Understanding Overrepresentation of Indigenous Children in Child Welfare Data

Abstract

Child welfare data collected for administrative purposes are often used as a source of information for understanding the population impact of child abuse and neglect (CA/N). This study used administrative data linked at the individual level for a cohort of Aotearoa New Zealand (NZ) children to follow and extend a model developed by Drake et al. In this investigation, we aimed to build an understanding of the high representation of indigenous NZ children in administratively sourced measures of CA/N. Variation in rate ratios (RRs) within infant mortality and birth outcomes considered as possible proxies for actual CA/N RRs leaves open a range of interpretations. Our findings indicate that a more nuanced interpretation of the overrepresentation of indigenous children in administratively recorded maltreatment statistics is required. Rather than considering risk and bias as competing explanations, we suggest an acknowledgment of the impact of colonization and the existence of systemic bias generating increased risk as key drivers. As linked administrative data are increasingly used for research and evaluation, and considered for use in supporting decision making, there is a need for a deeper understanding of the drivers of administratively recorded CA/N in order to effectively address the needs of indigenous populations.

Keywords

child maltreatment ethnic minority populations child welfare services

Introduction

There are inherent difficulties in determining the true prevalence and incidence of child abuse and neglect (CA/N). CA/N most frequently occurs within the confines of a child’s own home and begins before the age that the child has the required social and verbal skills to understand and articulate what is occurring. As a result, CA/N is underreported, leading to uncertainty about whether those cases that are reported accurately reflect the distribution of harm within the community (Brownwell & Jutte, 2013; Gilbert et al., 2011). Despite this, child welfare data collected for administrative purposes are used in many developed countries as one of the main sources of information for understanding the population impact of CA/N (Brownwell & Jutte, 2013; Gilbert et al., 2011). There is growing interest in linking child welfare data with data from health and other administrative systems for research and program evaluation and improved decision making (Brownwell & Jutte, 2013; Putnam-Hornstein, Wood, Fluke, Yoshioka-Maxwell, & Berger, 2013). With this, comes a need to understand the degree to which overrepresentation of some population groups in these data is proportionate to real differences in their exposure to risk and experience of harm. If child welfare data overstate exposure, for example, then their use in decision making has the potential to feed a cycle of bias in reports and substantiations of maltreatment (Fluke, Shusterman, Hollinshead, & Tyuan, 2008; Harcourt, 2006).

Indigenous Peoples in Child Welfare Data

The overrepresentation of indigenous people in administratively recorded CA/N is an international phenomenon. Indigenous peoples in Canada, the United States, and Australia have had similar experiences to Maori in Aotearoa New Zealand (NZ). In each case, indigenous children and young people are overrepresented among those found to be maltreated and those subsequently removed from their familial home (UNICEF, 2013). Indigenous people argue that this is the result of a combination of factors, including assimilationist policies of colonial governments leading to the fragmentation of families, inequitable distribution of the goods and resources of society (e.g., employment, housing, and wealth), systemic racism of a child welfare protection system imposing white middle-class notions of family and child-rearing upon indigenous families (Blackstock & Trocmé, 2005; Grier, 2005), and racial bias in reporting of maltreatment and in child welfare agency decision making (Cram, 2012; Trocmé, Knoke, & Blackstock, 2004). This is not to ignore the maltreatment of children within some indigenous families and contexts. Rather, it is an acknowledgment that “[p]rotection for Indigenous children is complex: they require the same protective measures as non-Indigenous children but also require culturally-sensitive support” (Woolley, 2009, p. 400).

The issue of indigenous CA/N fits the description of a “wicked problem” (Rittel & Webber, 1973). It has many etiological drivers and thus requires a multilayered response (Cram, 2012). Complex explanations and structural responsiveness are, however, in tension with testing straightforward hypotheses for indigenous and minority group disparities within child welfare systems. In an investigation to generate new knowledge about the nature of and response to maltreatment of First Nations children in Canada, Sinha and colleagues compared details of investigations involving First Nations and non-Aboriginal children included in the Canadian Incidence Study (2008; Sinha et al., 2011). The investigators found that the protection from immediate, severe harm was not the central concern for most First Nations children. Instead, difficulties facing many of the families required programs offering longer term, comprehensive services designed to help them address factors (including poverty and social isolation) which posed chronic challenges to their ability to ensure the well-being of their children (Sinha et al., 2011). The World Health Organization (WHO) has also underlined the importance of political, social, and economic forces in driving health and social outcomes. In order to address the social determinants of health, WHO recommend the development of policies and programs that cut across the whole of society, by understanding the problems of inequities and tackling resource redistribution (Commission on Social Determinants of Health, 2008).

Empirical Explanations for Disparities

A number of empirical investigations to understand disparities have been U.S. based, looking at Black children’s engagement with and passage through the child welfare system. However, information about the vulnerability of these children may not be applicable to indigenous (Native American) children and families in the United States, let alone indigenous children and families in other countries (Trocmé et al., 2004). In one study, Drake and colleagues (2011) sought to examine racial differences in victimization rates in the United States, examining disproportionate risk (the “risk model”) and bias in reporting and in the child welfare system (the “bias model”) as competing explanations. Drake et al. tested the risk and bias models by conducting an ecological investigation—comparing rates of victimization from official child welfare organizations with rates of key public health outcomes from health sources.

Drake’s “risk” model states that official statistics accurately report CA/N rates for minority group children. It is the higher exposure to risk factors for maltreatment that accounts for their heightened levels of administratively reported CA/N (see Figure 1; Drake et al., 2011; Joelving, 2011). In Aotearoa NZ, unequal exposure to risk is clearly evident. Maori bears a higher share of social harms that are risk factors for CA/N (Child Youth and Family [CYF], 2006; Cooper & Wharewera-Mika, 2009; Cram, 2012), the most pervasive of which is poverty (Trocmé et al., 2004). Using an after housing costs measure, 34% of Maori children are estimated to live in poverty compared to 16 percent of European children (Perry, 2013). Maori children are 4 times more likely to experience persistent deprivation than non-Maori, non-Pacific children (Imlach Gunasekara & Carter, 2012).

Figure 1.

The risk model (Drake et al., 2011, p. 473).

Drake’s “bias” model states that bias among those reporting and investigating suspected CA/N results in hyper-surveillance and discriminatory treatment by the child welfare system (see Figure 2). Historically, this view has been a strong motivator for change in the Aotearoa NZ care and protection system. Almost 30 years ago, a Ministerial Advisory Committee identified institutional racism as a key issue. “At the heart of the issue is a profound misunderstanding or ignorance of the place of the child in Māori society and its relationship with whānau [Māori families], hapū [sub-tribe], iwi [tribe] structures” (Ministerial Advisory Committee on a Maori Perspective for the Development of Social Welfare, 1988, p. 7). The Committee recommended tackling cultural racism, eliminating deprivation, enabling Iwi to make decisions about child welfare services for Maori, and making the child welfare agency an intervention of last resort.

Figure 2.

The bias model (Drake et al., 2011, p. 473).

To test whether increased risk or bias was driving the overrepresentation of Black children, Drake and colleagues compared rates of infant mortality and poor birth outcomes with administratively recorded CA/N. Their assumption was that infant mortality and birth outcomes could be viewed as sensitive to the same risk factors as CA/N but not sensitive to bias. Differences between groups in rates (rate ratios [RRs]) of infant mortality and birth outcomes were used as a proxy for actual differences in rates of maltreatment. Larger RRs of administratively recorded CA/N relative to RRs for infant mortality and poor birth outcomes would indicate bias, while comparable or smaller RRs would indicate risk. The authors reported that, although their findings could not preclude the possibility of bias, bias in reporting and investigating suspected maltreatment was not a large-scale driver of the overrepresentation of Black children in the U.S. child welfare system (Drake et al., 2011). Importantly, the Drake study was based on ecological investigation, therefore group-level associations do not necessarily translate at the individual level.

The aim of this investigation was to follow and extend Drake et al.’s study using administrative data linked at the individual level for a cohort of Aotearoa NZ children to build an understanding of the high representation of Maori children in administratively sourced measures of CA/N.

Method

Study Population

The study population included all live births from July 1, 2004, to June 30, 2007 (N = 180,794), where there was either (i) a birth registration completed by the time the child reached 5 years of age or, where this did not exist (approximately 1% of live births are registered more than 2 years after the birth of the child, Statistics NZ, 2010), (ii) a birth event recorded in the Ministry of Health (MoH) Maternity Collection. Child–parent relationships were as recorded in birth registration data (where present) or as recorded in the maternity data.

Data

We probabilistically linked data for children and their parents held in national systems for administering: (i) birth notifications and registrations (Statistics NZ, 2013), (ii) publicly funded (universal) maternity services (MoH, 2011), (iii) public welfare benefits (Ministry of Social Development, 2014), (iv) child welfare (care and protection) services (CYF, 2014), (v) publicly funded hospitalizations (MoH, 2009), (vi) corrections sentences (Statistics NZ, 2014), and (vii) mortality records (MoH, 2013). Ethics approval was granted by the Central Region Health and Disability Ethics Committee. Details of the linkage method and assessments of accuracy are in the Online Appendix.

Variables

Details of variable definitions are included in Online Appendix Table 1.

Table 1.

Measures of Risk and Outcomes for Maori and non-Maori, non-Pacific Children.

	Maori (n = 52,783)				non-Maori, non-Pacific (n = 108,621)
	n	Rate (per 1,000 live births)	LCI	UCI	n	Rate (per 1,000 live births)	LCI	UCI
Risk
Mother supported by welfare benefit for 4+ of the last 5 years	16,202	306.95	302.23	311.68	5,704	52.51	51.15	53.88
Mother with prison sentence in last 5 years	869	16.46	15.37	17.56	293	2.70	2.39	3.01
Father with prison sentence in last 5 years	1,799	34.08	32.51	35.66	975	8.98	8.41	9.54
Child supported by welfare benefit in first 3 months of life	23,990	454.50	448.75	460.25	12,565	115.68	113.65	117.70
Mortality and birth outcomes not subject to substantial classification error
Birth weight <2.5 kg	3,765	71.33	69.05	73.61	6,487	59.72	58.27	61.17
Birth at <32 weeks	959	18.17	17.02	19.32	1,370	12.61	11.94	13.28
Birth at <37 weeks	4,530	85.82	83.32	88.32	8,216	75.64	74.00	77.27
Infant mortality—neonatal	200	3.79	3.26	4.31	272	2.50	2.21	2.80
Infant mortality—postneonatal	212	4.02	3.48	4.56	129	1.19	0.98	1.39
Mortality measures with higher apparent potential for classification error
SUDI	117	2.22	1.81	2.62	38	0.35	0.24	0.46
Nonmotor vehicle accident mortality <3	55	1.04	0.77	1.32	33	0.30	0.20	0.41
Violent death mortality <3	13	0.25	0.11	0.38	3	0.03	0.00	0.06
Administratively sourced CA/N rates and indicators of concern
Notification or police family violence contact record <3	11,639	220.51	216.50	224.51	6,230	57.36	55.93	58.78
Investigation or CFA <3	7,845	148.63	145.34	151.92	4,013	36.94	35.80	38.09
Substantiated finding of maltreatment <3	4,031	76.37	74.01	78.73	1,727	15.90	15.15	16.65
Substantiated finding of sexual abuse <3	23	0.44	0.26	0.61	16	0.15	0.08	0.22
Substantiated finding of physical abuse <3	319	6.04	5.38	6.71	170	1.57	1.33	1.80
Substantiated finding of neglect <3	1,628	30.84	29.34	32.34	783	7.21	6.70	7.71
Substantiated finding of emotional abuse <3	2,879	54.54	52.55	56.54	1,062	9.78	9.19	10.37
In care of chief executive <3	973	18.43	17.28	19.59	573	5.28	4.84	5.71
Hospitalization for intracranial injury <1	63	1.19	0.90	1.49	43	0.40	0.28	0.51
Hospitalization for fracture of long bones injury <1	65	1.23	0.93	1.53	66	0.61	0.46	0.75
Hospitalization for intentional injury <1	92	1.74	1.39	2.10	33	0.30	0.20	0.41
Hospitalization for intentional injury <3	162	3.07	2.60	3.35	71	0.65	0.50	0.81

Note. SUDI = sudden unexpected death in infancy; CFA = Child and Family Assessment; CA/N = child abuse and neglect; LCI = lower confidence interval; UCI = upper confidence interval.

Ethnic group

Ethnicity of the child was derived from tick box and free text data captured on birth registration forms according to the NZ Statistical Standard for Ethnicity (Statistics NZ, 2013). Parents are invited to indicate the ethnic group(s) to which the child belongs. Where there was no birth registration for the child, MoH maternity data on the ethnicities of the child were used. For the purposes of this investigation, where the data indicated that the child belonged to Maori and other ethnic groups, the child was categorized as Maori.

Pacific peoples have also been recorded with higher levels of family violence than non-Maori, non-Pacific peoples (Fanslow, Robinson, Crengle, & Perese, 2010). As such, we compared Maori children with non-Maori, non-Pacific children.

Administratively sourced CA/N rates and indicators of concern

Substantiated findings of CA/N by type prior to the child turning 3 were obtained from CYF data. CYF data were also used to examine reports of concern (notifications), further action taken in respect of the case (investigation or Child and Family Assessment [CFA]), or the child being taken into the care of the chief executive.

Hospital events associated with CA/N prior to 3 years of age were identified from publicly funded hospital discharge data (National Minimum Dataset).

Risk

Household income levels were not available from the linkage. Therefore, maternal duration of welfare benefit receipt in the 5 years prior to the birth of the child was used as a proxy for poverty persistence. Parental history of imprisonment in the 5 years prior to the birth of the child (Statistics NZ, 2014) was also included as a marker of risk.

Mortality and birth outcomes not subject to substantial classification error

Mortality outcomes not subject to substantial classification error (and without potential for racial bias to affect reporting or measurement) were obtained from MoH Mortality Collection records and included neonatal and postneonatal infant mortality (MoH, 2013).

Mortality outcomes with higher potential for classification error

Mortality outcomes with higher potential for classification error (where, e.g., racial bias might impact the determination of the cause of death) were also examined. These were obtained from the mortality collection (MoH, 2013) and included violent death mortality, unintentional injury (excluding motor vehicle traffic) mortality and sudden unexpected death in infancy (SUDI).

Analysis

Following Drake et al., we calculated RRs—the rate of an outcome for Maori children in the cohort divided by the rate of the same outcome for non-Maori, non-Pacific children—for each variable. Selected RRs were stratified by time on benefit, as a proxy for poverty persistence, and maternal age. We compared children with mothers aged less than 25 years with those with mothers aged 25 years and over. For ease of interpretation, we examined the extremes of benefit dependency and present results for those with mothers who had been on the benefit for 4 or more of the 5 years prior to the child being born and those whose mothers had not been on the benefit in that time. Of all children in the study population with substantiated findings of maltreatment by age 3, 53.3% had a mother who had been supported by benefit for 4 or more of the 5 years prior, and 8.5% had a mother who had not been supported by benefit in that time. For the purposes of this investigation children born to young mothers who had formerly been in care and appeared to have been supported by benefit for no time in the 5 years prior to the birth were excluded. This is a highly disadvantaged group who would have been supported outside of the welfare benefit system by a foster care allowance or through residential institutions.

Tests of statistical significance were not applied as we were interested in whether differences existed rather than the strength of the evidence against the null hypothesis (Vittinghoff, Glidden, Shiboski, & McCulloch, 2004). We considered RRs whose 95% confidence intervals (CIs) did not include 1.0 to be indicative of a significant difference comparing Maori and non-Maori, non-Pacific children. RRs whose CIs did not overlap were considered indicative of differences comparing RRs across different risk and outcome measures examined for the two groups. While it is tempting to build a statistical model around this data to seek to explain as much of the variance as possible, the measures used in this investigation do not fully capture the concept of either “risk” or “bias.” As such, it is possible that modeling would produce spurious results, leading to inappropriate conclusions being drawn. Indeed, the aim of this investigation was to follow and expand on the Drake et al. model, furthering our understanding of the high representation of Maori children in administratively sourced maltreatment data.

Results

Crude RR Comparisons

The rate of fatal and nonfatal outcomes for Maori and non-Maori, non-Pacific children born between July 1, 2004, and June 30, 2007, are presented in Table 1.

RRs were in excess of 1.0 for all measures considered (Figure 3a). RRs for welfare and justice measures of risk were in excess of 3.0. RRs for birth outcomes were lower than 1.50, while the infant mortality RR was significantly higher for deaths in the postneonatal period (3.38, 95% CI [2.72, 4.21]) than in the neonatal period (1.51, 95% CI [1.26, 1.83]).

Figure 3.

(a) Rate ratios (RRs) classified according to Drake et al.’s system and (b) RRs classified according to proximity of outcome to birth. * Upper 95%CI: 31.29.

RRs for mortality outcomes with greater potential for classification error ranged from 3.67 (95% CI [2.45, 5.49]) for unintentional injury (excluding motor vehicle traffic) mortality before age 3 to 8.93 (95% CI [2.54, 31.29]) for early childhood violent death mortality. RRs for CYF CA/N outcomes were generally over 3.0. For injury hospitalization, they ranged from 2.03 (95% CI [1.44, 2.85]) for fracture of long bones in infancy to 4.70 (95% CI [3.55, 6.21]) for hospitalization for intentional injury by age 3.

Figure 3b presents an alternative view of the outcome measures. In this figure, we have distinguished measures by proximity to birth (in time) rather than the potential for classification error. While it appears that with increasing length of time from birth there are increasing RRs, further exploration of CYF notifications and substantiations revealed no significant difference between RRs at 1 year and RRs at 3 years (data not shown).

Stratified RR Comparisons

RRs stratified by duration on benefit and maternal age are presented in Table 2. After stratification, RRs for all measures of risk and outcomes were attenuated. The majority of RRs for birth outcomes and neonatal infant mortality were not significantly different from 1.0, with the exception of birth at <32 weeks and neonatal infant mortality for mothers aged under 25 years who had spent no time on the benefit.

Table 2.

RRs Maori: Non-Maori, Non-Pacific Children by Age of Mother and Mother’s Time Supported by Welfare Benefits in Previous 5 Years.

	Mother aged 25+						Mother aged under 25
	≥4 of the last 5 years			0 of the last 5 years			≥4 of the last 5 years			0 of the last 5 years
	Maori, n = 4,583			Maori, n = 11,052			Maori, n = 5,129			Maori, n = 2,462
	non-Maori non-Pacific, n = 2,249			non-Maori non-Pacific, n = 72,000			non-Maori non-Pacific, n = 1,501			non-Maori non-Pacific, n = 5,650
		95% CI			95% CI			95% CI			95% CI
	RR	LCI	UCI	RR	LCI	UCI	RR	LCI	UCI	RR	LCI	UCI
Risk
Child supported by welfare benefit in first 3 months of life	1.09	1.06	1.12	2.66	2.36	3.01	1.06	1.03	1.09	2.38	2.10	2.68
Mother with prison sentence in last 5 years	1.53	1.22	1.93	2.71	1.49	4.96	1.81	1.24	2.64	2.68	0.90	7.96
Father with prison sentence in last 5 years	1.90	1.54	2.34	2.85	2.34	3.47	1.32	1.02	1.70	1.59	1.08	2.34
Outcome
Birth weight <2.5 kg	1.0	0.89	1.13	0.93	0.85	1.01	1.07	0.91	1.26	0.95	0.8	1.13
Birth at <32 weeks	0.76	0.59	0.98	1.08	0.90	1.29	1.38	0.96	1.97	1.85	1.32	2.60
Birth at <37 weeks	0.99	0.88	1.10	0.98	0.91	1.05	1.11	0.95	1.30	1.04	0.88	1.22
Infant mortality—neonatal	0.50	0.28	0.88	1.36	0.94	1.98	1.57	0.71	3.50	2.87	1.49	5.52
Infant mortality—postneonatal	1.84	0.95	3.53	1.97	1.20	3.25	1.40	0.71	2.74	1.38	0.33	5.76
Notification or police family violence contact record <3	1.31	1.24	1.39	2.10	1.86	2.37	1.00	0.94	1.06	1.59	1.37	1.86
Investigation or CFA <3	1.32	1.23	1.41	2.01	1.71	2.37	0.93	0.86	1.00	1.40	1.12	1.75
Substantiated finding of maltreatment <3	1.59	1.43	1.77	2.24	1.69	2.97	0.98	0.87	1.10	1.45	1.03	2.03
Substantiated finding of physical abuse <3	1.62	1.05	2.49	1.74	0.80	3.79	0.88	0.56	1.37	0.76	0.21	2.82
Substantiated finding of neglect <3	1.30	1.11	1.52	3.33	2.02	5.50	0.72	0.61	0.84	1.43	0.75	2.73
Substantiated finding of emotional abuse <3	1.84	1.61	2.11	2.26	1.60	3.19	1.21	1.04	1.42	1.80	1.22	2.67
In care of chief executive < 3	1.16	0.96	1.39	1.89	1.12	3.20	0.49	0.40	0.60	1.15	0.54	2.45

Note. CFA = Child and Family Assessment; RR = rate ratio; CI = confidence interval; LCI = lower confidence interval; UCI = upper confidence interval.

In Table 3 we show the rates for outcomes by welfare benefit duration and compare RRs within the ethnic groups for those with a mother who had spent 4 of the 5 years prior to the child’s birth on the benefit compared with those with a mother who had spent no time on the benefit. The table shows that, compared with non-Maori, non-Pacific children, the RRs comparing the longer benefit duration and no benefit duration groups for Maori children are substantially lower. This indicates that, for the following risk factors, there are not the same improvements for mothers with no benefit receipt compared to those who received the benefit: Child supported by benefit in the first 3 months of life and father with a prison sentence in the previous 5 years of life. Similarly, reduced RRs are also observed for the following outcomes: notification or police family violence contact record, investigation or CFA, substantiated finding of maltreatment, substantiated finding of neglect, substantiated finding of emotional abuse, and in care of chief executive.

Table 3.

RRs for Maori and Non-Maori, Non-Pacific Children Comparing High Maternal Benefit Dependence With No Benefit Dependency Prior to Birth.

	4+ of the last 5 years supported by benefit				0 of the last 5 years supported by benefit				RR
	4+ of the last 5 years supported by benefit				0 of the last 5 years supported by benefit				(4+ of the last 5 years: 0 of the last 5 years)
	Maori, n = 16,198		non-Maori non-Pacific, n = 5,703		Maori, n = 13,512		non-Maori non-Pacific, n = 77,645		Maori		non-Maori non-Pacific
	n	Rate (per 1,000)	n	Rate (per 1,000)	n	Rate (per 1,000)	n	Rate (per 1,000)		95% CI		95% CI
Risk
Child supported by welfare benefit in first 3 months of life	11,878	733.30	3,810	668.07	815	60.32	1,314	16.92	12.16	[11.32, 13.07]	39.48	[37.07, 42.06]
Mother with prison sentence in last 5 years	534	32.97	121	21.22	22	1.63	42	0.54	20.25	[13.23, 32.59]	39.22	[27.40, 57.13]
Father with prison sentence in last 5 years	804	49.64	172	30.16	184	13.62	384	4.95	3.64	[3.10, 4.30]	6.10	[5.06, 7.32]
Outcome
Birth weight <2.5 kg	1,510	93.22	524	91.88	744	55.06	4,472	57.60	1.69	[1.55, 1.85]	1.60	[1.45, 1.75]
Birth at <32 weeks	341	21.05	124	21.74	193	14.28	885	11.40	1.47	[1.23, 1.77]	1.91	[1.58, 2.30]
Birth at <37 weeks	1,706	105.32	589	103.28	992	73.42	5,757	74.15	1.43	[1.33, 1.55]	1.39	[1.28, 1.52]
Infant mortality—neonatal	67	4.14	29	5.09	53	3.92	174	2.24	1.05	[0.72, 1.54]	2.27	[1.48, 3.38]
Infant mortality—postneonatal	100	6.17	21	3.68	23	1.70	71	0.91	3.63	[2.28, 5.98]	4.03	[2.35, 6.63]
Notification or police family violence contact record <3	6,410	395.73	1,898	332.81	599	44.33	1,435	18.48	8.93	[8.21, 9.72]	18.01	[16.81, 19.30]
Investigation or CFA <3	4,614	284.85	1,410	247.24	304	22.5	791	10.19	12.66	[11.27, 14.27]	24.27	[22.23, 26.51]
Substantiated findings of maltreatment <3	2,533	156.38	679	119.06	118	8.73	273	3.52	17.91	[14.88, 21.72]	33.86	[29.38, 39.11]
Substantiated findings of physical abuse <3	182	11.24	51	8.94	11	0.81	39	0.50	13.80	[7.53, 28.13]	17.80	[11.50, 27.74]
Substantiated findings of neglect <3	1,091	67.35	377	66.11	38	2.81	69	0.89	23.95	[17.34, 34.03]	74.39	[57.42, 97.61]
Substantiated findings of emotional abuse <3	1,793	110.69	404	70.84	87	6.44	180	2.32	17.19	[13.86, 21.57]	30.56	[25.57, 36.63]
In care of chief executive < 3	650	40.13	282	49.45	28	2.07	82	1.06	19.36	[13.27, 29.37]	46.82	[36.49, 60.63]

Note. CFA = Child and Family Assessment; RR = rate ratio; CI = confidence interval.

Discussion

The investigation presented in this article followed and extended Drake et al.’s (2011) investigation by using administrative data linked at the individual level for a cohort of children. By stratifying our results by maternal age and duration of benefit receipt (as a proxy for poverty), we sought to further understand the role of these factors in explaining differences in rates of adverse child outcomes for Maori versus non-Maori, non-Pacific children. We also stratified our results according to proximity to when the child was born. We were unable to identify a consistent effect for more distal outcomes.

Our results suggest a more nuanced interpretation than competing risk and bias models as suggested by Drake and colleagues. The Drake model primarily relies on the assumption that infant mortality and birth outcome RRs serve as a useful proxy for actual CA/N RRs, based on the suggestion that infant mortality and birth outcomes could be viewed as sensitive to the same risk factors as CA/N but not sensitive to bias. Drake asserted that RRs for administratively recorded CA/N that were substantially higher than RRs for infant mortality and poor birth outcomes would strongly support the bias model. In contrast, RRs for administratively recorded CA/N rates that were consistent with those for these negative child outcomes would support the risk model (Drake et al., 2011).

Where RRs for birth outcomes were comparable to RRs for measures of risk for Black children in the Drake et al. study, the difference between RRs for birth outcomes and risk for Maori children in this investigation was comparable to that for the Hispanic children included in Drake’s investigation (high RRs for risk, RRs for birth outcomes close to 1). Among other factors, Drake et al. attributed the results for Hispanic children to strong cultural support for maternity within Hispanic families (Drake et al., 2011). While for Hispanic children RRs were also comparatively low for administratively recorded CA/N, this is not apparent for the Maori children included in the current investigation. We suggest that low RRs for birth outcomes and neonatal mortality for Maori children reflect the dual importance of adequate maternal service provision as well as the Maori world view of the importance of pregnant Maori women (Makowharemahihi et al., 2014).

System-Related Factors and Services

Potential moderating factors for infant mortality and birth outcomes and for children’s well-being post-birth are wide ranging. In addition to “extended family supports and/or cultural emphasis on the maternal role” (Drake et al., 2011, p. 472) highlighted in the Drake et al. study, they include system-related factors and access to services and the quality of the services able to be accessed. The NZ Human Rights Commission uses the “interrelated and essential elements” outlined in the United Nations International Covenant on Economic, Social and Cultural Rights, General Comment 14 (United Nations, 2000) to assess the promotion and protection of the right to health. These four elements might be adapted, as below, to assess services:

Availability: a sufficient number of functioning services, facilities, and programs.

Accessibility: available without discrimination, physically accessible and affordable, and people should be aware of their existence.

Acceptability: must respect ethics and confidentiality and be culturally appropriate.

Quality: evidence informed and of good quality (adapted from the Human Rights Commission, 2004).

The provision of services is therefore multifaceted and cannot be taken for granted merely by their operation within a community.

For the cohort studied, the moderating impact of system-related factors and services appears to have differed between the neonatal and postneonatal periods (Figure 3a and b). In the health domain, free, universal, maternity services, while not without deficits (Paterson et al., 2012; Perinatal and Maternal Mortality Review Committee, 2013), have narrowed neonatal mortality RRs (Child and Youth Mortality Review Committee, 2004; Pool, Boddington, Cheung, Didham, & Amey, 2009). A contributor to high RRs for postneonatal mortality for birth cohorts including the children studied in this investigation was inadequate tailoring of SUDI prevention messages and their delivery for Maori (Child and Youth Mortality Review Committee, 2009; Tipene-Leach et al., 2010). More recent cohorts have been offered new SUDI prevention messages, and resources that allow culturally important shared sleeping behaviors to continue in a safer way (Best Practice Journal, 2013). These changes have been accompanied by a significant reduction in SUDI rates for Maori babies between 2002 and 2012 (NZ Mortality Review Group, 2013; Whakawhetu, 2014).

In the maltreatment domain, there is broad consensus that preventive services and systems need to be strengthened (Health Select Committee, 2013; Maori Reference Group for the Taskforce for Action on Violence within Families, 2013; NZ Government, 2012), and that evidence on program acceptability, accessibility and effectiveness, both overall, and for Maori is required (Social Policy Evaluation and Research Unit, 2014). Blackstock and Trocmé (2005) note that the removal of indigenous children from Canadian reservations occurs within the context of very little, if any, prevention or family support, and that child removal has become the only response to poverty-induced neglect in these impoverished communities. Walker (2004) attributes an increase in Maori maltreatment to the devolution of Maori Affairs programs to mainstream government departments and community agencies at the end of the 1980s. The loss of Maori Affairs programs meant the loss of the Maori welfare officers who had been pivotal with Maori urban communities, working alongside them for the health and well-being of whānau.

Unfortunately, there were gaps in social welfare provision from these new providers and the state agency now called Child, Youth and Family. Child abuse in dysfunctional whānau that wasn’t detected in time sometimes culminated in homicide … At the heart of this problem is anomie arising out of social dislocation and the loss of whānau and hapū support structures. (Walker, 2004, p. 289)

Osterling and colleagues suggest that the quality of child welfare services for indigenous children is disrupted by system-related factors. A lack of organizational leadership, high staff caseloads, under-resourcing and crisis-driven, stressful work environments negatively affect staff satisfaction and service quality (Osterling, D’Andrade, & Austin, 2010, p. 26). These factors, in turn, contribute to overrepresentation. Similar themes are outlined in the Decision Making Ecology (Baumann, Dalgleish, Fluke, & Kern, 2011) where a range of case, external, organizational, and individual factors combine to influence the child welfare referrals and case worker decisions that can contribute to disproportionality in outcomes (Fallon et al., 2013). Unfortunately, with the data available for this investigation, these themes could not be explored.

When we stratified presentation of the results by key risk factors for early childhood adversity (young maternal age and a proxy for poverty persistence), RRs for birth outcomes and neonatal infant mortality were no longer significantly different from 1.0, while RRs for welfare and justice risk factors and administratively sourced maltreatment outcomes were attenuated but continued to be elevated above 1.0 (with the exception of mothers aged 25 years or under who had spent at least 4 of the last 5 years receiving a benefit). The similarity of these elevated RRs may suggest common drivers in the welfare, justice, and maltreatment domains. Putnam-Hornstein, Needell, et al. (2013) have investigated the effect of adjusting for a range of demographic, socioeconomic and health factors on referral, substantiation, and foster care placement rates among Californian children covered by public health insurance at birth as a marker for poverty. Adjusting for these factors resulted in RRs significantly below 1.0 for Black children compared to the White reference group and accentuated the substantially lower RRs for Latino children born to foreign-born mothers (Putnam-Hornstein, Needell, King, & Johnson-Motoyama, 2013).

Our findings of attenuation after stratification by a marker of poverty persistence generally align with those of Putnam-Hornstein et al. who suggested that focusing on “poverty and its correlates” when attempting to address the overrepresentation of indigenous children in administratively recorded maltreatment may effect more change than focusing on the attitudes of those who come in contact with children and their families (Putnam-Hornstein, Needell, King, et al., 2013). Indeed, in this investigation, the results for mothers aged under 25 years who had spent more than four of the previous five years on the benefit generally showed no significant difference in the RRs for administratively recorded maltreatment despite there being RRs in excess of 1.0 for welfare and justice risk factors. It is possible that, for these young women, the financial certainty that came with receipt of the benefit diminished the relative likelihood of a maltreatment event that came to the notice of CYF.

An Alternative Model

An alternative to the risk and bias models is a more complex schema that reflects a colonization theory of indigenous CA/N (Daoud, Smylie, Urquia, Allan, & O’Campo, 2013), highlights the moderating role of access to services (Blackstock & Trocmé, 2005) as well as protective cultural practices (Jenkins & Harte, 2011), and broadens the conception of the possible form and role that bias might play to include systemic factors (Dettlaff, 2013; Figure 4). The model acknowledges the historical and contemporary disenfranchisement and marginalization of Maori as a root cause of disparities in poverty and inadequate access to services (Cram, 2012). Reparations made under Treaty settlements and moves toward decolonization are part of redressing upstream, historically generated risk factors. Solutions to child poverty are called for to alleviate the burden placed upon young people growing up in financially deprived circumstances (Children’s Commissioner’s Expert Advisory Group on Solutions to Child Poverty, 2012; Health Select Committee, 2013). The need for strong leadership and the strengthening of collaborative relationships with communities complements this as means to ensure the cultural responsiveness of services and the child welfare system for Maori whānau. Promotion of traditional positive parenting practices (Jenkins & Harte, 2011) plays a role in preventing exposure to risk factors and in reducing the impact of risk factors on children’s outcomes.

Figure 4.

An alternative model showing possible relationships.

With this model, the relative similarity of RRs across measures of risk and CYF outcomes considered here does not necessarily provide assurance of an absence of bias in the maltreatment domain. One interpretation is that real differences in risks to children play out in real differences in harm. But another is that measures of risk themselves reflect historical inequities, system-related factors, and service deficits that see more Maori have contact with welfare benefit and correction systems and overstate the real risks faced by Maori children. Given that the inequalities that were identified in this analysis were least marked for birth outcomes and neonatal mortality, it appears that the services and support structures that are in place at the time of birth do not have a lasting effect on the life of the child. These findings highlight the inequitable distribution of risk and outcomes for Maori children and reinforce the need for coherent policy development to address these issues (Commission on Social Determinants of Health, 2008). Effective and meaningful support structures that reduce the risk of child maltreatment include consideration of distal as well as proximal determinants of risk, including the political, social, cultural, and economic structures within which families live (Robson, Cormack, & Cram, 2007).

Strengths and Limitations

Where Drake et al. provide an ecological comparison of data derived from a number of different sources, our study uses administrative data linked at the individual level for a cohort of children. This allows information specific to the parents of the cohort to be included. It also allows a single and consistent definition of the ethnic groups of the children to be applied across the analysis. Against the considerable advantages of a study drawing on linked administrative data (Brownwell & Jutte, 2013; Putnam-Hornstein, Needell, & Rhodes, 2013), there are limitations. Combining data from across social sector administrative systems requires linking on name, date of birth, and other identifying information. A clerical review of a sample of linkages found accuracy to be good overall for the children, but found lower accuracy for adults (see the Online Appendix). Results that include estimates of parental history should therefore be treated with some caution. Administrative data captures only information collected or generated in the process of administering services. Direct measures of poverty or socioeconomic status (income, education level, and employment) were not included in the data set used for this investigation. Instead, we used the duration of maternal welfare benefit receipt prior to the birth as a proxy for poverty persistence as parental benefit receipt has a strong association with child poverty (Perry, 2013) and measured material deprivation (Perry, 2009). The results of our analysis, however, indicate that it is an imperfect proxy for vulnerability to hardship post-birth. Some of the mothers included in this investigation progressed into hardship after the birth. The increased RR for Maori children being supported by a benefit in the first 3 months of life suggests that this transition occurred at a higher rate for Maori children than for non-Maori, non-Pacific children whose mothers had no pre-birth benefit receipt (Table 2). Whether it is correlates of this nature or poverty itself that intensifies maltreatment risk for children remains unsettled empirically. Of note is the fact that, included in the follow-up period for children born in 2005 and later, was the global financial crisis. It is possible that this higher rate of transition captures increased vulnerability to job losses. Between June 2008 and December 2009, the Maori unemployment rate rose from 6.3% to 14.8%. At the same time, the unemployment rate for European New Zealanders rose from 2.8% to 4.6% (Mills, 2010).

We examined one birth cohort and tracked outcomes to age 3. Our findings may not be generalizable to other birth cohorts or to older age ranges. Given the confines of the data linkage, our analysis excludes the experience of children who arrive as migrants and fails to capture outcomes for children who leave the country through outward migration. To the extent that outward migration varies systematically by ethnic group, biases in the results cannot be ruled out. Finally, we presented our results stratified by benefit duration and maternal age. A comprehensive regression analysis of the linked data set, adjusting for a variety of confounders (e.g., administratively sourced proxies for parental poverty, criminality, maltreatment exposure, and partnership status), may further explain variations in measures of risk and outcomes. However, even with such an approach, interpretation of results can be uncertain if the available controls partly embody the effects of historical inequities or system biases, and overstate the real risks for minority children (Putnam-Hornstein, Needell, King, et al., 2013; Trocmé et al., 2004).

Conclusion

In this study of Maori children’s overrepresentation in the Aotearoa NZ child welfare system, variation in RRs within the set of infant mortality and birth outcomes considered as possible proxies for actual CA/N RRs leave open a range of interpretations. Our findings indicate that a more nuanced interpretation of the overrepresentation of indigenous children in administratively recorded maltreatment statistics is required. Rather than considering risk and bias as competing explanations, we suggest an acknowledgment of the impact of colonization and the existence of systemic bias generating increased risk as key drivers. As linked administrative data are increasingly used for research and evaluation, and considered for use in supporting decision making, a deeper understanding of the drivers of administratively recorded CA/N is required in order to effectively address the needs of indigenous populations. Systemic factors and service gaps that reach beyond the child welfare system and inequities that have deep historic and structural roots all potentially contribute to administratively recorded CA/N. We recommend a multi-strand preventive approach that addresses a wide range of potential drivers, including material hardship and access to services, and builds the evidence base about what works for Maori.

Footnotes

Acknowledgments

The authors would like to thank Dr Beverley Lawton (Director and Senior Research Fellow, Women’s Health Research Centre, University of Otago) and Dr Paula King (Principal Advisor Population Maori Health at the MoH) who reviewed earlier drafts and made helpful suggestions for improvement. Ministry of Social Development officials who participated in a discussion of draft results also provided helpful comments and suggestions. The assistance of officials at the Department of Internal Affairs, MoH, Office of the Privacy Commissioner, and Corrections Department is gratefully acknowledged.

Authors’ Note

Any errors or omissions in this report are the responsibility of the authors. The views expressed may not reflect those held by the reviewers, do not accord with the views of all participants in discussion of results, and do not represent the position of the Ministry of Social Development.

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 Ministry of Social Development provided funding for the study. Wilson and Ota undertook the study as employees of the Ministry of Social Development. Cram was contracted by the Ministry of Social Development to contribute to the study. The Ministry of Social Development reimbursed Cram and Gulliver for travel and meeting attendance related to the study. No other financial disclosures were reported by the authors of this article.

Supplemental Material

The online appendices are available at .

References

Baumann

D. J.

Dalgleish

Fluke

Kern

H. D.

(2011). The decision making ecology. Washington, DC: American Humane Association.

Best Practice Journal. (2013). Upfront—Sudden unexpected death in infancy: Where are we now? Best Practice Journal, 56, 3–7.

Blackstock

Trocmé

(2005). Community-based child welfare for Aboriginal children: Supporting resilience through structural change. Social Policy Journal of New Zealand, 24, 12–33.

Brownwell

M. D.

Jutte

D. P.

(2013). Administrative data linkage as a tool for child maltreatment research. Child Abuse and Neglect, 37, 120–124.

Child and Youth Mortality Review Committee. (2004). First report to the Minister of Health. 1 January 2002 to 30 June 2003. Wellington, New Zealand: Author.

Child and Youth Mortality Review Committee. (2009). Fifth report to the Minister of Health: Reporting mortality 2002–2008. Wellington, New Zealand: Author.

Child Youth and Family. (2006). Children at increased risk of death from maltreatment and strategies for prevention. Wellington, New Zealand: Ministry of Social Development.

Child Youth and Family (Producer). (2014, 3 31). Key Statistics and information for media. Retrieved from http://www.cyf.govt.nz/about-us/who-we-are-what-we-do/information-for-media-backup.html#FrequentlyAskedQuestions10

Children’s Commissioner’s Expert Advisory Group on Solutions to Child Poverty. (2012). Solutions to child poverty in New Zealand: Evidence for action. Wellington, New Zealand: Office of the Children’s Commissioner.

10.

Commission on Social Determinants of Health. (2008). Final report. Executive summary closing the gap in a generation: Health equity through action on the social determinants of health. Geneva, Switzerland: World Health Organisation.

11.

Cooper

Wharewera-Mika

(2009). Maori child maltreatment: A literature review. Auckland, New Zealand: Te Kahui Mana Ririki.

12.

Cram

(2012). Safety of subsequent children—Maori children and Whanau. A review of selected literature. Wellington, New Zealand: Families Commission.

13.

Daoud

Smylie

Urquia

Allan

O’Campo

(2013). The contribution of socio-economic position to the excesses of violence and intimate partner violence among Aboriginal versus non-Aboriginal women in Canada. Canadian Journal of Public Health, 104, e278–e283.

14.

Dettlaff

A. J.

(2013). The evolving understanding of disproportionality and disparities in child welfare. In Korbin

J. E.

Krugman

R. D.

(Eds.), Handbook of child maltreatment (pp. 149–168). New York, NY: Springer.

15.

Drake

Jolley

J. M.

Lanier

Fluke

Barth

R. P.

Johnson-Reid

(2011). Racial bias in child protection? A comparison of competing explanations using national data. Pediatrics, 127, 471–478.

16.

Fallon

Chabot

Fluke

Blackstock

MacLaurin

Tonmyr

(2013). Placement decisions and disparities among Aboriginal children: Further analysis of the Canadian incidence study of reported abuse and neglect. Part A: Comparisons of the 1988 and 2003 surveys. Child Abuse & Neglect, 37, 47–60.

17.

Fanslow

Robinson

Crengle

Perese

(2010). Juxtaposing beliefs and reality: prevalence rates of intimate partner violence and attitudes to violence and gender roles reported by New Zealand women. Violence Against Women, 16, 812–831.

18.

First Nations Component of the CIS (FNCIS-2008). Fallon

Trocmé

MacLaurin

Sinha

Hélie

Turcotte

Daciuk

Chabot

Black

(2010). Canadian incidence study of reported child abuse and neglect - 2008: Study methods. McGill University. Retrieved from http://cwrp.ca/sites/default/files/publications/en/CIS_2008_Methods_March_2013_0.pdf

19.

Fluke

Shusterman

Hollinshead

TYuan

Y. T.

(2008). Longitudinal analysis of repeated child abuse reporting and victimisation: multistate analysis of associated factors. Child Maltreatment, 13, 76–88.

20.

Gilbert

Fluke

O’Donnell

Gonzalez-Izquierdo

Brownell

Gulliver

P.,

… Sidebotham

(2011). Child maltreatment: variation in trends and policies in six developed countries. Lancet, 378. doi:10.1016/ S0140-6736(11)61087-8

21.

Grier

(2005). Aboriginal children in limbo: A comment on Re R.T. Saskatchewan Law Review, 68, 435–453.

22.

Harcourt

B. E.

(2006). Against prediction: Profiling, policing and punishing in an actuarial age. Chicago, IL: University of Chicago.

23.

Health Select Committee. (2013). Health select committee inquiry into improving child health outcomes and preventing child abuse with a focus from pre-conception until 3 years of age. Wellington: New Zealand Parliament.

24.

Human Rights Commission. (2015). Your right to health. Retrieved from http://www.hrc.co.nz/your-rights/social-equality/your-rights/. Accessed 30 March 2015.

25.

Imlach Gunasekara

Carter

(2012). Dynamics of income in children in 2002-2009. A descriptive analysis of the Survey of Family, Income and Employment (SoFIE) public health monograph. Wellington, New Zealand: Department of Public Health, University of Otago.

26.

Jenkins

Harte

H. M.

(2011). Traditional Maori parenting: An historical review of literature of traditional Maori childrearing practices pre-European times. Auckland, New Zealand: Te Kahui Mana Ririki.

27.

Joelving

(2011, 2 10). No racial bias at child protective services: Study, Reuters. Retrieved from http://www.reuters.com/article/2011/02/11/us-no-racial-bias-idUSTRE71A09520110211

28.

Makowharemahihi

Lawton

B. A.

Cram

Ngata

Brown

Robson

(2014). Initiation of maternity care for young Maori women under 20 years of age. New Zealand Medical Journal, 127, 52–61.

29.

Maori Reference Group for the Taskforce for Action on Violence within Families. (2013). Programme of action for addressing family violence: 2013–2018. Wellington, New Zealand: Ministry of Social Development.

30.

Mills

(2010). Health, employment and recession: The impact of the global crisis on health inequities in New Zealand. Policy Quarterly, 6, 53–59.

31.

Ministerial Advisory Committee on a Maori Perspective for the Development of Social Welfare. (1988). Tuao-Te-Ata-Tu (Daybreak). Wellington, New Zealand: Ministerial Advisory Committee on a Maori Perspective for the Development of Social Welfare.

32.

Ministry of Health (Producer). (2009, 1 27, 2011). National minimum dataset (Hospital Events) (NMDS). Retrieved from http://www.moh.govt.nz/moh.nsf/indexmh/dataandstatistics-collections-nmds#Collection@@

33.

Ministry of Health (Producer). (2011, 3 31, 2014). National Maternity Collection. National Collections and Surveys. Retrieved from http://www.health.govt.nz/nz-health-statistics/national-collections-and-surveys/collections/national-maternity-collection

34.

Ministry of Health (Producer). (2013, 3 31, 2014). Mortality Collection. National Collections and Surveys. Retrieved from http://www.health.govt.nz/nz-health-statistics/national-collections-and-surveys/collections/mortality-collection

35.

Ministry of Social Development (Producer). (2014, 3 31, 2014). Benefit fact sheets. Statistics. Retrieved from http://www.msd.govt.nz/about-msd-and-our-work/publications-resources/statistics/benefit/index.html

36.

New Zealand Government. (2012). White paper for vulnerable children. Wellington: New Zealand Government.

37.

New Zealand Mortality Review Group. (2013). 9th data report 2008–2012. Wellington, New Zealand: Child and Youth Mortality Review Committee.

38.

Osterling

K. L.

D’Andrade

Austin

M. J.

(2010). Understanding and addressing racial/ethnic disproportionality in the front end of the child welfare system. In Austin

M. J.

(Ed.), Evidence for child welfare practice (pp. 9–30). New York, NY: Routledge.

39.

Paterson

Candy

Lilo

McCowan

Naden

O’Brien

(2012). External review of maternity care in the Counties Manukau District. Auckland, New Zealand: Counties Manukau District Health Board.

40.

Perinatal and Maternal Mortality Review Committee. (2013). Seventh Annual Report of the Perinatal and Maternal Mortality Review Committee: Reporting mortality 2011. Wellington, New Zealand: Health Quality and Safety Commission.

41.

Perry

(2009). Non-income measures of material wellbeing and hardship: First results from the 2008 New Zealand Living Standards Survey, with international comparisons. Wellington, New Zealand: Ministry of Social Development.

42.

Perry

(2013). Household incomes in New Zealand: Trends in indicators of inequality and hardship 1982–2012. Wellington, New Zealand: Ministry of Social Development.

43.

Pool

Boddington

Cheung

Didham

Amey

(2009). Differential trends in the compression of mortality: Assessing the antecedents to current gaps in health expectancy in New Zealand. New Zealand Population Review, 35, 23–53.

44.

Putnam-Hornstein

Needell

King

Johnson-Motoyama

(2013). Racial and ethnic disparities: A population-based examination of risk factors for involvement with child protection services. Child Abuse and Neglect, 37, 33–46.

45.

Putnam-Hornstein

Needell

Rhodes

A. E.

(2013). Understanding risk and protective factors for child maltreatment: The value of integrated, population-based data. Child Abuse & Neglect, 37, 116–119.

46.

Putnam-Hornstein

Wood

J. N.

Fluke

Yoshioka-Maxwell

Berger

R. P.

(2013). Preventing severe and fatal child-maltreatment: making the case for the expanded use and integration of data. Child Welfare, 92, 59–75.

47.

Rittel

H. W.

Webber

M. M.

(1973). Dilemmas in a general theory of planning. Policy Sciences, 4, 155–169.

48.

Robson

Cormack

Cram

(2007). 3. Social and economic indicators. In Robson

Harris

(Eds.), Hauora: Maori standards of health IV: A study of the years 2000–2005 (pp. 21–32). Wellington, New Zealand: Te Ropu Rangahau Hauora a Eru Pomare.

49.

Sinha

Trocmé

Fallon

MacLaurin

Fast

Thoms Prokop

S.,

… Richard

(2011). Kiskisik Awasisak: Remember the children. Understanding the over-representation of first nations children in the child welfare system. Ontario, Canada: Assembly of First Nations.

50.

Social Policy Evaluation and Research Unit. (2014). Effective parenting programmes: A review of parenting programmes for parents of vulnerable children. Wellington, New Zealand: Families Commission.

51.

Statistics New Zealand. (2010). Late birth registrations Vitals Articles. Wellington: Statistics New Zealand.

52.

Statistics New Zealand (Producer). (2013, 5 7, 2014). 2013 Censust QuickStats about Maori. Retrieved from http://www.stats.govt.nz/Census/2013-census/profile-and-summary-reports/quickstats-about-maori-english.aspx

53.

Statistics New Zealand (Producer). (2014, 4 2, 2014). Criminal conviction and sentencing statistics. Retrieved from http://www.stats.govt.nz/tools_and_services/nzdotstat/justice-statistics-fiscal-year.aspx

54.

Tipene-Leach

Hutchison

Tangiora

Rea

White

Stewart

Mitchell

(2010). SIDS-related knowledge and infant care practices among Maori mothers. New Zealand Medical Journal, 123, 88–96.

55.

Trocmé

Knoke

Blackstock

(2004). Pathways to the overrepresentation of Aboriginal children in Canada’s child welfare system. Social Science Review, 78, 577–600.

56.

United Nations. (1966). International Covenant on Economic, Social and Cultural Rights. General Assembly resolution 2200A (XXI) 16 December 1966.

57.

UNICEF. (2013). Innocenti Digest No 11: Ensuring the rights of Indigenous children. Florence, Italy: Innocenti Research Centre and UNICEF.

58.

Vittinghoff

Glidden

D. V.

Shiboski

S. C.

McCulloch

C. E.

(2004). Basic statistical methods. In Dietz

Gail

Krickeberg

Tsiatis

(Eds.), Regression methods in biostatistics: Linear, logistic, survival and repeated measures models (pp. 29–67). New York, NY: Springer.

59.

Walker

(2004). Ka Whawhai Tonu Matou: Struggle without end (Rev. ed.). Auckland, New Zealand: Penguin.

60.

Whakawhetu (Producer). (2014, 1 16, 2014). Baby and young people deaths down, but complacency unwise. Retrieved from http://www.whakawhetu.co.nz/news/child-and-youth-mortality-review-committee

61.

Woolley

S. L.

(2009). The rights of indigenous children around the world—Still far from a reality. Archives of Diseases in Childhood, 94, 397–400.