The fact that net undercoverage of young children has been relatively high in censuses compared with older children and adults is well documented (see Coale 1955, Siegel 1974, U.S. Census 2012a, Hogan et al. 2013, and U.S. Census Bureau 2014). Many hypotheses have been offered including the unstable living arrangements precipitated by having young children (e.g., the need to move in with the child’s grandparents), time stress because of the care needs of the child, and the possibility that the respondents, including proxies, may not remember the child (U.S. Census Bureau 2014). The younger the child, the more salient most of the hypotheses for omission seem.

In this report, we compare census counts to counts of births from vital statistics by month. We show that the census counts actually exceeded the vital statistics counts for January and February 2010, implying a net overcount for these months. The 2010 Census records for children age 0 with a year of birth of 2010 and missing a date of birth (DoB) or having an invalid DoB were assigned a DoB for the months of January, February, and March 2010. It is likely that many of these records corresponded to children born after Census Day, children who should not have been included in the 2010 Census count. This helps explain the lower net undercount for children age 0 shown in Figure 1. Without these imputations, children age 0 would have had net coverage errors consistent with, or greater than, those for children ages 1 and 2.