Note: Positives and negatives are reversed from the source reports to reflect the intuition that undercounts are more easily grasped as negative numbers. To construct these numbers, the bureau assigned people to one of seven mutually exclusive race/Hispanic origin categories. This method allowed comparisons of undercount data across the 1990, 2000, and 2010 Censuses.
*Mule concluded that these percent net undercounts were “statistically significant from zero.”
† Hogan et al. reported no available undercount data for American Indians off Reservation in 1990. As Mule noted, American Indians off Reservation were included in the Non-Hispanic White domain; for that reason, Mule treated the percent net undercount for that domain as equal to the Non-Hispanic White undercount.