Demographic, Economic, Household, and Health Profile of Grandparents Responsible for Grandchildren

Abstract

The objective of this brief report was to outline the demographic, economic, household, and health profile of “grandparents responsible for grandchildren” (GRfGs) in the United States (US). Public Use Microdata Sample (PUMS) from the 2009-2011 American Community Survey (ACS) 3-year file was used to investigate characteristics of GRfGs by race-ethnic groups and the geographical distribution over the US mainland (contiguous states). The 9,177 actual units in the microdata are estimated to represent about 306,122 GRfGs. Population profiles are presented in tables for GRfGs of any age and the characteristics for those aged 65 and over are discussed. Amongst GRfGs aged 65 and over (n=33,168): 27% have ambulatory difficulty; 33% own their home free and clear; 77% have resided in their current residence for more than four years; 41% care for grandchild(ren) with the parent being absent; 61% are married; 36% have a college education; 77% are not in the labor force; and the majority (55%) are Non-Latino-Whites. Qualitative comparisons from descriptive statistics suggest race-ethnic minority GRfGs may be more economically and socially vulnerable than Non-Latino-Whites. Research on GRfGs and efforts on understanding how best to assist them should continue.

Introduction

When the Adoption Assistance and Child Welfare Act of 1980 was enacted, “relative caregivers” became the most desirable option for the placement of children in need of alternate caregivers (Goodman and Silverstein, 2002). Prior to this law, the placement of children entering the custody of Child Protective Services (CPS) was sought from available foster care parents. The 1980 law was extended and narrowed in on grandparents as the preferred relative caregivers with the Personal Responsibility and Work Opportunity Reconciliation Act of 1996—Grandparents Responsible for Grandchildren (GRfGs) became a special population of interest with this law. In order to understand the fiscal and social cost of grandparents as alternate caregivers, the 1996 legal act mandated the United States (US) Census Bureau to collect information on GRfG prevalence in the US population (Mutchler and Baker, 2004). As a response, the US Census Bureau developed questions for quantifying the prevalence of GRfGs living in the same household as the grandchild(ren) and included the questions in the 2000 decennial census questionnaire. Following legal acts in the 1980s and 1990s, the National Family Caregiver Support Act of 2000 reauthorized the Older American Act which created the National Family Caregiver Support Program (NFCSP). Under the latter, US states are legally permitted to utilize 10% of their NFCSP federal funds in the assistance of GRfGs—where priority can be given to those deemed with the greatest social and fiscal need (see Minkler and Fuller-Thomson, 2005). More generally, legislation by the 104^th US Congress initiated the collection of GRfGs to help the US Department of Health and Human Services (DHHS), under the Temporary Assistance to Needy Families (TANF) Program, determine the yearly allocation of $200 million dollars to top 10 performing states under the High Performance Bonus Awards Program.

Since the American Community Survey (ACS) entered full implementation in 2005—as a replacement to the “long form” previously administered during decennial censuses—the timely assessment of GRfGs prevalence has been possible using larger scale and more generalizable data than that available from alternate sources like the National Survey of Families and Households or the Health and Retirement Study. The ACS is the main source of information on GRfGs in the US and influences the distribution of billions of dollars each year. For example, in 2008, data from the ACS was used to determine the allocation of $261.1 billion to the DHHS for the Medical Assistance Program (Reamer, 2010). In addition to informing on the prevalence of GRfGs, population estimates derived from the ACS sample are used to inform which governmental entities are eligible for funds and how much should be allocated to them. This implies that population estimates derived from the ACS sample have the potential to exert significant policy, social, and economic impacts at the structural- and individual-level.

The specific aim of this brief report is to investigate the racial-ethnic-specific demographic, economic, household, and disability profiles of GRfGs during the 2009-2011 survey-period using ACS microdata—as it is the primary source for the estimation of GRfG prevalence. The GRfG profile of those aged 65 and over is also provided for contrast to the ‘any age GRfG estimates’ provided in the tables below. The main goal of the project, discussing the profile of any age and older adult GRfGs, is complimented by highlighting why their small sample size should prohibit any attempt to investigate the statistical significance of difference by race-ethnic group without regard to the potentially large confidence intervals around prevalence estimates. In agreement with others (Mutchler and Baker, 2004), the argument here is that inferring statistical significance from their small sample size requires great caution. We now turn our attention to how ACS derived GRfG race-ethnic-specific profiles can help us paint a blurry but informative picture of their population in the US.

Method

Participants

The sample of GRfGs in this analysis comes from the Public Use Microdata Sample (PUMS) 3-year (2009-2011) ACS file. It includes any GRfGs at or over the age of 21 and who reside in the US mainland (contiguous states). A total of 9,177 actual survey respondents (“unweighted count”) are used in the analysis. When the population weight is applied to them (“weighted count”), it is said that the 9,177 individuals represent 306,122 GRfGs in the US mainland over the age of 21 during the 2009-2011 ACS survey-period. The tables provide the unweighted and weighted counts of each race-ethnic group and only present weighted estimates across the groups to create their profiles. Please note advice on the careful use of ACS derived population estimates is not an argument for the high fallibility of ACS data—the data source is transparent, high quality, large scale, and the most readily available information source on the GRfG population.

Procedures

The ACS is a population-based survey administered by the US Census Bureau headquarters in Suitland, MD, USA. Data from the ACS helps the US federal government determine how hundreds of billions of dollars are distributed each year (Siordia, 2014a). As explained elsewhere, ACS PUMS files are available to anyone with an internet connection (Siordia, 2014b). The ACS has been previously argued to be a high quality and transparent data source (Siordia, 2014c). Technical details on the collection of data for the ACS are readily available through the internet. The use of this secondary and de-identified data source does not require IRB approval.

Measures

GRfG refers to anyone who responded with a “yes” to the following question: Is this grandparent currently responsible for most of the basic needs of any grandchild(ren) under the age of 18 who live(s) in this house or apartment? The question has some limitations. In the eloquent words of Mutchler and Baker (2004): “indicators involving the reporting of responsibility for grandchildren may be highly subjective… reports are subject to a higher level of reporting variability than occurs for more objective indicators” (Pg. 364). In particular, the phrase ‘responsible for most basic needs’ may introduce considerable differences between respondents in that it allows individuals to self-define the meaning of “responsible” and leisurely interpret what qualifies as “basic needs.” Because laws requiring the US Census Bureau to collect data on the prevalence of GRfGs do not specify that a spouse-GRfG-dyad be treated as a caregiving unit, this analysis estimates the prevalence of GRfGs without aggregating these dyads—an approach seen by others as “double-counting” (Mutchler and Baker, 2004). The fact that both individuals in a spouse-GRfG-dyad claim responsibility for the basic needs of a grandchild(ren) has implications for legal and funding matters. The “double-counting” approach is important as almost 7 out of 10 GRfGs are married (shown in Table 1). If the concern is to not multi-count individuals responsible for a grandchild(ren), then a more complex logic and set of variables would be required to determine when a grandparent is truly responsible in the presence of the parent in the same home on an alternate grandparent also claiming responsibility for meeting the basic needs of a grandchild(ren). This too is important since 8 out of 10 GRfGs reside in a house where at least one of the child's parents is present.

Table 1. Housing characteristics of grandparents responsible for grandchildren by race-ethnicity.

	NLW1	%	NLB2	%	NLO3	%	MEX4	%	NML5	%	Total6	%
Unweighted count7	5,133		1,568		669		1,315		492		9,177
Weighted count8	159,489		56,835		20,723		49,683		19,392		306,122
Inflation & Mode
Person Inflation Ratio9	31		36		31		38		39		33
Data by mail	80,222	50%	20,217	36%	10,627	51%	11,232	23%	5,655	29%	127,955	42%
House Ownership
Paying mortgage	90,976	57%	21,503	38%	9,732	47%	21,658	44%	8,264	43%	152,135	50%
Owns free and clear	28,378	18%	5,238	9%	3,106	15%	7,265	15%	1,517	8%	45,505	15%
Renting	37,761	24%	29,045	51%	7,545	36%	19,941	40%	9,433	49%	103,727	34%
Rent free	2,315	1%	1,049	2%	34 0	2%	819	2%	178	1%	4,701	2%
Length of Residence
< 12 months	21,253	13%	10,803	19%	3,092	15%	9,193	19%	3,264	17%	47,606	16%
13 to 23 months	7,066	4%	2,795	5%	1,279	6%	1,883	4%	954	5%	13,977	5%
2 to 4 years	25,657	16%	12,538	22%	4,268	21%	9,068	18%	4,369	23%	55,901	18%
5 to 9 years	29,391	18%	9,447	17%	4,579	22%	11,777	24%	3,933	20%	59,128	19%
10 to 19 years	41,745	26%	12,043	21%	5,319	26%	11,736	24%	4,165	21%	75,009	25%
20 to 29 years	18,879	12%	4,450	8%	1,493	7%	3,695	7%	1,931	10%	30,448	10%
> 30 years	15,498	10%	4,915	9%	693	3%	2,331	5%	776	4%	24,213	8%
HH10 Composition
Grandparent no parent	39,691	25%	13,621	24%	3,468	17%	6,064	12%	2,299	12%	65,144	21%

¹Non-Latino-Whites;

²Non-Latino-Blacks;

³Non-Latino-Others;

⁴Mexican-Latinos/as;

⁵Non-Mexican-Latinos/as;

⁶Total across all race-ethnic groups;

⁷Actual number of subjects in microdata;

⁸Count of subjects when population weights applied;

⁹PIR = (weighted count ÷ weighted total population);

¹⁰HH=household

GRfGs' home ownership status is captured with the following: paying mortgage; owns free and clear; renting; and resides in a ‘rent free’ place. To capture their residential stability, their ‘length of residence’ at the current address is broken down into 7 month-and year-categories. To understand the composition of the family unit in the household, the number of GRfG homes with parent present is also provided (Mutchler and Baker, 2004). Because the person population prevalence of GRfGs is the focus, house population weights are not used to estimate these counts.

GRfGs are divided into the following racial-ethnic groups: Non-Latino-White (NLW: the racial-ethnic majority group in the US); Non-Latino-Black (NLB); Non-Latino-Others (NLO); Mexican-Latino/a (MEX); and Non-Mexican-Latino/a (NML). NLWs are treated as the reference group. Tabulations, stratified over these five racial-ethnic groups, include: age (if ≥ 65 then =1); sex (female=1); nativity (native born=1); educational attainment (some college or more=1); language spoken at home (only speaks English, “mono English”=1); marital status (3 groups).

Several economic factors are included. Labor force participation is broken down into: civilian or armed forces “employed and at work” (works=1); unemployed; and not in labor force. To account for the overall economic status of the family unit, the GRfG's poverty status is measured as those “in-poverty” (poverty score from 0 to 100) and those “near-poverty” (poverty score from 101 to 150). The poverty measure is created by the US federal government—varies by year but not by geography. For example, in 2011, a family of three that includes a child under 18 would be under the poverty threshold if the family reported a yearly income ≤ $18,123. The poverty scores ranging from 0 to 500 in the ACS PUMS data are ratios that provide a simple measure of economic vulnerability.

All the six disability items in the ACS are included. They refer to individuals reporting having difficulty with: self-care; independent-living; ambulatory; cognitive; hearing; and vision. The report refers to these items as measuring “disability” because the ACS refers to them as such. Detailed disability questions are displayed in Appendix A.

ACS PUMS data allows individual units to be geographically referenced to Public Use Microdata Area (PUMA). Although not shown here, the prevalence of GRfGs was examined at the PUMA level and across all the US mainland states. This report provides the 20 states with the highest GRfG concentrations and identifies the two counties containing the two highest GRfG concentrated PUMAs.

Data Analyses

The main goal is to discuss racial-ethnic-specific profiles. Race-ethnicity-specific weighted and unweighted counts of the sample are presented at the top of each table. Please note PUMS data slightly differ from internal files at the US Census Bureau because public microdata are a sub-sample of the internal ACS file (see Keathley, Navarro, & Asiala, 2010). Weighted numbers are used to display demographic characteristics. Percentages are calculated as follows: (W^c_ij ÷ TW^c_j); where W^c_ij is the weighted count of “i^th” demographic factor for “j^th” race-ethnic group and TW^cj is the weighted total count of “j^th” race-ethnic group. The average number of GRfGs represented by each actual respondent is represented by the Person Inflation Ratio (PIR): (W^c_j ÷ uW^c_j); where W^c_j is the weighted count for “j^th” race-ethnic group and uW^c_j is the unweighted count for “j^th” race-ethnic group. Please note only qualitative comparisons between race-ethnic groups are made. The small number (9,177) of GRfGs being used to infer the characteristics of 306,122 GRfGs, stratified by racial-ethnicity, suggest the potential for the confidence intervals around population estimates to be heterogeneous, large, and geographically unstable. Quantitative comparisons are avoided under the assumption that population estimates may differ in their precision for capturing the ‘true’ GRfG population characteristic and that inferring statistical significance is not required to inform social policy.

Results

Within GRfGs, only 11% (n=33,168) are aged 65 and over. From Table 1, we see that when compared to NLWs and NLOs (where PIR=31) the following have a larger PIR: NMLs (39); MEXs (38); and NLBs (36). We also see about 50% of NLW GRfGs participated with the survey via mail and that all other groups (except NLO) had lower rates of mail participation. NLW GRfGs have the largest (57%) number who owns home free and clear— while both NLB and NML GRfGs have the highest rates of renting (51% and 49% respectively). For those aged 65 and over, about 33% own their home free and clear. With regards to residential stability, NLW GRfGs have about 48% of their group with 10 or more years at current residence, compared to: NLBs at 38%; NLOs and MEXs at 36%; and NML at 35%. About 77% have resided in their current residence for more than four years in the sub-population of GRfGs aged 65 and over. Skipped-generation households (no parent present) are most prevalent amongst NLWs (25%) than the other racial-ethnic groups: NLBs at 24%; NLOs at 17%; and MEXs and NMLs at 12%. Amongst those aged 65 and over, 41% belong to skipped-generation households.

From Table 2, we see that most GRfGs are females—particularly amongst NLBs (72%). Within racial-ethnic groups, older adult GRfGs are most prevalent amongst NLOs (19%) followed by: NLWs at 12%; NLBs and NMLs at 9%; and then MEXs at 8%. Foreign-born GRfGs are most frequent amongst MEXs (62%). Amongst those aged 65 and over, 78% are native. With regards to educational attainment, MEX GRfGs have the lowest rates of college educated followed by NLMs. Almost one-fourth (36%) have a college education amongst those age ≥ 65. The lowest rates of only-English speaking households are found amongst NMLs (13%), MEXs (14%), and NLOs (38%). Please note the rates of married GRfGs: NLWs and MEXs at 72%; NLOs at 67%; NMLs at 58%; and NLBs at 47%. About 61% are married in the older adult GRfG sup-population. More than half of all GRfGs work and only NLOs have 47% of their GRfGs not in the labor force. In those aged 65 and over, 77% are not in the labor force. Poverty rates are highest in MEXs and NLBs (30%) followed by NMLs (29%), NLOs (23%), and NLWs (16%). Where age ≥ 65: 30% of NLWs; 33% of NLOs; 39% of MEXs; 44% of NLOs; and 47% of NLBs are in- or near-poverty.

Table 2. Demographic characteristics of grandparents responsible for grandchildren by race-ethnicity.

	NLW1	%	NLB2	%	NLO3	%	MEX4	%	NML5	%	Total6	%
Unweighted count7	5,133		1,568		669		1,315		492		9,177
Weighted count8	159,489		56,835		20,723		49,683		19,392		306,122
Demographics
Female	94,259	59%	41,068	72%	13,289	64%	28,791	58%	12,922	67%	190,332	62%
Age ≤ 65	18,377	12%	5,310	9%	3,839	19%	3,872	8%	1,170	9%	33,168	11%
Native born	155,202	97%	52,218	92%	9,619	46%	19,071	38%	8,413	43%	244,526	80%
College educated	73,272	46%	26,097	46%	8,587	41%	10,293	21%	5,841	30%	124,092	41%
Mono English9	152,368	96%	53,959	95%	7,882	38%	7,003	14%	2.601	13%	221,215	72%
Married	114,200	72%	26,847	47%	13,903	67%	35,976	72%	11,178	58%	202,107	66%
Widow	9,497	6%	3,963	7%	2,023	10%	2,104	4%	1,119	6%	18,706	6%
D,S,NM10	35,792	22%	26,025	46%	4,797	23%	11,603	23%	7,095	37%	85,313	28%
Economics
Works11	97,306	61%	31,885	56%	9,586	46%	30,945	62%	12,005	62%	181,729	59%
Unemployed	9,318	6%	4,315	8%	1,027	5%	3,608	7%	1,323	7%	19,591	6%
Not in labor force	50,334	32%	19,619	35%	9,837	47%	14,231	29%	5,807	30%	99,829	33%
In-poverty12	25,557	16%	16,795	30%	4,838	23%	15,003	30%	5,532	29%	67,726	22%
Near poverty13	21,486	13%	9,378	17%	3,182	15%	9,885	20%	3,394	18%	47,326	15%

¹Non-Latino-Whites;

²Non-Latino-Blacks;

³Non-Latino-Others;

⁴Mexican-Latinos/as;

⁵Non-Mexican-Latinos/as;

⁶Total across all race-ethnic groups;

⁷Actual number of subjects in microdata;

⁸Count of subjects when population weights applied;

⁹Only speak English;

¹⁰Divorce, separated, or never married;

¹¹Civilian or armed forces employed and at work;

¹²Poverty score ≤ 100;

¹³Poverty score from 101 to 150

Table 3 displays the prevalence of disability. Ambulatory difficulty has the highest rates (e.g., NLBs at 17%) and self-care difficulties the lowest (e.g., NLWs, NLOs, and MEX at 3%). Within those aged 65 and over: 27% have ambulatory difficulties; 16% have difficulty with hearing; 13% have difficulty with independence; 9% have cognitive difficulties; 7% have difficulty with vision; and 6% have difficulty with self-care. Most of the disable older adults are concentrated amongst the racial-ethnic minority groups.

Table 3. Health characteristics of grandparents responsible for grandchildren by race-ethnicity.

	NLW1	%	NLB2	%	NLO3	%	MEX4	%	NML5	%	Total6	%
Unweighted count7	5,133		1,568		669		1,315		492		9,177
Weighted count8	159,489		56,835		20,723		49,683		19,392		306,122
Disability
Self-Care	5,164	3%	2,088	4%	599	3%	1,590	3%	815	4%	10,256	3%
Independent-living	9,966	6%	4,520	8%	1,501	7%	2,169	4%	930	5%	19,086	6%
Ambulatory	20,676	13%	9,546	17%	2,290	11%	5,089	10%	2,291	12%	39,893	13%
Cognitive	10,253	6%	3,935	7%	1,165	6%	1,852	4%	939	5%	18,144	6%
Hearing	9,428	6%	2,120	4%	841	4%	1,872	4%	514	3%	14,775	5%
Vision	6,316	4%	3,234	6%	919	4%	3,016	6%	516	3%	14,001	5%

¹Non-Latino-Whites;

²Non-Latino-Blacks;

³Non-Latino-Others;

⁴Mexican-Latinos/as;

⁵Non-Mexican-Latinos/as;

⁶Total across all race-ethnic groups;

⁷Actual number of subjects in microdata;

⁸Count of subjects when population weights applied;

Table 4 shows the top 20 states with the highest concentrations of GRfGs (sorted on the “total” column). Texas and California have the highest concentrations of GRfGs. For most states, NLWs are the predominant group. For example, they make up about 84% of GRfGs in Indiana. In contrast, MEXs have the highest concentrations in Texas and California (47% and 45% respectively). Although not shown here, a list of PUMAs was also explored to identify highly concentrated sub-state geographical areas. From this analysis, the two most highly GRfG-concentrate PUMAs are found in Ector County, Texas and Riverside County, California. Creating race-ethnicity stratified demographic profiles for GRfGs at anything lower than the national level should be avoided. State and sub-state counts should only be used to give some indication on the geographical distribution of the GRfG population. Less than 50 racial-ethnic-minority actual survey participants may be used for some states to create population estimates. Splitting these small size sub-GRfG-groups by demographic characteristics could produce tabulations where some table cells are informed by the characteristics of one or two people. For example, there are only 492 NML GRfGs in the US mainland and only one for the whole state of Louisiana. State-specific profiles for older adult GRfGs should be used with great caution.

Table 4. Top 20 states with the most grandparents responsible for grandchildren by race-ethnicity.

	Total1	% N2	NLW3	%S4	NLB5	%S4	NLO6	%S4	MEX7	%S4	N ML8	%S4
Texas	37,685	12.3%	11,975	31.8%	4,962	13.2%	1,145	3.0%	17,723	47.0%	1,880	5.0%
California	34,523	11.3%	9,470	27.4%	2,453	7.1%	4,391	12.7%	15,476	44.8%	2,733	7.9%
Florida	17,881	5.8%	8,733	48.8%	4,742	26.5%	813	4.5%	830	4.6%	2,763	15.5%
New York	14,264	4.7%	5,265	36.9%	3,389	23.8%	1,976	13.9%	318	2.2%	3,316	23.2%
Georgia	12,668	4.1%	6,299	49.7%	4,761	37.6%	296	2.3%	812	6.4%	500	3.9%
Ohio	11,037	3.6%	8,685	78.7%	2,127	19.3%	65	0.6%	70	0.6%	90	0.8%
Michigan	10,014	3.3%	6,716	67.1%	2,436	24.3%	390	3.9%	440	4.4%	32	0.3%
Illinois	9,793	3.2%	3,931	40.1%	3,162	32.3%	551	5.6%	1,790	18.3%	359	3.7%
North Carolina	8,869	2.9%	4,474	50.4%	2,794	31.5%	838	9.4%	390	4.4%	373	4.2%
Pennsylvania	8,610	2.8%	5,400	62.7%	1,411	16.4%	381	4.4%	160	1.9%	1,258	14.6%
Arizona	8,588	2.8%	3,855	44.9%	437	5.1%	1,092	12.7%	3,194	37.2%	10	0.1%
Tennessee	8,543	2.8%	6,715	78.6%	1,370	16.0%	93	1.1%	188	2.2%	177	2.1%
Indiana	7,734	2.5%	6,479	83.8%	602	7.8%	437	5.7%	147	1.9%	69	0.9%
Alabama	6,853	2.2%	3,980	58.1%	2,595	37.9%	129	1.9%	85	1.2%	64	0.9%
Louisiana	6,589	2.2%	3,317	50.3%	2,964	45.0%	205	3.1%	97	1.5%	6	0.1%
Washington	6,570	2.1%	3,928	59.8%	357	5.4%	1,407	21.4%	761	11.6%	117	1.8%
Mississippi	6,464	2.1%	3,510	54.3%	2,795	43.2%	140	2.2%	0	0.0%	19	0.3%
South Carolina	6,098	2.0%	3,678	60.3%	2,044	33.5%	73	1.2%	103	1.7%	200	3.3%
Kentucky	5,983	2.0%	4,972	83.1%	715	12.0%	83	1.4%	0	0.0%	213	3.6%
Virginia	5,818	1.9%	3,207	55.1%	1,858	31.9%	406	7.0%	25	0.4%	322	5.5%

¹Total across all race-ethnic groups;

²Percent from nation = (weighted state count ÷ weighted total count) × 100;

³Non-Latino-Whites;

⁴Percent from state = (weighted race-ethnic count ÷ weighted state count) × 100;

⁵Non-Latino-Blacks;

⁵Non-Latino-Others;

⁶Mexican-Latinos/as;

⁷Non-Mexican-Latinos/as

Discussion

This brief report outlines the demographic, economic, household, and disability profile of GRfGs over the age of 21 who resided in the US mainland during the 2009-2011 survey-period of the ACS. The profile of GRfGs suggest racial-ethnic minorities may be more economically and socially vulnerable than their NLW counterparts. Policy should continue to advance the collection of high quality data on GRfGs and target efforts in the formation of transitional family units for children needing adult care. The proper treatment of sample derived population estimates for the formation of social policy has real consequences for future generations.