After the 2010 Census, Demographic Analysis (DA) estimated a net undercount of about 4.6 percent for young children (Hogan et al. 2013). This suggests a net undercount of about 1 million children under the age of 5. Several recent Census Bureau reports have provided additional information about the undercount of young children in the 2010 Census. One report summarized the analysis of data from the 2010 Census Coverage Measurement (CCM) program, focusing on characteristics of young children who were enumerated in the CCM but could not be matched to a child in the 2010 Census (U.S. Census Bureau 2017a). It also provided information about the types of nonmatches that involved young children. This data allowed us to determine:

• If children were nonmatches because the housing unit they lived in was a nonmatch.
• If nonmatching young children were the only nonmatches in a household.
• If they were nonmatches along with other household members.
  

In this report, we present further analyses of the CCM data. One recommendation from U.S. Census Bureau (2017a) was to create cross-tabulations of the CCM nonmatch data to gain greater insights into possible reasons for certain nonmatches. Our initial research showed important distinctions for young children based on their relationship to the householder and the size of the household. That research also found that young children living in a complex household had a greater likelihood of an enumeration error. Important differences were also found by race and Hispanic origin and by mode of data collection. This report will study the distributions of types of nonmatches by relationship to the householder and type of complex household. In addition, this report analyzes cross-tabulations of the CCM nonmatch rates for relationship to the householder by race and Hispanic origin and household size.

U.S. Census Bureau (2017a) also examined the nonmatch rates for young children by the mode of enumeration. That research showed higher nonmatch rates for young children enumerated in Nonresponse Followup (NRFU), especially when the NRFU interview was completed with a proxy respondent. In this report, we study the distribution of types of nonmatches by mode of enumeration. Our goal was to understand the reasons why young children might be missed in self-response and NRFU, and how these error sources may differ. Knowing these differences is important for taking measures to address the errors in future censuses and surveys.

In another analysis, we compare the CCM nonmatch rates for children in three age groups—0 to 4, 5 to 9, and 10 to 17. U.S. Census Bureau (2017b) showed that young children, age 0 to 4, were more likely than older children to have certain characteristics that were associated with high CCM nonmatch rates, such as a relationship other than biological or adopted child of the householder. This report studies whether the pattern of CCM nonmatch rates for young children hold for older children.