cohabiting-couple households in this chapter. Generally, children may be raised in households that do not include either of their biological parents for a variety of reasons. Children may be raised in households maintained by their grandparents; West and Robinson (1999) cite research suggesting that as many as 2 million children lived in grandparent households in 1990. Depending on circumstances—such as the loss of parents or removal by incarceration— children may also be raised by other relatives or by family friends. To the extent that such arrangements are not “formalized”—e.g., the grandparents do not actually take legal custody of the children—it is possible that some respondents to a census or other questionnaire asking them to roster their households might not think it proper to include the children.

Census coverage errors among children may arise from respondent misunderstanding of (or resistance to) stated census residence rules, but there is reason to suspect that a more basic, structural feature of the census questionnaire itself plays some part. In 2000, the census questionnaire allowed for detailed entry of data for six people, with space to list only the names of six additional people. These large households were intended to be automatic cases for the coverage edit follow-up operation, but if the household could not be contacted during follow-up, the exact nature of persons 7–12 in the household—and certainly any information in households with 13 or more— would be unknown.

The reason that this structural feature may affect the counting of children is that listing people in descending age order may be a natural approach for the adult responding to the census form. Indeed, this type of ordering was explicitly instructed during the enumerator-contact era from 1850 through 1960.22 Likewise, a respondent may intuitively reverse-sort people in the roster by the degree to which they “belong” in the household; in cases where extended families and multiple generations live in the same structure, children of kin may simply fall lower in the respondent’s mental ordering of people in the household. Hence, they may fall in the section where only names are recorded, or not at all if no space is left. Using 1998 data from the Current Population Survey, O’Hare (1999:8) suggested that 5.6 million children lived in households of seven or more people; to the extent that age ordering is used in answering the census form, these children would be at risk of being missed in the census.


The “Be Counted” program of the 2000 census placed blank questionnaires in public places so that people who believed that they had not been


Per the 1850 enumerator instructions, “the names are to be written, beginning with the father and mother [or other adult head of family]; to be followed, as far as practicable, with the name of the oldest child residing at home, then the next oldest, and so on to the youngest, then the other inmates, lodgers and borders, laborers, domestics, and servants.”

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement