This paper summarizes patterns of omissions and erroneous enumerations in the census coverage resulting from two mechanisms of coverage error affecting the whole household or affecting individuals within household. (See Brownrigg and Wobus 1993.)

Demographic analysis profiles the population not included in the census: The net under-count of males is greater than for females (3.48 million males; 1.20 million females in 1990). The net undercount rate for Blacks is greater than for non-Blacks. The net undercount is high for certain sex/race/age groups such as Black males of ages 25-54 (Ahmed 1991; Robinson et. al 1991:18). Black males are the race/sex group with the highest net undercounts in the 1950, 1960, 1970, 1980 decennial censuses. Adult Black men experienced higher net under-counts than Black females, or non-Black males or females in the past 5 censuses (Robinson 1988). The 1990 PES estimated that males of Hispanic origin (of any race) had a slightly higher net undercount than Black men, and American Indians men experienced net under-coverage almost as high as Black men. (Hogan 1991, 1992; Robinson et al 1991). The profile of the undercounted from demographic analysis and the PES imply that more minority adult men are omitted than are erroneously enumerated.