Article body

Introduction

Parents involved with the child protection system may be asked to complete a Parenting Capacity Assessment (PCA), to assist child protection workers and courts with case decisions. These assessments carry substantial weight in child protection courts across Canada (Choate & Hudson, 2014; Lennings, 2002). PCAs have three core questions:

  1. Can this parent be ‘good enough’ for this child based upon the needs of the child?

  2. If not, what can be done to enhance the skills of the parent to become good enough? And,

  3. If that cannot be done, can this parent be involved in the life of this child? (Choate & Engstrom, 2014).

A few structured models exist for PCAs (Budd, Clark & Connell, 2011; Choate, 2009; Pezzot-Pearce & Pearce, 2004; Budd, 2005, 2001; Steinhauer 1993/4; 1991). They suggest a number of steps that would be typically taken which can be summarized as:

  • Obtain a good referral that clearly outlines the questions to be answered;

  • Review the background data, including material from child welfare that outlines the involvement with the family;

  • Conduct clinical interviews that review the personal and family history in detail;

  • Complete psychometric testing;

  • Observe the parent with the child;

  • Interview the child as appropriate;

  • Conduct collateral interviews; and

  • Prepare a written report with conclusions and recommendations.

This approach is widely used in developed countries including Canada. The models have been constructed using Euro-North America understandings of parenting focusing on the nuclear family. The typical PCA does not consider the larger family system as a parenting system. Muir and Bohr (2014) show that extended family has a central place within traditional Aboriginal parenting, as does the community.

Stairs and Bernhard (2002) note, “Euro-North American views of child development have far too long been privileged over Aboriginal visions and values” (p.310). The assessment literature has limited data on Aboriginal parenting, although there is parenting education and support literature which offers insight into elements of traditional Aboriginal parenting (Parenting Path, 2012; Fearn, 2006; Fox, n.d.; Kawkeka, 2004; NICWA, 2001; Bruyere, 1983). However, PCAs should reflect the cultural reality of the family being assessed (Choate, 2013).

The existing approach to PCAs has not been validated within minority populations. This approach has received criticism for this in the United States (Lee, Gopalan & Harrington, 2014), as well as Australia with respect to the Aboriginal population (Drew, Adams & Walker, 2010). There has also been some criticism in Canada (Mushquash & Bova, 2007). Little attention has been paid to the validity of the present PCA practices with the Aboriginal peoples of Canada. There is a strong argument to be made that the present approach, which is not culturally sensitive, repeats the power dynamics that have been part of the history of Aboriginal peoples interacting with the dominant culture and its regulatory institutions (Denison, Varcoe & Browne, 2013).

The use of psychometrics is an integral part of PCAs (Budd, Clark & Connell, 2011; Pezzor-Pearce & Pearce, 2004). Typically, a parent will complete a personality measure and others that are focused on parenting. They may also be asked to complete questionnaires related to specific issues such as addictions. The test battery might include the Minnesota Multiphasic Personality Inventory, version 2 (MMPI-2) (Butcher, Graham, Ben-Porath, Tellegen & Dahlstrom, 2001), the Personality Assessment Inventory (PAI) (Morey, 1991), Child Abuse Potential Inventory (CAPI) (Milner, 1986), Adult-Adolescent Parenting Inventory (AAPI) (Bavolek & Keene, 2001) Parenting Stress Index (PSI), v. 4 (Abidin, 2012). While there may be other instruments used, these are the most common (Budd, Clark & Connell, 2011; Pezzot-Pearce & Pearce, 2004).

Developing assessment measures applicable across cultures is challenging. DeDe Klerk (2008) identifies a multitude of issues that must be addressed including construct, method and item biases. For example, if a test is to be valid across cultures, it must be developed in a way that what it purports to measure in a population is actually measured. Thus, the items would also need to be presented in a way that is linguistically and contextually meaningful to the person being assessed.

A Canadian study by Catternich, Gibson and Cave (2001) looked at the assessment of mental capacity in Canadian Aboriginal seniors. Their review aids the present work with their conclusion that assessment overall, including psychometrics, needs to reflect the linguistic conventions as well as the social context of the traditional Aboriginal cultures. They add that there needs to be sensitivity to the dynamics that influence social interaction with Aboriginal peoples (p. 1477).

The goal of this review is to consider whether there has been sufficient inclusion of Aboriginal peoples to conclude that the use of common assessment measures should be used in PCAs with these parents.

Method

It is important to consider the impact of the inclusion or absence of Aboriginal peoples in test creation and test validation. Item bias can exist if the items are not relevant and meaningful to the population being assessed (de Klerk, 2008). A review was conducted of the test manuals for the MMPI-2 (Butcher et al., 2001), PAI (Morey, 1996,1991), CAPI (Milner, 1986), AAPI (Bavolek & Keene, 2001) and PSI-4 (Abidin, 2001). Each was considered for the degree of Aboriginal population inclusion in the norming. A literature search, using Academic Search Complete, EBSCO and PsychInfo, was also conducted looking for studies identifying norming efforts that specifically targeted Aboriginal populations.

Results

The literature review identified a number of efforts to provide cross-cultural validity to various measures, particularly the MMPI-2, PSI, PAI and CAPI. The literature review did not identify studies that focused on extending the norming of these measures to Aboriginal peoples, particularly in Canada.

Some efforts were made to examine the specific utility of the MMPI with a few American Aboriginal groups (Robin, Greene, Albaugh, Caldwell & Goldmanm 2003; Greene, Robin, Albaugh, Caldwell & Goldman, 2003).

Throughout this section we examine which populations were included in the norming. This identified a number of instances where Aboriginal populations were marginally included or not at all. We show the demographic breakdown identified by the test developers so the reader understands the populations that were included.

The Personality Assessment Inventory (PAI) and the Minnesota Multiphasic Personality Inventory (MMPI-2) are commonly used personality tests. They are both widely accepted by the courts (Mullen & Edens, 2008).

PAI

Morey (1991) reports that the sampling for the construction of the PAI was based upon a United States census-matched standardized sample subset of 1000 people from an original sample of 1462 people from 12 American states. The subsample was selected on the basis of gender, race and age from the 1995 census (Morey, 1991, p.47). It consisted of 41.1% Caucasian males, 5.4% Black males, 1.5% other males, 44% Caucasian females, 6.3% Black females, and 1.7% other females.

There were two other samples reported, clinical and college students. The clinical sample consisted of 1265 people from 69 sites. It consisted of 78.8% Caucasian, 12.6% Black and 8.6% other, although no specific breakdown is given of other. The college sample consisted of 1051 students. The samples were drawn from seven universities across the US; 92.5% Caucasian, 2.8% Black and 4.7% other, again with no breakdown of the latter.

The developers of the PAI do acknowledge that psychometrics often bias on variables of gender, culture or other demographics. They took steps to address some of these issues, although there is no indication that Aboriginal peoples were part of that effort

MMPI-2

The manual for the MMPI-2 (Butcher et al., 2001) offers more specific information on the inclusion of Native Americans (the term used in the manual). There appears to have been a targeted sample from a federal reservation in the Tacoma, Washington area. The overall sample consisted of 1138 males (0.5% Asian; 11.1% Black; 3.1 % Hispanic; 3.3% Native Americans; 82% Caucasian) and 1462 females (0.9% Asian; 12.9% Black; 2.6% Hispanic; 2.7% Native American; 81% Caucasian).

The MMPI manual states:

Although the proportions are quite comparable for blacks and whites, Hispanics and Asian-American subgroups are underrepresented in the normative sample. Native Americans are somewhat overrepresented in the normative sample

Butcher et al., 2001, p.3

As noted above, a few studies have been done in an attempt to determine the applicability of the MMPI-2 to Aboriginal populations. Robin et al., (2003) and Greene et al., (2003) demonstrated that some efforts could be made to find applicability, although they went to extensive efforts to build relationships with the test subjects They found this was essential to assist their norming efforts. It also illustrates the steps that need to be taken in working cross culturally. They demonstrated sensitivity to the economic and social hardship, trauma and violence that this population has endured (Robin et al., 2003, p. 356). While they tended towards supporting the use of the MMPI-2, their results still raised caution about the context of interpretation stating, “…the heterogeneity of American Indian tribal groups suggests caution in generalizing results from this community to other tribal groups” (Robin et al., 2003, p.357). Such efforts support the conclusion of de Klerk (2008) that successful inclusion of subpopulations in a norming sample are hard to achieve.

A more recent review of the MMPI-2 by Hill, Pace and Robbins (2010) saw this instrument as having contextual concerns. Their qualitative study looked at how question items were understood. It built on earlier work by Pace et al. (2006) that concluded that the MMPI-2 “is reductive in so far as it screens out behaviors and perspectives its questions cannot absorb” (p.330). The MMPI-2 was thought, from this research, to assume norms of behaviors that may not be meaningful within the traditions and historical experiences of the American Indian peoples (p.330).

Hill et al., (2010) concluded from their research:

From an Indigenous perspective, the MMPI-2 explicitly represents Western power and domination as an instrument that denies Indigenous peoples the right to psychological self-determination. Based upon the results of the current study, it is not difficult to conclude that the MMPI-2 is not an instrument that legitimates or even acknowledges Indigenous knowledges, worldviews, and experiences, but rather an instrument that legitimates and privileges hegemonic Western standards, norms, values, epistemology, and ontology

p. 24

Child Abuse Potential Inventory (CAPI)

The CAPI was validated in three studies. The data indicates that Aboriginal Americans are only marginally represented in the norming populations.

The first study was composed of 38 participants from social services departments in Northern Carolina (Milner, 1986). It was a homogeneous sample matched for comparable variables of location of residence, gender, age, ethnic background, education, marital status, number of children, and gender of children.

A second validity study reported in the manual (Milner 1986) involved 130 people. This sample was drawn from Northern Carolina (social service population) and Tulsa, Oklahoma (At-Risk Parent-Child Program). The sample was matched on variables of location of residence, gender, age, ethnic background, education, marital status, number of children, and gender of children. The sample was 83% Caucasian.

Milner (1986) reports a third sample, which started with 219 people from At-Risk Parent-Child Program, Juvenile Court in Tulsa, Oklahoma and the department of social services in North Carolina. There were 97 people excluded. This left 122 along with 110 matched controls. The sample population was 67% Caucasian, 23% Black and 10% consisting of a mix of Hispanic, Aboriginal American and other.

Adult Adolescent Parenting Inventory (AAPI)

Bavolek and Keene (2001) report a sample drawn from 53 agencies in 23 states. Racial data is not provided. The sample consisted of 713 adult parents without parent training; 198 adolescents without formal training; and 87 adolescent mothers. The sample was drawn from a variety of agencies and included child welfare and related agencies. They report additional discriminating validity for black high school students and Mexican-Americans. There is a Spanish language version to be used by people who read or speak Spanish only, or for people who comprehend Spanish better than English. Other research has raised questions about the use of the AAPI with child protection populations in general (Lawson, Alamedqa-Lawson & Byrnes, 2015; Conners, Whitseide-Mansell, Deere, Ledet & Edwards, 2006).

Parenting Stress Index (PSI) – 4th Edition

Abidin (2012) reports that the goal was to match the overall proportion of the United States population for education and ethnicity from 17 states. There were 1056 people in the sample that included 534 mothers (Caucasian 66%; Africa American 14%; Hispanic 14%; other 5%) and 522 fathers (Caucasian 68%; African American 12%; Hispanic 16%; and other 4%). There was an additional sample of 233 Hispanic parents and a subset of gay/lesbian parents from the standardized sample (N=27).

In the PSI-4 manual, and its predecessor for the 3rd edition, there is a section that speaks to the PSI being multicultural. The PSI has been studied in many cultures. This includes Chinese, European, Portuguese, French Canadian, Finnish, and Dutch (Abidin, 2012). The PSI has been published with norms and detailed psychometric data by publishers in seven countries. They have found the statistics to be comparable to what is listed in the PSI-3 and PSI-4 manuals. The PSI-3 is available in 28 languages and the PSI-3-Short Form is available in 12 languages. They assert:

. . . the PSI is a robust measure that maintains its validity with diverse non-English speaking cultures. Its ability to effectively survive translation and retain its usefulness with non-English-speaking populations suggests that it is likely to maintain its validity with a variety of U.S. populations

Abidin, 2012.pp. 5-6

Abidin, Austin and Flens (2013) report that no specific effort has been made to stratify the PSI based upon subsets of populations. These authors note that efforts to create such stratification are complex. Such work might have to look at gender, race, social positioning and geography, for example. The subdivisions, they feel, could be almost endless.

They state that the significant large amount of cross-cultural research that has been done on the PSI would support the notion that the parenting constructs that underpin the tool are robust. Fathers have not been strongly studied suggesting a possible gender weakness (Abidin, Austin & Flens, 2013, p.355). They also report on various studies that have tried to expand the base of the PSI studied populations, including parents with disabled children, behavioral problems, for example. Thus, we see efforts to broaden the cultural base of the PSI, although it appears that Aboriginal peoples have not been included in those efforts.

Discussion

For each psychometric test used in assessment, it is important to look at the demographic composition of each norming group and validation group. This allows the assessor to establish utility of a particular psychometric for a given population.

The results of this review indicate that the Aboriginal population within the norming groups of these assessment measures is small. There is little data reported on the nature of the Aboriginal grouping other than the MMPI-2 which indicates a sampling with an Aboriginal population in the Tacoma, Washington area, although details are not offered as to which specific population was included. Other Aboriginal inclusion in test development is either not reported or done so in a manner that offers little data about the nature of the population.

Mushquash and Bova (2007) note that researchers have ensured reliability and validity within the dominant populations while the instruments get used with cultural groups for which there is not proper norming or psychometric research. To change this, researchers need to develop constructs and norming representative of Aboriginal populations. Our review of the manuals, and the very small literature base ascertaining validity of tests within Aboriginal populations, indicates that these assessment measures do not have sufficient presence of Aboriginal peoples. We are not confident that the norms are representative of them. Musquash and Bova (2007) note that Canada’s Aboriginal peoples are far from homogeneous. They indicate, First Nations, Metis and Inuit peoples are living on over 2200 reserves, in 596 bands and a large number living off reserve in Canada. The tests do not seem reflective of this population complexity. Butchers, Derksen, Sloor and Sirigatti (2003) have shown that cross-cultural norming is possible but it does not appear to have been done with the various Aboriginal peoples of Canada. The issues can include determining what is normative to a culture, specific meanings of constructs and language, and determining how the original tests constructs can be used within the intended culture. Butcher (2004) indicates that previously developed, validated test items, provide a baseline to assess the factors that can be measured by the particular psychometric. He goes on to note that there is a need for culturally based research to ensure validity within the culture. The PAI test developers examined previously developed test items with other cultures. This revealed some constructs that were not appropriate cross culturally. The cross culture variability was thought to change the intended use of the test questions. The existing test questions served as a baseline, but pointed out that the items needed to be examined to ensure validity within different cultures (Morey, 1996).

Pace et al. (2006) make the cogent observation that the MMPI-2 “does not offer a bird’s eye view of the personality landscape, but rather it provides a presupposed form upon which personality may be displayed” (p. 329). We suggest that this is true of all the measures considered here. A determination is developed about the standard against which all who are tested will be measured. The results of an assessment measure compare the norming data with specific results for the client.

There are reasons to be quite cautious in the use of assessment measures with Aboriginal peoples. The lack of substantial Aboriginal presence in the norming population is one concern. Another is a lack of clarity as to whether constructs measured are relevant to this population.

We are also struck by the reality that none of the measures considered here included any Canadian sampling, with the exception of a project with French Canadian parents for the PSI French Canadian version (Abidin, Austin & Flens, 2013). Thus, the versions available on the market in Canada will be drawn from a non-Canadian, mainly American population.

Drew, Adams and Walker (2010) remind us that assessment is a socially and culturally mediated practice. The parenting tests reviewed in this work are developed using constructs of parenting as defined by the dominant culture, although Abidin et al., (2013) suggest that the PSI has been found valid across several cultures. However, Ambert (1994) describes crucial underpinnings to research in this field. She indicates that Western paradigms are being used to construct definitions of parenting creating an ethnocentric view. Muir and Bohr (2014) have made a strong argument that there are traditional Aboriginal child-rearing practices, which vary from the paradigms used to construct both individual and familial social meaning, and behavioral practices measured in the parenting assessment tools. The existing psychometrics to assess parenting were not built utilizing traditional Aboriginal approaches to parenting. Muir and Bohr (2014) illustrate that traditional child rearing practices are still in use in contemporary Aboriginal parenting. They show that traditional Aboriginal values are represented in the culture. Examples include placing value on the autonomy of the child, the role of extended family, different forms of attachment, different parenting roles for males and females, different views of developmental progress and discipline and the impact of spirituality and language.

If PCAs are going to become more culturally relevant, then not only will the psychometrics used need to be culturally normed, but so too will the foundational definitions of parenting—which is the focus of the PCA. Indeed, if the construct of parenting in a PCA is based upon the dominant cultural definition and the psychometrics are developed using the dominant cultural definitions, then the Aboriginal parent will be disadvantaged in both ways.

The psychometrics that are specific to parenting are rooted in Western definitions of family and family functioning. They are not built upon Aboriginal understandings. Thus, the cultural context of the questions is based upon structural beliefs that may be in contrast to Aboriginal beliefs (Stairs & Berhnard, 2002). Elliott and Smith (2014) have pointed out that there is concern by Aboriginal professionals, leaders and parents that many of the assessment tools being used with children are not culturally appropriate. We suggest the same argument can be made for adult assessment tools.

The Federal Court, in Ewert v. Canada (2015), has recently considered actuarial risk assessment tools used in criminal matters. Relying upon the testimony of Dr. Stephen Hart of Simon Fraser University, the court noted that actuarial tests are likely to be affected by cross-cultural variance because of the cultural differences between Aboriginal and non-Aboriginal Canadians (Paragraph 27). The Court accepted the conclusion that the tests were not “sufficiently predictably reliable for Aboriginals because of the cultural variance or bias of the tests” (paragraph 40). Referring to the Criminal Code of Canada, the Court concluded that the assessment tools were not being responsive to the special needs of Aboriginal people,, and the use of these tests violated the charter rights of the plaintiff in this case. We believe that the reasoning of the Court in Ewert (2015) is applicable to the psychometrics used in PCAs.

Others have argued that foundational constructs that inform PCAs are not supported within the Aboriginal cultures of Canada. For example, Neckoway, Brownlee and Castellan (2007) indicate that attachment theory, which is widely discussed in the PCA literature, is not interpreted in a way that is consistent with the family and cultural structure of Aboriginal peoples. This further undermines the present PCA constructs with this population.

Milner (1986) indicates that social service populations may be different than the majority population. We suggest that, an Aboriginal child protection population may be even more so given the presence of both cultural and child protection issues. We think that the issues of inter-generational trauma and cultural genocide identified by the Truth and Reconciliation Commission (2015) further reinforces the uniqueness of the Aboriginal population. Such factors require consideration in the development, norming and application of assessment measures to be used with Aboriginal clients.

Culturally, Indigenous Aboriginal peoples have a more collective view of parenting and family. We argue the current approach should be changed. The nuclear family system that underpins the assessment framework is not presently adapted to the broader worldview of Aboriginal collectivistic parenting and family perspectives. The Child and Youth Advocate of Alberta, in a recent investigative review, challenged the belief that Aboriginal parents can be effectively assessed using constructs developed within the dominant culture (OCYA, 2014).

Meaningful PCAs of Aboriginals requires the development of a PCA model rooted in Aboriginal culture Such an approach might include tools such as the Medicine Wheel, for example (Twigg & Hengen, 2009);); however, research is needed in this area to accomplish a valid, revised approach. Given the over-representation of Aboriginal children in Canada’s child protection system (Sinha, Trocme, Fallon and Maclaurin, 2013), there is an urgent need for a revised method.

If dominant cultural definitions of parenting are to be used, then parents from Aboriginal communities and families are at a disadvantage even before the assessment begins. Understanding the context of parenting, historically and culturally, is essential to approaching PCAs with Aboriginal parents (Muir & Bohr, 2014). Van de Sande and Menzies (2003) offer one of the few publications that has documented that there are identifiable differences between Aboriginal parenting and mainstream Canadian society. The ways Aboriginal families are defined and how they approach parenting are different and must be understood (BigFoot & Funderburk, 2011).

Not all Aboriginal parents will reflect a connection to traditional ways of parenting. Those who have been raised away from their culture may find a PCA approach rooted in Aboriginal cultural parenting approaches unfamiliar. Those parents may be more comfortable with the current approach, although that has not been researched. The parent who exhibits a strong cultural connection will find present approaches to PCAs less reflective of who they are and the culture to which they belong.

In this article, we have raised a number of concerns that should act as a call to action by child protection authorities and assessors to develop different, culturally appropriate methods to assess Aboriginal parents. Such efforts would be part of reforming the relationship between child protection and the Aboriginal peoples of Canada.

Limitations

This work extends the limited literature on PCAs within the Aboriginal community but raises a number of questions. It raises concerns about both the psychometric norming and the definitional foundation of parenting that are presently used to assess Aboriginal parents. However, there is much research to be done and this paper outlines problems. Further work is needed to extend these concerns towards solutions. This work also suggests that the structure of PCAs might warrant a careful examination beyond psychometrics, to determine the applicability to Aboriginal peoples. It is one thing to point out the problem, it is quite another to build a different approach. Future work will need to address this considering the ways in which PCAs might be constructed in a culturally meaningful manner to assist child protection and the courts in cases where such support is needed.

This research did not review the less commonly used tools. Assessors should approach all PCA psychometrics with caution determining whether or not there is a sufficient Aboriginal inclusion in the norming to suggest validity with their client.