Examining Foster Care Outcomes for American Indian Children in the Context of the Indian Child Welfare Act

Abstract

The history of the child welfare system and related institutions with American Indian children and families has been marked by numerous atrocities, including unnecessary separations, assimilation, and trauma. The Indian Child Welfare Act (ICWA) was enacted in 1978 to promote the stability and security of American Indian tribes and families. For children involved in the child welfare system, ICWA prioritizes the placement of American Indian children with family or tribal members. This paper examines placement outcomes for American Indian children using recent national data over 3 years from the Adoption and Foster Care Analysis and Reporting System. Multivariate regression analyses showed that American Indian children were significantly less likely to be placed with same-race/ethnicity caretakers than their non-American Indian counterparts. In addition, American Indian children were not more likely to be placed with relatives or have trial home placement compared to non-American Indian children. These findings suggest that ICWA is not reaching its intended objectives regarding placement goals specified in the law regarding American Indian children. These policy shortcomings have significant implications for American Indian children, families, and tribes in terms of well-being, family connection, and cultural loss.

Keywords

Indian child welfare act American Indian native American indigenous child welfare system foster care foster placement foster parents kinship care

Introduction

Historically, American Indian children have been overrepresented within the U.S. foster care system (U.S. House Report, 1978; Yi et al., 2020). This overrepresentation is exacerbated by the trend of placement and adoption of American Indian children into non-American Indian (typically White European American) families, which led to the loss of American Indian cultural and familial ties. In 1978, the Indian Child Welfare Act (ICWA) was enacted to redress the disproportionate number of American Indian children in out-of-home care, particularly in non-American Indian out-of-home placements, as well as adoptions. Despite a handful of studies showing some positive effects of ICWA, the intended objectives of the law have not been empirically investigated in recent years (Francis et al., 2023). Therefore, the purpose of this study is to compare foster care outcomes (i.e., placement with relative, placement with a same-race/ethnicity caretaker, and trial home placement) between American Indian children and non-American Indian children to evaluate the extent that ICWA is promoting positive placement outcomes for American Indian children. We use the term “American Indian” throughout this paper because it reflects the language used in the Indian Child Welfare Act and the data system we used for our secondary data analysis. We recognize that language regarding Indigenous peoples is contested for various reasons, including not being culturally-centered, and has shifted over time. We also recognize that Indigenous ethnic and cultural groups have been racialized, and race continues to be a salient social construct across the United States despite its roots in racism. Nonetheless, the designation of a child as American Indian under ICWA entails special rights due to their membership in a sovereign tribal nation.

Historical Overview of American Indians and U.S. Systems

American Indians have a long and traumatic history with the U.S. government and its founding. It is estimated that approximately 7 million Indigenous people lived in North America prior to European colonization (Thornton, 2005). Beginning in 1492, the Indigenous population was devastated over hundreds of years by European colonists and citizens of the United States through infection with epidemic diseases (e.g., smallpox), warfare and armed attacks, near eradication of the buffalo (a central animal to Indigenous livelihood), and occupation of land and forced removal (Brave Heart & DeBruyn, 1998). It is estimated that by 1900, the Indigenous population was reduced to only 375,000 (Thornton, 2005). Some tribes had become extinct and others carried the burden of multigenerational trauma (Brave Heart & DeBruyn, 1998). American Indians had incurred devastating losses, including the loss of loved ones and community members, loss of land and other resources, and cultural loss, which involved considerable grief, pain, and disempowerment.

Boarding and Day School Era

Efforts to use schools to assimilate American Indians (focusing on children) into White European American culture were conducted after the establishment of the United States in 1776 and lasted until 1969 (Adams, 1995; Brave Heart & DeBruyn, 1998). Through federal laws (e.g., Indian Civilization Fund Act), funding appropriations, government regulations, court rulings, and treaties, the boarding school movement grew during the 1800s. Up until 1969, 408 federal boarding schools across 37 states and territories were in operation (Newland, 2022). In total, over 1000 federal and non-federal institutions for American Indian schooling were identified, which included boarding schools, day schools, sanitariums, asylums, orphanages, and dormitories. Through these institutions, American Indian children were given English names, their hair was cut short, and they were given uniforms to wear. Schools indoctrinated children through reading, writing, and speaking English (preventing speaking of native languages); emphasizing Christianity (forbidding Indigenous religious/spiritual practices); and instilling values regarding private property, material wealth, and monogamous nuclear families (disregarding Indigenous values of shared use of natural resources and broad, inclusive views of family; Adams, 1995; Littlefield & Knack, 1996). In addition to schooling, children were also required to do manual labor and industrial training, which included domestic work for girls (e.g., cooking, cleaning, laundry, sewing), and physical work for boys (e.g., farming, livestock raising, blacksmithing; Newland, 2022). Physical, emotional, and sexual abuse of children occurred in boarding schools, as well as malnutrition, poor sanitation, and infectious disease (Newland, 2022).

Boarding schools were often located away from reservations to minimize children’s contact with their families and tribes (Adams, 1995). During the summer, many children were sent to live with non-American Indian families for continued assimilation into White European American culture. Government policies initiated in the late 1800s made attendance “compulsory” at boarding schools for American Indian children, which led to children being forcibly removed from their families (Hoxie, 1989). Government officials were authorized to withhold food, clothing, and annuities from American Indian parents who refused to send their children to boarding schools (Adams, 1995). Some parents were even arrested and jailed for refusing to send their children. It is estimated that over 100,000 American Indian children attended boarding schools up until the 1960s (Dawson, 2012).

Child Welfare System Separations

From the 1920s to the 1960s, the boarding school movement faded (Dawson, 2012); however, the disconnection of American Indian children from their parents, families, and tribes continued through the child welfare system. In 1958, the Indian Adoption Project was created by the Bureau of Indian Affairs in partnership with the U.S. Children’s Bureau and the Child Welfare League of America to promote the adoption of American Indian children by White families (Palmiste, 2011). The Association of American Indian Affairs (2022) completed two studies in 1969 and 1974 demonstrating that 25%–35% of American Indian children had been removed from their families of origin and placed in foster care, adoptive homes, or institutions; and 90% of these children were placed in non-American Indian homes. Reasons for removal of children from their homes typically included alleged neglect, deprivation, and impoverished housing; abuse was rarely found (Palmiste, 2011). It is estimated that thousands of children were removed from their families and adopted during the 1960s and 1970s. These efforts continued the legacy of removing American Indian children from their homes for assimilation into White European American culture. Lawsuits, research, advocacy, and storytelling by families, tribes, and organizations led to the enactment of federal legislation to correct inhumane practices of the child welfare system (Association of American Indian Affairs, n.d.; Palmiste, 2011).

The Indian Child Welfare Act

The Indian Child Welfare Act was enacted in 1978 in response to the disproportionately high numbers of American Indian children removed from their homes and families in conjunction with placements and adoptions outside familial and tribal contexts (Cross, 2014). The statute states that the federal government has a responsibility to address this critical child welfare situation and “that an alarmingly high percentage of Indian families are broken up by the removal, often unwarranted, of their children from them by nontribal public and private agencies and that an alarmingly high percentage of such children are placed in non-Indian foster and adoptive homes and institutions”(Indian Child Welfare Act of 1978, 1978). ICWA aimed to enhance the stability and integrity of American Indian families, tribes, and cultures. Today, there are 574 federally recognized tribes (Bureau of Indian Affairs, 2023) and 63 state recognized tribes (National Conference of State Legislatures, 2023), which include many different cultures and tribal nations.

Some of the directives of ICWA apply specifically to foster care placement (ICWA, 1978). ICWA mandates that out-of-home placement of American Indian children prioritize (1) placement with extended family, (2) a foster home approved by the child’s tribe, (3) an American Indian foster home, or (4) an institution approved or operated by the tribe. ICWA defines an “Indian child” as any unmarried person under 18 years of age who is a member of an American Indian tribe or eligible for membership in a tribe and is the biological child of a member of a tribe. ICWA requires that active efforts be made to prevent the breakup of the family and promote reunification of children with parents. Such efforts can include assessing for family challenges and strengths, connecting children and families with comprehensive services so that children can live safely in the home, and partnering with tribal communities for social network supports and cultural resources.

Evaluation of Placement Outcomes for American Indian Children

Despite the importance of ICWA given the historical context, there has been limited evaluation of its effectiveness in reaching the goals of the legislation. Only a handful of studies have recently investigated placement-related outcomes for American Indian children in the context of ICWA (Francis et al., 2023). Findings across these studies paint a mixed picture. For example, a study in a southwestern state found that 83% of children were placed using ICWA guidelines, and that among those children 55% were placed with extended family, 33% were placed in homes approved by tribes, and 13% were placed in American Indian licensed foster homes (Limb et al., 2004). In addition, court cases examined showed that active efforts were used to prevent the break-up of the family. In another study, Barth et al. (2002) found that rates of kinship care were higher for American Indian children than for non-American Indian children. Conversely, a large national study found that American Indian children were more often placed into group homes or residential homes and less often placed into kinship care compared to non-American Indian children (Carter, 2009).

Predictors and Correlates for Placement Outcomes of Interest

Kinship care placement is the most frequently used placement type for children in out-of-home care (Denby, 2015). The decision-making process for children in out-of-home placement is under-researched (Meiksans et al., 2015). This means the factors that lead some children to be placed with foster parents versus relatives are not yet fully understood. Few studies have examined this phenomenon. Beeman et al. (2000) examined child characteristics leading to kinship placement for 2000 children in a Midwestern urban county. They found older children, children without disabilities, children of color, children court-ordered into care, and children who entered care due to parental substance abuse were more likely to be placed in kinship care. Grogan-Kaylor (2000) found child age, race, health status, reason for removal, and Aid for Families with Dependent Children, and status of families were related to the type of placement children were placed in.

Predictors of foster care placement in general include age, gender, race, marital status of parents, the caregiver and child relationship, previous allegation history, and severity of most recent allegation (McDonald et al., 2001). Bhatti-Sinclair and Sutcliffe (2012) found that several child characteristics impacted the placement of children, including emotional, behavioral, and medical issues, as well as the child being under the age of 1.

Few studies have examined matching of racial/ethnic identity among children and foster caretakers in placements. LaBrenz et al. (2022) was the first study to examine this phenomenon and used the Adoption and Foster Care Analysis and Reporting national dataset to analyze same-race/ethnicity foster care placement and found that over two-thirds of children were placed in same-race/ethnicity homes upon entering foster care. It also found that White children had the highest same-race/ethnicity placement rates, followed by African American children, then Native Hawaiian/Pacific Islander. Same-race/ethnicity matching increased placement stability.

Current Study

Few studies have examined the child welfare system’s compliance with ICWA by evaluating outcomes for American Indian children, and no studies have analyzed foster placement outcomes by American Indian status in recent years (Francis et al., 2023). Therefore, this paper will analyze placement outcomes of American Indian children compared to their non-American Indian counterparts. This study seeks to answer several research questions: (1) Are American Indian children more likely to be placed with at least one same-race/ethnicity caretaker (who is also American Indian), compared to their non-American Indian counterparts? (2) Are American Indian children more likely to be placed with relatives than their non-American Indian counterparts? (3) Are American Indian children more likely to experience a trial home placement (i.e., a child who has been in out-of-home care is returned to the home they were removed from on a trial basis) than their non-American Indian counterparts? Given the additional measures and protections afforded to American Indians via ICWA, one could expect higher rates of same-race/ethnicity placement, placement with relatives, and trial home placement for American Indian children compared to non-American Indian children.

Methods

This study was a secondary data analysis. We used data from the Adoption and Foster Care Analysis and Reporting System (AFCARS). AFCARS is a quantitative dataset system documenting foster care experiences for children throughout the United States. AFCARS collects individual and state-level foster care and adoptive outcomes for child welfare-involved children. AFCARS is spearheaded by the National Data Archive on Child Abuse and Neglect (NDACAN) at Cornell University on behalf of the Children’s Bureau, Administration on Children, Youth, and Families, U.S. Department of Health and Human Services. All data are de-identified (NDACAN, n.d.). AFCARS data will be used to examine placement outcomes for American Indian children versus their non-American Indian counterparts using three of the most recent datasets available: 2017, 2018, and 2019.

Variables

Outcome Variables

Three placement-related outcomes were explored: placement with same-race/ethnicity foster parent, placement with relatives, and trial home placement. The racial/ethnic identity variables for children and foster caretakers were used to create a new binary variable: whether or not there was a match between a foster caretaker’s race/ethnicity and the child’s race/ethnicity. The current placement variable was originally coded as one variable with eight categories (i.e., pre-adoptive home, foster home family with relative, foster home family not related, group home, institution, supervised independent living, runaway, and trial home placement). This variable was recoded to examine outcomes specific to ICWA: current placement with relatives was coded as 1, trial home placement was coded as 2, and all other placements were coded as 0. In a trial home placement, a child who has been in out-of-home care is returned to the parent(s) or home they were removed from; the trial home placement is intended to become a permanent home for the child, but the child remains the child welfare system’s responsibility.

Independent Variables

This study’s primary independent variable of interest was a child’s American Indian status. It was measured as a dichotomous variable. American Indian status was coded 1, and non-American Indian status was coded 0.

Several control variables were included based on prior research on predictors/correlates identified in the literature that have been associated with the outcomes, as well as other exploratory covariates available in AFCARS. Demographic covariates included child age at entry (continuous variable), gender (coded as male [0] or female [1]), and clinical disability (coded as “yes” [0], “no” [1], or “not yet determined” [2]). The removal manner from the home was coded as voluntary or non-court-ordered via agreement between parents and the child welfare agency (0), court-ordered (1), or not yet determined (2). Removal reasons included 10 variables: physical abuse, sexual abuse, neglect, parent substance abuse, child substance abuse, child disability or behavioral problem, parent incarceration, parents unable to caretake, inadequate housing, and parent death. All removal reason variables were coded as 1 for the presence of the removal reason and 0 for the absence of the removal reason. The total number of times the child was removed from the home was a continuous variable. The rurality of the county was coded as a 9-point interval variable ranging from more than 1 million population (metropolitan) to less than 2500 (rural). The state variable was recoded into five regions: Pacific West (AZ, CA, HI, NV, OR, WA), Midwest (IL, IN, IA, MO, MI, MN, NE, ND, OH, SD, WI), South (AL, AR, FL, GA, KY, LA, MS, NC, PR, SC, TN, TX, VA, WV), Mountain-Plains (AK, CO, ID, KS, MT, NM, OK, UT, WY), and Northeast (CT, DE, DC, ME, MD, MA, NH, NJ, NY, PA, RI, VT).

Data Analysis

Multinomial logistic regression was used to evaluate the outcomes of current placements. Binary logistic regression was used to examine same-race/ethnicity of foster parents. Before running the regression models, various assumptions, diagnostics, and other data issues were examined, including independence of observations, multicollinearity, clustering, and missing data. Missing data was not an issue for the variables except same-race/ethnicity foster parent. Current placement had 0.5%–0.6% missing values. Same-race/ethnicity foster parent had 33.7%–34.0% missing; this missingness varied significantly by state. The child’s status as American Indian or not had very little missing data (1.3%–2.5%). And, all of the covariates had little missing data (0.0%–4.4%). Given the large sample sizes, a complete case analysis was used.

Another data issue that was investigated was the potential clustering of the data by state. Using the intraclass correlation coefficient (ICC) developed by Raudenbush and Bryk (2002), we examined the clustering effect of the variable state for each outcome for each year. The ICC is the total explained variation in an outcome attributable to differences between states. Results showed the ICC was low for each year for current placement and somewhat low for same-race/ethnicity placement (2017, 2018, and 2019 respectively): current placement (.057, .057, and .058), and same-race/ethnicity placement (.103, .130, and .146). Generally, multilevel modeling is unnecessary when the ICC is less than .10 (Heinrich & Lynn, 2001; Kreft, 1996; Lee, 2000; What Works Clearinghouse, 2020) because the results would most likely be no different than results from regression modeling. Therefore, multinomial logistic regression models were run for the current placement outcomes, and robust binary logistic regression models were run for same-race/ethnicity placement outcomes using the Huber-White sandwich estimator (Rogers, 1993) to account for clustering. Odds ratios will be reported for the robust binary logistic regression models, and relative risk ratios will be reported for the multinomial logistic regression models. Pseudo R² values for each model are reported in the regression tables; however, given the limitations of pseudo R², they may only be useful for comparing models with the same outcome in the same dataset. In addition, bivariate analyses (chi-square tests) were run between the primary independent variable (child’s status as American Indian) and all of the outcomes. The same control variables were used in each model; however, geographic region was not used in the same-race/ethnicity foster parent models because the state was used as a clustering variable in these analyses. Data were analyzed using Stata 15.

Results

The sample characteristics are presented in Table 1. The results of bivariate chi-square analyses for American Indian status and the outcomes (same-race/ethnicity foster parent and current placement) are presented in Tables 2 and 3. The results of the regression analyses are presented below for each outcome and in Tables 4 and 5.

Table 1.

Sample Characteristics for Children Involved in the U.S. Child Welfare System.

	2017 Sample (N = 689,778)	2018 Sample (N = 688,562)	2019 Sample (N = 675,936)
Characteristic	M (SD) or %	M (SD) or %	M (SD) or %
Age	6.42 (5.40)	6.34 (5.37)	6.36 (5.37)
American Indian status
American Indian	5.05%	4.89%	4.87%
Non-American Indian	94.95%	95.11%	95.13%
Gender
Male	51.59%	51.47%	51.45%
Female	48.41%	48.53%	48.55%
Disability status
Disabled	22.44%	21.57%	23.27%
No disability	64.11%	64.69%	63.59%
Disability status not determined	13.46%	13.74%	13.13%
Urban-to-rural county Area	2.42 (1.93)	2.45 (1.95)	2.26 (1.95)
Geographic region
Northeast	11.70%	11.59%	21.32%
Midwest	24.77%	25.11%	25.14%
South	32.85%	33.52%	33.77%
Mountain-Plains	8.62%	8.54%	8.43%
Pacific West	22.06%	21.24%	21.32%

Note. Urban-to-Rural County Area was coded as 1 = more than 1 million (metropolitan) to 9 = less than 2500 (rural).

Table 2.

Bivariate Analysis for Same-Race/Ethnicity Placement and American Indian Status.

Independent variable	Same-Race/Ethnicity Placement		χ²
Independent variable	Same-Race/Ethnicity	Different Race/Ethnicity	χ²
Child’s status (2017)			3.2e+03^a
American Indian	15,214 (62.6%)	9088 (37.4%)
Non-American Indian	335,877 (78.2%)	93,440 (21.8%)
Child’s status (2018)			3.5e+03^a
American Indian	14,805 (62.0%)	9085 (38.0)
Non-American Indian	338,815 (78.4%)	93,356 (21.6%)
Child’s status (2019)			3.6e+03^a
American Indian	14,599 (61.3%)	9187 (38.7%)
Non-American Indian	333,121 (78.1%)	93,468 (21.9%)

^ap < .001.

Table 3.

Bivariate Analysis of Current Placement and American Indian Status.

Independent variable	Current Placement			χ²
Independent variable	Relative foster home	Trial home placement	Other placement	χ²
Child’s status (2017)				373.98^a
American Indian	10,213 (30.2%)	4509 (13.4%)	19,064 (56.4%)
Non-American Indian	185,786 (29.3%)	65,619 (10.3%)	383,821 (60.4%)
Child’s status (2018)				457.24^a
American Indian	10,006 (30.4%)	4515 (13.7%)	18,384 (55.6%)
Non-American Indian	186,474 (29.2%)	66,364 (10.4%	386,530 (60.5%)
Child’s status (2019)				518.90^a
American Indian	9856 (30.5%)	4469 (13.8%)	17,999 (55.7%)
Non-American Indian	182,036 (28.9%)	65,209 (10.3%)	383,708 (60.8%)

^ap < .001.

Table 4.

Binary Logistic Regression Analyses Predicting Same-Race/Ethnicity Foster Parents.

Independent variable	2017 Model (N = 437,312)			2018 Model (N = 437,436)			2019 Model (N = 435,535)
Independent variable	OR	[95% CI]	SE	OR	[95% CI]	SE	OR	[95% CI]	SE
Child’s status as American Indian	0.37^a	[0.90, 0.94]	0.082	0.36^a	[0.24, 0.56]	0.080	0.37^a	[0.24, 0.57]	0.081
Gender of child (male)	0.98^a	[0.90, 0.94]	0.010	0.93^a	[0.91, 0.95]	0.009	0.92^a	[0.91, 0.94]	0.008
Age of child	0.98	[0.96, 0.99]	0.009	0.98	[0.96, 0.99]	0.009	0.98	[0.96, 1.00]	0.010
Disability status of child (disabled)
No disability	1.31	[1.04, 1.64]	0.152	1.20	[0.91, 1.60]	0.172	1.05	[0.80, 1.40]	0.159
Disability status not determined	0.78	[0.47, 1.30]	0.204	0.71	[0.44, 1.15]	0.174	0.69	[0.44, 1.06]	0.152
Removal manner (voluntary)
Court ordered	1.64	[1.08, 2.49]	0.349	1.62	[1.07, 2.47]	0.347	1.68	[1.07, 2.66]	0.393
Not yet determined	1.51	[0.94, 2.43]	0.367	1.40	[0.84, 2.33]	0.365	1.57	[0.88, 2.80]	0.464
Removal reason
Physical abuse	1.02	[0.88, 1.19]	0.080	1.02	[0.89, 1.18]	0.075	1.00	[0.86, 1.16]	0.077
Sexual abuse	1.12	[0.94, 1.33]	0.101	1.13	[0.95, 1.36]	0.106	1.06	[0.92, 1.22]	0.076
Neglect	0.92	[0.76, 1.13]	0.094	0.92	[0.74, 1.13]	0.099	0.84	[0.66, 1.08]	0.106
Parental substance abuse	1.44^a	[1.18, 1.75]	0.143	1.43	[1.16, 1.79]	0.160	1.31	[1.11, 1.55]	0.112
Child substance abuse	1.12	[0.81, 1.55]	0.184	1.08	[0.78, 1.50]	0.179	0.99	[0.73, 1.34]	0.153
Child disability or behavior problem	0.46^a	[0.34, 0.62]	0.071	0.44^a	[0.32, 0.60]	0.070	0.45^a	[0.33, 0.60]	0.068
Parent incarceration	1.09	[0.92, 1.29]	0.094	1.07	[0.89, 1.29]	0.103	1.06	[0.90, 1.26]	0.093
Parent unable to caretake	1.26	[1.05, 1.53]	0.122	1.22	[1.01, 1.47]	0.117	1.14	[0.96, 1.34]	0.098
Inadequate housing	1.17	[0.99, 1.39]	0.101	1.17	[0.99, 1.38]	0.100	1.08	[0.95, 1.23]	0.074
Parent death	1.57^a	[1.34, 1.83]	0.125	1.41	[1.13, 1.77]	0.161	1.29	[0.95, 1.75]	0.199
Total removals from the home	1.05	[0.95, 1.15]	0.052	1.07	[0.97, 1.19]	0.055	1.08	[0.97, 1.21]	0.062
Urban-to-rural continuum of the county	1.08	[1.03, 1.15]	0.033	1.09	[1.03, 1.15]	0.031	1.07	[1.01, 1.13]	0.032
	pseudo R² = 0.0419			pseudo R² = 0.0410			pseudo R² = 0.0319

Note. Gender was coded male = 1, female = 0. The reference group for disability status are children who have been clinically diagnosed with a disability. The reference group for removal manner are children whose parents agreed to voluntary placement in out-of-home care. The Northeast was the reference group for geographic region.

^ap < .001.

Table 5.

Multinomial Logistic Regression Analysis Predicting Current Placement Type.

Independent variable	2017 Model (N = 617,814)						2018 Model (N = 617,384)
	Reference Category: Other placement						Reference Category: Other placement
	Placement with a relative			Trial home placement			Placement with a relative			Trial home placement
	RRR	SE	[95% CI]	RRR	SE	[95% CI]	RRR	SE	[95% CI]	RRR	SE	[95% CI]
Child’s status as American Indian	1.02	0.014	[0.99, 1.04]	0.94	0.019	[0.91, 0.98]	1.04	0.015	[1.01, 1.07]	0.97	0.019	[0.93, 1.01]
Gender of child (male)	0.95*	0.005	[0.94, 0.96]	1.01	0.009	[0.99, 1.03]	0.96*	0.006	[0.95, .097]	1.00	0.009	[0.99, 1.02]
Age of child	1.00*	0.001	[0.99, 1.00]	1.01*	0.001	[1.01, 1.01]	1.00*	0.001	[0.96, 1.00]	1.01*	0.001	[1.01, 1.01]
Disability status of child (disabled)
No disability	1.80*	0.015	[1.77, 1.83]	1.69*	0.021	[1.65, 1.73]	1.86*	0.015	[1.83, 1.89]	1.73*	0.021	[1.69, 1.77]
Disability status not determined	1.72*	0.019	[1.68, 1.76]	2.00*	0.030	[1.95, 2.06]	1.71*	0.019	[1.68, 1.75]	2.00	0.031	[1.95, 2.07]
Removal manner (voluntary)
Court ordered	1.41*	0.035	[1.34, 1.48]	1.80*	0.071	[1.66, 1.94]	1.42*	0.036	[1.35, 1.50]	1.57*	0.060	[1.46, 1.69]
Not yet determined	1.47*	0.050	[1.37, 1.57]	1.25*	0.070	[1.12, 1.40]	1.50*	0.052	[1.40, 1.61]	1.07	0.061	[0.96, 1.20]
Removal reason
Physical abuse	1.07*	0.010	[1.05, 1.09]	1.31*	0.016	[1.27, 1.34]	1.08*	0.010	[1.06, 1.10]	1.30*	0.016	[1.27, 1.33]
Sexual abuse	0.78*	0.013	[0.76, 0.81]	0.95	0.021	[0.91, 1.00]	0.80*	0.013	[0.77, 0.84]	0.94	0.021	[0.90, 0.98]
Neglect	1.19*	0.008	[1.17, 1.20]	1.24*	0.012	[1.21, 1.26]	1.18*	0.008	[1.16, 1.19]	1.28*	0.012	[1.26, 1.31]
Parental substance abuse	1.66*	0.011	[1.64, 1.68]	1.38*	0.013	[1.36, 1.41]	1.61*	0.010	[1.59, 1.63]	1.47*	0.014	[1.44, 1.49]
Child substance abuse	1.03	0.022	[0.98, 1.07]	1.09	0.031	[1.03, 1.15]	1.02	0.021	[0.98, 1.06]	0.96	0.027	[0.91, 1.01]
Child disability or behavior problem	0.34*	0.005	[0.33, 0.35]	0.96	0.016	[0.94, 1.00]	0.35*	0.005	[0.34, 0.36]	0.98	0.016	[0.95, 1.01]
Parent incarceration	1.11	0.012	[0.98, 1.07]	1.08*	0.017	[1.05, 1.12]	1.05*	0.011	[1.04, 1.08]	1.06*	0.016	[1.03, 1.10]
Parent unable to caretake	0.86*	0.007	[0.85, 0.87]	0.59*	0.007	[0.57, 0.60]	0.84*	0.007	[0.83, 0.85]	0.60*	0.007	[0.58, 0.61]
Inadequate housing	0.77*	0.007	[0.76, 0.79]	0.83*	0.011	[0.81, 0.85]	0.77*	0.007	[0.76, 0.79]	0.80*	0.011	[0.78, 0.82]
Parent death	1.58*	0.048	[1.49, 1.68]	0.47*	0.033	[0.41, 0.54]	1.40*	0.041	[1.32, 1.48]	0.54*	0.034	[0.48, 0.61]
Total removals from the home	0.80*	0.005	[0.79, 0.81]	0.78*	0.007	[0.76, 0.79]	0.81*	0.005	[0.80, 0.82]	0.78*	0.007	[0.76, 0.79]
Urban-to-rural continuum of the county	0.98*	0.002	[0.98, 0.98]	1.09*	0.002	[1.08, 1.09]	0.99*	0.002	[0.98, 0.99]	1.08*	0.002	[1.08, 1.09]
Geographic region (Northeast)
Midwest	0.88*	0.009	[0.86, 0.90]	2.00*	0.033	[1.94, 2.07]	0.96*	0.010	[0.94, 0.98]	1.94*	0.033	[1.88, 2.00]
South	0.66*	0.007	[0.65, 0.68]	0.89*	0.015	[0.86, 0.92]	0.66*	0.007	[0.65, 0.67]	0.89*	0.015	[0.86, 0.92]
Mountain-Plains	0.70*	0.010	[0.68, 0.72]	1.55*	0.031	[1.49, 1.61]	0.70*	0.010	[0.68, 0.72]	1.52*	0.031	[1.46, 1.58]
Pacific West	0.96*	0.010	[0.94, 0.98]	0.93*	0.018	[0.90, 0.97]	0.96*	0.011	[0.94, 0.98]	0.96	0.018	[0.92, 1.00]
	pseudo R² = 0.0496			pseudo R² = 0.0496			pseudo R² = 0.0460			pseudo R² = 0.0460

Independent variable	2019 Model (N = 611,371)
	Reference Category: Other placement
	Placement with a Relative			Trial home placement
	RRR	SE	[95% CI]	RRR	SE	[95% CI]
Child’s status as American Indian	1.06*	0.015	[1.03, 1.09]	1.01	0.020	[0.97, 1.05]
Gender of child (male)	0.96*	0.006	[0.95, 0.97]	1.01	0.009	[0.99, 1.02]
Age of child	1.00	0.001	[1.00, 1.00]	1.01*	0.001	[1.01, 1.01]
Disability status of child (disabled)
No disability	1.84*	0.015	[1.81, 1.87]	1.66*	0.019	[1.62, 1.70]
Disability status not determined	1.64*	0.018	[1.61, 1.68]	1.86*	0.028	[1.81, 1.92]
Removal manner (voluntary)
Court ordered	1.45*	0.038	[1.38, 1.52]	1.38*	0.052	[1.29, 1. 49]
Not yet determined	1.68*	0.061	[1.57, 1.80]	1.06	0.060	[0.94, 1.18]
Removal reason
Physical abuse	1.07*	0.010	[1.05, 1.08]	1.29*	0.016	[1.26, 1.31]
Sexual abuse	0.79*	0.013	[0.76, 0.81]	0.96	0.021	[0.92, 1.00]
Neglect	1.12*	0.008	[1.11, 1.14]	1.28*	0.012	[1.26, 1.31]
Parental substance abuse	1.54*	0.010	[1.52, 1.56]	1.46*	0.014	[1.43, 1.47]
Child substance abuse	1.04	0.022	[0.99, 1.08]	1.05	0.030	[0.99, 1.11]
Child disability or behavior problem	0.35*	0.005	[0.34, 0.36]	0.95	0.016	[0.92. 0.98]
Parent incarceration	1.09*	0.012	[1.07, 1.12]	1.19*	0.018	[1.16, 1.23]
Parent unable to caretake	0.85*	0.007	[0.83, 0.86]	0.62*	0.008	[0.60, 0.63]
Inadequate housing	0.77*	0.007	[0.87, 0.89]	0.79*	0.010	[0.76, 0.81]
Parent death	1.56*	0.047	[1.47, 1.66]	0.41*	0.030	[0.36, 0.48]
Total removals from the home	0.81*	0.005	[0.80, 0.82]	0.79*	0.007	[0.78, 0.81]
Urban-to-rural continuum of the county	0.99*	0.002	[0.99, 0.99]	1.08*	0.002	[1.08, 1.08]
Geographic region (Northeast)
Midwest	0.91*	0.010	[0.87, 0.92]	1.78*	0.029	[1.72, 1.84]
South	0.61*	0.006	[0.60, 0.63]	0.83*	0.013	[0.80, 0.86]
Mountain-Plains	0.67*	0.009	[0.65, 0.69]	1.38*	0.027	[1.33, 1.44]
Pacific West	0.91*	0.010	[0.90, 0.93]	0.86*	0.017	[0.85, 0.92]
	pseudo R² = 0.0430			pseudo R² = 0.0430

*p < .001.

Same-Race/Ethnicity Placement

Results for the robust binary logistic regression analyses for same-race/ethnicity placement are in Table 4. Child status as American Indian was significant: American Indian children had 63%–64% lower odds of being placed with a same-race/ethnicity caretaker, compared to non-American Indian children. Boys were 2%–8% less likely to be placed with same-race/ethnicity caretakers than girls. Age, disability status, removal manner, total removals, and many removal reasons (physical abuse, sexual abuse, neglect, child substance abuse, parental incarceration, parent ability to caretake, and inadequate housing) were not significant predictors. Children removed for parental substance abuse had 44% higher odds of being placed with a same-race/ethnicity foster caretaker in 2017. Children with clinical disabilities or behavior problems were 54%–56% less likely to be placed with a same-race/ethnicity foster caretaker. Total removals was not a significant predictor of placement with same-race/ethnicity caretaker. Children with parental deaths were 57% more likely to be placed with same-race/ethnicity caretakers. Rurality was not significantly related to the outcome.

Current Placement

Placement with a Relative

Results for the multinomial logistic regression analyses for current placement are presented in Table 5. Child status as American Indian was only statistically significant in 2019: American Indian children were 6% more likely to be placed with relatives than non-American Indian children. Boys were 4%–5% less likely to be placed with relatives than girls. Children’s age was not related to placement. Children without disabilities were 80%–86% more likely to be placed with relatives than children with disabilities. Children with disability status not yet determined were 64%–72% more likely to be place with relatives than children with disabilities. Children with removal by court order were 41%–45% more likely to be placed with relatives than children who entered voluntarily. Children with removal status undetermined were 47%–68% more likely to be placed with relatives than children who entered voluntarily. Children removed for physical abuse were 7%–8% more likely to be living with relatives. Children removed for sexual abuse were 20%–22% less likely to be living with relatives. Children removed for neglect were 12%–19% more likely to be living with relatives. Children removed for parental substance abuse were 54%–66% more likely to be residing with relatives. Child substance abuse was not a significant predictor. Children with a clinical disability or behavioral concern were 64%–65% less likely to be placed with relatives. Children with incarcerated parents were 5% and 9% more likely to be living with relatives in 2018 and 2019 respectively; 2017 results were non-significant. Children with parents unable to care for them were 14%–16% less likely to be living with relatives. Children with inadequate housing were 23% less likely to be living with relatives. Children with parental deaths were 40%–58% more likely to be living with relatives. As the number of removals increase, children were 19%–20% less likely to be residing with relatives. As residence became more rural, children were 1%–2% less likely to be residing with relatives. Compared to children in the Northeast, children in other regions were less likely to live with relatives to varying levels: Midwest (4%–9%), South (34%–39%), Mountain-Plains (23%–30%), and Pacific West (4%–9%).

Trial Home Placement

Child status as American Indian was not significantly related to trial home placement (Table 5). Gender was not a significant predictor. As child age increased by 1, children were 1% more likely to be in a trial home placement. Children without disabilities were 66%–73% more likely to be in a trial home placement than their disabled counterparts. Children removed by the court system were 38%–80% more likely to be in a trial home placement than children that entered voluntarily. Children removed for physical abuse were 29%–31% more likely to be in a trial home placement than children removed for other reasons. Sexual abuse removal was not a significant predictor. Children removed for neglect were 24%–28% more likely to be in trial home placement than children removed for other reasons. Children removed for parental substance abuse were 38%–47% more likely to be in a trial home placement. Child substance abuse and child disability or behavioral problem were not significant predictors. Children with parental incarceration were 6%–19% more likely to be in a trial home placement. Children with caretakers unable to care for them were 38%–41% less likely to be in a trial home placement. Children with inadequate housing were 17%–21% less likely to be in a trial home placement. Children with parental deaths were 46%–59% less likely to be in a trial home placement. As the total number of removals increases, children were 21%–22% less likely to be in a trial home placement. As a child’s county of residence became more rural, the child was 8%–9% more likely to be in a trial home placement. Compared to children in the Northeast, children in the Midwest (78%–100%) and Mountain-Plains (38%–55%) regions were more likely to be in a trial placement. Compared to children in the Northeast, children in the South (11%–17%) and Pacific West (4%–14%) were less likely to be living in a trial home placement.

Discussion

This study aimed to evaluate the extent to which ICWA is reaching its stated goals by examining child placement outcomes for American Indian children and families in comparison to non-American Indians because ICWA provides specific protections and additional measures regarding placement outcomes for American Indian children. In this section, we will discuss the findings from same-race/ethnicity placement and current placement, study strengths and limitations, and directions for future research.

Same-Race/Ethnicity Placement Findings

American Indian children were significantly less likely to be placed with a same-race/ethnicity caretaker than their non-American Indian peers (61%–63% vs. 78%; OR = 0.36–0.37). Although this effect was small (Cohen, 1988), in the context of the population of tens of thousands of children, it is noteworthy. This suggests that ICWA is not reaching its intended goal of tribal and cultural preservation for children while children are in out-of-home care. The weight of this finding as a lack of policy success over three recent years should be underscored given a recent study conducted by LaBrenz et al. (2022), which found that same-race/ethnicity placements improve placement stability for children. When a child is removed from the home and placed in out-of-home placement, various factors come into play when determining the best placement, including racial/ethnic and cultural factors.

Current Placement Findings

Current placement findings are important because they show whether a child is placed with a relative or a non-relative foster parent, as well as trial home placement. American Indian children were not more likely to be placed with relatives (30% vs. 29%; RRR = 1.02–1.06) or have trial home placement (13%–14% vs. 10%; RRR = 0.94–1.01) compared to non-American Indian children. These findings suggest that ICWA may not be making a substantial difference for American Indian families regarding kinship placement and trial home placement. Kinship placements often have benefits for all children, including fewer placement disruptions, potentially fewer mental health and behavioral problems, and a lower likelihood of being adopted (Winokur et al., 2014). However, kinship placement is especially important for American Indian families to maintain family and cultural connections amid a history marked by attempted genocide, ethnic cleansing, and cultural destruction.

Covariate Findings

Among the covariate findings, we note that parental substance abuse as a removal reason showed a small but significant association with placement with a same-race/ethnicity foster caretaker or relative caretaker. It may be the case that parents with substance use problems have relatives or community members who are willing to foster their children while parents are in substance use treatment. This finding may be particularly relevant to American Indians, who face disproportionately higher rates of substance use problems, such as alcohol (Soto et al., 2022).

Implications for Policy and Practice

Study findings demonstrate that American Indian children are not being placed at high rates in homes that can preserve their familial integrity and cultural heritage. Findings suggest that the aspirations of ICWA are not being met. Current shortcomings should be addressed so that the objectives of ICWA can be reached. Agencies should recruit more American Indian foster parents to be available to American Indian children when they come into care. These foster parents should reflect the children’s tribes who enter foster care. This would help with placing American Indian children in culturally congruent placements. More support for relatives is also necessary to increase the number of relatives available to care for American Indian children in foster care. Guardianship assistance is one form of support that could help relatives provide care to American Indian children in out-of-home placement. Guardianship includes training and financial support. More services must be available to children so they can have a trial home placement. This could include in-home counseling or therapy for the family as a unit.

Furthermore, policy-related measures should be taken so that there are negative consequences for states that do not follow the ICWA standards. Compliance with ICWA can lead to positive outcomes for American Indian children. For example, one study found that compliance with ICWA resulted in more placements with relatives and tribal-approved placements for American Indian children (Limb et al., 2004). ICWA compliance along with public social services can play a vital role in increasing the number of relative placements by identifying relatives and providing sufficient resources to ensure the stability of the placement, such as financial support and training.

Study Strengths and Limitations

This study has several strengths, including using large recent national datasets across multiple years. The consistency in results across years enhances the confidence in the findings. In addition, the findings are representative of children and families involved in the U.S. child welfare system. One limitation of this study is the variables available in the AFCARS dataset because relevant control variables were identified in the literature but were not included in AFCARS. For example, research shows that certain parent variables (e.g., one parent vs. two-parent home, the severity of recent allegation) have been related to placement outcomes (McDonald et al., 2001); however, the AFCARS datasets did not include such variables. Another limitation of AFCARS data is that it does not collect tribal affiliation, so we cannot ascertain if a child was placed with their same tribe or not. The high volume of cases in the dataset resulted in many very small effect sizes that were statistically significant at the p < .05 level, which made it challenging to determine which differences and associations were meaningful or practically significant. AFCARS is dependent on data entry from states, and there can be a lag in data entry causing missing values among variables. There can also be differences in how thoroughly states collect data, which can lead to missing values. This was the case for the foster caretaker race/ethnicity variable that had a high number of missing values. Key variables with high levels of missing data, such as foster caretaker race/ethnicity, raise concern about the generalization of findings related to this outcome. Finally, the findings should be interpreted cautiously because they rest on the assumption that American Indian children should have more desirable placement outcomes than non-American Indian children because of ICWA. It may be that ICWA is providing some benefits to American Indian children, which would otherwise be much worse if the law did not exist.

Future Research

Adoption and Foster Care Analysis and Reporting System data collection should be modified to include tribal affiliation of children, parents, and foster caretakers, so that researchers can examine if American Indian children are placed with their tribal communities. This can be done by requiring case workers to collect this information and enter it into AFCARS. Such descriptive information could yield important results regarding placement outcomes for American Indian children. Future research should also examine the stability of placement into kinship care and reunification for American Indian children. This can be done by examining the placement history of children in care and the number of days spent in each placement. Mixed-methods research is needed involving the day-to-day decisions and actions for American Indian children and families involved in the child welfare system to ascertain the extent to which ICWA is being implemented as intended. This implementation data is essential to understand ICWA-related outcomes. Indeed, more work is needed to bolster the intended aims of ICWA and to improve outcomes for American Indian children and families.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The first author was supported with funding from the Sequoyah Fellowship for American Indian Research through the Graduate School at the University of North Carolina at Chapel Hill, as well as the Health Policy Research Scholars program from the Robert Wood Johnson Foundation.

ORCID iD

William J. Hall

References

Adams

D. W.

(1995). Education for extinction: American Indians and the boarding school experience (pp. 1875–1928). University Press of Kansas.

Association of American Indian affairs . (2022). Indian child welfare act, https://www.indian-affairs.org/icwa.html

Barth

R. P.

Webster

Lee

(2002). Adoption of American Indian children: Implications for implementing the Indian child welfare and adoption and safe families acts. Children and Youth Services Review, 24(3), 139–158. https://doi.org/10.1016/s0190-7409(02)80002-0

Beeman

S. K.

Kim

Bullerdick

S. K.

(2000). Factors affecting placement of children in kinship and nonkinship foster care. Children and Youth Services Review, 22(1), 37–54. https://doi.org/10.1016/s0190-7409(99)00072-9

Bhatti-Sinclair

Sutcliffe

(2012). What determines the out-of-home placement of children in the USA? Children and Youth Services Review, 34(9), 1749–1755. https://doi.org/10.1016/j.childyouth.2012.05.004

Brave Heart

M. Y.

DeBruyn

L. M.

(1998). The American Indian holocaust: Healing historical unresolved grief. American Indian and Alaska Native Mental Health Research: Journal of the National Center, 8(2), 56–78. https://doi.org/10.5820/AIAN.0802.1998.60

Bureau of Indian Affairs . (2023). Indian entities recognized by and eligible to receive services from the United States bureau of Indian affairs. https://www.federalregister.gov/documents/2021/01/29/2021-01606/indian-entities-recognized-by-and-eligible-to-receive-services-from-the-united-states-bureau-of

Carter

V. B.

(2009). Comparison of American Indian/Alaskan natives to non-Indians in out-of-home care. Families in Society: The Journal of Contemporary Social Services, 90(3), 301–308. https://doi.org/10.1606/1044-3894.3895

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Academic Press.

10.

Cross

T. L.

(2014). The Indian child welfare act: We must still fight for our children. Reclaiming Children and Youth, 23(2), 23–24.

11.

Dawson

A. S.

(2012). Histories and memories of the Indian boarding schools in Mexico, Canada, and the United States. Latin American Perspectives, 39(5), 80–99. https://doi.org/10.1177/0094582x12447274

12.

Denby

R. W.

(2015). Kinship care: Increasing child well-being through practice, policy, and research. Springer Publishing Company.

13.

Francis

A. M.

Hall

W. J.

Ansong

Lanier

Albritton

T. J.

McMillan

(2023). Implementation and effectiveness of the Indian child welfare Act: A systematic review. Children and Youth Services Review, 146(3), 106799. https://doi.org/10.1016/j.childyouth.2022.106799

14.

Grogan-Kaylor

(2000). Who goes into kinship care?: The relationship of child and family characteristics to placement into kinship foster care. Deutsches Zentralinstitut für soziale Fragen/DZI.

15.

Heinrich

C. J.

Lynn

L. E.

(2001). Means and ends: A comparative study of empirical methods for investigating governance and performance. Journal of Public Administration Research and Theory, 11(1), 109–138. https://doi.org/10.1093/oxfordjournals.jpart.a003490

16.

Hoxie

(1989). A final promise: The campaign to assimilate the Indians: 1820–1920. Cambridge University Press.

17.

Indian Child Welfare Act of 1978 (1978). Public Law, No. 95-608, 92 Stat. 3069. https://www.govinfo.gov/content/pkg/STATUTE-92/pdf/STATUTE-92-Pg3069.pdf

18.

Kreft

I. G.

(1996). Are multilevel techniques necessary? An overview including simulation studies. California State University. Unpublished manuscript.

19.

LaBrenz

C. A.

Kim

Harris

M. S.

Crutchfield

Choi

Robinson

E. D.

Findley

Ryan

S. D.

(2022). Racial matching in foster care placements and subsequent placement stability: A national study. Child and Adolescent Social Work Journal, 39(5), 583–594. https://doi.org/10.1007/s10560-022-00831-x

20.

Lee

V. E.

(2000). Using hierarchical linear modeling to study social contexts: The case of school effects. Educational Psychologist, 35(2), 125–141. https://doi.org/10.1207/s15326985ep3502_6

21.

Limb

G. E.

Chance

Brown

E. F.

(2004). An empirical examination of the Indian child welfare act and its impact on cultural and familial preservation for American Indian children. Child Abuse and Neglect, 28(12), 1279–1289. https://doi.org/10.1016/j.chiabu.2004.06.012

22.

Littlefield

Knack

M. C.

(Eds), (1996). Native Americans and wage labor: Ethnohistorical perspectives. University of Oklahoma Press.

23.

McDonald

T. P.

Poertner

Harris

(2001). Predicting placement in foster care: A comparison of logistic regression and neural network analysis. Journal of Social Service Research, 28(2), 1–20. https://doi.org/10.1300/j079v28n02_01

24.

Meiksans

Iannos

Arney

(2015). Factors influencing decision making about the placement of children in care: Development of the child placement questionnaire. Children and Youth Services Review, 55, 71–83. https://doi.org/10.1016/j.childyouth.2015.05.002

25.

National Conference of State Legislatures . (2023). State recognition of American Indian tribes. https://www.ncsl.org/quad-caucus/state-recognition-of-american-indian-tribes

26.

Newland

(2022). Federal Indian boarding school initiative investigative report. Bureau of Indian Affairs, U.S. Department of the Interior.

27.

Palmiste

(2011). From the Indian adoption project to the Indian child welfare act: The resistance of native American communities. https://hal.univ-antilles.fr/hal-01768178/document

28.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Sage.

29.

Rogers

W. H.

(1993). Regression standard errors in clustered samples. Stata Technical Bulletin, 13, 19–23.

30.

Soto

West

A. E.

Ramos

G. G.

Unger

J. B.

(2022). Substance and behavioral addictions among American Indian and Alaska native populations. International Journal of Environmental Research and Public Health, 19(5), 2974. https://doi.org/10.3390/ijerph19052974

31.

Thornton

(2005). Native American demographic and tribal survival into the twenty-first century. American Studies, 46(3/4), 23–38.

32.

United States House Report . (1978) No. 1386, 95th congress, 2nd session. Establishing standards for the placement of Indian children in foster or adoptive homes. To prevent the breakup of Indian families, United States Government Printing Office.

33.

What Works Clearinghouse . (2020). Procedures handbook. version 4.1. https://ies.ed.gov/ncee/wwc/handbooks

34.

Winokur

Holtan

Batchelder

K. E.

(2014). Kinship care for the safety, permanency, and well‐being of children removed from the home for maltreatment: A systematic review. Campbell Systematic Reviews, 10(1), 1–292. https://doi.org/10.4073/csr.2014.2

35.

Edwards

F. R.

Wildeman

(2020). Cumulative prevalence of confirmed maltreatment and foster care placement for US children by race/ethnicity, 2011–2016. American Journal of Public Health, 110(5), 704–709, https://doi.org/10.2105/AJPH.2019.305554