Validation of the Child and Youth Resilience Measure (CYRM-28) on a Sample of At-Risk New Zealand Youth

Abstract

Purpose:

This article reports on an examination of the psychometric properties of the 28-item Child and Youth Resilience Measure (CYRM-28).

Methods:

Exploratory factor analysis, confirmatory factor analysis, Cronbach’s α, t-tests, correlations, and multivariate analysis of variance were applied to data collected via interviews from 593 at-risk adolescents (12–17 years) to identify the factor structure, internal consistency, test–retest reliability, construct validity, and floor and ceiling effects of the CYRM-28.

Results:

A four-factor structure was identified comprising two contextual factors, individual and family factors. The CYRM-28 and its factors show good internal reliability, stable test–retest properties, and no floor or ceiling effects. The measure also showed good construct validity.

Discussion:

The CYRM-28 shows good overall validity on this group of New Zealand youth, and researchers and social workers can have some confidence in its usefulness as a measure that can be used to assess resilience in youth from a range of ethnic backgrounds.

Keywords

scale validation Child and Youth Resilience Measure resilience vulnerable youth risk

Introduction

It has long been recognized that exposure to high levels of ongoing risk throughout childhood undermines the capacity of individuals to achieve to their potential in adulthood (Ai, Jackson Foster, Pecora, Delaney, & Wenceslao, 2013; Zolkoski & Bullock, 2012). For more than three decades, the goal of resilience research has been to explain how protective processes function to dampen the impact of high levels of risk thereby facilitating the achievement of better than expected outcomes (Alvord & Grados, 2005; Anthony, 1974; Brooks, 2006; Cicchetti, 2003; Luthar, 2006; Luthar, Cicchetti, & Becker, 2000; Madsen & Abell, 2010; Masten, 2007, 2011; Masten, Best, & Garmezy, 1991; Masten & Coatsworth, 1998; Pines, 1975; Rak & Patterson, 1996; Rutter, 2000; Ungar, 2011; Werner, 1986). Four waves of resilience research have been identified (see, e.g., Masten, 2007) since the pioneering works of Anthony (1974) and Pines (1975), who sought to describe the characteristics of the so-called “invulnerable” children. Briefly, the four waves have focused, respectively, upon resilience as an individual construct; defining developmental, ecological systems and resilience processes; intervention studies designed to enhance resilience through treatment; and finally understanding how resilience is shaped by multiple systems, levels, and the interactions between these multiple systems and levels (O’Dougherty Wright, Masten, & Narayan, 2013). Throughout this time, resilience has been variously conceptualized as:

a trait, a process, an outcome, a pattern of life course development, narrow or broad, multifaceted or unidimensional, short or long term … [and as] encompass[ing] recovery as well as resistance, internal as well as external adaptive functioning, and external as well as internal resources. (Masten, 2007, p. 924)

Despite variation in definitions, researchers now consistently argue that resilience is a multidimensional construct that varies with context: time, age, gender, cultural background, other life circumstances, and biological factors all influence resilience (see, e.g., Connor & Davidson, 2003; Liebenberg, Ungar, Armstrong, & Van de Vijver, 2012; Masten, 2011; Ungar, 2008). Furthermore, different factors and combinations of factors appear to come into play at different times and in response to the presence of different combinations of risks (Ungar, 2011). It is clear then that a broad focus must be adopted in studies of resilience because no single factor consistently appears to make the defining difference to outcomes (Daigenault, Dion, Hébert, McDuff, & Collin-Vézina, 2013; Gunnar, 2006). Accordingly, the “fourth wave” (Masten, 2007, p. 924) of resilience research now takes a wide remit examining the interactions between factors that range from the neurobiological and genomic to the community level. Masten (2007, p. 925) elaborates upon a range of “hot spots” that are ideally suited to multilevel analysis in this fourth wave. In terms of the focus of the current article, it is notable that many of these hot spots are clusters of relationships and supportive resources located in the family, peer, school, community, and cultural domains. The hot spots identified by Masten (2007) underscore the importance of understanding children’s development in an ecological context (Luthar, 2006; Masten, 1999) because different facets of resilience processes can offset the impact of risks and influence outcomes through this buffering effect. For instance, Ungar, Liebenberg, Armstrong, and Van de Vijver (2012) argue that culturally anchored resilience processes have the potential to mitigate both individual and contextual risks. When understood ecologically, as in the case of the current study, resilience becomes a “multi-dimensional process that mediates the effects of stressors and the achievement of positive outcomes” (Liebenberg et al., 2012, p. 291).

Regardless of how resilience is conceptualized, Ungar (2011, p. 6) explains that it is clear that “protective processes” matter greatly when trying to tip the odds in favor of positive outcomes for children exposed to high levels of risk throughout childhood. In this sense, resilience represents the scaffolding around vulnerable children that enables them to increase their capacity to do well despite exposure to significant adversity (Liebenberg et al., 2012). Better outcomes are achieved when the focus is not only on risk reduction but also upon the intentional enhancement of these resilience processes and resources (Ahern, Kiehl, Sole, & Byers, 2006).

Despite more than three decades of research into resilience, and an enduring concern with understanding how to intervene effectively to weight the odds in favor of better than expected developmental outcomes for children exposed to high levels of adversity, very few resilience measures have been developed and validated that focus explicitly on children and young people (Daigenault et al., 2013). Differences in the approach taken to conceptualizing and measuring resilience across studies have led to inconsistencies in the identification of potential risk factors and protective processes and also in estimates of prevalence (Windle, Bennett, & Noyes, 2011). This has made it difficult to achieve a broadly agreed definition of resilience. A recent methodological review of resilience scales (Windle et al., 2011) noted that while there were promising developments in the field, no measure currently met the range of acceptability criteria commonly agreed as required for a valid and reliable measure. Of the 15 measures reviewed, 9 focused exclusively at the individual level, meaning that no assessment was able to be made of the presence of relational, cultural, or contextual resources. Furthermore, the three measures that achieved the highest quality scores in relation to their psychometric properties focused on adults only.

The lack of a validated instrument for use with children and young people is a significant issue. It leaves social work practitioners struggling to quantify and understand the resources, assets, and needs of children and youth when assessing the impact which risks have upon their capacity to thrive. It further makes it difficult for social workers to be confident in their assessment of the ways in which the resources around youth may better be harnessed in their interventions. There is a pressing need for an easy to administer, contextually sensitive resilience measure which social workers can be confident will accurately identify the presence of protective resources located in the individual, relational, and contextual/cultural domains around children and youth.

The Child and Youth Resilience Measure-28 (CYRM-28) was specifically developed for assessing resilience in vulnerable adolescents in diverse sociocultural contexts. It has a particular concern with assisting researchers and social work practitioners to build a clear picture of the resources available to vulnerable youth, including those relating to their own individual characteristics, the presence of supportive people around youth, and their access to cultural and contextual resources that will support their positive development (Liebenberg et al., 2012; Ungar & Liebenberg, 2011). The CYRM-28 uses a 5-point Likert-type scale to assess social ecological factors related to resilience (i.e., individual, relational, and contextual). It was developed using an iterative, mixed-method strategy. Initially, a 58-item measure was developed using qualitative methods at 14 sites in 11 countries. This was followed by statistical testing with 1,451 youth aged 13–23 years (Ungar et al., 2008; Ungar & Liebenberg, 2011). Sites were selected to maximize the diversity of youth in terms of risks and community contexts and also to include a range of languages in addition to English (Liebenberg et al., 2012). Initial analysis suggested the 58 items clustered into a four-component structure comprising individual, relational, community, and cultural dimensions with Cronbach’s α values of .84, .66, .79, and 71, respectively (Ungar et al., 2008). This initial analysis however also involved reducing the CYRM to a 28 item (Ungar & Liebenberg, 2011).

Windle, Bennett, and Noyes, in their 2011 review of resilience measures, assessed the CYRM-28 as a promising measure due to its development in 11 different nations intentionally selected to maximize diversity of context and risk and because youth were directly involved in the selection of items included in the questionnaire. They thus scored the CYRM-28 highly on content validity and involvement of the target population in item selection. Windle and colleagues (2011) also noted that further work was required on the validation of the CYRM-28. Following this, an assessment of the validity of the CYRM-28 in relation to Atlantic Canadian youth has been reported in this journal (Liebenberg et al., 2012). This analysis identified three subscales relating to individual, relational, and contextual components of resilience. Further assessments of the measure in relation to French Canadian (Daigenault et al., 2013) and Mi’kmaq youth (Zahradnik et al., 2010) have been reported. The current article adds to this validation process by assessing it in a large study of at-risk New Zealand youth. The article addresses some of the outstanding validation issues (Windle et al., 2011). Specifically, it presents analyses relating to the factor structure, the internal consistency, the test–retest reliability, the construct validity, and any floor and ceiling effects of the CYRM-28 and its subscales. While other scales have been validated across multiple cultural contexts (see, e.g., the Resilience Scale for Adults; Friborg, Hjemdal, Rosenvinge, & Martinussen, 2003; Ruvalcaba-Romero, Gallegos-Guajardo, & Villegas-Guinea, 2015), most of these have originally been developed for use with adults and adapted for use with youth (see, e.g., Hjemdal, Friborg, Martinussen, & Rosenvinge, 2001). This is the only measure we are aware of that has been intentionally developed for use with children and youth, focused on resilience processes in contexts of adversity, where context was seen as the identifier of risk rather than clinical indicators of the individual.

In 2009, the CYRM-28 was integrated into a questionnaire administered as part of a larger study of risk among youth in New Zealand. New Zealand is often thought of as a relatively benign, safe place to raise children. However, a substantial number of children face high levels of poverty, ill health, and exposure to enduring risks in their families and communities. There are well-recognized patterns of disadvantage in New Zealand characterized by unequal distributions of both risks and resilience processes between different ethnic/cultural groups. These patterns suggest that indigenous youth (Māori) and Pacific Islands youth experience poorer outcomes than their white/European descent (Pākehā) counterparts (Education Counts, 2000–2009; Bovan, Harland, & Grace, 2011; Johnson, 2013). On the other hand, there are also arguments that the stronger connections to culture experienced by Māori and Pacific Islands youth may buffer the impact of risks upon them, including the cumulative generational impact of colonization in New Zealand and the Pacific diaspora (Macfarlane, 2004; Mila-Schaaf, 2010; Papoutsaki & Strickland, 2009). Child poverty, as assessed, for example, by income-based measures, has increased sharply over recent decades from 12% in 1988 to 35% by 2014 (Perry, 2014), compromising in the process the health, educational, safety, and well-being status of these children. These children can be expected to face adulthood with compromised capacity to achieve to their abilities and with increased risks of transmitting the disadvantages they confronted as children on to their own children (Zolkoski & Bullock, 2012). This pattern of intergenerational transmission of disadvantage is now established in New Zealand and therefore being able to confidently assess the resources these children have around them that can help to offset the risks their circumstances contain is pressing. The Successful Youth Transitions research program (the Youth Transitions Study; see www.youthsay.co.nz) emerged in this context of deteriorating life chances for a substantial number of New Zealand children. This is a longitudinal, mixed-methods investigation into the factors that influence successful transitions for at-risk youth seeking in particular to identify the ways in which risk and protective factors interact to shape outcomes for vulnerable young people.

Hypotheses

Six hypotheses are proposed in order to assess the validity of the CYRM-28 (see Research Design, for analytical procedures):

Hypothesis 1: That a three-factor solution similar to that observed by others (Daigenault et al., 2013; Liebenberg et al., 2012) would be found in the New Zealand data.

Hypothesis 2: That the CYRM-28 and any subscales identified in this study would exhibit a Cronbach’s α of at least the minimum value achieved in the original Canadian validation of α = .65 (Liebenberg et al., 2012).

Hypothesis 3: That there would not be any significant differences between two time points for the CYRM-28 overall nor for any subscales identified in Hypothesis 1.

Hypothesis 4: That the CYRM-28 and any subscales identified in Hypothesis 1 would positively associate with measures that assess general well-being and with measures of positive parent/caregiver relationships.

Hypothesis 5: That fewer than 15% (Terwee et al., 2007; Windle et al., 2011) of the study group sample would return highest (i.e., a score of 140) and lowest (i.e., a score of 28) possible scores.

Hypothesis 6: That there would be significant differences in youth scores on the CYRM-28 and its subscales depending on the gender, ethnicity, and age of youth.

Method

Participant Characteristics

The article draws upon data from 1,366 youth who were recruited into the New Zealand Youth Transitions research program between 2009 and 2010. The current article focuses in particular on the responses of a subset of 593 youth (the study group) from the larger study. For comparison purposes (see Sampling Procedures, subsequently), an additional group were selected from the larger data set comprising 593 youth who were matched to the study group based on age, gender, and ethnicity. Youth ranged in age from 12 to 17 years (M = 15.3, SD = 1.1), and 58.7% were male. The higher percentage of males is due to the inclusion of juvenile justice services as one of the services working with the study group. Juvenile justice services have a disproportionately high representation of males. Youth could list as many ethnicities as accurately described their sense of cultural identity. Youth predominantly identified with one ethnic group (n = 814, 69%), approximately one third (n = 336, 28%) identified with two ethnic groups, and a small number identified with three or four ethnic groups (n = 36, 3%). A system of prioritizing ethnicity that is used in analysis of social and health data in New Zealand was used to sort data (Cormack & Robson, 2010). This involved coding any youth who identified Māori (the indigenous population of New Zealand) as one of their ethnicities as Māori and youth who identified Pacific as Pacific, providing they did not also identify Māori. Finally, youth who did not identify either Māori or Pacific identities, but who did identify Pākehā (white New Zealander/other European), were coded as Pākehā. This coding system accounted for 1,163 (98%) youth, and the remaining youth were coded to the category “other ethnicity.” The prioritized approach has a good fit with other research in New Zealand focusing on ethnicity. By this system of coding, one half of the youth were Māori (n = 523, 44%), approximately one fifth (n = 253, 21%) were Pacific, and approximately one third were Pākehā (n = 387, 33%).

The descriptive data collected in the survey identifies that the study and comparison groups had markedly different lives. For instance, their living arrangements were significantly different; Fisher’s exact test: χ²(12, n = 1,186) = 267.572, p < .001, η² = .408. Study group youth lived in a diverse range of situations, while comparison group youth tended to be concentrated in households with either both birth parents or one birth parent (with or without a step parent). Just over half of the respondents from the comparison group (58.7%) were living with both birth parents, whereas this was the case for only 19.4% of the study group participants (standardized residual score = 7.7). It is also of note that study group youth who were not living with both parents were significantly less likely (2.7% study group and 12.3% comparison group, standardized residual score = −2.6) to spend equal time with custodial and nonresident parents and more likely to have no contact with the nonresident parent than was the case for the comparison group. While 22.8% of the study group youth had living arrangements that did not contain family members as caregivers, this was the case for only 1.7% of comparison group youth (standardized residual score = 7.3).

Significantly more youth from the comparison group (95.6%) than study group (79.1%, standardized residual score = 2.2) were enrolled in an educational program at the time of completing the questionnaire, χ²(1, n = 1,186) = 73.297, p < .001, η² = .249. It should be noted that youth did not have to be attending a mainstream school to answer yes to this question. There were also significant differences between the groups, t(962.147) = −12.665, p < .001, η² = .143, in terms of the number of schools they had attended over their lifetimes with the study group attending significantly more schools (M = 5.14, SD = 2.89) than comparison group youth (M = 3.39, SD = 1.70). Youth from the study group were also less likely to be attending a mainstream school at the time of the interview, χ²(10, n = 1,186) = 809.957, p < .001, η² = .826; study group 18.2% and comparison group 99.5%, standardized residual score = −12.8).

The questionnaire also explored the value youth attached to education, the sense of belonging they felt to the last mainstream school they had attended (or to their current school if they were still attending a mainstream school), and a global assessment of the nature of the last mainstream school they attended (or to their current school if they were still attending a mainstream school). There were significant differences between the two groups on each of these questions: sense of belonging: χ²(2, n = 1,186) = 200.108, p≤ .001, η² = .409; value of education: χ²(2, n = 1,186) = 133.442, p≤ .001, η² = .331; and global assessment of school: χ²(2, n = 1,186) = 239.308, p < .001, η² = .449. Youth from the comparison group were more likely to report feelings of belonging in relation to school (comparison group—74.5% and study group—39%, standardized residual score = 5.8), to place a higher importance on education (comparison group—89.2% and study group—60.9%, standardized residual score = 4.0), and comparison group youth (77.6%) were more likely to report positively about their school than were study group youth (37.3%, standardized residual score = 6.5). These differences were statistically significant. Alongside this, youth reported vastly different experiences of exclusion from school, being held back, self-exclusion such as wagging (skipping, bunking, or not attending school) and not being able to attend school due to circumstances beyond their control. There were significant differences between the two groups on each of these questions: stood down: χ²(1, n = 1,186) = 410.533, p < .001, η² = .588; suspended: χ²(1, n = 1,186) = 454.663, p < .001, η² = .619; excluded: χ²(1, n = 1,186) = 368.642, p < .001, η² = .558; and held back: χ²(1, n = 1,186) = 57.092, p < .001, η² = .219. Study group youth reported higher rates of harsher penalties than did the comparison group. Sixty-eight percent of study group youth had been stood down (comparison group youth—10.1%, standardized residual score = 11.2) and 64.9% had been suspended (comparison group—5.7%, standardized residual score = 12.1), while 50.1% of study group youth had been expelled or excluded from school (comparison group—1.3%, standardized residual score = 11.7). The study group also reported significantly greater risk exposure, t(1,032.809) = −27.732, p < .001, η² = .427, and lower resilience, t(1,172.647) = 8.803, p < .001, η² = .062, than comparison group youth.

Sampling Procedures

This validation of the CYRM-28 focuses on the responses of a subset of 593 youth (the study group) from the larger youth Transitions Study (N = 1,366). The sampling plan specified a minimum of 480 participants as this would generate a minimum of 120 young people in each of the four service systems (see subsequently). Recruitment continued to 593 participants to ensure adequate geographic distribution of young people across rural and urban communities. Youth were purposively selected into the study group from four sites across New Zealand based on their risk of not graduating from high school or because they were clients of one or more of the major service systems which work with high risk youth in New Zealand. The service systems were juvenile justice, child welfare, education additional to mainstream classroom programming, and mental health. Based on published research, it was reasoned that these characteristics made the youth vulnerable to poor psychosocial outcomes through adolescence and into adulthood (McLean, Wood, & Breen, 2013; Sanders, Munford, Liebenberg, & Ungar, 2014). A “community saturation” approach was taken to recruiting the study group (Bowen, 2008). This involved negotiating and securing support of all or most of the service providers who worked with the target population of youth in each of five geographical areas that were selected to ensure a good mix of rural and urban settings. Researchers then worked from the largest to the smallest organization in each locality to identify youth who met the selection criteria. This process was continued with each organization until no new names were generated. All eligible youth were approached first by agency staff to gain permission from the youth to meet a researcher and then consent for participation was secured with the youth by a member of the research team. In all cases, organizations were not aware of which of the youth from their caseloads eventually completed questionnaires.

To confirm that the 593 young people in the study group were exposed to elevated risks of poor psychosocial outcomes, a second subsample was taken from the larger Youth Transitions Study. This comprised 593 young people who were matched to the study group based on age, gender, and ethnicity (see Table 1). These young people (the comparison group) came from the same communities as the study group and were recruited from schools and community organizations that provide community-based normative developmental opportunities for youth such as sport and cultural activities.

Table 1.

Age, Gender, and Ethnicity of Study Group and Comparison Group Youth.

Demographic characteristics	Study Group		Comparison Group		p
Demographic characteristics	n	Percentage	n	Percentage	p
Gender					.906
Female	245	41.30	247	41.70
Male	348	58.70	346	58.30
Age					.068
15 years and under	327	55.10	358	60.40
16+ years	266	44.90	235	39.60
Ethnicity					.140
Māori	281	47.40	242	40.80
Pacific	117	19.70	136	22.90
Pākehā	185	31.20	202	34.10
Other	10	1.70	13	2.20
Total	593	100.00	593	100.00

The research was approved by the authors’ University Ethical Review Board prior to commencement. Ethics protocols ensured that research processes and measures were culturally appropriate, sensitive, and respectful that young people were well informed about the study and gave their own written consent to participate. Only those youth who consented to participate were included in the study. Young people were given a NZD$20 gift voucher of their choice after completion of the survey and food to a maximum value of NZD$10 was taken to each interview. The refusal rates for this study were 2.5% for the study group and 12% for the comparison group. Further details about ethical protocols and procedures adopted in the research can be found at www.youthsay.co.nz. Interviews were completed by trained interviewers who administered the questionnaire. Youth were able to complete the questionnaire confidentially and had access to trained researchers who could answer their questions and provide assistance as required.

Measures

The following measures were used in the validation process reported on in this article. In addition to describing the CYRM-28, this section details the measures used to assess the construct validity of the CYRM-28. Because resilience processes have been theoretically, clinically, or empirically associated with improved well-being and prosocial functioning (see, e.g., Ungar, 2011), the Satisfaction with Life and the Strengths and Difficulties Questionnaire (SDQ) Prosocial scales were used to assess the construct validity of the CYRM-28 in terms of its capacity to accurately reflect general well-being and positive functioning. These two measures were chosen because they have been extensively used in a wide range of locations globally and have been found to be effective in capturing cultural diversity. Positive relationships with parents/caregivers have also been linked to greater resilience (Daigenault, et al., 2013), and for this reason, two measures that assess the quality of the caregiving relationship were also used to assess construct validity of the CYRM-28.

Resilience

As noted above, the CYRM-28 (Liebenberg et al., 2012; Ungar & Liebenberg, 2011) is a 28-item ecological measure of resilience. The CYRM-28 asks youth to identify the extent to which 28 statements describe them, for example, “I get on with people around me.” All items are rated on a 5-point scale from 1 = does not describe me at all to 5 = describes me a lot. Higher scores indicate greater presence of resilience processes. The reliability coefficient in the current study (α = .86) was strong indicating preliminary support for Hypothesis 2.

Prosociality

This was assessed using the SDQ Prosocial behavior subscale (α = .66; Goodman, 1997, 2001), which assessed youth capacity for kindness, sharing, and concern for others. Positive social interaction was measured on a 3-point scale from 0 = not true to 2 = certainly true. Higher scores indicate greater level of prosocial behavior. The α coefficient for the scale was .61.

Life satisfaction

The Satisfaction with Life Scale (SWLS; α = .87; Diener, Emmons, Larsen, & Griffin, 1985) was used to capture levels of well-being. This scale assesses overall levels of current life satisfaction using 5 items all rated on a 7-point scale, from 1 = strongly disagree to 7 = strongly agree. The scale includes items such as “The conditions of my life are excellent,” “So far, I have gotten the important things I want in life,” and “If I could live my life over, I would change almost nothing.” Higher scores are indicative of greater levels of satisfaction. In this study, response options were reduced to a 5-point scale (1 = strongly disagree to 5 = strongly agree). The reliability coefficient for this study was .85.

Quality of relationship with caregivers/parents

Youth were also asked two sets of questions about the quality of the relationship they experienced with two parental/caregiving figures (a mother figure and a father figure). The first asked them to rank on a 4-point scale how close they felt to each of these parent figures where 1 = did not have this parent figure, 2 = not close at all, and 4 = very close. The second question asked youth to identify how much affection they received from these parent figures where 1 = did not have this parent figure, 2 = none at all, and 5 = a lot of affection. Responses were summed to give a score for each youth relating to (a) how close they felt to their parent figures and (b) how much affection they received from these people. Higher scores indicate higher levels of closeness to and affection from parents/caregivers.

Research design

The analysis presented in this article has six goals that together seek to add to current knowledge concerning the validity of the CYRM-28 by assessing its psychometric properties in relation to a sample of vulnerable New Zealand youth (the study group). The following hypotheses guide this process of psychometric assessment:

Hypothesis 1: To identify and confirm the factor structure of the CYRM-28 in relation to a large group (n = 593) of ethnically diverse, at-risk New Zealand youth.

Hypothesis 1 proposed that a three-factor solution similar to that observed by others (Daigenault et al., 2013; Liebenberg et al., 2012) would be found in the New Zealand data. A combination of exploratory factor analysis (EFA) conducted on approximately half of the sample (n ₁ = 296; 50%) and a confirmatory factor analysis (CFA) on the second half of the sample (n ₂ = 297; 50%) was used to test this hypothesis. The EFA was used to identify the smallest number of meaningful factors to include later in the CFA. The CFA in turn assessed whether there was a sufficiently strong rationale for including the specified factors from the EFA in the overall resilience measure. Two indices, Tucker–Lewis index (TLI) and comparative fit index (CFI), were used to test model fit, and the root mean square error of approximation (RMSEA) was used to test the closeness of fit.

Hypothesis 2: To assess the internal consistency of the CYRM-28 on a New Zealand sample of ethnically diverse, at-risk youth (n = 593).

Hypothesis 2 proposed that the CYRM-28 and any subscales identified in this study would exhibit a Cronbach’s α of at least .65, the minimum value achieved in the original Canadian validation (Liebenberg et al., 2012). The CYRM-28 has been designed to be sensitive to differences in culture and setting, and thus in this diverse group of New Zealand youth, we expected that both the global scale and any subscales identified in the data set would exhibit internal consistency coefficients of a similar level to other validations (see, e.g., Daigenault et al., 2013).

Hypothesis 3: To assess the test–retest reliability of the CYRM-28 (n = 38).

Hypothesis 3 proposed that there would not be any significant differences between two time points for the CYRM-28 overall nor for any subscales identified in Hypothesis 1. In order to assess the scale’s reliability, it was administered twice to a subgroup of 38 youth (22 males and 16 females; 20 aged 15 years or under and 18 aged 16+ years) on two occasions separated by approximately 1 month. Paired-samples t-tests were used to test differences in scores between these two administrations of the measure.

Hypothesis 4: To establish the construct validity of the CYRM-28 (n = 593).

Hypothesis 4 proposed that the CYRM-28 and any subscales identified in Hypothesis 1 would positively associate with measures that assess general well-being and with measures of positive parent/caregiver relationships. To test this hypothesis, convergent validity was assessed by calculating correlations between the CYRM-28 (overall and subscales) and two measures of well-being: the SDQ Prosocial measure (Goodman, 1997, 2001) and the SWLS developed by Diener, Emmons, Larsen, and Griffin(1985) and two measures which assessed the quality of youth relationships with their parents/caregivers.

Hypothesis 5: To determine floor and ceiling effects of the CYRM-28.

Hypothesis 5 proposed that fewer than 15% (Terwee et al., 2007; Windle et al., 2011) of the study group sample would return the highest (i.e., a score of 140) and lowest (i.e., a score of 28) possible scores on the CYRM-28. Floor and ceiling effects were assessed by evaluating the percentage of participants who had the highest and lowest possible scores on the total scale.

Hypothesis 6: To explore the functioning of resilience processes among different groups of youth.

Hypothesis 6 proposed that there would be significant differences in youth scores on the CYRM-28 and its subscales depending on the gender, ethnicity, and age of youth. This hypothesis enables the argument that resilience processes operate differently depending on culture and context to be assessed (Liebenberg et al., 2012; Ungar, 2008). To achieve this, multivariate analysis of variance (MANOVA) was undertaken on the study group sample (n = 593) with gender (two levels: male and female), age (two levels: youth 15 years and younger and 16 and older), and ethnicity (four levels: Māori, Pacific Island, Pākehā, and Other) as independent variables and the CYRM-28 and any subscales identified in Hypothesis 1 as the dependent variables.

IBM SPSS Statistics 22 and AMOS 22 for Windows were used to undertake these analyses.

Results

Prior to any analyses, missing data analysis was conducted; 1.68% of the data were missing and these data were missing at random. Missing data were replaced using maximum likelihood estimation (Enders, 2010). Continuous variables were examined for normality, linearity, and homoscedasticity, and all were within acceptable limits.

Hypothesis 1: Identify and confirm the factor structure of the CYRM-28.

Prior to performing the EFA, the suitability of the data for factor analysis was assessed. Inspection of the correlation matrix revealed the presence of many coefficients of 0.3. The Kaiser–Meyer–Oklin value was 0.820 (indicating that just over 80% of variance in the measured variables had common variance). This exceeded the recommended sampling adequacy value of 0.6 (Hutcheson & Sofroniou, 1999; Kaiser, 1970, 1974). Bartlett’s test of Sphericity (Bartlett, 1954) reached statistical significance, χ²(378) = 2,301.112, p < .001, supporting the factorability of the correlation matrix. The EFA results showed that the percentage of variance explained by these components ranged between 3.6% and 20.5%.

There is debate about how many factors should be applied when conducting EFA. Any component with an eigenvalue exceeding 1 is considered an important factor (Tabachnick & Fidell, 2007). Further, the more factors included, the better the fit and the greater the percent total of variance explained in the data. The EFA using principle components extraction and oblique rotation (direct oblimin) revealed eight components with eigenvalues exceeding 1 suggesting an eight-factor structure. Examination of the scree plot, however, showed four distinct eigenvalues above the point of inflection indicating that a four-factor structure should be retained (Cattell, 1966). The four-factor structure collectively explained 50.59% of the variance. The rotated solution revealed the presence of a simple structure (Thurstone, 1947) with a number of strong loadings on each of the four components (context: social/cultural, family, individual, and context: spiritual/community) and all variables loaded substantially on only one component (Table 2).

Table 2.

Pattern Matrix of the Four-Factor Solution of the CYRM-28.

CYRM Item	Components
CYRM Item	Context: Social/Cultural	Family	Individual	Context: Spiritual/Community
Q1: I cooperate with people around me			.673
Q2: I aim to finish what I start	.229		.355
Q3: People think I am fun to be with			.677
Q4: I solve problems without drugs or alcohol			.551
Q5: I am aware of my own strengths			.281
Q6: Spiritual beliefs are a source of strength for me				.740
Q7: I think it is important to serve my community	.241		.247	.496
Q8: I feel supported by my friends	.314		.594	−.234
Q9: My friends stand by me during difficult times	.264		.519	−.356
Q10: My caregiver(s) watch me closely		−.534		.272
Q11: My caregiver(s) know a lot about me		−.680
Q12: I eat enough most days		−.643
Q13: I talk to my caregiver(s) about how I feel		−.635
Q14: My caregiver(s) stand by me during difficult times		−.818
Q15: I feel safe when I am with my caregiver(s)		−.806
Q16: I enjoy my caregiver(s)’ cultural and family traditions		−.536		.250
Q17: Getting an education is important to me	.366			.256
Q18: I feel I belong at my school	.394
Q19: I have people I look up to	.627
Q20: I know how to behave in different social situations	.652
Q21: I am given opportunities to become an adult	.496
Q22: I know where to go to get help	.694
Q23: I have opportunities to develop job skills	.515
Q24: I am proud of my ethnic background	.551			.214
Q25: I am treated fairly in my community	.593
Q26: I participate in organized religious activities				.743
Q27: I enjoy my community’s traditions	.223			.626
Q28: I am proud to be a Kiwi	.447	−.266
Total number of items in each component	10	7	7	4
α	.772	.805	.662	.746

Note. Boldface values indicate on which factor items were retained. Extraction method: principal component analysis. Rotation method: oblimin with Kaiser normalization. CYRM = Child and Youth Resilience Measure.

The absolute values of the factor loadings for the four subscales reached an acceptable threshold (Pallant, 2007) with most items having values above 0.3. The exception was Question 5 “I am aware of my own strengths” (Component 3, Table 2) which had factor loading of 0.28, falling just below the threshold. Most item loadings were in line with the original Canadian validation of the CYRM’s factor structure (Liebenberg et al., 2012). However, 5 items were found to fit in different components than was the case in the Canadian study (Table 3). The current analysis suggested that there were two contextual subscales in New Zealand: one (context: social/cultural) which drew on items from both the contextual and individual subscales in the Canadian analysis. The other component (context: spiritual/community) included Question 27: I enjoy my community’s traditions; this fitted with the cultural component in the Canadian study, but for New Zealand youth, it associated more strongly with spiritual and community aspects of resilience.

Table 3.

Items in the New Zealand Analysis That Formed Into Different Components Than Indicated in the Canadian Analysis.

New Zealand	Canada
Context: Social/cultural
Q20: I know how to behave in different social situations	Individual—social skills
Q21: I am given opportunities to become an adult	Individual—social skills
Q22: I know where to go to get help	Individual—social skills
Q23: I have opportunities to develop job skills	Individual—social skills
Context: Spiritual/community
Q27: I enjoy my community’s traditions	Context cultural

A CFA (see Figure 1) was then conducted on the four-factor structure to confirm the interrelationships between the variables suggested by the EFA. The measurement model tested comprised the four latent variables (context: social/cultural, family, individual, and context: spiritual/community) found in the EFA (Table 2). These latent variables were allowed to covary. Modification indices suggested covariance between several items. For example, between Question 17: “getting an education is important to me” and Question 18: “I feel I belong at my school.” Once the relationships identified in the modification indices were added, an adequate fit was obtained, χ²(293, n = 297) = 331.52, p = .060; CFI = .982; TLI = .977; RMSEA = .021 (Hu & Bentler, 1999; Schreiber, Nora, Stage, Barlow, & King, 2006). The four-factor structure identified in the EFA is therefore supported by the results of the CFA. Specifically, all four latent variables showed a significant positive correlation with each other. The standardized coefficients of the item loadings for all factors were between 0.31 and 0.78, the standard errors were all positive and ranged from 0.04 to 0.12 and all were significant. Thus, the analysis here does not support the three-factor structure proposed in Hypothesis 1, rather a four-factor structure appears to more accurately represent the New Zealand data.

Figure 1.

Confirmatory factor analytic model of the Child and Youth Resilience Measure-28. All pathways are significant at p < .050 and the dotted lines represent negative relationship.

Hypothesis 2: Identify Cronbach’s α for the CYRM and any subscales identified in the EFA/CFA.

As can be seen in Table 2, the Cronbach’s α coefficients for the four components identified by the CFA/EFA analyses ranged from .66 to .81 and were therefore acceptable in all cases, providing support for Hypothesis 2.

Hypothesis 3: Assess the test–retest reliability of the CYRM-28.

Test–retest correlation coefficients at the 1-month interval (n = 38) were calculated for the overall CYRM-28 and for the four dimensions identified in the EFA/CFA. There were no significant differences between Time 1 and Time 2 for the CYRM overall nor for the four components (see Table 4) and accordingly Hypothesis 3 is supported by the data.

Hypothesis 4: Establish the construct validity of the CYRM-28.

Table 4.

Paired-Sample t-Test for Four CYRM-28 Components Between Time 1 and Time 2.

Measure	t	df	p	Time 1		Time 2
Measure	t	df	p	M	SD	M	SD
CYRM-28	.819	37	.418	106.53	16.56	105.03	15.17
Context: social/cultural	.544	37	.590	41.21	5.72	40.79	5.11
Family	.205	37	.839	26.02.	7.65	25.90	7.00
Individual	1.204	37	.236	28.29	3.61	27.47	4.13
Context: spiritual/community	.376	37	.709	11.00	3.35	10.87	3.13

Note. CYRM = Child and Youth Resilience Measure.

Table 5 presents correlations between the CYRM-28 overall along with the four subscales (identified in EFA/CFA) and four measures: SDQ Prosocial, Satisfaction with Life, caregiver closeness, and caregiver affection. Scores of prosociality and life satisfaction had significant positive associations with the CYRM-28 and the subscales. The strongest association was between the Satisfaction with Life measure and the context: social/cultural (r = .368). A similar association was also noted with the context: spiritual/community subscale (r = .340) and the SDQ Prosocial scale. The SWLS is an overall measure of general well-being and the SDQ Prosocial a measure of social functioning. These results suggest that the two CYRM-28 contextual subscales have good construct validity. Furthermore, because the wording in the SWLS has a strong individual focus asking questions such as “I am satisfied with my life” the strong association with the individual subscale of the CYRM-28 (r = .327) suggests good construct validity of this subscale as well. The two caregiving relationship quality measures showed positive associations with the CYRM-28 overall. They also had a strong positive association with the family subscale, suggesting construct validity for this subscale. These results confirm Hypothesis 4 and support the construct validity of the CYRM-28.

Hypothesis 5: Determine if the CYRM-28 has any floor or ceiling effects.

Table 5.

Correlations Between CYRM-28 Subscales and Total Score With SDQ Prosocial and Satisfaction With Life and Relationship With Caregiving Figures.

Measure	Context: Social/Cultural	Family	Individual	Context: Spiritual/Community	Total CYRM-28
SDQ Prosocial	.352**	.197**	.327**	.340**	.411**
Life satisfaction	.368**	.225**	.327**	.235**	.401**
Caregiver closeness	.265**	.351**	.119**	.211**	.346**
Caregiver affection	.257**	.294**	.155**	.181**	.321**

Note. CYRM = Child and Youth Resilience Measure.

**p = <.001.

No floor or ceiling effects were identified with the CYRM-28 total scale. No youth scored the lowest (28) or highest (140) possible scores, a similar result to previous reports of the CYRM-28 (Liebenberg et al., 2012). This result provides support for Hypothesis 5.

Hypothesis 6: Identify any significant differences in CYRM and subscale scores depending on age, gender, and ethnicity of youth.

A MANOVA was conducted using the study group sample (n = 593) to establish if there were significant differences in youth responses in terms of ethnicity, gender, and age. No significant multivariate main effects were found for age or gender, but there was a significant main effect for ethnicity, Wilks’ λ = .859, F(12, 1,545.41) = 6.821, p ≤ .001, η² = .049. The univariate test showed that the four ethnic groups were similar in terms of their scores on the context: spiritual/community and individual subscales. They differed significantly in regard to their scores on the areas of context: social/cultural and family subscales. Bonferroni post hoc comparisons showed that the groups that reached a significant level for each subscale were:

Pākehā youth reported lower context: social/cultural scores (M = 3.74, SD = .65) than both Māori (M = 3.96, SD = .59, p = .002) and Pacific Island youth, M = 4.06, SD = .70, p = ≤.001, mean difference = .319, 95% confidence interval (CI) = [0.123, 0.517].

Pākehā youth also reported lower family scores (M = 3.61, SD = .95) than Pacific Island youth, M = 3.95, SD = .81, p = .011, mean difference = −.340, 95% CI = [.553, .627].

These results suggest partial support for Hypothesis 5, in that certain aspects of resilience processes were connected to the cultural or ethnic identity of participants.

Overall, then the present findings lend support for content, construct, and criterion validity of the CYRM-28 for measuring the resilience of at-risk youth in New Zealand. The EFA, supported by the CFA, identified a four-scale structure, namely,

context: social/cultural (10 items),

family (7 items),

individual (7 items), and

context: spiritual/community (4 items).

Reliability analyses demonstrated that the four subscales were internally consistent and stable over time and no floor or ceiling effects were detected. While no systematic differences in scores across the four subscales could be found in terms of gender or age of youth, differences were noted in regard to the youth cultural background. This suggested that ethnicity of youth played a larger role in differentiating resilience resources youth draw on than did gender or age. This pattern was consistent with the findings from other validations (Daigenault et al., 2013; Liebenberg et al., 2012).

Discussion and Applications to Practice

While the construct of resilience has received considerable attention at a theoretical level, as others have observed (Daigenault et al., 2013; Liebenberg et al., 2012; Windle et al., 2011), measures that will allow for rigorous assessment of resilience processes, particularly in relation to at-risk youth, have yet to be fully validated. This article documents the continued validation process for the CYRM-28, building on groundwork undertaken by the measure’s developers (Liebenberg et al., 2012; Ungar & Liebenberg, 2011). Results from the current analysis suggest that overall the scale represents a stable, valid, and reliable measure of resilience among youth. The analysis points to some interesting patterns.

The family subscale identified in the Canadian data can be mapped exactly onto the New Zealand data set. Further, two of the three clusters identified in the Canadian individual subscale remain the same in the New Zealand data. However, in New Zealand, the social skills cluster of questions relate more closely to cultural resources than to individual aspects of resilience than was the case in the Canadian validation of the CYRM-28. There appear to be two different contextual subscales in this sample of New Zealand at-risk youth. One contextual subscale is composed of social resources that support youth capacity to function in social situations along with cultural resources that also support their development. The other contextual subscale relates to spiritual connections and community belonging. This deviation from the original Canadian validation does not compromise the overall functionality of the scale. Rather, it suggests that in some contexts, access to resources that build social skills may be more strongly associated with the presence of cultural resources that support resilience than they are with individual aspects of the resilience process. Indeed, the original work on the CYRM-58, by Ungar and colleagues (2008), identified a four-factor solution and this configuration featured two contextual factors: one that clustered around cultural dimensions and one around community dimensions. The New Zealand data for the CYRM-28 appears to have a somewhat similar pattern, although the connection between social skills and cultural resources is new.

The CFA shows a strong positive correlation between the two contextual components and also strong correlations between the individual and contextual components. Lower correlations were observed between the family subscale and the other three components. This is similar to the pattern observed by Liebenberg, Ungar, Armstrong, and Van de Vijver (2012) in their validation of the CYRM-28 with Atlantic Canadian youth. This pattern confirms the importance of contextual resources in the overall resilience of youth and of the way in which contextual resources interact positively with the individual resilience processes of young people. The moderate but positive correlations between the quality of caregiving relationships and the individual and the two contextual components may indicate that, for vulnerable youth, caregiving relationships are less stable than other types of resilience resources. As a result, at-risk youth find themselves drawing more on their own resources and those located in the cultural, social, spiritual, and community domains than they do within the caregiving environment.

The clustering of cultural adherence dimensions with those relating to access to resources that support development of social skills in the New Zealand data is interesting and suggests that for vulnerable youth in New Zealand, cultural aspects of resilience are strongly related to the presence of resources that help build social skills. Indeed others have noted the importance of sensitivity to different contexts when exploring resilience and that these items might cluster differently within the overall CYRM-28 is to be expected (Daigenault et al., 2013). This finding reinforces the argument of the differential functioning of resilience processes (Ungar, 2008) and highlights the importance of site-specific assessments being conducted on the ways in which the CYRM-28 items cluster as part of wider analyses of resilience. The set of social skills that clustered with cultural resources make sense in the New Zealand context because many of the cultural supports available to at-risk youth also have a major focus upon building social skills. For instance, many of the culturally specific programs for vulnerable Māori and Pacific youth encourage development of a sense of connection to culture at the same time, as they build youth capacity to act appropriately in different settings and encourage the building of work-related skills (Māori Youth Council, 2011; Oliver, 2001). That this is the way in which youth interpreted these questions is reinforced by the fact that the significantly lower context: social/cultural resilience resources identified in the MANOVA for Pākehā youth compared to both Māori or Pacific youth. It may be that when Māori and Pacific youth are vulnerable, they have greater access to culturally anchored supportive resources than do similarly vulnerable Pākehā youth.

In terms of age, gender, and ethnicity differences in CYRM-28 scores, the results are somewhat different to those observed in the Canadian validation which found a significant main effect for gender suggesting diminished resilience resources around boys (Liebenberg et al., 2012). However, those authors also reported that the size of this difference, though significant, was minor, and elsewhere no gender-based differences have been found for the CYRM-28 (Daigenault et al., 2013). As with the Canadian study, the key difference in resilience resources is seen when considering ethnic status of youth. Indeed, as with the Canadian sample, in the New Zealand population, ethnic majority youth (in the New Zealand case being Pākehā or White New Zealanders) reported lower overall resilience resources than either indigenous youth (Māori) or Pacific Island youth. Pākehā youth also reported fewer resilience resources from relationships with their caregivers than Pacific youth. These patterns are somewhat different to established patterns of disadvantage in New Zealand, where Māori and Pacific groups are consistently shown to face disadvantage compared to their Pākehā counterparts across a wide range of indicators, and it might have been expected that across all four subscales, these youth would have scored lower than Pākehā youth (Bovan et al., 2011; Education Counts, 2000–2009).

These findings confirm that, as others have suggested, resilience is intimately bound up with culture and context (Masten, 2007). This finding has implications for the assessment practices and interventions of social workers. It indicates that resilience and risk combine together in ways that are particular to each young person. This calls for carefully targeted and relevant interventions that specifically address the manner in which risks and resilience resources appear in each young person’s life. The sensitivity of the CYRM-28 to culturally anchored differences appears to be a particular strength of the measure. This finding suggests that the CYRM-28 subscales have particular value as an assessment and intervention planning tool because social workers can confidently use it to identify strengths, resources, and capacities within minority populations that can be effectively drawn upon in treatment.

The relatively stronger relationships observed in the CFA between the two contextual subscales and the individual subscale and the relatively weaker association between these three factors and the family subscale has important implications for social work practice with at-risk youth in that it calls for caution in assuming that youth will necessarily find support from their caregivers. It also suggests a need to closely investigate the resources located across all four domains because the different types of contextual resources may compensate for reduced capacity in caregivers or ruptures in support processes within families (Liebenberg et al., 2012). Compensatory processes young people may activate, such as support from friends, neighbors, and other social networks, need to be taken into account by social workers when planning interventions. The CYRM-28 provides a mechanism for assessing all of these different aspects of resilience and from there to developing more nuanced and subtle interventions that specifically target those aspects of youth lives that are most likely to produce overall benefits for youth. For instance, in the New Zealand context, there were clear differences in the resilience profiles of Pākehā, Māori, and Pacific youth suggesting a somewhat different focus for social worker attention with youth from different ethnic groups. The CYRM-28 allows social workers to identify those aspects of youth ecologies that are making a positive contribution to their resilience, such as connections with cultural practices, and then integrating this knowledge into their practice.

Although the sample size for the current analysis was large, participants were not randomly selected. Rather they were identified by service providers and community organizations as being at risk of not completing school and/or as clients of the welfare, juvenile justice, mental health, and alternative education systems. Given the narrow sample included in the study and the similarity in this regard with the Atlantic Canadian sample, cutoff scores and predictive validity also still need to be established (Terwee et al., 2007). However, the current study does add confidence concerning the suitability of the CYRM-28 as an ecological measure of resilience among vulnerable youth both in New Zealand and globally. It adds confidence that the CYRM-28 is a culturally sensitive measure of resilience that can assess multiple aspects of resilience resources. It has value as a tool that is particularly suitable to social work practice because it produces a multifaceted picture of the resilience resources around vulnerable youth which social workers can use to design interventions that build on strengths and directly address areas where resources are missing.

Footnotes

Acknowledgments

The researchers gratefully thank all the young people who made a commitment to this study, who responded so positively to our regular contact with them and for their generosity in sharing the details of their lives with us. We also acknowledge the contribution of the Donald Beasley Institute, Kapiti Youth Support, Youthline NZ and all the researchers who have assisted with the study.

Declaration of Conflicting Interests

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

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: We would also like to thank the Ministry of Business, Innovation and Employment who has supported this research program by providing funding over many years.

References

Ahern

Kiehl

Sole

Byers

(2006). A review of instruments measuring resilience. Issues in Comprehensive Pediatric Nursing, 29, 103–125. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1080/01460860600677643

Jackson Foster

Pecora

Delaney

Wenceslao

(2013). Reshaping child welfare’s response to trauma: Assessment, evidence-based intervention, and new research perspectives. Research on Social Work Practice, 6, 651–668. doi:10.1177/1049731513491835

Alvord

Grados

(2005). Enhancing resilience in children: A proactive approach. Professional Psychology: Research and Practice, 36, 238–245. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1037/0735-7028.36.3.238

Anthony

E. J.

(1974). The syndrome of the psychologically invulnerable child. In Anthony

E. J.

Koupernik

(Eds.), The child and his family (Children at psychiatric risk , Vol. 3, pp. 529–544). New York, NY: Wiley.

Bartlett

(1954). A note on the multiplying factors for various chi square approximations. Journal of the Royal Statistical Society, Series B, 16, 296–298.

Bovan

Harland

Grace

(2011). More ladders, fewer snakes: Two proposals to reduce youth disadvantage. Auckland: The New Zealand Institute.

Bowen

(2008). Naturalistic inquiry and the saturation concept: A research note. Qualitative Research, 8, 137–152.

Brooks

(2006). Strengthening resilience in children and youths: Maximizing opportunities in the schools. Children and Schools, 28, 69–76. doi:10.1093/cs/28.2.69

Cattell

(1966). The Scree test for the number of factors. Multivariate Behavioural Research, 1, 245–276.

10.

Cicchetti

(2003). Forward. In Luthar

(Ed.), Resilience and vulnerability: Adaption in the context of childhood adversities (pp. xix–xxvii) . Cambridge, England: Cambridge University Press.

11.

Connor

Davidson

(2003). Development of a new resilience scale: The Connor–Davidson resilience scale (CD-RISC). Depression and Anxiety, 18, 76–82. doi:10.1002/da.10113

12.

Cormack

Robson

(2010). Classification and output of multiple ethnicities: Issues for monitoring Māori health. Wellington: Te Rōpū Rangahau Hauora a Eru Pōmare.

13.

Daigenault

Dion

Hébert

McDuff

Collin-Vézina

(2013). Psychometric properties of the Child and Youth Resilience Measure (CYRM-28) among samples of French Canadian youth. Child Abuse and Neglect, 37, 160–171. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1016/j.chiabu.2012.06.004

14.

Diener

Emmons

Larsen

Griffin

(1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71–75.

15.

Education Counts. (2000, 2003, 2006, 2009). Programme for international student assessment, various authors. Retrieved from http://www.educationcounts.govt.nz/publications/series/2543

16.

Enders

(2010). Applied missing data analysis. New York, NY: The Guildford Press.

17.

Friborg

Hjemdal

Rosenvinge

J. H.

Martinussen

(2003). A new rating scale for adult resilience: What are the central protective resources behind healthy adjustment? International Journal of Methods in Psychiatric Research, 12, 65–76.

18.

Goodman

(1997). The strengths and difficulties questionnaire: A research note. Journal of Child Psychology and Psychiatry, 38, 581–586. doi:10.1111/j.1469-7610.1997.tb01545.x

19.

Goodman

(2001). Psychometric properties of the strengths and difficulties questionnaire. Journal of the American Academy of Child and Adolescent Psychiatry, 40, 1337–1345. doi:https://dx-doi-org.web.bisu.edu.cn/10.1097/00004583-200111000-00015

20.

Gunnar

(2006). Social regulation of stress in early child development. In McCatney

Philips

(Eds.), Blackwell handbook of early childhood development (pp. 106–125). Malden, MA: Blackwell.

21.

Hjemdal

Friborg

Martinussen

Rosenvinge

(2001). Preliminary results from the development and validation of a Norwegian scale for measuring adult resilience. Journal of the Norwegian Psychological Association, 38, 310–317.

22.

Bentler

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1–55. doi:10.1080/10705519909540118

23.

Hutcheson

Sofroniou

(1999). The multi-variate scientist. London, England: Sage.

24.

Johnson

(2013). She’ll be right: A state of the nation report from the salvation army. Auckland, New Zealand: The Salvation Army, Social Policy and Parliamentary Unit.

25.

Kaiser

(1970). A second generation Little Jiffy. Psychometrika, 35, 401–415.

26.

Kaiser

(1974). An index of factorial simplicity. Psychometrika, 39, 31–36.

27.

Liebenberg

Ungar

Van de Vijver

(2012). Validation of the child and youth resilience measure-28 (CYRM-28) among Canadian youth. Research on Social Work Practice, 22, 219–226. doi:10.1177/1049731511428619

28.

Luthar

(2006). Resilience in development: A synthesis of research across five decades. In Cicchetti

Cohen

(Eds.), Developmental psychopathology: Vol. 3. Risk, disorder and adaptation (2nd ed., pp. 739–795). New York, NY: Wiley.

29.

Luthar

Cicchetti

Becker

(2000). The construct of resilience: A critical evaluation and guidelines for future work. Child Development, 71, 543–562.

30.

Macfarlane

(2004). Kia hiwa ra! Listen to Culture: Māori students plea to educators. Wellington, New Zealand: NZCER Press.

31.

Madsen

Abell

(2010). Trauma, resilience scale: Validation of protective factors associated with adaptation following violence. Research on Social Work Practice, 20, 223–233. doi:10.1177/1049731509347853

32.

Māori Youth Council. (2011). In their own words! The report of the Māori Youth Council. Wellington, New Zealand: The Māori Youth Council.

33.

Masten

(1999). Resilience comes of age: Reflections on the past and outlook for the next generation of research. In Glantz

Johnson

Huffman

(Eds.), Resilience and development: Positive life adaptations (pp. 282–296). New York, NY: Plenum.

34.

Masten

(2007). Resilience in developing systems: Progress and promise as the fourth wave rises. Development and Psychopathology, 19, 921–930. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1017/S0954579407000442

35.

Masten

(2011). Resilience in children threatened by extreme adversity: Framework for research, practice, and translational synergy. Development and Psychopathology, 23, 493–506. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1017/S0954579411000198

36.

Masten

Best

Garmezy

(1991). Resilience and development: Contributions from the study of children who overcome adversity. Development and Psychopathology, 2, 425–444.

37.

Masten

Coatsworth

(1998). The development of competence in favorable and unfavorable environments: Lessons from research on successful children. American Psychologist, 53, 205–220.

38.

McLean

Wood

Breen

(2013). Reflecting on a difficult life: Narrative construction in vulnerable adolescents. Journal of Adolescent Research, 28, 431–452. doi:10.1177/0743558413484355

39.

Mila-Schaaf

(2010). Polycultural capital and the Pasifika second generation: Negotiating identities in diasporic spaces (Doctoral thesis). Massey University, Albany, New Zealand. Retrieved from http://mro.massey.ac.nz/handle/10179/1713

40.

O’Dougherty Wright

Masten

Narayan

(2013). Resilience processes in development: Four waves of research on positive adaptation in the face of adversity. In Goldstein

Brooks

(Eds.), Handbook of resilience in children (2nd ed., pp. 15–37). New York, NY: Springer.

41.

Owen

(2001). Whanake rangatahi: Programmes and services to address Māori youth offending. Social Policy Journal of New Zealand, 16, 175–190.

42.

Pallant

(2007). SPSS survival manual. Crows Nest, Australia: Allen and Unwin.

43.

Papoutsaki

Strickland

(2009, 6). Diasporic islands: Communicating Pacific cultural identities in diaspora. Paper presented at the 5th International Conference on Small Island Cultures, Sado Island, Japan. Retrieved from http://hdl.handle.net/10652/1743

44.

Perry

(2014). Household incomes in New Zealand: Trends in indicators of inequality and hardship 1982 to 2012. Revised Tables and Figures 27 February 2014. Wellington, New Zealand: Ministry of Social Development.

45.

Pines

(1975, 12). In praise of “invulnerables.” APA Monitor, p. 7.

46.

Rak

Patterson

(1996). Promoting resilience in at-risk children. Journal of Counseling and Development, 74, 368–373.

47.

Rutter

(2000). Resilience reconsidered: Conceptual considerations, empirical findings, and policy implications. In Shonkoff

Meisels

(Eds.), Handbook of early intervention (2nd ed., pp. 651–681). New York, NY: Cambridge University Press.

48.

Ruvalcaba-Romero

N. A.

Gallegos-Guajardo

Villegas-Guinea

(2015). Validation of the resilience scale for adolescents (READ) in Mexico. Journal of Behaviour, Health & Social Issues, 6, 21–34.

49.

Sanders

Munford

Liebenberg

Ungar

(2014). Peer paradox: The tensions that peer relationships raise for vulnerable young people. Child and Family Social Work. doi:10.1111/cfs.12188

50.

Schreiber

Nora

Stage

Barlow

King

(2006). Reporting structural equation modelling and confirmatory factor analysis results: A review. The Journal of Educational Research, 99, 323–338.

51.

Tabachnick

Fidell

(2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn & Bacon.

52.

Terwee

Bot

de Boer

van der Windt

Knol

Dekker

… de Vet

(2007). Quality criteria were proposed for measurement properties of health status questionnaires. Journal of Clinical Epidemiology, 60, 34–42. doi:10.1016/j.jclinepi.2006.03.012

53.

Thurstone

(1947). Multiple factor analysis. Chicago, IL: University of Chicago Press.

54.

Ungar

(2008). Resilience across cultures. British Journal of Social Work, 38, 218–235. doi:10.1093/bjsw/bcl343

55.

Ungar

(2011). The social ecology of resilience: Addressing contextual and cultural ambiguity of a nascent construct. American Journal of Orthopsychiatry, 81, 1–17. doi:10.1111/j.1939-0025.2010.01067.x

56.

Ungar

Liebenberg

(2011). Assessing resilience across cultures using mixed methods: Construction of the child and youth resilience measure. Journal of Multiple Methods in Research, 5, 126–149. doi:10.1177/1558689811400607

57.

Ungar

Liebenberg

Boothroyd

Kwong

Lee

Leblanc

… Makhnach

(2008). The study of youth resilience across cultures: Lessons from a pilot study of measurement development. Research in Human Development, 5, 166–180. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1080/15427600802274019

58.

Ungar

Liebenberg

Dudding

Armstrong

Van de Vijver

(2012). Patterns of service use, individual and contextual risk factors, and resilience among adolescents using multiple psychosocial services. Child Abuse and Neglect. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1016/j.chiabu.2012.05.007

59.

Werner

(1986). The concept of risk from a developmental perspective. Advances in Special Education, 5, 1–23.

60.

Windle

Bennett

Noyes

(2011). A methodological review of resilience measurement scales. Health and Quality of Life Outcomes, 9. Retrieved from http://www.hqlo.com/content/9/1/8

61.

Zahradnik

Stewart

O’Connor

Stevens

Ungar

Wekerle

(2010). Resilience moderates the relationship between exposure to violence and posttraumatic reexperiencing in Mi’kmaq youth. International Journal of Mental Health and Addiction, 8, 408–420. doi:10.1007/s11469-009-9228-y

62.

Zolkoski

Bullock

(2012). Resilience in children and youth: A review. Children and Youth Services Review, 34, 2295–2303. Retrieved from https://dx-doi-org.web.bisu.edu.cn/10.1016/j.childyouth.2012.08.009