Appendix I.
Describing Kinship Care from Current Population Survey Data
(Supplemental Material to Section I.)
The analysis of children in kin care is based on data from the March Current Population Survey (CPS) for 1983 through 1994. "Children in kin care" or "kin-care children" refers to never-married children who do not live with their parents, but live with other adult relatives. Whether these children are in formal foster care arrangements cannot be ascertained from the CPS data. The CPS is a monthly survey of approximately 57,000 U.S. households.(1)
The March CPS includes detailed demographic and financial information. Since the number of children in kin care appearing in the CPS sample is small, we pooled data for three-year intervals to improve the reliability of the estimates. The analysis includes comparisons of children in kin care with children being cared for by their parents and, for 1989-91 and 1992-94, children in foster care. For all three groups, we limited our analysis to unmarried individuals aged 17 or younger. Other restrictions are noted below.
The Sample
In order to understand how we identified children in kin care and how we coded the characteristics of kin-care, parent-care, and foster children, it is necessary to understand how households and families are coded in the CPS. In each household in the CPS sample, someone, usually the individual who owns or rents the dwelling, is designated the household head. Anyone related to the household head is part of the "primary family." Within the primary family, there may also be "related subfamilies." A related subfamily contains a relative of the household head and the relative's spouse, or the relative's minor child, or both.(2)
Family groups not related to the household head are designated "unrelated subfamilies."(3)
Individuals unrelated to the household head or anyone else in the household are "unrelated individuals."
Identifying kin-care children, their caretakers, and their families. We defined kin-care children as children who were living with neither parent, but who were related to either the household head or the head of an unrelated subfamily.(4)
(Note that kin-care children cannot be members of related subfamilies.) For kin-care children in the primary family, we designated as kin-caretakers the head of the household and his or her spouse, if there was one. Family characteristics were based on the characteristics of the entire primary family, including members of related subfamilies. For kin-care children in an unrelated subfamily, we designated as kin-caretakers the head of the subfamily and his or her spouse, if there was one. Family characteristics were based on the characteristics of all individuals in the unrelated subfamily. To avoid tagging as kin-care children teenagers who were living with a slightly older sibling or another relative, we excluded children whose oldest putative caretaker was less than 18 years old or less than five years older than the child in question.
Identifying parent-care children, their parents,
and their families. We designated individuals as
being in parent care if they were the child of the head of
household, the head of a related subfamily, or the head of an
unrelated subfamily. The head of the household--or, if
appropriate, the head of the subfamily--and his or her spouse, if
there was one, were coded as the child's parents. For children in
the primary family--including children in related subfamilies--we
coded family characteristics using the characteristics of all
individuals in the primary family.(5)
Identifying foster children, their foster parents, and their families. Starting in 1988, the CPS identified foster children in the sample. Because we pooled data into three-year groups, our comparisons of kin-care and foster children begins in 1989. The CPS data do not allow us to determine whether foster children are also kin-care children; that is, related to other household members. (The CPS data also do not allow us to identify whether kin-care children are also foster children; that is, children who have been placed with the relative by the state foster care agency.) We designate the household head and his or her spouse, if there is one, as the foster parents. Family characteristics are coded using data on all individuals in the primary family, including members of related subfamilies.
Family-level variables
The coding of most variables is straightforward. In the following section, we discuss the variables we coded especially for this analysis; that is, variables not included in the CPS. In some cases, the variables are similar to those in the CPS, but were coded somewhat differently. We also discuss variables for which the CPS definitions changed over the periods analyzed.
Poverty status. Starting in 1988, the CPS includes a single measure of poverty status for all members of the primary family. Before then, poverty status was calculated separately for related subfamilies. We recoded poverty status for 1983-87 so that it was comparable to the later years.
Educational attainment of caregiver. The attainment of the best educated caregiver, or parent, is used. Before 1992, the CPS provides the highest year of schooling attended and flag indicating whether this year of schooling was completed. Whether an individual has earned a degree must be inferred. For example, we assumed that all individuals who had completed exactly twelve years of schooling were high school graduates, that individuals who had between 13 and 16 years of schooling, but had not completed the 16th year, had attended, but not completed college, etc. In later years, the CPS provides explicit data on whether an individual has completed a degree.
Labor force status of caregiver. The labor force status of the caregiver or parent most attached to the labor force was used. In order of attachment, from highest to lowest, the possible statuses are: employed (which includes with a job, but not at work), unemployed (which includes looking for a job or on layoff), not in the labor force, or other (which includes individuals under age 15 and those in the armed forces).
Metropolitan status. This CPS variable indicates whether an individual lives in a metropolitan area. For 1983-85, the CPS measure is based on SMSAs; thereafter it is based on MSAs.
Program participation variables and earned income. We coded the program participation variables using data on all individuals in the child's family. For children in the primary family, including children in related subfamilies, program use by any individual in primary family is included in the measure. For children in unrelated subfamilies, program use by any individual in the subfamily is included. The types of programs examined were: public assistance/welfare, supplemental security income (SSI), social security, disability (1988-94 only), unemployment compensation, workers' compensation, receipt of free school lunches (a household-level variable), residence in public housing, receipt of rent subsidy, receipt of food stamps (a household-level variable). (Receipt of school lunches and receipt of food stamps are reported at the household, not the family, level in the CPS.) Similarly, the earned income variable is coded to indicate whether anyone in the child's family had earned income. Within the primary family, distinctions are not made between the main family and related subfamilies.
Relationship to kin-caregiver. We could not ascertain the relationship between kin-care children and their caretakers--specifically, whether the kin-caretaker was a grandparent--until 1989. Before that year, an individual was identified as the grandchild of the householder only if his or her parents were also present in the household (Weyland, 23 May 1996).
Race and ethnicity. We created a single variable indicating both the race of the child and whether that child was Hispanic. Before 1988, the CPS provided only three race categories: white, black, and other. Thereafter, the "other" category was broken into three categories: Asian/Pacific Islander, American Indian/Eskimo, and other.
Notes on Personal Communications with Greg Weyland, U.S. Bureau of the Census, Demographic Surveys Division
May 23, 1996: He explained that prior to 1988, a person was identified as a grandchild of the householder ONLY if the child's parents were present. In other words, they would only know if a child was the grandchild of the householder if all three generations resided in the household.
June 4, 1996: He confirmed that from 1988 forward, there was no may to identify kin kids of an unrelated subfamily because they removed the classification "other relative of an unrelated subfamily reference person."
REFERENCES
U.S. Bureau of the Census (1994). Current Population Survey, March 1994 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1993). Current Population Survey, March 1993 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1992). Current Population Survey, March 1992 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1991). Current Population Survey, March 1991 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1990). Current Population Survey, March 1990 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1989). Current Population Survey, March 1989 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1988). Current Population Survey, March 1988 Rewrite [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1987). Current Population Survey, March 1987 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1986). Current Population Survey, March 1986 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1985). Current Population Survey, March 1985 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1984). Current Population Survey, March 1984 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1983). Current Population Survey, March 1983 [machine-readable data file]. Washington, D.C.
U.S. Bureau of the Census (1995). Statistical Abstract of the United States: 1995 (115th edition). Washington, D.C.: U.S. Government Printing Office.
Weyland, Greg. (1996a) Personal Communication on May 23, 1996 from Greg Weyland, Demographic Surveys Division, U.S. Bureau of the Census to Karen Maguire, Urban Institute.
Weyland, Greg. (1996b) Personal Communication on June 4, 1996
from Greg Weyland, Demographic Surveys Division, U.S. Bureau of
the Census to Karen Maguire, Urban Institute.
Obtaining Living Arrangement Categories from STF3 data tables.
(Supplemental Material to Section II)
The Census information used in this report was obtained from the Summary Tape Files -- Level 3A information that is tabulated and publicly distributed by the Bureau of the Census. Most of the data was obtained electronically from the web site maintained by the Census Bureau.
The information about child living arrangements is not available from one single table in the STF data, but was computed with information from several different tabulations. Our living arrangement counts were produced from the following tables:
P13. AGE (31).
Universe: Persons
P74 PRESENCE AND AGE OF CHILDREN(2) BY EMPLOYMENT STATUS OF PARENTS(8).
Universe: Own children under 18 in families and subfamilies.
P126 POVERTY STATUS IN 1989(2) BY FAMILY TYPE AND AGE(9).
Universe: Related Children under 18 years of age.
The reader should note here that the Census term 'related', as used in Table P126 is a broad category which includes "own-children" as well as other kinship relations. It should also be noted that the relationship data itself, except for the fact that a relation exists, is not comparable to the relation categories in P74. P126 categorized children by relation to the head of the household, and not by presence or absence of a parent.
Original Census Data Table Computational Elements Needed
Age and Total Child Population:
P13 1 Under 1 Year A. Children 0-5 = 13.1 + 13.2 + 13.3 + 3.4
2 1 and 2 years
3 3 and 4 years B. Children 6-17 = 13.5 + 13.6 +.....+ 13.12
4 5 years
5 6 years C. Children 0-17 = A + B
6 7 to 9 years
7 10 and 11 years
8 12 and 13 years
9 14 years
10 15 years
11 16 years
12 17 years
13-31 not used here
Own Children by Age and Living Arrangement:
P74 1 Under 6 years, 2 parents in labor force D. Own Child, two parents, 0-5
2 Under 6 years, 2 parents, father only in labor force = 74.1 + 74.2 + 74.3 + 74.4
3 Under 6 years, 2 parents, mother only in labor force
4 Under 6 years, 2 parents, neither in labor force E. Own Child, two parents, 6-17
5 Under 6 years, father only, in labor force = 74.9 + 74.10 + 74.11 + 74.12
6 Under 6 years, father only, not in labor force
7 Under 6 years, mother only, in labor force F. Own child, mother only, 0-5
8 Under 6 years, mother only, not in labor force = 74.7 + 74.8
9 6 to 17 years, 2 parents in labor force
10 6 to 17 years, 2 parents, father only in labor force G. Own child, mother only, 6-17
11 6 to 17 years, 2 parents, mother only in labor force = 74.15 + 74.16
12 6 to 17 years, 2 parents, neither in labor force
13 6 to 17 years, father only, in labor force H. Own child, father only, 0-5
14 6 to 17 years, father only, not in labor force = 74.5 + 74.6
15 6 to 17 years, mother only, in labor force
16 6 to 17 years, mother only, not in labor force I. Own child, father only, 6-17
= 74.13 + 74.14
Related Children by Age and Living Arrangement:
P126 1 Married-couple family, 0-4 years, above poverty J. Related Child 0-5
2 Married-couple family, 5 years, above poverty = 126.1 + 126.2 + 126.4
3 Married-couple family, 6-17 years, above poverty + 126.5 + 126.7 + 126.8
4 Male householder, no wife, 0-4 years, above poverty + 126.10 + 126.11 +126.13
5 Male householder, no wife, 5 years, above poverty + 126.14 + 126.16 + 126.17
6 Male householder, no wife, 6-17 years, above poverty
7 Female householder, no husband, 0-4 years, above poverty K. Related Child 6-17
8 Female householder, no husband, 5 years, above poverty = 126.3 + 126.6 + 126.9
9 Female householder, no husband, 6-17 years, above poverty + 126.12 + 126.15
10 Married-couple family, 0-4 years, below poverty + 126.18
11 Married-couple family, 5 years, below poverty
12 Married-couple family, 6-17 years, below poverty
13 Male householder, no wife, 0-4 years, below poverty
14 Male householder, no wife, 5 years, below poverty
15 Male householder, no wife, 6-17 years, below poverty
16 Female householder, no husband, 0-4 years, below poverty
17 Female householder, no husband, 5 years, below poverty
18 Female householder, no husband, 6-17 years, below poverty
Using the computational quantities derived from the raw Census
tables as described above, the living arrangement distributions
used in this section can be computed as follows:
0 - 5 years 6-17 years 0-17 years
Own Child, 2 parents D E D + E
Own Child, mother only F G F + G
Own Child, Father only H I H + I
Related Child, no parent J - ( D + F + H) K - ( E + G + I) (J + K) -
(D+E+F+G+H+I)
Unrelated Child ( A - J ) ( B - K) C - ( J + K )
Total Children A B C
As can be seen, own children indicators are obtained directly from combined counts, kinship indicators by subtracting own-children from "all related" children, and unrelated indicators by subtracting "all related" children from "all children".
Appendix 3
Extended County-Level Indicators for Four States,
California, Illinois, Missouri, and New York
[Addendum to Section III]
1. 0 Institutionalized individuals are excluded from the CPS sample.
2. 0 Examples of a related subfamily are (1) the daughter of the household head and the daughter's baby; (2) the nephew of the household head and the nephew's wife; and (3) the sister of the household head, the sister's husband, and the sister's child.
3. 0 Examples of an unrelated subfamily are: (1) a live-in maid and her daughter; (2) a man and his nephew who rent a room in another--unrelated--family's house; and (3) a divorced woman, her son, and her daughter who share a house with another divorced woman--who has been designated the household head by the Census Bureau--and the second woman's daughter. In the last case, the "Kate and Allie" example, both mothers may have their names on the lease or mortgage, but the CPS only allows one to be designated the household head.
4. 0 Starting in 1988, kin-care children in unrelated subfamilies could not be identified because the Census Bureau limited membership in unrelated subfamilies to the head of the unrelated subfamily, his or her spouse, and his or her children (Weyland, 4 June 1996).
5. 0 For example, children of the head of household who have children of their own are coded as heads of a related subfamily, even if they are minors. In such a circumstance, we coded the young mother as a parent-care child, assigned the household head and his or her spouse as parents, and used data on all members of the primary family to generate the family-level variables.