tracts the estimated number of children under 5 in poverty has a margin of error greater than the estimate. Uncertainty of this magnitude complicates the use of social data in policy making, research, and governance. This article develops a spatial optimization algorithm that is capable of reducing the margins of error in survey data via the creation of new composite geographies, a process called regionalization. Regionalization is a complex combinatorial problem. Here rather than focusing on the technical aspects of regionalization we demonstrate how to use a purpose built open source regionalization algorithm to post-process survey data in order to reduce the margins of error to some user-specified threshold.