It’s the Data We Have: What Can We Learn from 5-Year ACS Data about Segregation and Neighborhoods Given the Margins of Error?

Nancy A. Denton, University at Albany, State University of New York (SUNY)
Hui-Shien Tsao, University at Albany, State University of New York (SUNY)
Lee Jin-Wook, University at Albany, State University of New York (SUNY)
Wang Ruby, University at Albany, State University of New York (SUNY)
Jeffrey Napierala, University at Albany, State University of New York (SUNY)

While the replacement of the Census long-form with the ACS has led to much more current data, little attention has been paid to the margins of error (MOE) which now accompany every number provided by the Census Bureau. Using race/ethnic data from the five-year ACS (2006-2010) and the 2010 Census for census tracts and block groups in New York State, we focus on residential segregation to explore some the issues the MOE raises for substantive analyses and local officials. Though statistically it is clear that size of place and size of group will both affect the estimates, there has been no investigation of whether this is consistently true given the tremendous heterogeneity across groups and places in terms of other characteristics. As diversity spreads to smaller and smaller areas of the U.S. how can we accurately understand and be sure of what is happening using ACS data?