RESILIENT ENGLISH SPEAKING CANADIANS:
PREDICTORS FROM THE NATIONAL LONGITUDINAL SURVEY OF CHILDREN
AND YOUTH
by

Vivian Houg
B.A., The King' s University College, 2010

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
OF MASTER OF SCIENCE
IN
PSYCHOLOGY

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
December, 2015

© Vivian Houg, 2015

Abstract

Resilience research has been growing several decades but has not addressed a national Canadian
population. This study constructed a Structural Equation Model (SEM) of resilience using
secondary data from the National Longitudinal Survey of Children and Youth. The sample used
for analysis consisted of English speaking Canadians born in 1987 and 1988. The model
analyzed two groups: Primary (642; 332 female, 310 male) and Supplemental (298; 146 female,
152 male). The model contained measures for Biological Sex and Family Income (ages 7-8),
Praise, Quality Time, and Verbal Abuse (ages 9-10), Physical Abuse and Parental Alcohol Use
(ages 11-12), Community Size, close relationships, and anti-social behavior (ages 13-14), and
Optimism and Problem Solving (ages 21-22). Results suggest that resilience functions for
Canadians as it does for previously studied American, European, and Asian populations and
underscores the contributions of close relationships throughout childhood and positive Problem
Solving skills in adolescence for Resilience in young adulthood.

11

TABLE OF CONTENTS
Abstract
Table of Contents
List of Tables
List of Figures
Acknowledgment
Introduction
Literature Overview

11

lll
IV
V
Vl

1

Origins and Definitions
Theoretical Frameworks
Protective Factors: Individual, Family, and Community
Explained Variance
Summary

2
4
5
8
10

The National Longitudinal Survey of Children and Youth
(NLSCY)
Structural Equation Modeling (SEM)
Measurement
Latent Variable Indicators
Variables for Analysis
Weighting and Missing Data
Sample Population for the Present Study

17
18
22
22

Modeling Procedure

Listwise vs. Pairwise Covariance Matrices
NLSCY Original Cohort Complete Observations Model

24
25

Results

Primary Group (Complete Data) Model
Stacked Model
Model Strengths
Model Limitations

30
36
38
39

Discussion

Middle Childhood
Late Childhood
Early Adolescence
Emerging Adulthood
Implications for Future Research
Conclusion

43
44
47
50
52
54
83
98

Method

References
Appendix A
Appendix B

Lisrel Syntax Primary Group Non-Significant Model
Lisrel Syntax Primary and Supplemental Groups
Non-Significant Model

11

12
13

99
lll

List of Tables
Table 1

Key Protective Factors identified by Benzies and Myachesiuk (2009)

56

Table 2

Exogenous and Endogenous Variables used in calculating the
Covariance Matrix

57

Fixed Error Coefficients for the Primary Group and the Supplemental
Group

60

Table 4

Frequencies for Collapsed Exogenous and Endogenous Indicators

61

Table 5

Descriptive Statistics for the Primary Group and the Supplemental
Group

63

Direct Effects Endogenous Pathway Estimates for the Model
in Figure 6

64

Table 7

Direct Exogenous Pathway Estimates for the Model in Figure 6

65

Table 8

Total Indirect Endogenous Pathway Estimates for the Model
in Figure 6

66

Table 9

Total Indirect Exogenous Pathway Estimates for the Model in Figure 6

67

Table 10

Stacked Model Direct Endogenous Pathway Estimates for the
Model in Figure 7

68

Stacked Model Direct Exogenous Pathway Estimates for the
Model in Figure 7

69

Stacked Model Total Indirect Endogenous Pathway Estimates for the
Model in Figure 7

71

Stacked Model Total Indirect Exogenous Pathway Estimates for the
Model in Figure 7

72

Stacked Model Endogenous Common Metric Standardized Solution for
the Model in Figure 7

74

Stacked Model Exogenous Common Metric Standardized Solution for
the Model in Figure 7

75

Table 3

Table 6

Table 11

Table 12

Table 13

Table 14

Table 15

IV

List of Figures
Figure 1

Proposed SEM Model.

76

Figure 2

Proposed SEM model removal of Handle Problems variable.

77

Figure 3

Proposed SEM model with single indicators.

78

Figure 4

Proposed SEM model with Optimism as a cause of Resilience.

79

Figure 5

Proposed SEM model with additional pathways from Friendships to
Reactivity and from Quality Time to lmpulsivity.

80

Proposed SEM model with the additional pathway from Verbal Abuse
to Coping Skills (hostility) and reciprocal effects between Resilience
and Problem Solving held equal.

81

Proposed SEM model with two groups.

82

Figure 6

Figure 7

V

Acknowledgment and Dedication
I would like to express my gratitude to several individuals as without their support the
completion of my thesis would not have been possible. First and foremost, I would like to
express my appreciation for my supervisor Dr. Cindy Hardy for her ongoing guidance,
patience and support. I would also like to thank Dr. Leslie Hayduk for his guidance in the use of
Structural Equation Modeling. Thanks are also due to the other members of my supervisory
committee Dr. Sherry Beaumont and Dr. Russell Callaghan.
I would also like to express my sincere thanks to Statistics Canada for providing the
opportunity for students to analyze rich sources of data through their Research Data Centers
Program (RDC). The research and analyses are based on data from Statistics Canada and the
opinions expressed do not represent the views of Statistics Canada. I am also grateful to Irene
Wong at the RDC at the University of Alberta for her very friendly and helpful service.
Additionally, I must thank the Social Sciences and Humanities Research Council and
UNBC for their financial support.
Last but not least I thank my husband and best friend Marc Johnson for his love,
support and encouragement throughout my education and in life. I also express my sincerest
gratitude to Sarah and David Thomson and their dog Jack.

Vl

Introduction
The focus of this project was the relationships between variables considered to be
protective factors related to resilience when applied to Canadian children and youth. The present
study examined individual, family and community level variables in a longitudinal Canadian
sample. Previous longitudinal studies regarding resilience have focused on a limited number of
variables at a time (Benzies & Mychasiuk, 2009), whereas this study examines these variables
simultaneously. Additionally, research has focused on predominantly urban populations in major
American cities. Some studies have examined resilience within the Canadian Armed Forces
(Skomorovsky & Stevens, 2013; Lee, Sudom, & McCreary, 2011) however, currently no
literature exists in relation to resilience in Canada at the civilian population level.
It is inappropriate to assume that the Canadian population is analogous to the populations

studied in the American samples despite the many similarities between the two populations.
Krueger, Bhaloo, and Rosenau (2009) suggest that health lifestyles are actually more similar for
Canadian and American populations living in close proximity to the Canada-United States border
than within each national population. However, Canadians still demonstrate lower fertility rates
and lower death rates when compared to Americans as a whole (Krueger, Bhaloo, & Rosenau;
2009). Barbieri and Oullette (2012) note that almost 80% of the Canadian population lives within
150 kilometers of the border with the United States whereas the population in the United States
is comparatively more evenly distributed. Additionally, the Canadian population demonstrates a
higher percentage of immigrants than the United States, especially from Asia and Africa
(Barbieri & Oullette, 2012). Barbieri and Oullette (2012) also point out that the probability of
dying at ages 15 to 65 is almost 30% higher in the United States than in Canada. According to
Hardwick, Marcus, and Isaak (2010) Canadian national identity, as represented in Social Studies

1

curricula, places a greater emphasis on collaborative communal behavior, multiculturalism,
multilingualism, and the importance of becoming a global citizen than curricula in the United
States. This study provides much needed insight into how resilience works as part of a
developmental system (Cudeck & du Toit, 2009) within a Canadian context and informs the
direction of future Canadian studies.

Literature Overview
Origins and Definitions
Origins. Resilience research stems from inquiry into the negative developmental
outcomes for at risk children that began in the 1970s (Luthar, Cicchetti, & Becker, 2000). Emmy
Werner's initial work with the Hawaiian longitudinal study of children born in 1955 on the
island of Kauai (Burt & Paysnick, 2012; Werner, Bierman, & French, 1971) identified at risk
children on the basis of poverty, parental mental illness, and the death of a parent (Aldwin,
Cunningham, & Taylor, 2010). Norman Garmezy' s research at the University of Minnesota with
vulnerable children examined the effects of parental schizophrenia on child development.
Garmezy' s work led to his involvement in the Project Competence longitudinal study (Burt &
Paysnick, 2012). Michael Rutter's early work focused on the influence of childhood
psychopathology on adolescent and adult development (Rutter, Greenfeld, & Lockyer, 1967;
Rutter & Lockyer, 1967). Other longitudinal studies that have contributed to resilience research
include the Minnesota Longitudinal Study of Risk and Adaptation, the Iowa Youth and Families
Project, the Dunedin Health and Development Study, and the Christchurch Health and
Development Study (Burt & Paysnick, 2012).The data generated by these research initiatives
suggested that despite extreme disadvantage many of the children in these studies achieved

2

normative developmental outcomes (Aldwin, Cunningham, & Taylor, 2010; Garmezy, 1974;
Luthar et al., 2000; Rolf & Garmezy, 1974; Rutter et al. , 1967; Werner et al. , 1971).
To date the only longitudinal study of a Canadian population that approximates the
American studies is the Concordia Longitudinal Risk Project (CLRP) which was initiated in
1976 to study children from French language schools in Quebec (Stack, 2013). The CLRP is
ongoing and has focused on individuals from low socio-economic urban backgrounds. Findings
from the CLRP have demonstrated that early life experiences influence outcomes in early to midadulthood (Serbin et al., 2010).

Definitions. The 30 years following these initial observations have distinguished
resilience research from developmental psychopathology as a distinct field of study in
psychology. To date inquiry has primarily focused on identifying and operationalizing the
characteristics of resilient individuals in comparison to their less resilient peers (Benzies &
Mychasiuk, 2009; Masten & Obradovic, 2006; Olsson, Bond, Burns, Vella-brodrick, & Sawyer,
2003). However, the literature continues to identify a need for an overarching conceptual
definition for resilience. Currently researchers have a tendency to define resilience in relation to
their given study rather than in reference to a generally agreed upon construct.
Despite the absence of formal consensus, the literature suggests two requirements for a
definition of resilience. First, it is clear that in order for an individual to be considered resilient
he or she must demonstrate positive adaptation or functioning in response to some kind of
adverse life event or circumstance (Aven, 2011; Davis, Luecken, & Lemery-Chalfant, 2009;
Easterbrooks, Driscoll, & Bartlett, 2008; Greeff & Merwe, 2004; Greeff & Du Toit, 2009;
Masten & Tellegen, 2012; McElwee, 2007; Roosa, 2000; Rutter, 2006, 2012; Shiner & Masten,
2012; Stewart, 2011 ; Zautra, Arewasikporn, & Davis, 2010). Second, it is important to recognize

3

that resilience is a dynamic process that arises out of the interaction of adverse experiences and
individual and environmental protective factors (Burt & Paysnick, 2012; Easterbrooks et al.,
2008; Greeff & Du Toit, 2009; Harney, 2007; Roosa, 2000; Rutter, 2006, 2012; Sameroff &
Rosenblum, 2006; Ward, Martin, Theron, & Distiller, 2007).
Anne Masten and Auke Tellegen have recently proposed that resilience be defined as
"the capacity of a dynamic system to withstand or recover from significant threats to its stability,
viability or development" (Masten & Tellegen, 2012, p.348). This definition provides a way to
focus and categorize research within an ecological framework: "cellular, neural, whole organism,
family and larger social and cultural systems" (Masten & Tellegen, 2012, p.348). The complex
nature ofresilience is reiterated by Unger (2012) and is supported by the overarching themes in
the literature.
Theoretical Frameworks
Interactionism. As early as 1951 Kurt Lewin advocated for the consideration of the
complex interaction between the individual and the environment in social science research
(Unger, 2012). Pangallo, Zibarras, Lewis, and Flaxman (2015) note that the interactionist
approach is helpful because it is based on the assumption that the individual and environment are
in constant relationship: "(a) individuals are not randomly assigned to the environments in which
they live but select and create their own experiences, and (b) environments can maintain
personality characteristics that initially developed in response to earlier socialization
experiences" (Pangallo et al. , 2015 , p. 3).
Bioecological model. Bronfenbrenner and Ceci (1994) offer three propositions that
characterize human development within a system:

4

1. Human development can be characterized by the interaction between the human
organism and his/her environment. Frequent interactions are considered proximal
whereas rare interactions are considered distal. Proximal interactions tend to exert greater
influence on human development than distal interactions.
2. Proximal processes are a joint function of the human organism and the environment and
must be considered together when assessing developmental outcomes.
3. Proximal processes influence the ability of the human organism to achieve effective
psychological development through the expression of genetic potential due to the
constraints of the human-environment interaction.

Reciprocity. The reciprocal influence of the individual on the environment and the
environment on the individual is reiterated by Roisman et al. (2004), Harney (2007) and
Easterbrooks et al. (2008). Their work emphasizes that over time the relationship between the
individual and the environment changes and broadens. Nevertheless, the reciprocity of that
relationship remains at the core of human development and shapes the course of developmental
outcomes. Current developmental theories such as developmental systems theory, the
transactional model, and organizational theory support the conceptualization of the child as a
component of a larger system which both influences and is influenced by the child (Easterbrooks
et al., 2008).

Protective Factors: Individual, Family, and Community
Roisman, et al. (2004) underscore that "the best predictors of adult outcomes from
childhood appear to be broad developmental- adaptational [sic] attributes ... that reflect
cumulative, age-graded success or failure in adaptive behaviors supported by the environment"
(p. 131 ). Recent reviews of the literature (Benzi es & Mychasiuk, 2009; Masten & Obradovic,

5

2006; Olsson et al. , 2003) support the use of a socio-ecological framework to understand how
protective factors function in relation to the individual. The socio-ecological framework assumes
that the individual is embedded within his or her environmental context and that the
environmental context has multiple levels (Bronfenbrenner, 1977; Calkins, Blandon, Williford,
& Keane, 2007; Hamey, 2007; Stewart, 2011).

Benzies and Mychasiuk (2009) provide a comprehensive meta-analysis of the literature.
Their review examined 40 resilience publications from 2000 to 2008 and identified 24 key
factors related to resilience. In accordance with the socio ecological model (Bronfenbrenner,
1977) the authors divided these factors into three levels of influence. Each level represents a
unique set of variables that are most salient at one of three levels: the individual level, the family
level, or the community level (see Table 1).
Greeff and Du Toit (2009) identified similar protective factors in an independent study of
resilience in remarried families: "(1) supportive family relationships, (2) affirming and
supportive communication, (3) a sense of control over outcomes in life, (4) activities and
routines that help the family to spend time together, (5) a strong marriage relationship, (6)
support from family and friends, (7) redefining stressful events and acquiring social support, and
(8) spirituality and religion within the family" (p. 114).
There has been some focus on the importance of individual characteristics in relation to
resilience, especially in relation to personality and neuroscience (Fletcher & Sarkar, 2013;
Karatoreos & McEwen, 2013). Nasvytiene, Lazdauskas, and Leonaviciene (2012) confirm that
individual characteristics have a slightly stronger relationship to resilience. However, this is
possibly due to the prevalence of studies focusing on individual characteristics in the literature
compared to family and community factors .

6

Biological sex differences. Biological sex differences in resilience are frequently noted.
According to Boardmen, Blalock, and Button (2008) these differences are an example of geneenvironment interaction because biological sex implies, in addition to hormonal and genetic
differences, the experience of different social environments for men and women making it an
environmental moderator of latent genetic factors. This is corroborated by Waaktaar and
Torgersen in their 2012 twin study which found that both biological sex and shared genetics
were significant predictors of differences in trait resilience.

Socio-economic status. The discussion of socio-economic status - disease mechanisms
is prevalent in epidemiological literature. From a developmental perspective socioeconomic
adversity impacts cognitive and socioemotional development, parent-child social interaction, and
parental ability to provide monitoring and social support (Kroenke, 2008). Starfield, Riley, Witt,
and Robertson (2002) provide an overview of socio-economic research and discuss a follow up
study of American adolescents ages 11 to 17. They found significant differences between social
class (lower $7300 per household member; middle per household member $11300; higher
$17100 per household member) in resilience, family involvement, problem solving, emotional
discomfort, and self-esteem after controlling for age, sex and rurality (Starfield et al. 2002).

Attachment. Early childhood attachment is considered a protective factor against many
material and psychological risks (Black-Hughes & Stacy; 2013). Schore (2001) goes so far as to
claim that early secure attachment has a significant impact on right brain development and the
connections from the right brain to the limbic and autonomic nervous system. Efficient right
brain function is expressed though flexibility in coping with novelty and stress (Schore, 2001 ).
In a comparative study of female inmates, and their comparatively resilient non-incarcerated
siblings, Black-Hughes and Stacy (2013) demonstrated that, compared to the inmate, the non-

7

incarcerated sibling reported higher attachment scores for mother, father, friend, and other adult.
Additionally, Karreman and Vingerhoets (2012) found that attachment style had a small
significant direct effect on resilience in a Dutch sample aged 16 to 67 years.
Childhood maltreatment. It is the consensus of the literature that childhood
maltreatment produces externalizing and internalizing problems regardless of the population
being studied (Lansford et al. , 2006; Villoda, 2015). Although impaired functioning is not
necessarily an inevitable result of childhood maltreatment only 10-25% of maltreated children
ever achieve a level of functioning that can be considered resilient (Cicchetti, 2013). Using the
Chicago Longitudinal Study Topitzes, Mersky, Dezen, and Reynolds (2013) found that young
adults who had experienced childhood maltreatment demonstrated significantly worse outcomes
across several measures including resilient functioning. According to Afifi and MacMillan
(2011) the only protective factors consistently identified to improve resilient outcomes for
maltreated children are stable family environment and supportive relationships.
Problem solving. The literature consistently provides evidence of a relationship between
problem solving and resilience (Benzies & Myachesiuk, 2009). Frydenburg and Lewis (2009)
used a cross-sectional sample of Australian youth and exploratory factor analysis to validate the
relationship between problem solving efficacy and productive coping. According to Li and Yang
(2009) resilience produced a significant medium effect on problem solving in Taiwanese college
students. Another study (Li, Eschenauer, & Yang, 2013) of college students from China, Taiwan,
and the United States found that trait resilience was a significant predictor of problem solving
ability.
Explained Variance

8

Unger (2012) notes that use of specifically child-focused theories of development have
accounted for less than half of the variance in the literature. In fact, many studies do not report
explained variance and for those that do Unger's (2012) criticism appears justified:
•

Masten et al. (2004) accounted for 22% of the variance in social competence

•

Li and Yang (2009) accounted for 25% of the variance in problem solving, 6% of the
variance in social support seeking, and 11 % of the variance in avoidance coping
responses

•

Englund et al. (2011) accounted for 39.5% of the variance in adaptive functioning

•

Waaktaar and Torgersen (2012) claim that genetics accounted for almost 25% of the
variance in resilience

•

Karreman and Vingerhoets (2012) states that attachment style accounts for 32.8% of the
variance in resilience in the sample analyzed.
Shiner and Masten's (2012) work on personality and resilience suggest that individual

protective factors may be less malleable than family and community level factors because of the
relative stability of personality traits. Despite the stability of personality traits, the overall
contribution of individual factors to resilience is unclear. In fact, according to Stewart (2011)
family level factors related to stability and caring relationships overshadowed the effects of
individual psychological factors in her review of 27 publications. Harney (2007) and Afifi and
MacMillan (2011) also note this superseding trend of family and community factors over
individual factors in resilient outcomes. Tiet, Huizinga, and Byrnes (2009) even go so far as to
suggest a hierarchy of influence placing community factors over family and individual factors in
relation to resilience. Davis et al. (2009) recognize that family and community factors have a

9

significant influence on the development of the children they encompass especially in the context
of early childhood (Hamey, 2007).
Olsson and colleagues (2003) acknowledge that resilience comes about in a complex
system of risks and protective factors that interact with each other across the lifespan. Resilience
is not the result of one developmental path but can arise out of multiple developmental
interactions depending on the individual and the environment (Calkins et al., 2007; Rutter,
2012).
Summary
Resilience is a developmental construct that is rooted in the developmental systems
theory and has grown into an independent field of research in the last 30 years. Inquiry has
focused predominantly on urban American populations and has identified multiple factors that
increase or reduce the likelihood of resilience which is defined as positive adaptation in response
to some kind of adverse life event (Aven, 2011; Davis et al., 2009; Easterbrooks et al., 2008;
Greeff & Merwe, 2004; Greeff & Du Toit, 2009; Masten & Tellegen, 2012; McElwee, 2007;
Roosa, 2000; Rutter, 2006, 2012; Shiner & Masten, 2012; Stewart, 2011; Zautra et al., 2010).
To date there are no studies that address resilience with a Canadian population. The
present study addresses this deficiency in the literature by examining a longitudinal Canadian
cohort. Whereas most longitudinal studies employ repeated measures of a limited number of
variables, this study examined several different variables sampled during consecutive
developmental periods that are congruent with meta-analyses of the literature performed by
Greeff and Du Toit (2009) and Benzies and Mychasiuk (2009): (a) Biological Sex and Family
Income ( ages 7 to 8), (b) Verbal Praise, Quality Time, and Verbal Abuse (ages 9 to 10), (c)
Physical Abuse and Parental Alcohol Use (ages 11 to 12), (d) Community Size, Relationship

10

with Parents, Friendship, Impulsivity, Coping Skills (hostility), and Reactivity (ages 13 to 14),
and (e) Optimism and Problem Solving Skills (ages 21 to 22). This method of analysis
emphasizes the importance of the functional relationship of the individual, time, and
environment in explaining resilience as part of a larger developmental system (Cudeck & du
Toit, 2009).
Method
The National Longitudinal Survey of Children and Youth (NLSCY)
Description. The National Longitudinal Survey of Children and Youth is a longitudinal

study of Canadian children designed to address the deficiency of Canadian data regarding
children's social, emotional and behavioural development from birth through early adulthood.
The NLSCY was instituted in 1994/1995 with a sample of children ages O to 11 years (Statistics
Canada, 2010c). Sampling was achieved using the Labour Force Survey and is thought to be
representative of the non-institutionalized civilian population of Canada's 10 provinces at the
time. Children living on reservations and remote regions, including the territories, were not
included (Statistics Canada, 2010c). The NLSCY is now complete and consists of eight Cycles
of data collection. After the initial data collection, the original cohort participated in a follow up
interview every two years. The current study utilizes cycles l, 2, 3, 4, and 8.
Data collection. Participation in the NLSCY was voluntary and confidential. Data were

collected directly from participants using written response paper questionnaires and verbal
response computer-assisted interviewing (CAI). Prior to data collection questionnaires
underwent scrutiny by an expert advisory group and were tested in focus groups and pilot
surveys (Statistics Canada, 201 Oa). The questionnaire consisted of questions that elicited data
from the child or youth (youth component and self-completed questionnaire), the person most

11

knowledgeable about the child (PMK) (child component and adult component), and the spouse
of the PMK if applicable (adult component). In addition, a variety of cognitive tests were
administered to children and youth depending on their age (Statistics Canada, 2010a).
Structural Equation Modeling (SEM)
Experimental manipulation of the factors that explain resilience, specifically adverse
experiences including abuse and neglect, is not ethically appropriate (Canadian Psychological
Association, 2000). As such research relies on survey methodology which is inherently
observational and cannot address causality. However, according to Saris (1999), these problems
could be potentially overcome by employing the statistical technique of SEM; "[SEM] is widely
and intensively used to derive causal inferences using data originating from non-experimental
research" (p. 221 ). Saris (1999) presents a very optimistic evaluation of the capabilities of SEM.
However, by applying SEM to the NLSCY data this study attempted to develop evidence for a
theoretical statement about the mechanisms of resilience within this sample of the Canadian
population that extends beyond the capabilities of conventional survey methods (Greenhoot &
Dowsett, 2012).
Confirmatory or exploratory. According to Hayduk (1987), "if a theory, knowledge of
the data collection procedures, and the covariance go in, substantive findings come out" (p. xv).
He emphasizes the use of SEM as a confirmatory analysis where the researcher has intentionally
constructed a model to test the relationships between the concepts of interest (Hayduk, 1987).
This theory driven approach is helpful in its ability to identify the specific implications of the
theoretical statement a model makes (Hayduk, 1996). However, this view of SEM also entails
the possibility that the theoretical statement made by a model may not match what is observed.

In cases of failure to fit the data the model may bring into question the theory and the literature

12

that initially informed its creation (Hayduk, 1996). This is confirmed by Cudeck and du Toit's
(2009) assertion that mathematical modeling "makes a theory explicit, which minimizes
ambiguities and encourages a critical analysis of its parts, and that it can explain data, which
implies that it is a possible description of behavior" (p. 516).
The structural equation model, and its modifications, constructed for this study took
careful consideration of these inherent features of SEM and proceeded appropriately at each step
of analysis (see Procedure).
Measurement
Composite scores vs. individual items. The structural equation modelling (SEM)
literature is divided over the use of composite scores as indicators of latent variables (Hayduk &
Littvay, 2012; Hayduk, 1996; 1987). The literature identifies four main concerns regarding the
use of composite scores (Yang, Nay, & Hoyle, 2010). First, Marsh and O'Neill (1984) note that
composite scores can result in a loss of information about the individual items on a
questionnaire. Second, Bollen and Lennox (1991) link the potential loss of information to the
assumption that each item carries equal weight when combined into a composite score. Third,
Coanders, Satorra, and Saris (1997) warn that the use of composite scores could influence the
relationship between the indicator and the latent variable by transforming the linear or nonlinear
nature of the questionnaire item. Finally, Wright (1999) reasons that a composite score
calculated from binary or trichotomous questionnaire items would have a severely limited range
potentially biasing the variance and covariance estimates of the latent variable.
Some of the variables in the NLSCY are derived scores (i.e. Family Functioning Score;
Friends Score). These composite scores were calculated from individual questions according to
prior analysis. All scales used in the NLSCY were preexisting in the literature and were further

13

verified by a three step procedure prior to implementation: factor analysis, score calculation
based on factor structure, and reliability measurements. Consequently, the NLSCY provides a
Cronbach' s Alpha for each scale used in the survey (Statistics Canada, 201 Oc).
However, given the contention on the use of composite scores as indicators this study
used scores from individual questions on the NLSCY as indicators of latent variables.
Consequently, the Cronbach's alpha was not applicable to the individual items selected for the
present study. The use of individual items provided protection from the potential biasing effects
of composite scores (Hayduk, 1996) and allowed the model to make a clear theoretical statement
(Hayduk, 1987).

Single vs. multiple indicators. The SEM literature is also contentious regarding the use
of single or multiple indicators for latent variables (Hayduk & Littvay, 2012; Hayduk, 1987;
1996). Latent variables represent the "true score" free from measurement error and respondent
bias. Whether a latent has one indicator or many should not change the "trueness" of the latent
score. Consequently, Hayduk and Littvay (2012) argue against the excessive use of multiple
indicators (three or more indicators per latent) as "more indicators ... do not necessarily mean
better latents, they mean more entrenched latents -where the entrenchment is provided by the
indicators, with the possible sacrifice of appropriate latent-level causal connectivity" (p 10).
Observed data is unlikely to respond to the variance of a latent construct in equal or proportional
parts. Modelling latent variables with multiple indicators creates a mathematical expectation in
the estimated covariance matrix (L) that is rarely found in the real world (Hayduk 1987).
Minimizing the number of indicators for a latent allows the researcher to use only the strongest
of the available indicators. This allows for the exclusion of weaker indicators that could
potentially cause the model to fail (Hayduk, 1996).

14

Fixed error variance coefficients. According to Hayduk (1987) it is preferable to fix the
proportion of error variance for each indicator rather than allow it to be free. The practice
regarding error variances allows the researcher to include information about data measurement
quality that would otherwise be lost in the model. Additionally, indicators may or may not be
close representations of the latents that influence them. Fixing the error variance coefficient
allows the variance in the indicator thought to arise from sources other than the latent to be
partitioned out and explicitly represented in the model.
The assignment of error variance accounts for the best case scenario of error variance
(respondents were truthful, viewed the question the same way as the researcher, and were careful
to select the appropriate response on the survey; the surveyors made no data entry mistakes) and
the worst case scenario of error variance (respondents were less truthful, didn' t view the question
the same way the researcher did, and did not pay attention to which response they chose on the
survey; the surveyors made many data entry mistakes) in accordance with the "half-double" rule.
An intermediate proportion of error variance was estimated and used in the Lisrel model (See
Table 3).
Common Factors or Phantom Variables. The NLSCY does not provide a specific
measurement of Resilience to use as an indicator. It is possible, given the literature, to model
resilience as a composite of Impulsivity, Coping Skills (hostility), Problem Solving Skills and
Optimism all of which can be defined by individual questions pulled from the NLSCY. Benzies
and Mychasiuk (2009) emphasize the emotional regulation and effective coping skills of resilient
individuals and Masten and Tellegen (2012) call attention to the ability ofresilient individuals to
withstand or overcome obstacles.

15

However, modelling Resilience as a composite implies correlation between the indicators
of hnpulsivity, Coping Skills (hostility), Problem Solving Skills and Optimism because
Resilience is conceptualized as a common cause, which is not the intent of a composite variable
(Hayduk, 1987). Alternatively, SEM allows for latent variables to be modeled without indicators
(Hayduk, 1987) - sometimes called phantom variables (Hayduk, 1996) - to place Resilience
within the causal framework of the latent variables . Consequently, resilience can be modeled as
an intermediary with no mathematical requirement for correlation between the measured latent
variables. This study initially postulated that Resilience results from an individual ' s level of
hnpulsivity and Coping Skills (hostility) which then results in an individual ' s level of Optimism
and Problem Solving ability (see Figure 1).
However, because Optimism influences how events are perceived and interpreted it was
decided that Optimism fit in the model better as a cause, rather than an effect, of Resilience both
conceptually and statistically (see Figure 4).

16

Latent Variable Indicators
Measured variables were chosen as indicators of the latent variables based on two
criteria: content and variance. Variables were first chosen based on the relevance of their content
to the latent variable. Then the variable, or variables, with the greatest variance were selected to
act as an indicator for the latent variables in the model.

Collapsed categories. To preserve confidentiality some scales were collapsed due to low
cell counts before the data could be released by Statistics Canada (Statistics Canada, 2010b;
Statistics Act, R .S.C., 1985, c. S-19): Optimism, Problem Solving, Confidence, Physical Abuse,
Verbal Abuse, Income Adequacy, Verbal Praise, and Quality Time (see Table 4).
The indicators for Relationship with Mother and Relationship with Father allowed
participants to respond with a 6 indicating "not applicable". Category 6 demonstrated a low cell
count and was collapsed into category 3 to maintain an expression of no relationship in the
measure of parental closeness (see Table 4).

Error variance. The fixed error variances can be found in Table 3. It is generally
preferable to fix the proportion of error variance for each indicator rather than allow it to be free.
This allows the researcher to include information about data measurement quality that would
otherwise be lost in the model (Hayduk, 1987). Additionally, indicators may or may not be close
representations of the latents that influence them. Fixing the error variance coefficient allows the
variance in the indicator thought to arise from sources other than the latent to be partitioned out
and explicitly represented in the model.
In most cases it was assumed that the researcher understood the questions and response
options in the same way the respondents did. In accordance with the "half-double" rule (Hayduk,
1987) an intermediate proportion of error variance was estimated and used in the Lisrel model.

17

As a result, most of the fixed error variances are quite small. Exceptions to this are discussed
below.
Inverse scales. Some of the scales for variables in the NLSCY were measured, by
interviewers in the original data collection, in reverse in the sense that big numbers actually
indicate low amounts or frequencies and small numbers actually indicate high amounts or
frequencies (see Table 2). Scales' directions were not transformed for this study and inverse
scales are noted in the text where relevant.
Variables for Analysis
Exogenous or endogenous. An exogenous variable is "any latent variable that does not
have a [path] pointing to it from another latent variable" (Shumacker & Lomax, 2010, p. 181).
An endogenous variable is "any latent that is predicted by other latent variables in a structural
equation model" (Shumacker & Lomax, 2010, p. 180).
Abuse. Abuse is addressed in the NLSCY in two ways: Physical Abuse, both witnessed
and experienced, and verbal abuse. Initially both Physical and Verbal Abuse were used as
indicators for the latent variable abuse however, after the initial run of the model it was decided
that these indicators were not influenced by the same latent variable and were consequently
divided into two different latent variables (see Figure 2).
Physical Abuse. The Cycle 4 child questionnaire allows the participants to report on their
own experience of physical abuse: "My parents (or step parents or foster parents or guardians)
hit me or threaten to do so" (see Table 2) (Statistics Canada, 2004, p. 108) The fixed error
variance was set at 40% of the variance to account for the sensitive nature of the question and the
high likelihood of the participant to minimize the extent of their experience of abuse to the

18

interviewer (see Table 3) resulting in a poor measure of the true score associated with Physical
Abuse.

Verbal Abuse. The Cycle 2 parent questionnaire asks the PMK to report on the frequency
the participants experience verbal abuse: "How often do you tell <him/her> that <he/she> is bad
or not as good as others?" (Statistics Canada, 1998, p. 172)

Parental Alcohol Use. (inverse scale; see Table 2) The Cycle 3 parent questionnaire asks
the PMK to report the frequency of his or her own drinking: "During the past 12 months, how
often did <you/he/she> drink beer, wine, liquor or any other alcoholic beverage? PMK"
(Statistics Canada, 2000, p . 44) .
Sex. The Cycle 1 and Cycle 2 parent questionnaires ask the PMK to identify the
participant as either male or female (Statistics Canada, 1996, p. 6; Statistics Canada, 1998, p. 2).
This was converted into a numerical score with 1 indicating female and 2 indicating male.

Community size. (inverse scales; see Table 2) The NLSCY provides information on the
size of each participant's resident community based on the 1996 census counts (Statistics
Canada, 2004), the 2001 census counts (Statistics Canada, 2007), and the 2006 census counts
(Statistics Canada, 2010a). Participants were given a designation based on the place of residence
the participant occupied in two or more of the census Cycles.

Family income. Cycle 1 provides a measure of income adequacy based on how many
members are living in the household in relation to the overall household income (Statistics
Canada, 1996, p. 35).

Relationship with mother. (inverse scale; see Table 2) The My Parent(s) and Me
Section of Cycle 4 explicitly asks youth participants about the closeness of their relationship with
the mother figure they spend most time with (biological, adoptive, stepmother, or foster) :"

19

Overall, how would you describe your relationship with your mother?" (Statistics Canada, 2004,
p. 85).
Relationship with Father. (inverse scale; see Table 2) The My Parent(S) and Me
Section of Cycle 4 explicitly asks the respondents about the closeness of their relationship with
the father figure they spend most time with (biological, adoptive, stepmother, or foster):
"Overall, how would you describe your relationship with your father?" (Statistics Canada, 2004,
p. 93).
Friendships. (inverse scale; see Table 2) The Friends and Family Section of Cycle 4 asks
the child participants about the level of intimacy in their friendships: "How often do you share
your secrets and private feelings with your close friends?" (Statistics Canada, 2004, p. 12).
Supportive parenting. During the Child Component of the Cycle 2 NLSCY the PMK
answered 18 questions that were derived from an adaptation of the Parent Practices Scale used
by Strayhorn and Weidman with additional questions provided by Michael Boyle (ChedokeMcMaster Hospital) (Statistics Canada, 1995). Initially two indicators were used for the latent
variable Supportive Parenting however, after the initial run of the model it made more sense to
divide this latent into two different concepts: Verbal Praise and Quality Time.

Verbal praise. The Cycle 2 Child Component of the NLSCY asked the PMK to report the
frequency with which he or she provides positive verbal feedback to the participant: "How often
do you praise <name>, by saying something like 'Good for you!' or 'What a nice thing you did!'
or 'That' s good going!'?" (Statistics Canada, 1998, p. 171).

Quality Time. The Cycle 2 Child Component of the NLSCY asked the PMK to report the
frequency with which he or she engages in positive interactions with the participant: " How often

20

do you and <he/she> talk or play with each other, focusing attention on each other for five
minutes or more, just for fun?" (Statistics Canada, 1998, p. 171 ).
Impulsivity. The indicator for a participant' s level of impulsiveness comes from Cycle 4.
Cycle 4 participants were asked to rate their level of impulsiveness, "I am impulsive, I act
without thinking." (Statistics Canada, 2004, p. 56), using a three-point rating scale with 1
indicating "never or not true" and 3 indicating often or very true" (see Table 2).
Reactivity. Initially Reactivity was used as a secondary indicator for Coping Skills
(hostility), however after the initial run of the model it was decided that Reactivity would be
more useful in the model as a separate latent variable (see Figure 2). Cycle 4 participants were
asked to rate their level of reactivity, "I get into many fights" (Statistics Canada, 2004, p. 52).
Coping skills (hostility). In Cycle 4 respondents were asked about their behavior in
response to the actions of others: "When another kid accidentally hurts me I assume that he/she
meant to do it, and I react with anger and fighting" (Statistics Canada, 2004, p. 61). It is
important to note that this indicator does not capture the entire range of coping skills and is more
likely more accurately conceptualized as hostility. However, given the indicators available from
Cycle 4 of the NLSCY (Statistics Canada, 2004) the researcher determined it was better to
include an incomplete measure of coping than leaving the concept out of the model. The
implications of this decision are discussed in the Limitations section.
Optimism. The EQ-I (BarOn, 2004) was designed to measure emotional intelligence on
five dimensions : intrapersonal, interpersonal, adaptability, stress management and general mood.
The indicator for Optimism comes from a single question on the EQ-I measure used in Cycle 8.
Participants were asked to rate their level of optimism, " ... You're optimistic about most things
you do" (Statistics Canada, 201 Oa, p. 218).

21

Problem solving. An indicator of problem solving ability comes from the Abilities
section of the Cycle 8 questionnaire: "How would you rate your: ... ability to solve new
problems? For example, identifying problems and possible causes, planning strategies to solve
problems or thinking of new ways to solve problems [sic]" (Statistics Canada, 2010a, p. 208).

Confidence. Confidence was also probed on the Cycle 8 Abilities questionnaire: "Tell
me how you feel, think, or act most of the time in most situations: ... You believe in your ability
to handle most upsetting problems" (Statistics Canada, 2010a, p. 213).

Weighting and Missing Data
In order to ensure that the original longitudinal cohort remained representative of the
1994/1995 Canadian population Statistics Canada calculated funnel and non-funnel weights for
each participant in the NLSCY. Funnel weights were calculated for participants in the
longitudinal cohort that had responded to all Cycles of data collection. Non-funnel weights were
calculated for participants in the longitudinal cohort who had responded to the current Cycle of
data collection but not all of the previous Cycles (Statistics Canada, 2010a). This study employed
the Cycle 8 longitudinal funnel weights for analysis.
Some participants did not respond for a given Cycle. Additionally, of those who
responded not all provided complete information. When data was missing for a participant it was
imputed only for adult income, youth income, household income and Motor and Social
Development items. All other non-responses were coded as such (Statistics Canada, 2010a). The
only item used in the present study that may have been influenced by imputation is Income
Adequacy.

Sample Population for the Present Study

22

Data screening. The data underwent an initial screening to verify that the chosen
variables were appropriate for the use of SEM. Preference was given to indicators that
demonstrated the greatest amount of variance within the sample population. The appropriate
bootstrap weights were applied to each group using Wesvar (Westat, Inc., 2014) to ensure that
the sample remained representative of the Canadian population in 1994/1995.

Selection criteria. In order to avoid the confound of potential differences in the French
and English versions of the NLSCY questions only data from respondents who answered the
NLSCY questions in English was included in the analysis. Consequently, the results do not
represent the significant French speaking populations in Quebec and New Brunswick.
NLSCY response rates are highly variable between Cycles and questions. Only a small
subset of the longitudinal cohort was found to have responded to all the questions of interest to
the present study. According to Carillo, Kovacevic, and Wu (2006) the probability of dropping
out or missing one visit in the NLSCY dependended on variables including age, gender,
depression, level of school completed by the PMK, the region of residence, the urban-rural
status, the child ' s parent status (family status), the household income status, and the number of
hours in daycare. Given these factors it is possible that the subset used for the current study does
not represent the true breadth of the Canadian experience. The issue of missing data was
addressed by placing respondents into groups based on the number of complete responses
available for the variables of interest. The Primary Group was required to have valid responses
for all indicators used in the model and was used to find a model with acceptable fit before
attempting a stacked model.
The stacked model included a Supplemental Group. In order to maximize the number of
participants included in the Supplemental Group IBM SPSS 22 (IBM Corp. , 2013) was used to

23

produce frequencies for each variable. Participants were then sorted according to the name of the
variable that was missing. The participants selected for the Supplemental Group adhered to a
specific pattern of missing data (Wothke, 1993) and represent a balance between the number of
missing variables and number of qualifying participants. Supplemental Group participants were
missing data on nine variables: Parents Hit/Threaten, Share Secrets with Friends, How Close to
Mother, and How Close to Father, Difficulty Waiting, Get into fights, React Anger/Fighting, and
Handle Upsetting Problems. Calculating separate covariance matrices for each group allowed the
design of a multi-group model that includes incomplete observations and models missing data
processes as potential group differences in the model to be analyzed.

Primary group. The complete observations consisted of respondents who were all born
in 1987 and 1988. There was one exception, born in 1986, which was excluded. The final
respondent count for the Primary Group was 642 (332 female, 310 male).

Supplemental group. The subset of incomplete observations demonstrated greater
variability in birth year. This presented a concern for the analysis as the model was developed
under the assumption that the same causal world applies to both groups. To maintain
comparability between the groups the second group was limited to respondents born in 1987 and
1988. The final respondent count for the Supplemental Group was 298 (146 female, 152 male).
See Table 5 for socio-demographic characteristics of the Primary and Supplemental Groups.

Modeling Procedure
Listwise vs. Pairwise Covariance Matrices
IBM SPSS 22 (IBM Corp., 2013) was used to produce a listwise covariance matrix of the
observed variables. Listwise deletion was chosen over pairwise deletion to avoid computing a
matrix that is not positive definite and consequently unsuitable for SEM analysis (Wothke, 1993)

24

as recommended in Structural Equation Modeling with Lisrel: Essentials and Advances (Hayduk,
1987). This covariance matrix was then added to the Lisrel 8.9 (Joreskog & Sorbom, 2006)
syntax (see appendix A) to estimate and evaluate the model using maximum likelihood
estimation (MLE).
The analysis attempted to identify the strength of the causal relationships between the
latent variables based upon the proposed model (see Figure 1) using the statistical program Lisrel
8.8 (Joreskog & Sorbom, 2006). Testing the fit of the model, and the accuracy of the theoretical
statement it made, was first done using only the data from complete observations. When an
acceptable fitting model was found, the Supplemental Group was added to the Primary Group
with a unique covariance matrix.

NLSCY Original Cohort Complete Observations Model
Specification and identification. In order for a model to be identified its degrees of
freedom (df) must be equal to or greater than zero. The dfvalue is obtained by subtracting the
number of coefficients to be estimated from the number of variances and covariances in the S
matrix (the covariance matrix generated by the data). If the number of coefficients is equal to the
number of values in the S matrix, then the degrees of freedom are zero and the model is
saturated. A saturated model is almost always guaranteed to fit the data however it will not allow
for a x2 test of significance consequently eliminating the model's ability to make testable
predictions about the real world (Hayduk, 1987).

Estimation of free parameters. The initial model attempted to demonstrate a causal
pathway to Resilience through Impulsivity and Coping Skills (hostility) that would produce
Optimism and Problem Solving Skills. See Figure 1. The model converged after 105 iterations. It

25

is important to note that the covariance matrix was assessed as not positive definite and could not
be inverted without further transformation.
According to Ullman (2013) perfect, or extremely high, correlations between two
variables preclude the inversion of the covariance matrix and consequently cannot be analyzed
using SEM without modifying the covariance matrix. Lisrel was able to provide estimates by
employing a ridge option.
Closer inspection of the covariance matrix revealed that the variables for Handle
Problems and Optimism were collinear. Both of these items were taken from the EQ-i (BarOn,
2004) which Statistics Canada (2010b) reports as having a Cronbach ' s Alpha value of 0.836 for
respondents born in 1987 and 1988. A high Cronbach' s Alpha indicates a high degree of
correlation between all the items on this scale (Trobia, 2008). Consequently, the questions that
targeted Optimism and Handle Problems in the NLSCY initially elicited similar responses.
Additionally, before the data was released for analysis by Statistics Canada examination of
crosstabs identified that some of the response cells had low numbers that could compromise
respondent confidentiality. As a result, response categories were collapsed for both Handle
Problems and Optimism according to Statistics Canada Release Guidelines (Statistics Canada,
2009; Statistics Act, R.S.C., 1985, c. S-19). This could have caused an approximate collinearity
arising from "a perfect linear dependency, occluded by a very small variance due to rounding
error." (Wothke, 1993, p. 263; see Table 5).
As Handle Problems was both a dependent variable and a secondary indicator for
Problem Solving, it was decided that it could be removed from the model with little possibility of
misspecification of the model (Ullman, 2013 ; Wothke, 1993) (see Figure 2).

26

After removing the variable Handle Problems Lisrel was able to converge on a solution
after 44 iterations and did not provide any warnings about the covariance matrix.
Assessment of model and model fit. This model produced a covariance matrix that was
significantly different than the data, df = 68, x2=302 .168, p < 0.001. The significant x2 and
consequent poor fit was interpreted as a likely resulting from the dual indicators for Abuse
(TellsBad and HitParnt), Supportive Parenting (Praise and TalkPlay), and Coping Skills
(hostility) (ReactAng and ManyFigh). The problematic nature of the dual indicators is
demonstrated by the low squared multiple correlations (SMC) for the secondary indicators for
each of the above latents. According to Ullman (2013) "each SMC is interpreted as the reliability
of the measured variable in the analysis and the proportion of the variance in the variable that is
accounted for by the factor" (p.733).

Squared multiple correlations. The SMCs suggested that the second indicator for these
latent variables was not being influenced to the same degree as the primary indicator by the
latent variables.

Covariance Underestimation and Overestimation. The estimates generated for the pairs
of indicators for Abuse, Supportive Parenting, and Coping Skills (hostility) suggested an
underestimation of one covariance and an overestimation of the other covariance. This was
clearly indicated by the matched positive-negative standardized residuals found for each pair
(Hayduk, 1987; 1996).
Model modification.

Single indicators. The latents with two indicators were divided creating two distinct
variables. This modification preserved the full set of indicators while increasing the complexity
of the theory presented in the latent portion of the model (See Figure 3):

27

a) Abuse (Verbal Abuse: TellsBad SMC=0.95, Physical Abuse: HitParnt SMC=0.00)
became Verbal Abuse (TELLBADbprcq05collapse) and Physical Abuse
(HITDPMCcQ 1Pcollapse)
b) Supportive Parenting (Verbal Praise: Praise SMC=0.93, Quality Time: TalkPlay
SMC=0.22) became Verbal Praise (PRAISEbprcqOlcollapse) and Quality Time
(TALKPLAYbprcq02collapse)
c) Coping Skills (hostility) (Coping Skills: ReactAng SMC=0.07, Reactivity: ManyFigh
SMC=0.95) became Coping Skills (DFBCdOlX), and Reactivity (DFBCQOlG),

It is important to note that the change from multiple to single indicators was implemented
all at once, rather than one at a time so it is unclear whether it was a specific pair of indicators or
the combined effects of pairs of indicators that caused the initial problem. These changes
provided a marginally better fit, df=42, x2=211.11 , p < 0.001 , but did not solve the problem of
the significant

x2 value.

Re-conceptualized model A new model was conceptualized through careful examination
of the modification indices, which suggested possible ways to account for the statistical
relationships between the variables, and consideration of the reasonableness of the suggestions
based on the literature. This model increased the number of connections between the exogenous
and endogenous variables and altered the placement of the latent variable Optimism from result
to cause of Resilience. The new model asserted Optimism as a mediating mechanism for positive
adaptation in response to adversity through which the effects of Impulsivity, Reactivity, and
Coping Skills (hostility) traveled (see Figure 4). It must be noted that this change inherently
discounts the theoretical statement that the previous model attempted to make and could have
resulted in an artificially good fit because the changes may simply be mirroring the data. The

28

theoretical change was constructed with reference to the literature surrounding optimism
(Benzies & Mychasiuk, 2009; Greeff & Merwe, 2004; Greeff & Du Toit, 2009; Rutter, 2006,
2012; Shiner & Masten, 2012; Stewart, 2011 ; Zautra et al. , 2010) to limit the possibility that the
improvement in the model was a result of capitalizing on chance and is discussed further in the
Limitations section.
Lisrel did converge on a solution that was much closer to fitting the data but still
provided a significant x2 value, d/=36 , x2 =66.38,p <0.01. The modification indices were
assessed in light of the literature regarding resilience and two additional pathways were
simultaneously added to the theoretical framework: Friendships to Reactivity (DiGennaro Reed,
McIntyre, Dusek, & Quintero, 2011; Gaertner, Fite, & Colder, 2011 ; Geven, Weesie, & van
Tubergen, 2013; Sturaro, van Lier, Cuijpers, & Koot, 2011) and Quality Time to Impulsivity
(Russell, Londhe, & Britner, 2013; Tichovolsky, Arnold, & Baker, 2013; see Figure 5) These
additional pathways made sense from a theory standpoint and were also backed up by the current
literature. The direction of the pathways was established through the modification indices - one
possible way of accounting for the relationship between the two variables. This model provided a

x2 value that was lower than the previous model and was barely non-significant, df=34, x2=48.05,
p=0.06.
Final model. Within the modification indices Lisrel suggested the addition of a pathway
from Verbal Abuse to Coping Skills (hostility). According to Evans, Simons, and Simons (2012)
there is evidence of a relationship between these two concepts. Therefore, the pathway from
Verbal Abuse to Coping Skills (hostility) was included in the theoretical framework (see Figure
6).

29

Reciprocal effects. Finally, since resilience is an ongoing process, not a finite outcome, a

reciprocal relationship between Resilience and Problem Solving was included. Again these
pathways were added simultaneously which has the potential to limit the findings of the study.
Initially Lisrel demonstrated signs that the model was under identified by providing wildly
unreasonable estimates for the reciprocal coefficients. By constraining the reciprocal relationship
between Resilience and Problem Solving Skills to be equal, Lisrel was able to converge on a
solution, df=33, x2=38.19,p=0.25, with a non-significant x2 value (see Figure 6 and Appendix
A).

Results
The explanation of results will be limited to the final two models (see Figures 6 and 7)
which achieved non-significant x2 values and are therefore believed to accurately provide one
possible explanation of resilience in young English speaking Canadian adults.

Primary Group (Complete Data)
Significant coefficient estimates. Lisrel provides estimates of direct effects that
implicitly control for the effects of all the other predictors in the equation in which the effect
estimate appears. The implicit control of all other effects is similar to the language used in
multiple regression equations where all predictor variables except one are held constant.
However, this language becomes problematic in light of the requirement of SEM that an effect
must be able to produce a chain of changes within the model. Therefore, it is helpful to think of
the model as a stable system where variables that are not part of the causal chain between the
predictor and dependent variables remain untouched (Hayduk, 1987).
The estimated coefficients for the exogenous and endogenous pathways were considered
significant if the estimate was at least two times greater than its standard error. The interpretation

30

of effect estimates in unstandardized solutions is usually phrased in terms of the consequences of
one real unit of increase in the value of the causal latent variable which allows the
appropriateness of the sign of each effect to be easily assessed.

Direct effects.
lmpulsivity. According to the proposed model (see Tables 6 and 7 and Figure 6),
increased lmpulsivity at Cycle 4 (ages 13-14; participant report) was produced by high Physical
Abuse at Cycle 3 (ages 11-12; participant report), and being male (Biological Sex; Cycle 1; PMK
report). Decreased Impulsivity at Cycle 4 (ages 13-14; participant report) was produced by high
Verbal Abuse and high Quality Time at Cycle 2 (ages 9-10; PMK report). This suggests
parenting practices have an impact on children's Impulsivity at ages 13 to 14. Respondents who
experienced frequent Verbal Abuse as reported by PMK at ages 9 to 10 self-reported lower
levels of impulsive behavior at ages 13 to 14. The novel effect sign from Verbal Abuse to
lmpulsivity is addressed in the discussion section.

Reactivity. According to the model, increased Reactivity at Cycle 4 (ages 13-14
participant report) was produced by being male (Biological Sex; Cycle 1; PMK report), low
Family Income at Cycle 1 (ages 7-8; PMK report), and a lack of close friendships and high
Impulsivity at Cycle 4 (ages 13-14; participant report). These effects suggest that early
experiences of poverty and being male increase the risk of high reactivity at ages 13 to 14, as
does concurrent Impulsivity and few close Friendships.

Coping skills (hostility). According to the model decreased Coping Skills (hostility) at
Cycle 4 (ages 13-14; participant report), indicated by a greater frequency ofreported physical
fights, was produced by high Verbal Abuse at Cycle 2 (ages 9-10; PMK report), high Physical
Abuse at Cycle 3 (ages 11-12; participant report), and low Relationship with Mom, low

31

Relationship with Dad, high Impulsivity, and high Reactivity at Cycle 4 (ages 13-14; participant
report). These effects suggest that punitive parenting practices decreases the ability to cope at
ages 13 to 14, as does concurrent Impulsivity, Reactivity, and disconnected parental
relationships.

Optimism. According to the model increased Optimism at Cycle 8 (ages 21-22;
participant report) was produced by close Friendships at Cycle 4 (ages 13-14; participant report).
This effect suggests that early adolescent Friendships have an impact on Optimism at ages 21 to
22.

Resilience. According to the model, increased Resilience in early adulthood was
produced by being male (Biological Sex; Cycle 1; PMK report) and having high Problem
Solving Skills at Cycle 8 (ages 21-22; participant report). These effects suggest that being male
increases resilience in early adulthood, as does concurrent problem solving skills. It is important
to note that there is an effect from Optimism to Resilience. However, it is not interpretable
regarding strength or magnitude because Optimism was used as a scaling variable for Resilience
and consequently no coefficient was estimated.

Problem solving skills. According to the model increased Problem Solving Skills at
Cycle 8 (ages 21-22; participant report) was produced by high Resilience in early adulthood.
This effect, in light of the previously discussed effect of problem solving on resilience, suggests
that the there is a dynamic interaction between resilience and problem solving skills at ages 21 to
22. The interchange between these effects is represented in the model (see Figures 6 and 7) by
the reciprocal relationship between Problem Solving Skills and Resilience.

Total indirect effects. According to Hayduk (1987) an indirect effect is determined by
calculating the product of the coefficient estimates making up a series of paths linking two

32

variables; " ... adding the direct and indirect effects gives the total effect of X1 on X3 and is best
interpreted as the change in X3 produced to follow a unit change in X1 if all variables in the
model are left untouched except for the changes originating from the hypothetical unit change in

X1 " (p.247) (See Tables 8 and 9 and Figure 6). Indirect effects can be a way to think about
mediating variables which transform or influence the effects of stimuli on behavior (Baron &
Kenny, 1986).

Coping Skills (hostility). Coping Skills (hostility) was measured by the NLSCY item
about reacting with fighting and anger. This is functionally considered to be anti-social behavior
along with Impulsivity and Reactivity. According to the model, variations in Verbal Abuse at
Cycle 2 (ages 9-10; PMK report) produced changes at Cycle 4 (ages 13-14; participant report) in
Impulsivity and possibly Reactivity, and then changes in Coping Skills (hostility). Variations in
Biological Sex (Cycle 1; PMK report) produced changes at Cycle 4 (ages 13-14; participant
report) in Impulsivity and Reactivity, and then changes in Coping Skills (hostility). Variations at
Cycle 4 (ages 13-14; participant report) in Relationship with Mom produced changes in
lmpulsivity and Reactivity, and then changes in Coping Skills (hostility). Variations at Cycle 4
(ages 13-14; participant report) in Friendships produced changes in Reactivity, and then changes
in Coping Skills (hostility). Variations in Quality Time at Cycle 2 (ages 9-10; PMK report)
produced changes at Cycle 4 (ages 13-14; participant report) in Impulsivity and possibly
Reactivity, and then changes in Coping Skills (hostility). These effects suggest that parenting
practices have an impact on antisocial behavior at ages 13 to 14. Additionally, antisocial
behavior and relationship quality with parents at ages 13 to 14 are interrelated.

Reactivity. According to the model, variations in Biological Sex (Cycle 1; PMK report),
Verbal Abuse and Quality Time at Cycle 2 (ages 9-10; PMK report), and Physical Abuse at

33

Cycle 3 (ages 11-12; participant report), produced changes at Cycle 4 (ages 13-14; participant
report) in Impulsivity, and then changes in Reactivity. These effects suggest that parenting
practices and being male impact children's level oflmpulsivity and Reactivity at ages 13 to 14.

Resilience. According to the model, variations in Biological Sex (Cycle 1; PMK report)
produced variations at Cycle 4 (ages 13-14; participant report) in Impulsivity, Reactivity, and
Coping Skills (hostility). Variations at Cycle 4 (ages 13-14; participant report) in lmpulsivity
produced further changes in Coping Skills (hostility) and Reactivity. Variations at Cycle 4 (ages
13-14; participant report) in Reactivity also produced changes in Coping Skills (hostility), and
then variations in Optimism at Cycle 8 (ages 21-22; participant report). Variations at Cycle 8
(ages 21-22; participant report) in Optimism produced changes in Resilience. Because of the
reciprocal relationship between Resilience and Problem Solving Skills at Cycle 8 ( ages 21-22;
participant report) variations in Friendship at Cycle 4 (ages 13-14; participant report)
demonstrated an indirect effect on Resilience. Variations in Friendship at Cycle 4 (ages 13-14;
participant report) produced changes in Resilience, and then changes in Problem Solving Skills
at Cycle 8 (ages 21-22; participant report), which produced further changes in Resilience.

Problem solving skills. According to the model, variations in Resilience, as a result of
variations in Biological Sex (Cycle 1; PMK report), produced changes in Problem Solving Skills.
Repeated cycling through the reciprocal effects of Resilience on Problem Solving Skills and vice
versa also imply Resilience has an indirect effect on Problem Solving Skills as a consequence of
its direct and equality-constrained reciprocal effect.

Squared multiple correlations. The R 2 values for the endogenous variables indicate that,
with the exception of Resilience, the percent of variance in each endogenous variable explained
by the model is quite low (Impulsivity R 2 =0.128; Reactivity R 2 =0.151; Coping Skills (hostility)

34

R 2 = 0.121; Optimism R2 =0.008; Problem Solving R2=0.043; Resilience R2 =0.398). The
disparity between the explained variance of Resilience and the explained variance of Impulsivity,
Reactivity, Coping Skills (hostility), Optimism, and Problem Solving Skills is an artifact of the
error-variance specification for Resilience, as discussed in the Limitations section.
Stacked model using primary and supplemental groups. The stacked model provided
a way to include respondents that fit a specific pattern of missing data (Wothke, 1993) without
imputation. By only including participants that were missing data on specific variables (Parents
Hit/Threaten, Share Secrets, How Close to Mother, How Close to Father, Difficulty Waiting, Get
into fights, and React Anger/Fighting,) the variances and covariances calculated for the Primary
Group were added to the covariance matrix for the Supplemental Group (see Appendix B)
enabling Lisrel to analyze the model and provide estimates. Then the coefficients that related to
those substituted parts of the covariance matrix were constrained to be equal to the
corresponding coefficient estimates in the Primary Group. These constraints controlled the
components of the model for which the Supplemental Group had no data by using the estimates
calculated for the Primary Group and allowed Lisrel to calculate unique estimates for the free
coefficients in the Supplemental Group. The stacked model was also non-significant, df=85,

x2=85 .77, p=0.46 (see Figure 7 and Appendix B).
Group comparability. To be sure that the Primary and Supplemental Groups were indeed
comparable all effects in the Supplemental Group were constrained to be equal to the Primary
Group and the model was analyzed again, df=103, x2=92.73 , p=0.77. A x2 difference test
between the models demonstrated resulted in a p-value of 0.99, x2 diff = 6.96 = 92.73-85.77, dfaiff
= 18 = 103-85. This indicates that the stacked model that only constrained the coefficients
related to missing data in the Supplemental Group was not significantly different than the model

35

that constrained all pathways to be equal to the Primary Group. The findings confirm that the
two groups are comparable and operate within the same causal world, a conclusion that is further
validated by the similarity in descriptive statistics for the two groups (see Table 5).

Stacked Model
Significant coefficient estimates. The estimated coefficients for the exogenous and
endogenous pathways were considered significant if the estimate was at least two times greater
than its standard error. The significant effects for the Primary and Supplemental Groups were
generally the same as the previous model with several exceptions. The differences that arose
between the groups suggests that although they share a causal world the strength of the variables
to influence outcomes may not be equal in both groups.

Direct effects. Unlike the results of the previous model the path coefficients at Cycle 4
(ages 13-14; participant report) from Relationship with Mom to lmpulsivity and Reactivity
reached significance (coefficients constrained to be equal across groups). These effects suggest
that at ages 13 to14 maternal relationship quality impacts children's concurrent level of
lmpulsivity and Reactivity.
The path coefficient from Relationship with Dad Cycle 4 (ages 13-14; participant report)
to Resilience in early adulthood reached significance (coefficient constrained to be equal across
groups) unlike the previous model. This effect suggests that paternal relationship quality at 13 to
14 years of age impacts resilience in early adulthood. Additionally, the path coefficients from
Verbal Abuse at Cycle 2 (ages 9-10; PMK report) to Coping Skills (hostility) and Impulsivity at
Cycle 4 (ages 13-14; participant report) and from Biological Sex (Cycle l; PMK report) to
Resilience achieved significance in the Primary Group but not in the Supplemental Group (See
Tables 10 and 11 ). These effects suggest that parenting practices had a greater impact on

36

children's levels of anti-social behavior at ages 13 to 14 in the Primary Group than the
Supplemental Group. Additionally, being male had an impact on resilience in early adulthood for
the Primary Group but not the Supplemental Group.

Total indirect effects. The Primary Group demonstrated significant indirect effects that
were both congruent and incongruent with the previous model (See Table 12 and Table 13). Two
chains of effects that were significant for the Primary group were not significant for the
Supplemental Group: (a) the path to Reactivity at Cycle 4 (ages 13-14; participant report) from
Verbal Abuse at Cycle 2 (ages 9-10; PMK report) through Impulsivity at Cycle 4 (ages 13-14;
participant report) and (b) the path to Coping Skills (hostility) at Cycle 4 ( ages 13-14; participant
report) from Verbal Abuse at Cycle 2 (ages 9-10; PMK report) through lmpulsivity and
Reactivity at Cycle 4 (ages 13-14; participant report).
Three chains of effects that were not significant for the Primary group were significant
for the Supplemental Group: (a) the path to Resilience from Verbal Abuse at Cycle 2 (ages 9-10;
PMK report), through lmpulsivity, Reactivity, and Coping Skills (hostility) at Cycle 4 (ages 1314; participant report), and Optimism at Cycle 8 (ages 21-22; participant report), (b) the path to
increased Problem Solving Skills at Cycle 8 (ages 21-22; participant report) from Verbal Abuse
at Cycle 2 (ages 9-10; PMK report) through Impulsivity, Reactivity, and Coping Skills (hostility)
at Cycle 4 (ages 13-14; participant report), and Optimism and Resilience at Cycle 8 (ages 21-22;
participant report), (c) and the path to Problem Solving Skills at Cycle 8 (ages 21-22; participant
report) from Relationship with Dad at Cycle 4 (ages 13-14; participant report) through
Resilience.
These effects suggest that parenting practices did not have as strong an impact on Coping
Skills (hostility) at ages 13 to 14 in the Supplemental Group compared to the Primary Group.

37

However, parenting practices combined with children' s level of antisocial behavior and paternal
relationship quality at ages 13 to 14 did have an impact on Resilience and Problem-solving Skills
at ages 21-22 for the Supplemental Group.
Squared multiple correlations. As in the Primary Group model the percent of variance
explained in the endogenous variables is small with the exception of Resilience in the Primary
Group. In the Supplemental Group the percent of variance in Resilience explained by the model
was closer to the other variables (Impulsivity R2 =0.120; Reactivity R2=0.121; Coping Skills
(hostility) R2 = 0.147; Optimism R2 =0.018 ; Problem Solving Skills R2 =0.017; Resilience
R2 =0.120).
Model Strengths
The models produced in this analysis are noteworthy for several reasons. Despite arising
from the previously untested English Speaking Canadian population, the results demonstrate
congruency with previous studies of diverse populations. This study and similar studies from
other nations draw attention to the impact that child sex (male/female), early parenting practices,
and adolescent levels of antisocial behavior and close parental relationships, have on resilience
and problem solving in early adulthood.
Additionally, both the Primary Group model and the Primary and Supplemental Groups
model achieved a goodness of fit in non-significant x2 values. Use of the x2 statistic as a measure
of fit is fiercely contested in the literature because of its sensitivity and the difficulty of achieving
a non-significant value. The debate over use x2 test is best demonstrated by the responses
generated by Barrett' s (2007) article on measuring model fit (Hayduk, 2014; Hayduk et al. ,
2007; Hayduk & Glaser, 2000a; 2000b; Herting & Costner, 2000). Even when differences of

38

significance and non-significance were noted between the models and between the groups, the
direction of the effects remained consistent between the Primary and Supplemental Groups.

Model Limitations
It is important to acknowledge the limitations of the model before proceeding to
interpretation of effects. The model employed secondary data with substantial respondent
attrition, demonstrated estimation problems, underwent multiple modifications, and did not
explain a large proportion of the variance in the endogenous variables.
The difficulty encountered in estimating a solution indicates that the model may not
actually fit the data as well as the solution suggests. The accuracy of the estimates derived for the
reciprocal relationship between Resilience and Problem Solving Skills is a compromise of what
the estimates would be if each coefficient was free to vary. Optimism has an indicator which
means that the latent is reasonable well known through the indicator. However, Optimism is not
explained well by the rest of the model. Consequently, it may not have been appropriate to use
Optimism as a scaling variable for Resilience.

Removal of confidence latent. According to Tomarken and Waller (2005) the majority
of models omit variables because SEM only provides an approximation of reality. It is important
to note that the omission of a variable does not necessarily stop a model from fitting the data well
(Tomarken & Waller, 2005). Therefore, even though the removal of Confidence improved the
model fit (see Figure 2), Confidence cannot be wholly ruled out as a variable in the explanation
of Resilience.

Multiple model modifications. The modelling completed for this study involved 5
modifications. The modifications were consistent with theory and prior evidence but were made

39

after seeing feedback from other runs of the model. Such practices may raise concerns but were
used effectively in this study, as discussed in the next paragraphs.

Simultaneous addition of multiple pathways. By failing to add additional pathways one
at a time it is not possible to determine if it was a given pathway or an artifact of the combination
of the pathways added that improved the model fit. Compared to individual additions,
simultaneous addition of multiple pathways reduced the likelihood that observed changes in fit
were the result of chance.

Inflated R 2• As Resilience does not have an indicator within the model it is difficult to
provide a realistic R2 value. To achieve the R2 value of 0.803 the error variance for Resilience
was set at 0.00. Once the model was estimated the latent variable variance-covariance table was
examined and the variance for Resilience, 3.49, was used as the latent error variance and the
model was estimated a second time. Replacing the latent error variance reduced the R2 value
from 0.803 to 0.398. The same method was used in the stacked model; the variance for
Resilience (Primary Group 2.78; Supplemental Group 0.988) was used as the latent error
variance and the model was estimated a second time. Replacement reduced the R 2 value from
0.758 to 0.374 (Primary Group) and 0.251 to 0.120 (Supplemental Group).
Proportion of explained variance. The non-significant x2 value for the model is
contrasted by the relatively small proportion of explained variance for the endogenous variables
Impulsivity, Reactivity, Coping Skills (hostility), Optimism, and Problem Solving Skills. There
is currently no standard way to address the issue of error variance in phantom variables. For this
reason, the explained variance for Resilience should be treated as undetermined and interpreted
with caution.

40

Because SEM provides for the prospect of equivalent models (Hayduk, 1996) it is
possible that there is another model that would fit just as well but provide a more complete
explanation of the variance for the endogenous variables (Hayduk & Glaser, 2000a).
Optimism as a scaling variable. Optimism is the key connection to Resilience in the
Primary Group model. However, less than 1% of the variance in Optimism is explained by the
model. This is especially problematic because not only does Optimism act as a bridge between
Resilience and the rest of the model, it is specified as a scaling variable for Resilience. If
Optimism is not explained by the model, then Resilience may not be explained by the model.
Secondary data. The data analyzed for the present study was not collected by the
researcher. This may be problematic because the measures used may not have been the best tool
for the current study had the researcher been able to control this aspect (Greenhoot & Dowsett,
2012). Further, the present research was limited in its inclusion of variables to those available in
the NLSCY data set. Other variables (e.g., genetic differences, resting cortisol levels and,
personal spirituality,) are known to contribute to the development of resilience but could not be
included here because they are not in the NLSCY.
Additionally, the attrition of respondents across time is not random (Carillo, Kovacevic,
& Wu, 2006) and likely resulted in the exclusion of those most at risk to experience adversity.

This brings into question the NLSCY's ability to truly represent the Canadian experience.

Coping skills (hostility). The measure chosen from the NLSCY used for Coping Skills
was only informative regarding the use of anger and fighting as a coping response. This seriously
limits the model's ability to make a strong theoretical statement about other, more productive,
styles of coping in early adolescence. The limited informative value of the Coping Skills

41

(hostility) measure may also explain why the model failed to estimate a significant coefficient
between Coping Skills (hostility) and Optimism.
As a result of these limitations the accuracy of this model in representing Resilience in
the Canadian context should be interpreted with caution. Even with these limitations, the
NLSCY is unique in its representation of the development of Canadian children and the present
findings are confirmation that NLSCY data are worthy of secondary analysis.
Discussion

The present study makes a unique contribution to the literature on resilience. The cyclic
collection of data by the NLSCY lends itself to a developmental interpretation; however, this
study does not present a traditional repeated measures longitudinal study. Instead, by employing
a nationally representative sample of Canadian children with repeated measurements from
childhood to early adulthood, the present study places resilience within a system of normative
development that provides a broad view of protective factors at different stages of development
(see Figure 1). Findings confirm that, in terms of factors and pathways that contribute to the
development of resilience, the English-speaking Canadian population is developmentally
analogous to American, Australian, Taiwanese, Polish, and other industrialized populations. The
findings underscore the importance of relationships with parents and peers that are close, without
dictating what these relationships must look like, throughout childhood and positive problemsolving skills, which are effective for the individual, in adolescence for resilience in young
adulthood.
In the following paragraphs findings from the present study are discussed in the context
of the existing literature as a means of drawing conclusions about resilience from this study.

42

Where relevant, effect sizes are interpreted from the standardized solution (see Tables 14 and 15)
according to the recommendations by Durlak (2009) and Cohen (1992).
Middle Childhood
Biological Sex. Gender differences have consistently been identified in the resilience
literature. The Kauai Longitudinal Study identifies males as being at higher risk during the first
10 years of life however females are considered more vulnerable throughout adolescence
(Werner, 1989). Analysis of the Project Competence data demonstrates that gender is a
significant predictor of outcomes in social competence (Masten, et al., 2004). The present study
found that biological sex, had a significant small effect on Impulsivity and Reactivity in early
adolescence (boys were more impulsive and reactive than girls) and a significant large effect on
Resilience in emerging adulthood (girls were less resilient than boys). In relation to the Kauai
Longitudinal study these results suggest that the higher risk attributed to boys up to age 10 may
be an indication of less developed emotional regulation. Whereas the risk attributed to girls
during adolescence may be an indication of internalizing behaviors that were not observed or
measured in childhood. Rey Pefia and Pacheco (2012) found that there was a significant
difference between boys and girls ages 11 to 18 regarding cognitive emotion regulation
strategies; girls were more likely to use rumination and catastrophizing which informed
depressive symptoms whereas boys were more likely to use higher levels of physical-verbal
aggression. Lyons, Otis, Huebner, and Hills (2014) found boys to be more sensitive to the effects
of differing levels of life satisfaction in early adolescence than girls. Given these results
biological sex and age should be taken into consideration when planning interventions for
problem behaviors.

43

Family income. Regarding the data collected through the Project Competence Study
Garmezy, Masten and Tellegen (1984) posit that SES should mediate the effects of stressful life
events so that children from a higher SES, given equal life stress, should display fewer disruptive
behaviors than children from a lower SES. The present study identified a significant small effect
from Family Income in middle childhood to Reactivity in early adolescence - respondents from
lower income families demonstrated more reactive behavior. In relation to the findings of the
Project Competence Study the present findings suggest that early experiences of poverty or
affluence may influence access to coping resources including access to structured activities
(Bennett, Lutz, & Jayaram, 2012) and parental availability (Chien & Mistry, 2013). In a 2011
study Kraus, Horberg, Goetz, and Keltner found that low Socioeconomic Status and Social Class
elicited greater reactivity to the social environment. They interpreted this reactivity as a result of
greater threat vigilence and previous experience of hostile emotion. SES is a multifaceted factor
that provides many avenues for the development or methods to reduce long term reactivity
starting in early childhood.
Late Childhood
Quality Time. The Kauai Longitudinal Study identifies the importance of affectionate
family relationships to counteract stressful events through emotional support (Werner, 1989).
Granger et al. (1998) suggested that interpersonal difficulties are more prevalent in children who
demonstrate low levels of arousal in response to conflict. These interpersonal difficulties can
lead to stress and insensitivity in their parents (Granger, et al., 1998). Additionally, other studies
demonstrate that supportive maternal presence in early childhood predicted levels of
externalizing behaviors in early adolescence (Raby, Roisman, Fraley, & Simpson, 2015; Wang,
Christ, Mills-Koonce, Garrett-Peters, & Cox, 2013). Englund, et al. (2011), using the Minnesota

44

Longitudinal Study of Risk and Adaptation, determined that early attachment security had an
indirect influence on future relationships and overall adult functioning. Additionally, the
Concordia Longitudinal Risk Project, (Pougnet, Serbin, Stack, & Schwartzman, 2011) identified
paternal involvement in middle childhood as a significant predictor of children's cognitive and
behavioural competence in early adolescence.
The present study identified a significant small effect from Quality Time in late
childhood to Impulsivity in early adolescence - respondents who experienced more one on one
time with parents in childhood later demonstrated lower levels of reactive behavior. In relation to
the Kauai Longitudinal Study, the Minnesota Longitudinal Study, and the Concordia
Longitudinal Risk Project the effect of early Quality Time on later Impulsivity suggests that
setting aside dedicated time for parent-child interaction may be more important than the specific
nature of the interaction. Cicchetti, Rogosch, and Toth (2006) support this possibility by noting
that secure attachment can be fostered through intervention that supports mothers in spending
more quality time with their children. Increased secure attachment is achieved either through
therapeutic improvement of the parent-child relationship or assistance in building a social
support network that allows for dedicated parent-child time (Cicchetti, Rogosch, & Toth, 2006).

Verbal and Physical Abuse. The Christchurch Health and Development Study identifies
middle childhood as a key window for addressing conduct problems and self-control issues as
these risk factors have implications for later legal, academic, and occupational outcomes
(Fergusson & Horwood, 2001). Childhood maltreatment is consistently linked to externalizing
problems in middle childhood and early adolescence (Cicchetti & Valentino, 2006). The present
study identified significant small effects from Physical Abuse and Verbal Abuse in late
childhood to Impulsivity and Coping Skills (hostility) in early adolescence. Respondents who

45

experienced frequent Physical Abuse demonstrated higher levels of impulsive behavior and were
more likely to respond with hostility. Additionally, the experience of frequent Verbal Abuse
resulted in less effective coping strategies. According to Villoda et al. (2015) the pattern of
externalizing problems as a result of maltreatment is consistent regardless of the specificity or
generality of the population being studied. This suggests that interventions aimed at reducing
punitive parenting practices may be a main way to reduce children's levels of anti-social
behavior in early adolescence.
Contrary to the literature, the present study demonstrated that high levels of Verbal
Abuse resulted in lower levels of impulsive behavior. The size of this effect is small and only
applies to the Primary Group. It is possible that the NLSCY item interpreted as Verbal Abuse
was actually a measure of authoritarian parenting. Impulsivity tends to be stable across the
lifespan (Niv, Tuvblad, Raine, Wang, & Baker, 2012). Parents of children who demonstrated
more impulsive behaviors tend to be less permissive and more authoritarian than parents of less
impulsive children (Gau, & Chang, 2013; Moghaddam et al., 2013). The Verbal Abuse reported
by the PMK may have been an attempt at controlling concurrent impulsive behavior. Predictable
and frequent feedback has been demonstrated to be helpful for children who are easily frustrated
(Kiff, Lengua, & Zalewski, 2011; Lengua, 2008). The frequent verbal involvement reported by
the PMK could have been effective at modifying concurrent Impulsivity so that later impulsive
behavior was reduced.
Another possibility is that the measure for Impulsivity was, in fact, a measure of selfawareness. Children who are more impulsive tend to be less self-aware and may have answered
the question without thinking. These children would also likely elicit more Verbal Abuse. If this
is the case then high levels of Verbal Abuse would result, or co-occur, with low levels of self-

46

awareness. These alternative interpretations of Verbal Abuse and Impulsivity could explain the
unexpected direction of the effect from Verbal Abuse to Impulsivity in the present model.
Early Adolescence
Externalizing behaviors. The present study identified a significant small effect within
early adolescence from Impulsivity to Reactivity and Coping Skills (hostility) as well as a
significant small effect from Reactivity to Coping Skills (hostility) - respondents who
demonstrated more impulsive behavior also demonstrated more reactive behavior and less
effective coping strategies. A 2008 study by Wittman, Arce, and Santisteban notes that
impulsivity influences verbal and physical aggression especially when the individual engages in
instant gratification leisure activities such as video games and television viewing. However,
Rutter (2003) argues that adolescent anti-social behavior is a phenomenon that is often limited to
the period of adolescent development and not an indicator of long term anti-social tendencies.
The Kauai Longitudinal Study found that by early adolescence resilient children were not overly
impulsive and engaged in cooperative activities and hobbies successfully (Werner, 1996).
Considered as a whole this information suggests that engagement in anti-social behavior in early
adolescence may have an impact on resilience in early adulthood but the relationship deserves
further investigation.
Relationships. The Kauai Longitudinal study identified supportive familial and extrafamilial relationships to be key predictors of resilience (Werner, 1995; 1989). The importance of
supportive relationships was also found in the Project Competence Study (Masten, et al., 2004;
Masten & Tellegen, 2012), the Minnesota Longitudinal Study of Risk and Adaptation (Englund,
I-Chun Kuo, Puig, & Collins, 2011), and the Christchurch Health and Development Study
(Newton-Howes, Horwood, & Mulder, 2015). The present study identified that early adolescent

47

Friendships, Relationship with Mom, and Relationship with Dad had small but significant effects
in emerging adulthood.
Friendships. The present study identified a significant small effect within early

adolescence from Friendships to Reactivity and from Friendships in early adolescence to
Optimism in emerging adulthood - respondents who experienced closeness in their friendships
demonstrated less reactive behavior and were more optimistic. This suggests that close
friendships in adolescence have the ability to influence concurrent behavior and emotional affect
in early adulthood. Newman, Lohman, and Newman (2007) demonstrated that when adolescents
experience a higher sense of group belonging externalizing problems decrease regardless of
gender. Additional evidence from a 2015 study suggested that friendship groups may influence
individual's experiences of hope through modeling and reinforcement regarding problem solving
strategies and attitudes (Parker, et al., 2015). Close friendships should be considered an
important resource when planning interventions for problem behaviors in adolescence.
Relationship with mom. The present study identified a significant direct small effect

within early adolescence from Relationship with Mom to Reactivity, Impulsivity, and Coping
Skills (hostility) - respondents that had close relationships with their mothers demonstrated less
reactive and impulsive behavior and more effective coping strategies. Bradley and Corwyn
(2013) found that maternal sensitivity was consistently significant in the prediction of
externalizing behavior. Additionally, analysis of the NICHD Study of Early Child Care and
Youth Development revealed that regular parental monitoring in early adoescence was
associated with a decrease in externalizing problems and that self-control mediated the
relationship between maternal sensitivity and externalizing behaviors (Bradley & Corwyn,
2013). Furthermore, German et al. (2013) found, in a group of Mexican American adolescents,

48

that high maternal warmth acted as a protective factor for externalizing problems even when
harsh discipline was used. This suggests that a close maternal relationship in early adolescence
provides an external source of emotional regulation that reduces anti-social behaviors and

.
.
improves copmg.
Relationship with dad. The present study identified a significant small effect within early
adolescence from Relationship with Dad to Coping Skills (hostility) and from Relationship with
Dad in early adolescence to Resilience in Emerging adulthood - respondents who had close
relationships with their fathers demonstrated more effective coping strategies and were more
resilient.
Coping skills (hostility) and aggression. Murray et al. (2014) examined the influence of
father-child and mother-child relationships on aggression in adolescence. Results indicated that
lower levels of female aggression were associated with higher quality parental relationships;
however, this was not found to be true for male aggression in their sample. Casselman and
Rosenbaum (2014) determined that perceived father rejection had significant direct and indirect
effects on male aggression in late adolescence and emerging adulthood. This suggests that as
with maternal relationships close paternal relationships in early adolescence provides an external
source of emotional regulation that reduces anti-social behaviors and improves coping.
Resilience. Zhang, Zhao, Ju, and Ma (2015) determined a clear direct effect from father
involvement to resilience in adolescence that did not differ by gender (Zhang et al., 2015). Taken
together this evidence on the effects of relationships in early adolescence suggests that because
executive function and the ability to self regulate internally are still developing (Zelazo &
Carlson, 2012) external sources of emotional regulation are invaluable. Interventions targeting

49

antisocial behaviors in early adolescence should consider including parents and close friends in
the intervention process.

Emerging Adulthood
Optimism. Results from the present study failed to demonstrate a significant effect from
Coping Skills (hostility) to Optimism. This is problematic as the absence of this effect removes a
key link in the model's ability to explain Resilience. Additionally, this zero effect causally
disconnects lmpulsivity, Reactivity, and Coping Skills (hostility) from Optimism, Resilience and
Problem Solving Skills. As discussed in the Limitations section the measure used for Coping
Skills (hostility) was incomplete at best. The insignificant effect from Coping Skills (hostility) to
Optimism can be interpreted in two different ways.
First, it is possible that the Impulsivity and aggression measured in early adolescence
reflected adolescence-limited antisocial behavior, characterized by onset in early adolescence
and recovery in early adulthood (Rutter, 2003; Moffitt, 1993). If true, Optimism in emerging
adulthood would not be explained by period limited anti-social behavior in adolescence.
Second, it is possible that a significant effect would have been present had a wider range
of coping skills been assessed by the measure for Coping Skills (hostility). In fact, a 2013 study
by Pietruska and Armony found that anger, temperament and reaction (Spielberger, 1996), did
not predict optimism despite its strong correlation with Impulsivity which did predict Optimism.
Puskar et al. (1999) did demonstrate a significant correlation between optimism and expressed
anger but the size of the correlation was considered very small. These results suggest that the
measure used for Coping Skills (hostility) in the present study may not have been an appropriate
choice and could explain the disconnect between Impulsivity, Reactivity, and Coping Skills
(hostility), and Optimism, Resilience, and Problem Solving Skills in the model.

50

Further investigation is needed to determine the implications of this part of the model. In
the literature coping skills cover a wide range of behaviors: (a) solving the problem which is
characterized by working at a problem while remaining optimistic, fit, relaxed and socially
connected, reference to others which is characterized by turning to others for support whether
they be peers, professionals or deities, non-productive coping which is characterized by
avoidance strategies (Frydenberg et al. , 2004), (b) task oriented coping, distraction oriented
coping, and disengagement oriented coping (Nicholls, Levy, & Perry, 2015), and (c) problem
solving, information seeking, negotiation, social support, positive cognitive restructuring,
emotional regulation, distraction, rumination, avoidance, helplessness, social withdrawal, and
opposition (PsychTests AIM. Inc. , 2011). Follow-up analysis would ideally include multiple
measures of coping skills to determine the true relationship between coping skills and optimism.
Problem solving. The present study found that Problem Solving Skills demonstrated a

significant small effect on Resilience. Additionally, Resilience demonstrated a significant
medium effect on Problem Solving Skills. Respondents who were more resilient demonstrated
more effective problem solving strategies and vice versa. Rutter (1985) noted that Resilience
involves the ability to use multiple problem solving approaches. The importance of problem
solving skills was also emphasized by the Kauai Longitudinal Study (Werner, 1989). Fruiht
(2015) links problem solving and hope as a reciprocal relationship; students who have trusted
adults to act as mentors are exposed to collaborative problem solving, and hopeful students are
more likely to seek out mentorship. Cooperative problem solving skills were also identified as a
key factor in the development of resilient communities (McKay, 2011) and a 2009 study by Li
and Yang of Taiwanese college students demonstrated a significant standardized effect from
Resilience to Problem Solving similar to that reported in the present study (Standardized= .292)

51

(p.168). Collectively this line of evidence suggests that the ability to approach adversity from a
solution focused perspective may be central to optimal developmental outcomes.
Implications for Future Research
Explained variance and effect size. The current study' s findings are congruent with the
literature regarding resilience in both results and predictive power. Although the present study
demonstrates pathways to resilience similar to those seen in prior research the predictive power
is low. The relatively low amount of explained variance and effect sizes in the present study and
the literature in general may be due to the emphasis on analysis at a population level especially
regarding longitudinal studies. Future research may be able to improve the predictive power of
by modeling resilience for distinctive groups that are exposed to specific rather than general risk
factors .
Intervention. A recent examination of the Big Brothers Big Sisters School-Based
Mentoring program showed that one of the most significant outcomes for program participants
was the benefit of a continued relationship with a trusted adult to act as an emotional support and
mentor (Herrera, Grossman, Kauh, & McMaken, 2011 ). This finding mirrors the assertion by
Fruiht (2015) that mentorship modelling collaborative problem solving increases hope and the
present observation on the importance of close relationships in the development of problem
solving skills and resilience. Together these results make a strong case for further investigation
of mentorship programs by both family and non-family members to increase resilient outcomes
for populations considered at risk.
Self-regulation. Self-regulation, the deliberate use of coping skills to modify thoughts,
affect, behavior, or attention, has demonstrated utility in predicting resilience in youth from low
income and homeless families (Buckner, Mezzacappa, & Beardslee, 2003). Chen (2013) argues

52

that within the developmental system parent modeling of emotional regulation during
emotionally charged situations created a zone of proximal development (Bronfenbrenner & Ceci,
1994) for the child's own emotional regulation. Parents are an integral part of the child's
environment in the development of emotional regulation skills (Guo et al., 2015).
Martel et al. (2007) demonstrated a reciprocal relationship between resilience and
reactive control during development which supported the development of problem solving skills.
During early adolescence children are continuing to develop the ability to regulate their emotions
and even demonstrate an increase in dysregulations during middle adolescence (Zimmermann, &
Iwanski, 2014) likely due to the increased emotional reactivity and salience of social interactions
during this period (Collins, 2003). Maturation of the lateral prefrontal cortex also takes place
during adolescence and impacts cognitive emotion regulation strategies (Vijayakumar et al.,
2014).
In light of these findings the results of the present study only emphasize the importance

of positive parental involvement and close friendships in early adolescence. Without these
relationships to support the development of emotional regulation, through mentorship and social
support, it is unrealistic to expect mature use of coping skills in early adulthood. It is possible
that by exploring this relationship between resilience and self-regulation, from a longitudinal
perspective future research can more clearly explain the underpinnings of resilience with regards
to specific relational factors.
Executive function. Additionally, resiliency and self-regulation have been linked to

Executive Functioning (EF) processes: arousal regulation, response delay and inhibition, multistep planning, and shifting mental set as requirements of a situation change (Martel et al., 2007).
The majority of resilience research has been descriptive in nature. Many protective and risk

53

factors identified do not lend themselves to experimental manipulation. However, problem
solving skills and close relationships deserve further investigation. By including EF in future
investigations an experimental design may be more easily achieved because of the prevalence of
laboratory tasks and formal assessment procedures available for investigating executive
functioning (Barkley, 2012; Delis, Kaplan, & Kramer, 2001 ; Gioia, Espy, & Isquith, 1996;
Gioia, Isquith, Guy, & Kenworthy, 1996; Guy, Isquith, & Gioia, 1996; N aglieri, & Goldstein,
2013; Roth, Isquith, & Gioia, 2005; Schretlen, 2010). This would likely improve the predictive
power of future models of resilience.
Conclusion
Results from this Canadian sample suggest that resilience and its identified risk and
protective factors function within this Canadian English speaking populations in a way that is
congruent with prior studies from the United States, Switzerland, Taiwan, China, and New
Zealand:
•

Biological sex has long term implications for resilience in early adulthood (Chapple
& Johnson, 2007; Masten, et al., 2004; Ribeaud & Eisner, 2010)

•

Early attachment has long term developmental implications for adaptive functioning
in adulthood (Englund et al. , 2011)

•

Punitive parenting practices have long term developmental implications regarding
antisocial behavior (Villodas, et al., 2015)

•

SES has a significant impact on antisocial behaviors and coping in early adolescence
(Garmezy et al. , 1984; Kraus et al., 2011)

•

Close relationships with parents and friends in early adolescence are key to the
development of problem solving skills (Casselman & Rosenbaum, 2014; Fergusson &

54

Horwood, 2001 ; German et al., 2013; Masten & Tellegen, 2012; Murray et al., 2014;
Newman, et al., 2007; Newton-Howes et al. , 2015; Raby et al. , 2015; Wang et al.,
2013; Werner, 1996; 1989; Zhang et al., 2015)
•

Mentorship though close relationships in early adolescence is crucial to the
development of problem solving skills (Fruiht, 2015; Li & Yang, 2009; McKay,
2011)

Given the observational nature of the present study and limited explained variance, the
findings should be generalized with caution. However, it appears that resilience is a
developmental construct that is relevant for many different populations including English
speaking Canadians. As the literature on resilience suggests it is a complex topic that has been
explained imprecisely at best (Englund et al., 2011; Li & Yang, 2009; Masten & Tellegen, 2012).
Nevertheless, the available evidence collectively makes a strong case for the importance of close
relationships and problem solving skills in developing resilient individuals in English speaking
Canada. Longitudinal studies that target specific risk factors and protective factors with the
intention of building structural equation models are required to improve the explicative power of
studies that address resilience.

55

Tables
Table 1
Key Protective Factors identified by Benzies and Myachesiuk (2009).
Individual
Family
internal
locus
of
family
structure
•
•
•
control
• intimate partner
relationship stability
• emotional regulation
•
• self-efficacy
• family cohesion
•
• effective coping
• supportive parent child
•
skills
interaction
•
• stimulating
• increased education,
skills and training
environment
•
health
social
support
•
•
•
• temperament
• family of origin
influences
• gender
stable
and adequate
•
mcome
• adequate housing

Community
involvement in the
community
peer acceptance
supportive mentors
safe neighborhoods
access to quality
schools
childcare,
access to quality health
care

Benzies, K., & Mychasiuk, R. (2009). Fostering family resiliency: a review of the key protective factors. Child
and Family Social Work, 103-114. doi : 10.1111/j .1365-2206.2008.00586.x

56

Table 2
Exogenous and Endogenous Variables used in calculating the Covariance Matrix.
Latents and
Rating Scale
Indicator Description
Indicators
1 = never, rarely
Physical Abuse
2 = sometimes
My parents (or step parents
3 = often
or foster parents or
4 = always
Abusea, b
guardians) hit me or
96 = not applicable, don't know, refusal,
(HITDPMCcQlPco threaten to do so.
not stated
llapse
Verbal Abuse
1 = never, about once a week or less, a
TELLBADbprcq05
few times a week
collapse)
2 = one or two times a day
How often do you tell
him/her that he/she is bad or 3 = many times each day
96 = not applicable, don't know, refusal,
not as good as others?
not stated
1 =everyday
2 = 4-6 times a week
During the past 12 months,
3 = 2-3 times a week
Parental Alcohol
4 = once a week
how often did you/he/she
Usec
drink beer, wine, liquor or
5 = 2-3 times a month
(PMKDRINKCHL
any other alcoholic
6 = once a month
PcQ05collapse)
7 = less than once a month
beverage? PMK
96 = not applicable, don't know, refusal,
not stated
Biological Sexd
Gender of the Child
1 = female
(SEXReportedSex) Reported by the PMK
2 = male
1 = Rural area
Size of area of residence in
2 = Urban, population< 30,000
Community Sizee
which the child lives,
3 = Urban, population 30,000 to 99,999
(RESIDENCEResi
according to 1996, 2001, or 4 = Urban, population 100,000 to 499,999
denceSize)
2006 Census counts.
5 = Urban, population 500,000 or over
9 = Not stated
1 = Lowestf OR Lower middleg
Family Incomed
2 = Middleh
(INCOMEAINHDO Income adequacy
3 = Upper middlei
?collapse)
4 = Highestj
Relationship with
1 = very close
Overall, how would you
Mo the~
2 = somewhat close
describe your relationship
(MOMDPMCcQ06
3 = not very close, not applicable
with your mother?
collapse)
9 = not stated
Relationship with
1 = very close
Overall, how would you
Fathe~
2 = somewhat close
describe your relationship
(DADDPMCcQ09c
3 = not very close, not applicable
with your father?
ollapse)
9 = not stated
~~~~~~~~~~~

(table continues)

57

Table 2
Exogenous and Endogenous Variables used in calculating the Covariance Matrix.
Latents and
Rating Scale
Indicator Description
Indicators
1 = all of the time
2 = most of the time
How often do you share
Friendships3
3 = some of the time
your secrets and private
(DFFCcQ4A)
feelings with your close
4 = rarely
friends?
5 = never
96 = not applicable, not stated
Verbal Praise
1 = never, about once a week or less
2 = a few times a week
How often do you praise
3 = one or two times a day
name, by saying something
4 = many times each day
like 'Good for you!' or
Supportive
'What a nice thing you did!' 96 = not applicable, don ' t know, refusal,
Parentingb
not stated
or 'That's good going!'?
(PRAISEbprcqO 1co
Quality Time
llapse,
- - - - - - - - - - - - 1 = never, about once a week or less
How often do you and
2 = a few times a week
TALKPLA Ybprcq
he/she talk or play with
02collapse)
3 = one or two times a day
each other, focusing
4
attention on each other for
= many times each day
96 = not applicable, don't know, refusal,
five minutes or more, just
not stated
for fun?
1 = never or not true
Impulsivity
2 = sometimes or somewhat true
I am impulsive, I act
3 = often or very true
Impulsivity3
without thinking.
6 = not applicable, not stated
(DFBCQOlS,
Reactivity
1 = never or not true
DFBCQOlG)
2 = sometimes or somewhat true
3 = often or very true
I get into many fights.
6 = not applicable, not stated
When another kid
1 = never or not true
Coping Skills
accidentally hurts me I
2 = sometimes or somewhat true
3
(hostility)
assume that he/she meant to
3 = often or very true
(DFBCdOlX)
do it, and I react with anger
6 = not applicable, not stated
and fighting.

Optimismk,I
(OPTIMISMHEQY
FQ 1Ocollapse)

Tell me how you feel, think,
or act most of the time in
most situations: ... You're
optimistic about most things
you do

(table continues)
1 = very seldom true or not true, seldom
true
2 = sometimes true
3 = often true
4 = very often true or true
6 = valid skip, don 't know, refusal, not
stated
(table continues)

58

Table 2
Exogenous and Endogenous Variables used in calculating the Covariance Matrix.
Latents and
Rating Scale
Indicator Description
Indicators
Problem solving
How would you rate your:
1 = poor, fair
... ability to solve new
2 = good
problems? For example,
3 = very good
identifying problems and
Problem Solving
4
= excellent
possible
causes,
planning
Skillsk,I
strategies to solve problems 6 = don't know, refusal, not stated
(PROB SOLVEHA
or thinking of new ways to
BYHQ5collapse,
solve problems.
CONFIDENCEHE
Confidence
1 = very seldom true or not true, seldom
QYFQ 1Ocollapse)
Tell me how you feel, think, true
2 = sometimes true
or act most of the time in
3 = often true
most situations: ... You
4 = very often true or true
believe in your ability to
6= valid skip, don't know, refusal, not
handle most upsetting
stated
problems

Note. The Covariance Matrices were calculated using the collapsed categorizations as presented above.
• Statistics Canada. (2004). National Longitudinal Survey of Children and Youth 10 -17 Year old Self- Completed
Questionnaires - Cycle 4, Release 2.
b Statistics Canada. (1998). National Longitudinal Survey of Children and Youth, 1996-1997 [Canada]: Cycle 2,
Primary File .
c Statistics Canada. (2000). National Longitudinal Survey of Children and Youth, 1998-1999 [Canada]: Cycle 3,
Primary File.
d Statistics Canada. (1996). National Longitudinal Survey of Children and Youth, 1994-1995 [Canada]: Cycle 1,
Release 2, Primary File.
• Statistics Canada. (2007). National Longitudinal Survey of Children and Youth, Cycle 6, 2004 Master file
(Longitudinal).
• Statistics Canada. (2005). National Longitudinal Survey of Children and Youth Cycle 4, Primary File - Version
Two.
fLowest: Household income is< 10,000 and household size is 1 - 4 persons; or Household income is < 15,000 and
household size is 5 or more persons
8 Lower middle: Household income is 10,000 - 14,999 and household size is 1-2 persons; or Household income is
10 ,000 -19,999 and household size is
3-4 persons; or House hold income is 15,000-29,999 and household size is 5 or more persons
h Middle: Household income is 15,000-29,999 and household size is 1 -2 persons; or Household income is 20,00039,999 and household size is 3 -4
persons; or Household income is 30,000 -59,999 and household size is 5 or more persons
; Upper middle: Household income is 30,000 - 59,999 and household size is 1-2 persons; or Household income is
40,000 -79,999 and household size is 3-4 persons; or House hold income is 60,000-79,999 and household size is 5
or more persons
i Highest: Household income is 60,000 or more and household size is 1-2 persons; or Household size is 80,000 or
more and household size is 3 or more persons
k Statistics Canada. (2010). National Longitudinal Survey of Children and Youth, Cycle 8, 2008 Master File
(Youth).
1
Some categories have been collapsed to protect confidentiality due to low cell counts. Statistics Canada. (2009).
Microdata User Guide National Longitudinal Survey of Children and Youth Cycle 8 September 2008 to July 2009.

59

Table 3
Fixed Error Coefficients for the Primary Group and the Supplemental Group.

Cl)

::,
0

c::
(1)

b1)

0

><

Jil

Cl)

::,
0

c::(1)

b1)

0

]

Jil

Primary Group
Fixed Error
Coefficient
Var.

Latent Concept

Indicator

Assessed
Error
Variance

Physical Abuse
Verbal Abuse
Parental Alcohol Use
Biological Sex
Community Size
Family Income
Relationship with Mother
Relationship with Father
Friendships
Verbal Praise
Quality Time

HITDPMCcQlPcollapse
TELLBADbprcq05collapse
PMKDR1NKCHLPcQ05collapse
SEXReportedSex
RES IDEN CEResidenceS ize
INCOMEAINHD07 collapse
MOMDPMCcQ06collapse
DADDPMCcQ09collapse
DFFCcQ4A
PRAISEbprcqO 1collapse
TALKPLA Ybprcq02collapse

0.40
0.05
0.07
0.01
0.12
0.12
0.02
0.09
0.05
0.07
0.07

0.511
0.230
3.128
0.248
2.175
0.810
1.003
0.235
0.419
0.688
0.663

0.205
0.012
0.219
0.003
0.261
0.097
0.024
0.021
0.050
0.048
0.046

Impulsivity
Reactivity
Coping Skills (hostility)
Optimism
Problem Solving Skills
Confidence

DFBCQOlS
DFBCdOlX
DFBCQOlG
OPTIMISMHEQ YFQ 1Ocollapse
PROBSOLVEHABYHQ5collapse
CONFIDENCEHEQYFQ 1Ocollapse

0.12
0.07
0.05
0.05
0.05
0.05

0.389
0.257
0.207
0.709
0.671
0.709

0.047
0.018
0.010
0.035
0.035
0.035

Supplemental Group
Fixed Error
Coefficient
Var.
0.239
3.338
0.251
2.282
0.831

0.012
0.234
0.003
0.274
0.100

0.619
0.543

0.043
0.038

0.656
0.687
0.656

0.033
0.034
0.033

Note. Fixed Error Coefficients were not calculated for the missing variables in the Supplemental Group and are indicated by - -.

60

Table 4
Frequencies for Collapsed Exogenous and Endogenous Indicators.
Indicators
NLSCY Response Options
Collapsed Response Options a, b
Physical Abuse
1 = never
I = never, rarely
2 = sometimes
2 = rarely
3 = sometimes
3 = often
4 = often
4 = always
5 = always
Verbal Abuse
I = never
1 = never, about once a week or less, a few times
2 = about once a week or less
a week
3 = a few times a week
2 = one or two times a day
3 = many times each day
4 = one or two times a day
5 = many times each day
Family Income
l = Lowest
1 = Lowest OR Lower middle
2 = Lower middle
2 = Middle
3 = Middle
3 = Upper middle
4 = Upper middle
4 = Highest
5 = Highest
Relationship with I = very close
1 = very close
Mother
2 = somewhat close
2 = somewhat close
3 = not very close
3 = not very close, not applicable
6 = not applicable
Relationship with I = very close
I = very close
Father
2 = somewhat close
2 = somewhat close
3 = not very close
3 = not very close, not applicable
6 = not applicable
Verbal Praise
1 = never
1 = never, about once a week or less
2 = about once a week or less
2 = a few times a week
3 = a few times a week
3 = one or two times a day
4 = one or two times a day
4 = many times each day
5 = many times each day

Collapsed Frequencies
491
85
46
20
556
57
29
84
221
23 7
I 00
4 73
150
19
3 88
199
56
19
142
233
248
(table continues)

61

Table 4
Frequencies for Collapsed Exogenous and Endogenous Indicators.
Indicators
NLSCY Response Options
Collapsed Response Options a, b
Quality Time
l = never
1 = never, about once a week or less
2 = about once a week or less
2 = a few times a week
3 = one or two times a day
3 = a few times a week
4 = one or two times a day
4 = many times each day
5 = many times each day
Optimism
l = very seldom true or not true
1 = very seldom true or not true, seldom true
2 = seldom true
2 = sometimes true
3 = often true
3 = sometimes true
4 = often true
4 = very often true or true
5 = very often true or true
Problem Solving
l = poor
l = poor, fair
Skills
2 = fair
2 = good
3 = good
3 = very good
4 = very good
4 = excellent
5 = excellent
l = very seldom true or not true, seldom true
Confidence
l = very seldom true or not true
2 = sometimes true
2 = seldom true
3 = sometimes true
3 = often true
4 = often true
4 = very often true or true
5 = verr often true or true

Collapsed Frequencies
40
174
310
118
33
103

270
236
49

279
226
88

33
103

270
236

Note. Categories were collapsed according to Statistics Canada Release Guidelines., h
•statistics Canada. (2009). Statistics Canada Quality Guidelines. Retrieved from http://www.statcan.gc.ca/pub/ l 2-539-x/ I 2-539-x2009001-eng.htm
hStatistics Act, Revised Statues of Canada ( 1985, c. S-19). Retrieved from http://laws-lois.justice.gc.ca/eng/acts/S-19/F ullText.html

62

Table 5

Descriptive Statistics for the Primary Group and the Supplemental Group.
Supplemental Group (n=298)

Primary Group (n=642)
%

SD

1987

45.44

1988
Sex

%

SD

C.I. (95%)
Upper

Lower

4.647

42.37

60.6

48.51

4.647

39.4

57.63

60 .59

49.10

4.567

40.14

58.06

39.41

51.09

50.90

4.567

41.94

59.86

1.586

8.08

14.31

12.64

2.424

7.89

17.4

29.21

2.679

23.95

34.46

31.76

4.44

23.05

40.48

Upper middle

41.48

3.195

35.21

47 .75

39.46

4.962

29 .72

49.19

Highest

18.12

2.693

12.83

23.4

16.14

3.761

8.76

23 .52

1 l.89

1.315

9.31

14.47

14.25

2.714

8.92

19.57

< 30,000

30.09

2.711

24.77

35.41

25.02

3.467

18.22

31.83

30,000 - 99,999

9.41

1.624

6.22

12.6

8.00

1.867

4.34

11 .66

100,000 to
499,999

15.79

1.835

12.19

19.39

17.80

2.847

12.21

23.39

> 500,000

32.82

3.469

26.01

39.62

34.93

5.621

23.9

45.96

Yes

88.88

2.267

84.43

93.33

89.36

2.199

85.05

93 .68

No

11.12

2.267

6.67

15.57

10.64

2 .199

6.32

14.95

Child Variables
Birth Year

C.I. (95%)
Upper

Lower

3.017

39.52

51.36

51.49

54.56

3.017

48 .64

60.48

Female

54.75

2.975

48.91

Male

45.25

2.975

Lowest or Low

11.19

Middle

Family Variables
Income Adequacy

Community Size

Rural
Urban

Spouse in the home

63

RESILIENT CANADIANS: PREDICTORS FROM THE NLSCY
Table 6
Direct Effects Endogenous Pathway Estimates for the Model in Figure 6.
Coping
Impulsivity
Skills
Reactivity Optimism
(hostility)
Est.
Impulsivity
SE
Sig.
Coping
Skills
(hostility)

Est.
SE
Sig.

0.139*
0.033
4.184

Reactivity

Est.
SE
Sig.

0.166*
0.037
4.511

Optimism

Est.
SE
Sig.

Resilience

Est.
SE
Sig.

Problem
Solving
Skills

Est.
SE
Sig.

64

Resilience

Problem
Solving

0.168*
0.038
4.466

0.066
0.077
0.857

0.099*
0.038
2.563

1.000

0.099*
0.038
2.563

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
* p < .05

64

Table 7
Direct Exogenous PathwaJ:._ Estimates [gr the Model in Figure 6.
Verbal Physical
PMK
Biological
Abuse
Abuse
Alcohol
Sex
0.226*
-0.004
Est. -0.123*
0.180*
Impulsivity
0.016
SE
0.054
0.059
0.051
Sig.
-2.268
-0.232
3.838
3.549
Coping
Skills
(hostility)

Est.
SE
Sig.

Reactivity

Est.
SE
Sig.

Optimism

Est.
SE
Sig.

Resilience

Est.
SE
Sig.

Problem
Solving
Skills

Est.
SE
Sig.

--

-

Community
Size
-0.007
0.020
-0.363

Family
Income
-0.011
0.035
-0.326

Relate
Mom
0.116
0.067
1.727

Relate
Dad
-0.018
0.048
-0.383

Friendship

--

0.123*
0.043
2.902

--

-0.016
0.038
-0.419

--

--

0.075
0.047
1.588

--

0.148*
0.042
3.501

--

-0.060*
0.024
-2.509

0.080
0.044
1.813

--

0.067*
0.021
3.124

-0.049
0.073
-0.670

--

--

--

--

--

--

--

-0.071 *
0.035
-2.031

--

--

--

2.966*
1.377
2.155

--

--

--

-1.024
0.663
-1.546

-0.077
0.368
-0.209

0.112*
0.037
3.052

--

0.161 * 0.067*
0.046
0.033
3.505
2.046

-0.015
0.019
-0.796

Verbal
Praise

--

Quality
Time
-0.100*
0.033
-2.997

-0.018
0.025
-0.728

0.038
0.026
1.473

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
* p < .05

65

Table 8
Total Indirect Endogenous Pathway Estimates for the Model in Figure 6.
Coping
Reactivity Optimism
Impulsivity
Skills
(hostility)
Est.
Impulsivity
SE
Sig.

Resilience

Problem
Solving

Coping
Skills
(hostility)

Est.
SE
Sig.

Reactivity

Est.
SE
Sig.

Optimism

Est.
SE
Sig.

0.011
0.013
0.845

Resilience

Est.
SE
Sig.

0.011
0.013
0.845

0.067
0.078
0.857

0.011
0.008
0.842

0.010
0.013
1.269

0.010
0.008
1.269

0.001
0.001
0.849

Problem
Solving
Skills

Est.
SE
Sig.

0.001
0.001
0.801

0.007
0.008
0.812

0.001
0.001
0.799

0.099
0.040
2.513

0.001
0.001
0.849

0.010
0.008
1.269

0.028*
0.009
3.172

0.011
0.013
0.842

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
* p < .05

66

Table 9
Total Indirect Exog_enous PathwaJ!._ Estimates jj_Jr the Model in Figure 6.
Verbal Physical
PMK
Biological Community
Abuse
Alcohol
Size
Abuse
Sex
Est.
---Impulsivity
SE
Sig.

Family
Income

Relate
Mom

----·--

Relate
Dad

Friendship

Verbal
Praise

Quality
Time

Coping
Skills
(hostility)

Est.
SE
Sig.

-0.020*
0.010
-2.058

0.050*
0.014
3.569

-0.001
0.003
-0.232

0.055*
0.014
4.017

-0.001
0.003
-0.362

-0.012
0.007
-1.611

0.033*
0.014
2.317

-0.003
0.008
-0.382

0.011 *
0.004
2.568

--

-0.017*
0.007
-2.543

Reactivity

Est.
SE
Sig.

-0.020*
0.010
-2.026

0.038*
0.013
2.985

-0.001
0.003
-0.232

0.030*
0.011
2.779

-0.001
0.003
-0.362

-0.002
0.006
-0.326

0.019
0.012
1.616

-0.003
0.008
-0.381

--

--

-0.017*
0.007
-2.486

Optimism

Est.
SE
Sig.

0.006
0.008
0.809

0.011
0.014
0.839

0.000
0.000
-0.224

0.003
0.004
0.651

0.000
0.000
-0.334

-0.001
0.001
-0.757

0.013
0.015
0.839

0.004
0.005
0.781

0.000
0.001
-0.193

-0.001
0.002
-0.555

0.001
0.002
0.592

Resilience

Est.
SE
Sig.

-0.044
0.074
-0.591

0.012
0.014
0.839

0.000
0.000
-0.224

0.032*
0.014
2.282

0.000
0.000
-0.334

-0.001
0.001
-0.757

0.013
0.015
0.839

-0.006
0.008
-0.678

-0.072*
0.035
-2.048

-0.00 l
0.002
-0.555

0.001
0.002
0.592

Problem
Solving
Skills

Est.
SE
Sig.

-0.004
0.007
-0.576

0.001
0.001
0.796

0.000
0.000
-0.223

0.295*
0.070
4.204

0.000
0.000
-0.331

0.000
0.000
-0.725

0.001
0.002
0.796

-0.101
0.052
-1 .943

-0.015
0.036
-0.406

0.000
0.000
-0.542

0.000
0.000
0.576

Note. The table reads from the column variable to the row variable regarding causality.
Note . Estimates are Unstandardized .
* p < .05

67

Table 10
Stacked Model Direct Endog_enous PathwaJ:'._ Estimates for the Model in Figure 7.
Coping Skills
Impulsivity
Reactivity
(hostilit )
Primary
Supplemental
Primary
Supplemental
Primary
Supplemental
Group
Est.
Impulsivity SE
Sig.

--

--

--

0.070
0.062
1.124

0.070
0.062
1.124

--

--

--

--

--

--

Coping
Skills
(hostility)

Est.
SE
Sig.

0.138*
0.025
5.408

0.138*
0.025
5.408

Reactivity

Est.
SE
Sig.

0.161 *
0.028
5.702

0.161*
0.028
5.702

Optimism

Est.
SE
Sig.

--

Resilience

Est.
SE
Sig.

Problem
Solving
Skills

Est.
SE
Sig.

Optimism

Resilience

Primary

Supplemental

Primary

Supplemental

0.169*
0.030
5.602

0.169*
0.030
5.602

--

--

--

1.000

1.000

--

--

--

--

--

--

--

0.109*
0.036
3.024

0.212*
0.050
4.212

Problem Solving
Skills
Primary

Supplemental

0.109*
0.036
3.024

0.212*
0.050
4.212

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
Note. Italics indicate coefficient estimates in the Supplemental Group that were constrained to be equal to the Primary Group.

* p < .05

68

Table 11
Stacked Model Direct Exogenous Pathway Estimates for the Model in Figure 7.
Verbal Abuse
Physical Abuse
PMKAlcohol
Supplemental
Primary
Primary
Supplemental
Primary
Supplemental
Group

Biological Sex

Community Size

Family Income

Primary

Supplemental

Primary

Supplemental

Primary

Supplemental

-0.007
0.020
-0.340

-0.001
0.027
-0.037

-0.013
0.036
-0.356

0.000
0.045
-0.008

--

--

-0.060*
0.024
-2.511

-0.070*
0.034
-2.072

Impulsivity

Est.
SE
Sig.

-0.126*
0.055
-2.287

-0.058
0.073
-0.801

0.164*
0.038
4.375

0.164*
0.038
4.375

-0.002
0.016
-0.136

0.020
0.021
0.924

0.190 *
0.051
3.726

0.178*
0.069
2.597

Coping Skills
(hostility)

Est.
SE
Sig.

0.110*
0.037
2.978

0.099
0.051
1.927

0.091 *
0.027
3.405

0.091*
0.027
3.405

--

--

-0.009
0.037
-0.247

-0.001
0.050
-0.025

Reactivity

Est.
SE
Sig.

--

--

0.057
0.030
1.909

0.057
0.030
1.909

--

--

0.152*
0.042
3.592

0.167*
0.059
2.812

Optimism

Est.
SE
Sig.

-0.050
0.072
-0.689

-0.180
0.105
-1.710

Resilience

Est.
SE
Sig.

--

--

--

--

2.556*
1.058
2.417

0.275
0.497
0.554

Problem Solving
Skills

Est.
SE
Sig.

(table continues)

69

Table 11
Stacked Model Direct Exogenous Pathway Estimates for the Model in Figure 7.
Relate Mom
Relate Dad
Friendship
Primary
Supplemental
Primary
Supplemental
Primary
Supplemental
Group
Est.
0.111 * 0.111*
-0.012
-0.012
--Impulsivity
0.038
SE
0.052
0.052
0.038
Sig.
2.112
2.112
-0.322
-0.322
Coping Skills
(hostility)

Est.
SE
Sig.

0.156*
0.036
4.351

0.156*
0.036
4.351

0.072*
0.026
2.749

0.072*
0.026
2.749

-0.017
0.015
-1.109

-0.017
0.015
-1.109

Reactivity

Est.
SE
Sig.

0.080*
0.035
2.296

0.080*
0.035
2.296

--

--

0.066*
0.022
3.031

0.060*
0.029
2.050

Optimism

Est.
SE
Sig.

--

--

--

--

-0.071 *
0.034
-2.061

-0.064
0.049
-1.311

Resilience

Est.
SE
Sig.

--

--

-0.683*
0.321
-2.126

-0.683*
0.321
-2.126

0.110
0.196
0.560

0.110
0.196
0.560

Problem Solving
Skills

Est.
SE
Sig.

Verbal Praise

Quality Time

Primary

Primary

Supplemental

-0.110*
0.033
-3.272

-0.134*
0.053
-2.534

0.034
0.026
1.300

0.028
0.042
0.665

Supplemental

--

--

-0.020
0.025
-0.811

-0.027
0.039
-0.680

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
Note. Italics indicate coefficient estimates in the Supplemental Group that were constrained to be equal to the Primary Group.
* p < .05

70

Table 12
Stacked Model Total Indirect Endogenous Pathway Estimates for the Model in Figure 7.
Coping Skills
Impulsivity
Reactivity
Optimism
(hostility)
Primary
Supplemental
Primary
Supplemental
Primary
Primary
Supplemental
Supplemental
Group
Est.
Impulsivity SE
Sig.

Resilience

Problem Solving
Skills

Primary

Supplemental

Primary

Supplemental

0.027*
0.007
4.013

Coping
Skills
(hostility)

Est. 0.027*
SE 0.007
Sig. 4.013

Reactivity

Est.
SE
Sig.

Optimism

Est.
SE
Sig.

0.012
0.010
1.107

0.012
0.010
1.107

--

--

0.012
0.011
1.102

0.012
0.011
1.102

Resilience

Est.
SE
Sig.

0.012
0.011
1.107

0.012
0.011
1.107

0.071
0.063
1.124

0.073
0.065
1.123

0.012
0.011
1.102

0.012
0.011
1.102

0.012
0.008
1.494

0.047*
0.017
2.799

0.012
0.008
1.494

0.047*
0.017
2.799

0.001
0.001
1.000

0.010
0.005
1.894

Problem
Solving
Skills

Est.
SE
Sig.

0.001
0.001
1.037

0.003
0.002
1.084

0.008
0.007
1.050

0.015
0.014
1.100

0.001
0.001
1.032

0.003
0.002
1.079

0.110*
0.037
2.952

0.221 *
0.041
5.357

0.001
0.001
1.000

0.010
0.005
1.894

0.012
0.008
1.494

0.047*
0.017
2.799

Note . The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
Note. Italics indicate coefficient estimates from in the Supplemental Group that were constrained to be equal to the Primary Group.
* p < .05

71

Table 13
Stacked Model Total Indirect Exogenous Pathway Estimates for the Model in Figure 7.
Biological Sex
Physical Abuse
PMK Alcohol
Verbal Abuse
Supplemental
Primary
Supplemental
Primary
Supplemental
Primary
Supplemental
Primary
Group

Community Size

Family Income

Primary

Supplemental

Primary

Supplemental

Impulsivity

Est.
SE
Sig.

Coping
Skills
(hostility)

Est.
SE
Sig.

-0.021 *
0.010
-2.153

-0.010
0.009
-1.042

0.037*
0.009
4.115

0.037*
0.009
4.115

0.000
0.003
-0.136

0.003
0.003
1.167

0.057*
0.013
4.471

0.058*
0.013
4.498

-0.001
0.003
-0.340

0.000
0.003
-0.048

-0.012
0.007
-1.671

-0.012
0.007
-1.606

Reactivity

Est.
SE
Sig.

-0.020*
0.010
-2.124

-0.009
0.009
-1.040

0.027*
0.008
3.499

0.027*
0.008
3.499

0.000
0.003
-0.136

0.003
0.003
1.161

0.031*
0.010
3.116

0.029*
0.010
2.983

-0.001
0.003
-0.340

0.000
0.003
-0.048

-0.002
0.006
-0.356

0.000
0.006
-0.010

Optimism

Est.
SE
Sig.

0.006
0.006
1.014

0.006
0.006
1.013

0.009
0.008
1.093

0.009
0.008
1.093

0.000
0.000
-0.135

0.000
0.000
0.809

0.003
0.004
0.842

0.004
0.004
0.898

0.000
0.000
-0.325

0.000
0.000
-0.048

-0.001
0.001
-0.933

-0.001
0.001
-0.921

Resilience

Est.
SE
Sig.

-0.044
0.073
-0.604

-0.182*
0.075
-2.410

0.009
0.008
1.093

0.009
0.009
1.093

0.000
0.000
-0.135

0.000
0.000
0.809

0.035*
0.013
2.686

0.015
0.048
0.315

0.000
0.000
-0.325

0.000
0.000
-0.048

-0.001
0.001
-0.933

-0.001
0.001
-0.921

Problem
Solving
Skills

Est.
SE
Sig.

-0.005
0.008
-0.592

-0.038*
0.017
-2.200

0.001
0.001
1.025

0.002
0.002
l.071

0.000
0.000
-0.135

0.000
0.000
0.800

0.287*
0.066
4.326

0.053
0.234
0.227

0.000
0.000
-0.323

0.000
0.000
-0.048

0.000
0.000
-0.889

0.000
0.000
-0.908

(table continues)

72

Table 13
Stacked Model Total Indirect Exogenous Pathway Estimates for the Model in Figure 7.
Relate Dad
Friendship
Verbal Praise
Relate Mom
Primary
Supplemental
Primary
Supplemental
Primary
Supplemental
Primary
Supplemental
Group
-Est.
Impulsivity
SE
Sig.

Quality Time
Primary

Supplemental

Coping
Skills
(hostility)

Est.
SE
Sig.

0.032*
0.011
2.890

0.032*
0.011
2.890

-0.002
0.006
-0.321

-0.002
0.006
-0.321

0.011 *
0.004
2.666

0.010*
0.004
2.473

--

--

-0.018*
0.006
-2.911

-0.022*
0.007
-3.389

Reactivity

Est.
SE
Sig.

0.018
0.009
1.983

0.018
0.009
1.983

-0.002
0.006
-0.321

-0.002
0.006
-0.321

--

--

--

--

-0.018*
0.006
-2.838

-0.022*
0.007
-3.271

Optimism

Est.
SE
Sig.

0.013
0.012
1.097

0.013
0.012
1.097

0.005
0.005
1.032

0.005
0.005
1.032

0.000
0.001
-0.358

0.000
0.001
-0.412

-0.001
0.002
-0.658

-0.002
0.002
-0.772

0.001
0.002
0.527

0.000
0.002
0.212

Resilience

Est.
SE
Sig.

0.013
0.012
1.097

0.014
0.012
1.097

-0.003
0.007
-0.399

-0.025
0.016
-1.566

-0.071 *
0.035
-2.027

-0.062
0.037
-1.667

-0.001
0.002
-0.658

-0.002
0.003
-0.772

0.001
0.002
0.527

0.000
0.002
0.212

Problem
Solving
Skills

Est.
SE
Sig.

0.001
0.001
1.028

0.003
0.003
1.075

-0.071
0.036
-1.991

-0.143*
0.066
-2.169

0.005
0.021
0.241

0.012
0.042
0.278

0.000
0.000
-0.642

0.000
0.001
-0.764

0.000
0.000
0.519

0.000
0.000
0.211

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Unstandardized.
Note. Italics indicate coefficient estimates in the Supplemental Group that were constrained to be equal to the Primary Group.
* p < .05

73

Table 14
Stacked Model Endogenous Common Metric Standardized Solution for the Model in Figure 7.
Coping Skills
Optimism
Impulsivity
Reactivity
(hostility)
Primary
Primary
Primary
Supplemental
Supplemental
Primary
Supplemental
Supplemental
Group
Impulsivity

Est.

Coping
Skills
(hostility)

Est.

0.185a

0.185a

Reactivity

Est.

0.195a

0.195 a

Optimism

Est.

Resilience

Est.

Problem
Solving
Skills

Est.

0 . 187 a

0.039

Resilience
Primary

Supplemental

Problem Solving
Skills
Primary

Supplemental

0.041

0.080

O.J8r

0.039

0.381b

0.381b

0.290a

0.563 C

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Standardized.
Note. Italics indicate coefficient estimates from the Supplemental Group that were constrained to be equal to the Primary Group.
Note. Effect sizes are •small (.10), bmedium (.30), and <large (.50) Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155-159. doi: I0.1037/00332909.112. l.l 55
74

Table 15
Stacked Model Exogenous Common Metric Standardized Solution for the Model in Figure 7.
Verbal Abuse
Physical Abuse
PMK Alcohol
Biological Sex
Group Primary Supplemental Primary Supplemental Primary Supplemental Primary Supplemental
Impulsivity

Est.

-0.099

-0.046

0.167 a

0.16?8

Coping Skills (hostility)

Est.

0.116 a

0.104 a

0.124 a

Reactivity

Est.

--

--

0.070

Optimism

Est.

-0.029

-0.104 a

Resilience

Est.

Problem Solving Skills

Est.
Relate Mom
Primary Supplemental

0.157 a

0.148 a

0.124 a

-0.010

-0.001

0.070

0.152 a

0.167 a

0.606c

0.055

-0.006

0.056

Friendship
Primary Supplemental

--

--

Impulsivity

Group
Est.

0.086

0.086

Relate Dad
Primary Supplemental
-0.013 a -o.013 a

Coping Skills (hostility)

Est.

0.163 a

0.163 a

0.102 a

o.1or

-0.037

-0.037

Reactivity

Est.

0.076

0.076

--

--

0.131 a

0.119 a

Optimism

Est.

--

--

--

--

-0.086

-0.077

Resilience

Est.

--

--

0.054

0.054

Problem Solving Skills

Est.

-0.194 a -o.194 a

Community Size
Primary Supplemental

Family Income
Primary Supplemental

-0.016

-0.018

-0.002

0.000

-o.103 a -o.12o a

Verbal Praise
Primary Supplemental

--

--

Quality Time
Primary Supplemental
-o.139 a -0.170 a

-0.036

-0.047

0.058

0.047

Note. The table reads from the column variable to the row variable regarding causality.
Note. Estimates are Standardized.
Note. Italics indicate coefficient estimates from the Supplemental Group that were constrained to be equal to the Primary Group.
Note. Effect sizes are •small (.10), hmedium (.30), and <Jarge (.50) Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1 ), 155-159. doi: 10.1037/0033-2909.l 12. l.155

75

Figures

Family
Income

-.----

Coping
Skill s
-.-Use

Fri endship

1

. 1.0

1.0l
Income

Sex
CYCLE I
1994/95
Age:7-8

1.0

I

1.0

1.0

I"lkPlay
Praise
CYCLE2
1996/97
Age: 9-10

1.0
1.0

DrinkPMK

HitParnt
CYCLE 3
1998/99
Age : 11-12

11

ClosoMom

I ResSize I

1.0

1.0

~

Friends

j
ReactAng

CYCLE4
2000/01
Age : 13-14

1.0

1.0

1.0

Many Figh

I DiffWait

SolvePro

HandlePr

Optimism

CYCLES
2008/09
Age:21-22

Figure 1. Proposed SEM Model

76

Fa mily
Income

--.--

Coping
Skill s

---

Use

_j_

1

1.0

Sex

1.0

Income

I

CYCLE 1
1994/95
Age:7-8

,U_Jo

t,~Play/
Prai se

[ TellsBa~

CYCLE 2
1996/97
Age: 9-10

I . 1.0_]

_l

1.0

1.0

Dn4MKI
-

HitParnt
CYCLE3
1998/99
Age : 11-12

CloseMom

I ResSize I

I

I

l

1.0

_j

1.0

I
1.0

\

\

1.0

11

I

1.0

\

1.0

friends

I CloseDad
CYCLE4
2000/01
Age : 13-14

CYCLE8
2008/09
Age :21-22

Figure 2. Proposed SEM model removal of Handle Problems Variable

77

io
log
i
c
a
S
ex

I

-

I

:Suppo
r
t
iv
eP
a
r
en
t
ing:
V
e
rba
]

I -

I
I
I

'
'

I
.I

~

~

T
ime

1_
_
_
_
_

I
P7 .

:----~---~----:

~

,
;
:J
I

Cop
ingSk
il
l
s

•)

I

,

U
s
e

~

r
-

:

I

Abu
s
e

~
·
-'
I

I
~

I
I
I
I

I

1
·
-·
-·
-·
-

J
.O

1
.0

1
.0

i

ln
com
e

P
c
,
;
,
e

I
~~

\

i

1
.
0

1
.
0

1
11Tdl,B,dI \D,iru<l'MK

S
ex

I
T
a
lkP
l
ay

CYCLE I
1
9
9
4
/
9
5
Ag
e
:7
-8

CYCLE2
1
9
9
6/
9
7
Ag
e
:9
-10

jj

,
.
o
~

I
H
i
tP
a
rn
t
11

jI

CYCLE3
1
9
9
8/
9
9
Ag
e:11
-12

i

1
.
0
1
.
0

I
1
.
0

~

-1L___
,C
l
o
s
eM
:
:
1
:
r
i
,
n
d
s
R
e
sS
i
z
e

Cl
o
s
e
D
ad

I .1.0

1
.
0

1
.0

!

1
.
0

1
.
0

I/I
M
o
o
yF
i
g,
h
~
R
e
a
c
tAng

CYCLE4
2000/
01
Ag
e
: 13
-14

D
i
f
tW
a
i
t

i

1
.
0

ISo
lv
eP
ro

Op
t
im
ism

CYCLE8
2008/
09
Ag
e:
21
-22

F
igu
r
e3
.P
ropo
s
edSEMmod
e
lw
i
ths
ing
l
eind
i
c
a
to
r
s
.
7
8

Bio logica l
Sex
Family
Income

Verbal

Optimism

1.0

...............

--

Prob lem
Solvin

CYCLE I
1994/95
Agc:7-8

CYCLE2
1996/97
Agc : 9-10

CYCLE 3
1998/99
Agl': I 1-12

CYCLE 4
2000/0 1
Age: 13-14

CYCLE 8
2008/09
Agc:2 1-22

Figure 4. Proposed SEM model with Optimism as a cause of Resilience.

79

Biologica l
Sex
Family
Inco me

CYCLE I
1994/95
Age:7-8

CYCLE2
1996/97
Age: 9- 10

CYCLE 3
1998/99
Age: 11 - 12

CYCLE4
2000/01
Age: 13-14

CYCLE 8
2008/09
Agc:2 1-22

Figure 5. Proposed SEM model with additional pathways from Friendships to Reactivity and from Quality Time to Impulsivity.

80

Tliolog il·al Sl:x

Family Income

Vcrbul Prui sc

Use

Optimi sm

Commun ity
Size

CYCLE I
1994/95

Age:7-8

Problem
Solving

CYCLE 2

CYCLE 3

CYCLE 4
2000/0 1

Age: 9-10

Age: 11- 12

Age: 13- 14

1996/97

199X/99

CYCLES
2008/09
Age:Z l-22

Fig ure 6. Proposed SEM model with the additional pathway from Verbal Abuse to Coping Skills and reciprocal effects between
Resilience and Problem Solving held equal.

81

CYCI. I' I
1994/95
Age:7-8

CYCLE 2
1996/97
Age: 9-1 11

CYCLE 3

CYCLE 4

CYl'I.E 8
2!Xl8!09
Age:2 1-22

CYCI.F.4
2000/01
Age: 13- 14

C'YCLL 8
2008/09

2000/01
Age: 13-1 4

1998/99
Age: 11 - 12
Primar) Group N=642

CYCLE 1
1994195
Agc:7-8

CYCLL: 2
1996!97
Age: 9- 10

CYCLE 3

1998!99
Age: 11 - 12

Agc:21-22

Supplernen Lal Grou p :-J~ 298

Figure 7. Proposed SEM model with two groups.

82

References
Afifi, T., & MacMillan, H. (2011). Resilience Following Child Maltreatment: A Review of
Protective Factors. The Canadian Journal of Psychiatry, 56(5), 266-272. Retrieved from
http: //www.ncbi.nlm.nih.gov/pubmed/21586 l 92
Aldwin, C., Cunningham, M., & Taylor, A. (2010). Resilience Across the Life Span: A Tribute
to Emmy E. Werner. Research in Human Development, 7(3), 159-163.
doi: 10.1080/ 15427609.2010.504502
Aven, T. (2011). On Some Recent Definitions and Analysis Frameworks for Risk, Vulnerability,
and Resilience. Risk Analysis, 31(4), 515-522. doi: 10.1111/j.1539-6924.2010.01528.x
Barbieri, M., & Oullette, N. (2012). The Demography of Canada and the United States from the
1980s to the 2000s A Summary of Changes and a Statistical Assessment. Population-£,
67(June 2012), 177-280. doi: 10.3917/pope.1202.0177
Barkley, R. A. (2012). Barkley Deficits in Executive Functioning Scale-Children and
Adolescents.
BarOn, R. (2004). The Bar-On Emotional Quotient Inventory (EQ-i): Rationale, description, and
summary of psychometric properties. In G. Geher (Ed.), Measurement of emotional
intelligence: Common ground and controversy (pp. 111-142). New York: Nova Science
Publishers.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of
Personality and Social Psychology, 51(6), 1173-1182. doi:10.1037/0022-3514.51.6.1173
Barrett, P. (2007). Structural equation modelling: Adjudging model fit. Personality and
Individual Differences, 42(5), 815-824. doi:10.1016/j.paid.2006.09.018
Bennett, P. R., Lutz, A. C., & Jayaram, L. (2012). Beyond the Schoolyard: The Role of Parenting
Logics, Financial Resources, and Social Institutions in the Social Class Gap in Structured
Activity Participation. Sociology of Education, 85(2), 131-157. doi:
10.1177/0038040711431585
Benzies, K., & Mychasiuk, R. (2009). Fostering family resiliency: a review of the key protective
factors. Child and Family Social Work, 103-114. doi:10.1111/j.1365-2206.2008.00586.x
Bijttebier, P., & Goethals, E. (2006). Parental drinking as a risk factor for children's
maladjustment: The mediating role of family environment. Psychology ofAddictive
Behaviors, 20(2), 126-130. doi: 10.1037/0893-164X.20.2.126

83

Black-Hughes, C., & Stacy, P. D. (2013). Early childhood attachment and its impact on later life
resilience: A comparison of resilient and non-resilient female siblings. Journal of
Evidence-Based Social Work, 10(5), 410-420. doi:10.1080/15433714.2012.759456
Boardman, J. D., Blalock, C . L., & Button, T. M. (2008). Sex differences in the heritability of
resilience. Twin Research and Human Genetics, 11(1), 12-27. doi: 10.1375/twin.11.1.12
Bollen, K., & Lennox, R. (1991). Conventional wisdom on measurement: A structural equation
perspective. Psychological Bulletin, 110(2), 305-314.
Bradley, R.H., & Corwyn, R. (2013). From parent to child to parent ... : Paths in and out of
problem behavior. Journal ofAbnormal Child Psychology, 41(4), 515-529.
doi: 10.1007/sl 0802-012-9692-x
Bronfenbrenner, U. (1977). Toward an Experimental Ecology of Human Development.
American Psychologist, (July), 513-531.
Bronfenbrenner, U., & Ceci, S. J. (1994). Nature-Nurture Reconceptualized in Developmental
Perspective: A Bioecological Model. Psychological review, 101(4), 568-586. Retrieved
from http://www.ncbi.nlm.nih.gov/pubmed/7984 707
Buckner, J.C., Mezzacappa, E., & Beardslee, W.R. (2003). Characteristics ofresilient youths
living in poverty: The role of self-regulatory processes. Development and
Psychopathology, 15(1 ), 139-162. doi: 10.1017/S0954579403000087
Burt, K. B., & Paysnick, A. a. (2012). Resilience in the Transition to Adulthood. Development
and Psychopathology, 24(2), 493-505. doi: 10.1017/S0954579412000119
Calkins, S. D., Blandon, A. Y., Williford, A. P., & Keane, S. P. (2007). Biological, Behavioral,
and Relational Levels Of Resilience in the Context of Risk for Early Childhood Behavior
Problems. Development and Psychopathology, 19(3), 675-700.
doi: 10.1017/S095457940700034X
Canadian Psychological Association. (2000). Canadian Code ofEthics for Psychologists (3rd
ed.). Ottawa, Ontario: Canadian Psychological Association.
Carillo, I., Kovacevic, M., & Wu, C. (2006). Analysis of longitudinal survey data with missing
observations: An application of weighted gee to the National Longitudinal Survey of
Children and Youth (NLSCY). 2006 Annual Meeting in London (pp. 1-11 ). London:
Statistical Society of Canada. Retrieved from
http: //www.ssc.ca/swvey/documents/SSC2006 Ivan Carillo.pdf

84

Casselman, R. B., & Rosenbaum, A. (2014). Fathers, Sons, and Aggression: A Path Model.
Journal ofAggression, Maltreatment & Trauma, 23(5), 513-531.
doi: 10.1080/ 10926771.2014.904464
Chapple, C. L., & Johnson, K. A. (2007). Gender Differences in Impulsivity. Youth Violence and
Juvenile Justice, 5(3), 221-234. doi: 10.1177/1541204007301286
Chien, N . C., & Mistry, R. S. (2013). Geographic Variations in Cost of Living: Associations with
Family and Child Well-Being. Child Development, 84(1), 209-225 . doi: 10.1111/j.14678624.2012.01846.x
Cicchetti, D. (2013). Annual research review: Resilient functioning in maltreated children-Past,
present, and future perspectives. Journal of Child Psychology and Psychiatry, 54(4), 402422.
Cicchetti, D., & Valentino, K. (2006). An ecological-transactional perspective on child
maltreatment: Failure of the average expectable environment and its influence on child
development. In D. Cicchetti, & K. Valentino (Eds.), Developmental psychopathology:
Vol. 3. Risk, disorder, and (2nd ed., pp. 129-201). Hoboken, NJ: Wiley.
Cicchetti, D., Rogosch, F. A., & Toth, S. L. (2006). Fostering secure attachment in infants in
maltreating families through preventive interventions. Development and
Psychopathology, 18(3 ), 623-649. doi: 10.10170S0954579406060329
Coanders, G., Satorra, A., & Saris, W. E. (1997). Alternative approaches to structural equation
modeling of ordinal data: A Monte Carlo study. Structural Equation Modeling: A
Multidiciplinary Journal, 4( 4), 261-282. doi: 10.1080/10705519709540077
Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155-159. doi:10.1037/00332909.112 .1.155
Collins, W .A. (2003). More than myth: the developmental significance of romantic relationships
during adolescence. Journal of Research on Adolescence, 13(1 ), 1-24. doi: 10.1111/15327795.1301001
Cudeck, R., & du Toit, S. (2009). General structural equation models. In R. Millsap, & A.
Maydeu-Olivares (Eds.), The SAGE handbook of quantitative methods in psychology (pp.
515-540). London: SAGE Publications Ltd.
doi:http://dx.doi.org/10.4135/9780857020994.n21
Davis, M. C., Luecken, L., & Lemery-Chalfant, K. (2009). Resilience in Common Life:
Introduction to the Special Issue. Journal of Personality, 77(6), 1637-44.
doi: 10.1111/j.1467-6494.2009.00595.x
85

Delis, D. C., Kaplan, E., & Kramer, J. H. (2001). Delis-Kaplan Executive Function System.
DiGennaro Reed, F. D., McIntyre, L. L., Dusek, J., & Quintero, N. (2011). Preliminary
assessment of friendship, problem behavior, and social adjustment in children with
disabilities in an inclusive education setting. Journal ofDevelopmental And Physical
Disabilities, 23(6), 477-489. doi:10.1007/s10882-011-9236-2
Dudak, J. A. (2009). How to select, calculate and interpret effect sizes. Journal of Pediatric
Psychology, 34(9), 917-928 . doi: 10.1093/jpepsy/jsp004
Easterbrooks, M.A., Driscoll, J. R., & Bartlett, J. D. (2008). Resilience in Infancy: A Relational
Approach. Research in Human Development, 5(3), 139-152.
doi: 10.1080/15427600802273987
Englund, M. M., I-Chun Kuo, S., Puig, J., & Collins, W . A. (2011). Early roots of adult
competence: The significance of close relationships from infancy to early adulthood.
International Journal of Behavioral Development, 35(6), 490-496.
doi:10.1177/0165025411422994
Evans, S. Z., Simons, L. G., & Simons, R. L. (2012). he effect of corporal punishment and verbal
abuse on delinquency: Mediating mechanisms. Journal Of Youth And Adolescence, 41(8),
1095-1110. doi: 10.1007/sl 0964-012-9755-x
Fergusson, D. M., & Horwood, L. J. (2001). The Christchurch Health and Development Study:
Review of findings on child and adolescent mental health. Australian And New Zealand
Journal OfPsychiatry, 35(3), 287-296. doi:10.1046/j.1440-1614.2001.00902.x
Fletcher, D., & Sarkar, M. (2013). Psychological resilience: A review and critique of definitions,
concepts, and theory. European Psychologist, 18(1), 12-23. doi:10.1027/ 10169040/a000124
Ford, M. E. (1982). Social cognition and social competence in adolescence. Developmental
Psychology, 18(3), 323-340. doi:10.1037/0012-1649.18.3.323
Fruiht, V. M. (2015). Supportive Others in the Lives of College Students and Their Relevance to
Hope. Journal of College Student Retention: Research, Theory & Practice, 17( 1), 64-87.
doi:10.1177/1521025115571104
Frydenberg, E., Lewis, R., Bugalski, K., Cotta, A., McCarthy, C ., Luscombe-Smith, N ., & Poole,
C. (2004). Prevention is better than cure: Coping skills training for adolescents at
school. Educational Psychology In Practice, 20 (2), 117-134.
doi: 10.1080/02667360410001691053

86

Gaertner, A. E., Fite, P. J., & Colder, C.R. (2011). Moderating effect of intimate exchange on
delinquent socialization processes. The Journal Of Early Adolescence, 31(2), 258-270.
doi: 10.l 177/0272431610363159
Garmezy, N . (1974). The study of competence in children at risk for severe psychopathology. In
E. J. Anthony & C. Koupernik (Eds.), The child in his family : Vol. 3. Children at
psychiatric risk (pp. 77-97). New York: Wiley.
Garmezy, N., Masten, A. S., & Tellegan, A. (1984). The Study of Stress and Competence in
Children: A Building Block for Developmental Psychopathology. Child Development,
55, 97-111.
Gau, S.S., & Chang, J.P. (2013). Maternal Parenting Styles and Mother-Child Relationship
among Adolescents with and without Persistent Attention-Deficit/Hyperactivity
Disorder. Research in Developmental Disabilities: A Multidisciplinary Journal, 34(5),
1581-1594.
German, M., Gonzales, N. A., Bonds McClain, D., Dumka, L., & Millsap, R. (2013). Maternal
warmth moderates the link between harsh discipline and later externalizing behaviors for
Mexican American adolescents. Parenting: Science and Practice, 13(3), 169-177.
doi: 10.1080/15295192.2013. 756353
Geven, S., Weesie, J., & van Tubergen, F. (2013). The influence of friends on adolescents'
behavior problems at school: The role of ego, alter and dyadic characteristics. Social
Networks, 35(4), 583-592. doi:10.1016/j.socnet.2013.08.002
Gioia, G. A., Espy, K. A., & Isquith, P. K. (1996). Behavior Rating Inventory of Executive
Function--Preschool Version.
Gioia, G. A., Isquith, P. K., Guy, S. C ., & Kenworthy, L. (1996). Behavior Rating Inventory of
Executive Function.
Granger, D. A., Serbin, L.A., Schwartzman, A., Lehoux, P., Cooperman, J., & Ikeda, S. (1998).
Children's salivary cortisol, internalising behaviour problems, and family environment:
Results from the Concordia Longitudinal Risk Project. International Journal of
Behavioral Development, 22(4), 707-728. doi: 10.1080/016502598384135
Greeff, A., & Merwe, S. Van Der. (2004). Variables Associated with Resilience in Divorced
families. Social Indicators Research, 68(1), 59-75 . Retrieved from
http://www.springerlink.com/index/R62064405687 5451.pdf
Greeff, A. P., & Du Toit, C. (2009). Resilience in Remarried Families. The American Journal of
Family Therapy, 37(2), 114-126. doi:10.1080/01926180802151919
87

Greenhoot, A. F., & Dowsett, C. J. (2012). Secondary Data Analysis: An Important Tool for
Addressing Developmental Questions. Journal Of Cognition & Development, 13(1), 218. doi: 10.1080/ 15248372.2012.646613
Guy, S. C., Isquith, P. K., & Gioia, G. A. (1996). Behavior Rating Inventory of Executive
Function--Self-Report Version.
Hardwick, S. W., Marcus, R., & Isaak, M. (2010). Education and national identity in a
comparative context. National Identities, 12(3), 253-268.
doi: 10.1080/ 14608941003727932
Hamey, P. A. (2007). Resilience Processes in Context: Contributions and Implications of
Bronfenbrenner' s Person-Process-Context Model. Journal ofAggression, Maltreatment
& Trauma, 14(3), 73-87. doi:10.1300/J146v14n03
Hayduk, L. (1987). Structural Equation Modeling with LISREL: Essentials and Advances.
Baltimore, Maryland: JHU Pres.
Hayduk, L. (1996). LISREL Issues, Debates and Strategies. Baltimore, Maryland: JHU Press.
Hayduk, L. (2014). Shame for disrespecting evidence: the personal consequences of insufficient
respect for structural equation model testing. BMC Medical Research
Methodology, 14(1), 1-17. doi:10.1186/1471-2288-14-124
Hayduk, L., Cummings, G., Boadu, K., Pazderka-Robinson, H., & Boulianne, S. (2007). Testing!
Testing! One, two, three--Testing the theory in structural equation models!. Personality
And Individual Differences, 42(5), 841-850. doi:10.1016/j.paid.2006.10.001
Hayduk, L., & Glaser, D. (2000a). Jiving the Four-Step, Waltzing Around Factor Analysis, and
Other Serious Fun. Structural Equation Modeling, 7(1), 1-35.
doi:10.1207/S15328007SEM0701 01
Hayduk, L., & Glaser, D. (2000b). Doing the four-step , right-2-3, wrong-2-3: A briefreply to
Mulaik and Millsap; Bollen; Bentler; and Herting and Costner. Structural Equation
Modeling, 7(1), 111-123. doi: 10.1207/Sl 5328007SEM0701_06
Hayduk, L., & Littvay, L. (2012). Should researchers use single indicators, best indicators, or
multiple indicators in structural equaltion models? BMC Medical Research Methodology,
12(159), 1471-2288. doi:10.1186/1471-2288-12-159
Herting, J. R., & Costner, H. L. (2000). Another perspective on 'the proper number of factors'
and the appropriate number of steps. Structural Equation Modeling, 7(1 ), 92-110.
doi:10.1207/S15328007SEM0701 05
88

IBM Corp. (2013). IBM SPSS Statistics for Windows, Version 22 .0. Armonk, New York.
Joreskog, K ., & Sorbom, D. (2006). LISREL 8.8 for Windows [Computer software]. Skokie, IL:
Scientific Software International, Inc.
Karatoreos, I. N ., & McEwen, B. S. (2013). Annual research review: The neurobiology and
physiology of resilience and adaptation across the life course. Journal of Child
Psychology and Psychiatry, 54(4), 337-347. doi:10.1111 /jcpp.12054
Karreman, A. , & Vingerhoets, A. J. J.M. (2012) Attachment and Well-being: The Mediating
Role of Emotion Regulation and Resilience. Personality and Individual Differences, 53,
821-826.
Kiff, C. J. , Lengua, L. J. , & Zalewski, M. (2011). Nature and Nurturing: Parenting in the Context
of Child Temperament. Clinical Child and Family Psychology Review, 14(3), 251-301.
doi: 10.1007/s10567-011-0093-4
Kraus, M . W., Horberg, E. J., Goetz, J. L. , & Keltner, D . (2011). Social Class Rank, Threat
Vigilance, and Hostile Reactivity. Personality and Social Psychology Bulletin, 3 7(10),
1376-1388. doi: 10.1177/0146167211410987
Kroenke, C. (2008) . Socioeconomic status and health: Youth development and neomaterialist
and psychosocial mechanisms. Social Science & Medicine, 66(1 ), 31-42.
doi: 10.1016/j.socscimed.2007.07 .018
Krueger, P. M ., Bhaloo, T. , & Rosenau, P. V. (2009) . Health Lifestyles in the U.S. and Canada:
Are We Really So Different? Social Science Quarterly, 90(5) , 1380-1402.
doi: 10.1111/j.1540-6237.2009.00660.x
Lansford, J. E., Malone, P. S. , Stevens, K. I. , Dodge, K. A., Bates, J. E., & Pettit, G . S. (2006).
Developmental trajectories of externalizing and internalizing behaviors: Factors
underlying resilience in physically abused children. Development and
Psy chopathology, 18(1), 35-55 . doi :10.1017/S0954579406060032
Lee, J. C. , Sudom, K. A. , & McCreary, D.R. (2011). Higher-order model of resilience in the
Canadian forces. Canadian Journal of Behavioural Science/Revue Canadienne Des
Sciences Du Comportement, 43(3), 222-234. doi: 10.1037/a00244 73
Lengua, L. J., Wolchik, S. A. , Sandler, I. N ., & West, S. G. (2000). The additive and interactive
effects of parenting and temperament in predicting problems of children of
divorce. Journal of Clinical Child Psychology, 29(2) , 232-244.
doi: 10.1207IS 15374424jccp2902_9

89

Li, M., Eschenauer, R., & Yang, Y. (2013). Influence of efficacy and resilience on problem
solving in the United States, Taiwan, and China. Journal ofMulticultural Counseling and
Development, 41(3), 144-157. doi: 10.1002/j .2161-1912.2013 .00033 .x
Li, M., & Yang, Y. (2009). Determinants of problem solving, social support seeking, and
avoidance: A path analytic model. International Journal of Stress Management, 16(3 ),
155-176. doi:10.1037/a0016844
Luthar, S., Cicchetti, D., & Becker, B. (2000). The Construct of Resilience: A Critical evaluation
and Guidelines for Future Work. Child Development, 71(3), 543-562.
Lyons, M. D., Otis, K. L., Huebner, E. S., & Hills, K. J. (2014). Life Satisfaction and
Maladaptive Behaviors in Early Adolescents. School Psychology Quarterly, 29(4), 553566. doi: 10.1037/spq0000061
Marsh, H. W., & O'Neill, R. (1984). Self Description Questionnaire III: The construct validity of
multidimensional self-concept ratings by late adolescents. Journal of Educational
Measurement, 21(2), 153-174. doi:10.2307/ 1434540
Martel, M. M., Nigg, J. T. , Wong, M. M., Fitzgerald, H . E. , Jester, J.M., Puttler, L. I. , & ...
Zucker, R. A. (2007). Childhood and adolescent resiliency, regulation, and executive
functioning in relation to adolescent problems and competence in a high-risk
sample. Development and Psychopathology, 19(2), 541-563 .
doi: 10.1017/S0954579407070265
Masten, A. S., & Tellegen, A. (2012). Resilience in developmental psychopathology:
Contributions of the Project Competence Longitudinal Study. Development and
Psy chopathology, 24, 345-361. doi: 10.1017/S095457941200003X
Masten, A. S., Burt, K. B ., Roisman, G. I. , Obradovic, J. , Long, J . D., & Tellegen, A. (2004).
Resources and resilience in the transition to adulthood: Continuity and change.
Development and Psychopathology, 16, 1071-1094. doi: 10.10170S0954579404040143
Masten, Ann, & Obradovic, J. (2006). Competence and Resilience in Development. Annals of
the New York Academy ofSciences, 1094, 13-27. doi: 10.l 196/annals.1376.003
McElwee, N . (2007). A Focus on the Personal and Structural: Resilience Explored. Child and
Youth Services (pp. 57-69). doi:10.1300/J024v29n01
McKay, C. (2011 ). The resilient community: Implications for out of school programming. Child
& Adolescent Social Work Journal, 28(5), 357-373. doi: 10.1007/s10560-011 -0240-2

90

Moffitt, T. E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A
developmental taxonomy. Psychological Review, 100(4), 674-701. doi:10.1037/0033295X.100.4.674
Moghaddam, M. F., Assareh, M., Heidaripoor, A., Eslami Rad, R., & Pishjoo, M. (2013). The
study comparing parenting styles of children with ADHD and normal children. Archives
of Psychiatry and Psychotherapy, 15(4), 45-49. doi:10.12740/APP/19375
Murray, K. W., Dwyer, K. M., Rubin, K. H., Knighton-Wisor, S., & Booth-Laforce, C. (2014).
Parent-child relationships, parental psychological control, and aggression: Maternal and
paternal relationships. Journal of Youth and Adolescence, 43(8), 1361-1373 .
doi: 10.1007/sl 0964-013-0019-1
Naglieri, J. A., & Goldstein, S. (2013). Comprehensive Executive Function Inventory.
Nasvytiene, D., Lazdauskas, T., & Leonaviciene, T. (2012). Child's resilience in face of
maltreatment: A meta-analysis of empirical studies. Psichologija, 46, 7-26.
Newman, B. M., Lohman, B. J., & Newman, P.R. (2007). Peer Group Membership and a Sense
of Belonging: Their Relationship to Adolescent Behavior Problems. Adolescence (San
Diego): An International Quarterly Devoted To The Physiological, Psychological,
Psychiatric, Sociological, And Educational Aspects Of The Second Decade OfHuman
Life, 42(166), 241-263.
Newton-Howes, G., Horwood, J., & Mulder, R. (2015). Personality characteristics in childhood
and outcomes in adulthood: Findings from a 30 year longitudinal study. Australian &
New Zealand Journal of Psychiatry, 49(4), 377-386. doi: 10.1177/0004867415569796
Nicholls, A. R., Levy, A. R., & Perry, J. L. (2015). Emotional maturity, dispositional coping, and
coping effectiveness among adolescent athletes. Psychology of Sport and Exercise, 173239. doi: 10.1016/j.psychsport.2014.11.004
Niv, S., Tuvblad, C., Raine, A., Wang, P., & Baker, L.A. (2012). Heritability and longitudinal
stability of impulsivity in adolescence. Behavior Genetics, 42(3), 378-392.
doi: 10.1007/sl 0519-011-9518-6
Olsson, C. A., Bond, L., Bums, J.M., Vella-brodrick, D. A., & Sawyer, S. M. (2003).
Adolescent Resilience: A Concept Analysis. Journal ofAdolescence, 26, 1-11.
Pangallo, A., Zibarras, L., Lewis, R., & Flaxman, P. (2015). Resilience through the lens of
interactionism: A systematic review. Psychological Assessment, 27(1), 1-20.
doi: 10.103 7/pas0000024

91

Parker, P. D., Ciarrochi, J., Heaven, P., Marshall, S., Sahdra, B., & Kiuru, N. (2015). Hope,
Friends, and Subjective Well-Being: A Social Network Approach to Peer Group
Contextual Effects. Child Development, 86(2), 642-650. doi: 10.1111/cdev.12308
Pietruska, K., & Armony, J. L. (2013). Differential effects of trait anger on optimism and risk
behaviour. Cognition and Emotion, 27(2), 318-325. doi: 10.1080/02699931.2012. 703130
Pougnet, E., Serbin, L.A., Stack, D. M., & Schwartzman, A. E. (2011). Fathers' influence on
children's cognitive and behavioural functioning: A longitudinal study of Canadian
families. Canadian Journal of Behavioural Science/Revue Canadienne Des Sciences Du
Comportement, 43(3), 173-182. doi:10.1037/a0023948
Puskar, Kathryn R., Susan M. Sereika, Jacqueline Lamb, Kathleen Tusaie-Mumford, and Teena
McGuinness. 1999. "Optimism and its relationship to depression, coping, anger, and life
events in rural adolescents." Issues in Mental Health Nursing 20, no. 2: 115130. Psyc!NFO, EBSCOhost (accessed November 3, 2015).
PsychTests AIM, I. (2011). Coping Skills Assessment.
Raby, K. L., Roisman, G. I., Fraley, R. C., & Simpson, J. A. (2015). The Enduring Predictive
Significance of Early Maternal Sensitivity: Social and Academic Competence Through
Age 32 Years. Child Development, 86(3), 1467-8624. doi: 10.1111/cdev.12325
Rey Pena, L., & Pacheco, N. E. (2012). Physical-verbal aggression and depression in
adolescents: The role of cognitive emotion regulation strategies. Universitas
Psychologica, 11(4), 1245-1254.
Ribeaud, D., & Eisner, M. (2010). Risk factors for aggression in pre-adolescence: Risk domains,
cumulative risk and gender differences - Results from a prospective longitudinal study in
a multi-ethnic urban sample. European Journal of Criminology, 7(6), 460-498 .
doi: 10.1 l 77/1477370810378116
Reisman, G. I., Masten, A. S., Coatsworth, J. D., & Tellegen, A. (2004). Salient and emerging
developmental tasks in the transition to adulthood. Child development, 75(1), 123-33.
Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/l50l5679
Rolf, J., & Garmezy, N. (1974). The school performance of children vulnerable to behavior
pathology. In D. F. Ricks, A. Thomas, & M. Roff (Eds.), Life history research in
psychopathology (III.). Minneapolis, MN.
Roosa, M. W. (2000). Some Thoughts about Resilience Versus Positive Development, Main
Effects Versus Interactions, and the Value of Resilience. Child Development, 71(3), 567575. Retrieved from http: //www.ncbi.nlm.nih.gov/pubmed/10953925
92

Roth, R. M., Isquith, P. K., & Gioia, G. A. (2005). Behavior Rating Inventory of Executive
Function--Adult Version.
Russell, B. S., Londhe, R., & Britner, P.A. (2013). Parental contributions to the delay of
gratification in preschool-aged children. Journal Of Child And Family Studies, 22(4),
471-478. doi: 10.1007/sl 0826-012-9599-8
Rutter, M., Greenfeld, D., & Lockyer, L. (1967). A Five to Fifteen Year Follow-up Study of
Infantile Psychosis II: Social and Behavioural Outcome. The British Journal of
Psychiatry, 113(504), 1183-1199.
Rutter, M., & Lockyer, L. (1967). A five to fifteen year follow-up study of infantile psychosis I:
Description of sample. The British Journal of Psychiatry, 113(504), 1169-1182.
Rutter, M. (1985). Resilience in the face of adversity: Protective factors and resistance to
psychiatric disorder. The British Journal Of Psychiatry, 147(6), 598-611 .
doi:10.l 192/bjp.147.6.598
Rutter, M. (2003). Commentary: Causal Processes Leading to Antisocial Behavior.
Developmental Psychology, 39(2), 372-378.
Rutter, Michael. (2006). Implications of Resilience Concepts for Scientific Understanding.
Annals OfTheNew York Academy Of Sciences, 1094, 1-12. doi:10.1196/annals.1376.002
Rutter, Michael. (2012). Resilience as a Dynamic Concept. Development and Psychopathology,
24(2), 335-344. doi:10.1017/S0954579412000028
Sameroff, A. J., & Rosenblum, K. L. (2006). Psychosocial Constraints on the Development of
Resilience. Annals Of The New York Academy Of Sciences, 1094, 116-124.
doi:10.l 196/annals.1376.010
Saris, W. (1999). Structural Equation Modelling. In H.J. Ader, & G. J. Mellenbergh (Eds.),
Research Methodology in the Social, Behavioural and Life Sciences (pp. 221-224).
London, England: SAGE Publications, Ltd.
doi:http://dx.doi.org.login.ezproxy.library. ualberta.ca/10.4135/9780857029027 .d 124
Schore, A. N. (2001 ). Effects of a secure attachment relationship on right brain development,
affect regulation, and infant mental health. Infant Mental Health Journal, 22(1-2), 7-66.
doi: 10.1002/1097-0355(200101/04)22: 1<7::AID-IMHJ2> 3.0.C0;2-N
Schretlen, D. J. (2010). Modified Wisconsin Card Sorting Test.

93

Serbin, L. a., Temcheff, C. E. , Cooperman, J . M., Stack, D. M., Ledingham, J., & Schwartzman,
a. E. (2010). Predicting Family Poverty and Other Disadvantaged Conditions for Child
Rearing from Childhood Aggression and Social Withdrawal: A 30-Y ear Longitudinal
Study. International Journal of Behavioral Development, 35(2), 97-106.
doi: 10.1177/01650254103 72008
Shiner, R. L. , & Masten, A. S. (2012). Childhood Personality as a Harbinger of Competence and
Resilience in Adulthood. Development and Psychopathology, 24(2), 507-528.
doi: 10.1017/S0954579412000120
Shumacker, R. E., & Lomax, R. G. (2010). A Beginner's Guide to Structural Equation Modeling
(3rd ed.). New York: Routledge.
Skomorovsky, A. , & Stevens, S. (2013) . Testing a resilience model among Canadian forces
recruits. Military Medicine, 178(8), 829-83 7. doi: 10. 7205/MILMED-D-12-003 89
Spielberger, C. D. (1996). State-Trait Anger Expression Inventory [1996 Edition].
Stack, D. (2013, May 1). CRDH - Concordia University. Retrieved from Stack Infant and Child
Studies Lab: http://crdh.concordia.ca/StackLab/Stack-Lab files/pageOOO l .htm
Statistics Act, Revised Statues of Canada (1985, c. S-19). Retrieved from http://lawslois.justice.gc.ca/eng/acts/S-19/FullText.html
Statistics Canada. (1995). Microdata User Guide National Longitudinal Survey of Children and
Youth Cycle 2 September 1994 to July 1995. doi:doi:10.1093/deafed/ens058
Statistics Canada. (1996) . National Longitudinal Survey of Children and Youth, 1994-1995
[Canada}: Cycle 1, Release 2, Primary File. Statistics Canada.
Statistics Canada. (1998) . National Longitudinal Survey of Children and Youth , 1996-1997
[Canada]: Cy cle 2, Primary File. Statistics Canada.

Statistics Canada. (2000) . National Longitudinal Survey of Children and Youth, 1998-1999
[Canada]: Cycle 3, Primary File. Statistics Canada.
Statistics Canada. (2004) . National Longitudinal Survey of Children and Youth 10-17 Year old
Self- Completed Que stionnaires - Cy cle 4, Release 2. Statistics Canada.
Statistics Canada. (2005a). Microdata User Guide National Longitudinal Survey of Children and
Youth Cycle 6 September 2004 to June 2005.

94

Statistics Canada. (2005b) . National Longitudinal Survey of Children and Youth Cy cle 4,
Primary File - Version Two. Statistics Canada.
Statistics Canada. (2007). National Longitudinal Survey of Children and Youth, Cy cle 6, 2004
Master file (Longitudinal) . Statistics Canada.
Statistics Canada. (2009) . Statistics Canada Quality Guidelines. Retrieved from
http: //www.statcan.gc.ca/pub/12-539-x/12-539-x2009001 -eng.htm
Statistics Canada. (2010a). National Longitudinal Survey of Children and Youth, Cy cle 8, 2008
Master File (Youth) . Statistics Canada.
Statistics Canada. (2010b). National Longitudinal Survey of Children and Youth (NLSCY).
Retrieved from
http: //www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&Survid=4450&SurvVer
= 1&Instaid= 16044&InstaVer=8&SDDS=4450&lang=en&db=imdb&adm=8&dis=2
Statistics Cananda. (2010c). Microdata User Guide National Longitudinal Survey of Children
and Youth Cy cle 8 September 2008 to July 2009. Retrieved from Statistics Canada:
http ://www23.statcan.gc.ca/imdb-bmdi/document/4450 D4 T9 V8-eng.pdf
Stewart, D . E. (2011). Toward Understanding Resilient Outcomes. Canadian Journal of
Psy chiatry., 56(5), 256-257. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/21586190
Sturaro, C., van Lier, P. C. , Cuijpers, P ., & Koot, H. M . (2011). The role of peer relationships in
the development of early school-age externalizing problems. Child Development, 82(3),
758-765 . doi:10.l 111/j .1467-8624.2010.01532.x
Tichovolsky, M. H. , Arnold, D. H. , & Baker, C. N. (2013). Parent predictors of changes in child
behavior problems. Journal OfApplied Developmental Psychology, 34(6), 336-345 .
doi : 10.1016/j.appdev.2013 .09.001
Tiet, Q. Q. , Huizinga, D., & Byrnes, H . F. (2009). Predictors of Resilience Among Inner City
Youths. Journal of Child and Family Studies, 19(3), 360-378 . doi: 10.1007/s10826-0099307-5
Tomarken, A. J. , & Waller, N. G. (2005). Structural Equation Modeling: Strengths, Limitations,
and Misconceptions. Annual Review of Clinical Psy chology(!), 2.1-2.35 .
doi: 10.l 146/annurev.clinpsy.1.102803 .144239
Topitzes, J. (2013). Adult resilience among maltreated children: A prospective investigation of
main effect and mediating models. Children and y outh services review, 35(6), 937-949.
doi: 10.1016/j .childyouth.2013 .03 .004
95

Trobia, A. (2008). Cronbach's alpha. In P. Lavrakas (Ed.), Encyclopedia of survey research
methods (pp. 169-171). Thousand Oaks, CA: SAGE Publications, Inc.
doi:http: //dx.doi.org/10.4135/9781412963947.nl 17
Ullman, J. B. (2013). Structural Equation Modeling. In B. G. Tabachnick, & L. S. Fidell, Using
Multivariate Statistics (6th ed., pp. 681-785). Boston: Pearson.
Unger, M. (2012). The Social Ecology of Resilience: Adressing Contextual and Cultural
Ambiguity of a Nascent Concept. American Journal Of Orthopsychiatry, 81 (1 ), 1-17.
doi: 10.1111/j.1939-0025.2010.01067.x
Vijayakurnar, N ., Whittle, S., Yiicel, M., Dennison, M., Simmons, J., & Allen, N . B. (2014).
Thinning of the lateral prefrontal cortex during adolescence predicts emotion regulation
in females . Social Cognitive and Affective Neuroscience, 9(11), 1845-1854.
doi: 10.1093/scan/nstl 83
Villodas, M . T. , Litrownik, A. J., Thompson, R., Jones, D., Roesch, S. C., Hussey, J. M., . ..
Dubowitz, H. (2015). Developmental transitions in presentations of externalizing
problems among boys and girls at risk for child maltreatment. Development and
Psychopathology, 27(1), 205-219. doi: 10.1017/S0954579414000728
Waaktaar, T., & Torgersen, S. (2012). Genetic and environmental causes of variation in trait
resilience in young people. Behavior Genetics, 42(3), 366-377. doi:10.1007/s10519-0119519-5
Wang, F., Christ, S. L., Mills-Koonce, W. R., Garrett-Peters, P., & Cox, M. J. (2013).
Association between maternal sensitivity and externalizing behavior from preschool to
preadolescence. Journal ofApplied Developmental Psychology, 34(2), 89-100.
doi:http: //dx.doi.org/l 0.1016/j.appdev.2012.11.003
Ward, C., Martin, E., Theron, C., & Distiller, G. (2007) . Factors Affecting Resilience in Children
Exposed to Violence. South African Journal of Psychology, 37(1), 165-187.
Werner, E. , Bierman, J., & French, F. (1971). The Children ofKauai Honolulu: a Longitudinal
Study from the Prenatal Period to Age Ten. Honolulu, Hawaii: University of Hawaii
Press.
Werner, E. E. (1989). High-risk children in young adulthood: A longitudinal study from birth to
32 years. American Journal Of Orthopsychiatry, 59(1 ), 72-81. doi: 10.1111/j .19390025 .1989. tbO 163 6.x
Werner, E. E. (1995). Resilience in Development. Current Directions In Psychological Science,
4(3), 81-85. doi: 10.1111/1467-8721.ep10772327
96

Werner, E. E. (1996). Vulnerable but invincible: High risk children from birth to adulthood.
European Child & Adolescent Psychiatry, 5(Suppl. 1), 45-51 .
Werner, W. (1993). Nonpositive Definite Matrices in Structural Equation Modeling. In K. A.
Bollen, & J. S. Long (Eds.), Testing Structural Equation Models (pp. 256-293). Newbury
Park: Sage Publications.
Westat, Inc. (2014). Wesvar. Rockville, MD.
Wittmann, M., Arce, E., & Santisteban, C. (2008). How impulsiveness, trait anger, and
extracurricular activities might affect aggression in school children. Personality and
Individual Differences, 45(7), 618-623 . doi:10 .1016/j.paid.2008 .07.001
Wothke, W. (1993). Nonpositive Definite Matrices in Structural Equation Modeling. In K. A.
Bollen, & J. S. Long (Eds.), Testing Structural Equation Models (pp. 256-293). Newbury
Park, California: SAGE Publications, Inc.
Wright, B. D . ( 65-104). Fundamental measurement for psychology. In E. S. E., & S. L.
Hershberger (Eds.), The new rules of measurement: What every psychologist and
educator should know (p. 1999). Mahwah, NJ: Lawrence Erlbaum.
Yang, C., Nay, S., & Hoyle, R.H. (2010). Three Approaches to Using Lengthy Ordinal Scales in
Structural Equation Models: Parceling, Latent Scoring, and Shortening Scales. Applied
Psychological Measurement, 34(2), 122-142. doi: 10. l 177/0146621609338592
Zautra, A. J., Arewasikporn, A., & Davis, M. C. (2010) . Resilience: Promoting Well-Being
Through Recovery, Sustainability, and Growth. Research in Human Development, 7(3),
221-238. doi: 10.1080/15427609.2010.504431
Zelazo, P. D. , & Carlson, S. M. (2012). Hot and cool executive function in childhood and
adolescence: Development and plasticity. Child Development Perspectives , 6(4), 354360.
Zhang, B., Zhao, F., Ju, C., & Ma, Y. (2015). Paternal Involvement as Protective Resource of
Adolescents' Resilience: Roles of Male Gender-Role Stereotype and Gender. Journal of
Child and Family Studies, 24(7), 1955-1965. doi: 10.1007/s10826-014-9995-3

97

Appendix A
Lisrel Syntax Primary Group Non-Significant Model
DA NJ;J6 N0; 642 MA;CM NG; J
CMSY
0.2295
0.00210.5111
0.1139 0.0826 3.1278
0.0275 0.0335 -0.0135 0.2481
-0.0838 -0.0189 -0.2348 0.0271 2.1745
-0.0590 -0.0288 -0.5404-0.0 173 0.3697 0.8102
0.0285 -0.0094 -0.0896 0.1840 0.0045 -0.0526 1.0034
0.0073 0.0398 0.1394 0.0007 -0.0395 -0.0328 -0.0095 0.2353
0.0 160 0.0373 0.0913 -0.0565 0.0002 -0.0448 -0.0785 0.1512
0.4 194
-0.0065 -0.0861-0.1204 -0.00900.1957 0.06 12 -0.0266-0.0322 0.0159 0.6883
-0.0493 -0.0931 -0.1485 -0.0 179 0.1918 0.0375 -0.0288 -0.0311 0.0097 0.3036 0.6626
-0.01510.08920.0287 0.0515 -0.0310-0.0185 -0.0032 0.0335
0.0076 -0.0187 -0.0863 0.3893
0.0120 0.0466 0.0245 0.0608 -0.0242 -0.0570 0.0927 0.0270
0.0043 -0.0332 -0.0306 0.0749 0.2573
0.0242 0.0646 0.0778 0.0137 -0.0355 -0.0413 -0.0075 0.0586
0.0613 -0.025 1 -0.02090.0716 0.0593 0.2074
-0.0108 0.02 18 0.0009-0.0051 -0.00840.0449 -0.0699-0.0097 0.0262 -0.0422 -0.0229 0.0260 0.00670.0 124 0.7086
-0.0245 -0.0 178 -0.0439 0.0763 -0.0272 -0.0356 0.0473 0.0018 0.0596 -0.0444 0.0055 0.0043 0.0308 -0.0235 0.0685 0.6710
LA
'TellsBad' 'HitPamt' 'DrinkPMK' 'Sex' 'ResSize' 'Income' 'Friends'
'CloseMom' 'CloseDad' 'Praise' 'TalkPlay' 'DiflWait' 'ReactAng'
'ManyFigh' 'Optimism' 'SolvePro'
SE
'DiflWait' 'ManyFigh' 'ReactAng' 'Optimism' 'SolvePro'
'TellsBad' 'HitPamt' 'DrinkPMK' 'Sex' 'ResSize' 'Income'
'CloseMom' 'CloseDad' 'Friends' 'Praise' 'TalkPlay'/
MO NY;5 NX;J J NE;6 NK;I I LY;FU,FI LX;FU,FI
BE;FU,FI GA;FU,FI C
PH;SY,FR PS;SY,FI TE;SY,FI TD; SY,FI
VA 1.0 LY(l,I) LY(2,2) LY(3 ,3) LY(4,4) LY(5,6) LX( l , I)
LX(2,2) LX(3 ,3) LX(4,4) LX(5 ,5) LX(6,6) LX(7 ,7) LX(8 ,8)
LX(9,9) LX( I0,10) LX(l 1,11)

FR GA(5 ,4) GA(5 ,8) GA(5,9)
ST 0.02 GA( l ,I) GA(l ,2) GA( l ,3) GA(l ,4) GA(l,5) GA( l ,6)
GA( l ,7) GA( l ,8) GA(l ,11)
ST 0.03 GA(2, I) GA(2 ,2) GA(2 ,4) GA(2 ,7) GA(2,8) GA(2,9)
GA(2, 10) GA(2, I I)
ST 0.02 GA(3 ,2) GA(3 ,4) GA(3 ,6) GA(3 ,7) GA(3 ,9)
ST0.02 GA(4,9)
ST 0.02 GA(5,4)
ST 0.1 GA(5,8)
ST 0.00 1 GA(5,9)
STO.I GA(4,I)
FR PS(l , I) PS(2 ,2) PS(3 ,3) PS(4,4) PS(6,6)
VA 3.49 PS(5,5)
ST 0.2 PS(3 ,3)
ST 0.5 PS(4,4)
ST 0.5 PS(6,6)
ST 0.09 PS( I , I) PS(2,2)
VA0.0467 TE(l , I)
VA 0.0180 TE(3 ,3)
VA 0.0104 TE(2,2)
VA 0.0336 TE(5,5)
VA 0.0354 TE(4,4)
VA 0.0115 TD( l , I)
VA 0.2045 TD(2,2)
VA 0.2189 TD(3 ,3)
VA 0.0025 TD(4,4)
VA 0.2609 TD(5 ,5)
VA 0.0972 TD(6 ,6)
VA 0.0235 TD(7 ,7)
VA 0.02 10 TD(8,8)
VA 0.0502 TD(9,9)
VA 0.0482 TD( I0,10)
VA 0.0464 TD(l 1,11)
PD
OU ML ALL AD;OFF ND;3

FR BE(2, I) BE(2,3)
FR BE(3,I)
FR BE(4,2)
FR BE(6,5)
FR BE(5,6)
EQ BE(6,5) BE(5,6)
ST 0.05 BE(2, I)
ST 0.0 1 BE(3,I)
ST 0.1 BE(2,3)
ST 0.1 BE(4,2)
ST 0.1 BE(6,5) BE(5,6)
VA 1.0 BE(5,4)
FR GA( l ,I) GA(l ,2) GA(l ,3) GA(l ,4) GA(l ,5) GA(l,6) GA(l ,7)
GA(l ,8) GA(l , 11 )
FR GA(2, I) GA(2 ,2) GA(2 ,4) GA(2,7) GA(2 ,8) GA(2,9)
GA(2 , IO) GA(2, I I)
FR GA(3 ,2) GA(3 ,4) GA(3,6) GA(3,7) GA(3 ,9)
FR GA(4, I) GA(4,9)

98

Appendix B
Lisrel Syntax Primary and Supplemental Groups Non-Significant Model
Stacked Model
Primary Group
DA Nl=l6 N0=642 MA=CM NG=2
CMSY
0.2295
0.00210.5111
0.1139 0.0826 3.1278
0.0275 0.0335 -0.0135 0.2481
-0.0838 -0.0189 -0.2348 0.0271 2.1745
-0.0590 -0.0288 -0.5404-0.0173 0.3697 0.8102
0.0285 -0.0094 -0.0896 0.1840 0.0045 -0.0526 1.0034
0.0073 0.0398 0.1394 0.0007 -0.0395 -0.0328 -0.0095 0.2353
0.0160 0.0373 0.0913 -0.0565 0.0002 -0.0448 -0.0785 0.1512
0.4194
-0.0065 -0.0861-0.1204-0.00900.1957 0.0612-0.0266 -0.0322 0.0159 0.6883
-0.0493 -0.0931-0.1485-0.01790.1918 0.0375 -0.0288 -0.0311 0.0097 0.3036 0.6626
-0.01510.08920.0287 0.0515 -0.0310-0.0185 -0.0032 0.0335
0.0076 -0.0187 -0.0863 0.3893
0.0120 0.0466 0.0245 0.0608 -0.0242 -0.0570 0.0927 0.0270
0.0043 -0.0332 -0.0306 0.0749 0.2573
0.0242 0.0646 0.0778 0.0137 -0.0355 -0.0413 -0.0075 0.0586
0.0613 -0.0251 -0.0209 0.0716 0.0593 0.2074
-0.0 108 0.0218 0.0009-0.0051-0.00840.0449-0.0699 -0.0097 0.0262 -0.0422 -0.0229 0.0260 0.0067 0.0124 0.7086
-0.0245 -0.0178 -0.0439 0.0763 -0.0272 -0.0356 0.0473 0.0018 0.0596 -0.0444 0.0055 0.0043 0.0308 -0.0235 0.0685 0.6710
LA
'TellsBad' 'HitPamt' 'DrinkPMK' 'Sex' 'ResSize' 'Income' 'Friends'
'CloseMom' 'CloseDad' 'Praise' 'TalkPlay' 'DiffWait' 'ReactAng'
'ManyFigh' 'Optimism' 'SolvePro'
SE
'DiffWait' 'ManyFigh' 'ReactAng' 'Optimism' 'SolvePro'
'TellsBad' 'HitPamt' 'DrinkPMK' 'Sex' 'ResSize' 'Income'
'CloseMom' 'CloseDad' 'Friends' 'Praise' 'TalkPlay'/
MO NY=5 NX=l l NE=6 NK=l l LY=FU ,FI LX=FU,FI
BE=FU,FI GA=FU,FI C
PH=SY,FR PS=SY,FI TE=SY,FI TD=SY,FI
VA 1.0 LY(l,I) LY(2,2) LY(3,3) LY(4,4) LY(5 ,6) LX(l,l)
LX(2,2) LX(3,3) LX(4,4) LX(5,5) LX(6,6) LX(7,7) LX(8,8)
LX(9,9) LX(I0,10) LX(l l,11)
FR BE(2, l) BE(2,3)
FR BE(3,l)
FR BE(4,2)
FR BE(5,6)
FR BE(6,5)
EQ BE(5,6) BE(6,5)
ST 0.1 BE(2,I) BE(3,1) BE(2,3)
ST 0.1 BE(4,2)
ST 0.1 BE(6,5)
ST 0.1 BE(5,6)
VA 1.0 BE(5,4)
FR GA(l,l) GA(l ,2) GA(l ,3) GA(l ,4) GA(l ,5) GA(l,6) GA(l ,7)
GA(l,8) GA(l ,11)
FR GA(2,l) GA(2,2) GA(2,4) GA(2,7) GA(2 ,8) GA(2,9)
GA(2,10) GA(2,I I)
FR GA(3,2) GA(3,4) GA(3,6) GA(3 ,7) GA(3,9)
FR GA(4,l) GA(4,9)

FR GA(5,4) GA(5,8) GA(5,9)
ST 0.02 GA(l,l) GA(l ,2) GA(l,3) GA(l,4) GA(l,5) GA(l,6)
GA(l ,7) GA(l,8) GA(l,11)
ST 0.03 GA(2, I) GA(2,2) GA(2,4) GA(2, 7) GA(2,8) GA(2,9)
GA(2, I 0) GA(2, 11)
ST 0.02 GA(3,2) GA(3,4) GA(3,6) GA(3 ,7) GA(3,9)
ST 0.02 GA(4,9)
ST 0.02 GA(5,4) GA(5 ,8)
ST 0.1 GA(5,9)
STD.I GA(4,I)
FR PS(l,l) PS(2,2) PS(3,3) PS(4,4) PS(6 ,6)
VA 2.78 PS(5,5)
ST 0.05 PS(l,l) PS(2,2) PS(3 ,3)
ST 0.04 PS(4,4)
ST 0.5 PS(6,6)
VA 0.0467 TE(l,I)
VA 0.0180 TE(3,3)
VA 0.0104 TE(2,2)
VA 0.0354 TE(5,5)
VA 0.0354 TE(4,4)
VA 0.0115 TD(l,I)
VA 0.2045 TD(2,2)
VA 0.2189 TD(3 ,3)
VA 0.0025 TD(4,4)
VA 0.2609 TD(5,5)
VA 0.0972 TD(6,6)
VA 0.0235 TD(7,7)
VA 0.0210 TD(8,8)
VA 0.0502 TD(9,9)
VA 0.0482 TD(I0,10)
VA 0.0464 TD(l l,11 )
OU
Supplemental Group
DA Nl=16 N0=289 MA=CM
CMSY
0.2392
0.0021 0.5111'
0.0763 0.0826 3.3385
0.0020 0.0335 -0.0612 0.2507
0.0519 -0.0189 -0.1482 -0.0 145 2.2820
0.0173 -0.0288 -0.1666 0.0275 0.3292 0.8307
0.0285 -0.0094 -0.0896 0.1840 0.0045 -0.0526 1.0034 a
0.0073 0.0398 0.1394 0.0007 -0.0395 -0.0328 -0.0095 0.2353'
0.0160 0.0373 0.0913 -0.0565 0.0002-0.0448-0.0785 0.1512
0.4194'
0.0404 -0.0861 0.2622 -0.0355 0.0856 0.0740 -0.0266 -0.0322 0.0159 0.6188
0.0350 -0.0931 0.3283 -0.0 162 0.0982 0.0718-0.0288-0.0311 0.0097 0.2831 0.5425
-0.01510.08920.0287 0.0515 -0.0310 -0.01 85 -0.0032 0.0335
0.0076 -0.0187 -0.0863 0.3893'
0.0120 0.0466 0.0245 0.0608 -0.0242 -0.0570 0.0927 0.0270
0.0043 -0.0332 -0.0306 0.0749 0.2573'
0.0242 0.0646 0.0778 0.0137 -0.0355 -0.0413 -0.0075 0.0586
0.0613 -0.025 1-0.02090.0716 0.0593 0.2074
-0.0405 0.0218 0.0325 0.0562 -0.0066 -0.0430 -0.0699 -0.0097 0.0262 -0.0169 -0.0183 0.0260 0.0067 0.0124 0.6559

99

-0.0575 -0.0178 -0.2193 0.0375 0.0417 0.0685 0.0473 0.00180.0596 -0.0711 -0.0682 0.0043 0.0308 -0.0235 0.1487 0.6868
LA
'TellsBad' 'XHitPamt' 'DrinkPMK' 'Sex' 'ResSize' 'Income'
'XFriends' 'XCloseMom' 'XCloseDad' 'Praise' 'TalkPlay'
'XDiflWait' 'XReactAng' 'XManyFigh' 'Optimism' 'SolvePro'
SE
'XDiflWait' 'XManyFigh' 'XReactAng' 'Optimism' 'SolvePro'
'TellsBad' 'XHitPamt' 'DrinkPMK' 'Sex' 'ResSize' 'Income'
'XCloseMom' 'XCloseDad' 'XFriends' 'Praise' 'TalkPlay'/
MO NY=5 NX=I I NE=6 NK=l 1 LY=FU,Fl LX=FU,FI
BE=FU,FI GA=FU,FI C
PH=SY,FR PS=SY,FI TE=SY,FI TD=SY,FI
VA 1.0 LY(l , I) LY(2,2) LY(3 ,3) LY(4,4) LY(5 ,6) LX(l , I)
LX(2,2) LX(3 ,3) LX(4,4) LX(5 ,5) LX(6,6) LX(7 ,7) LX(8 ,8)
LX(9,9) LX(I0,10) LX(I 1,11)
FR BE(2, I) BE(2,3)
FRBE(3, I)
FR BE(4,2)
FR BE(5,6)
FR BE(6,5)
EQ BE(5,6) BE(6,5)
VA 1.0 BE(5,4)
ST 0.1 BE(2,1) BE(3 , I) BE(2,3)
STO.I BE(4,2)
ST 0.01 BE(6,5)
ST O.DI BE(5,6)
FR GA(l ,1) GA(l ,2) GA(l ,3) GA(l ,4) GA(l ,5) GA(l ,6) GA( l,7)
GA(l ,8) GA(l ,11 )
FR GA(2, I) GA(2,2) GA(2,4) GA(2,7) GA(2,8) GA(2,9)
GA(2, I 0) GA(2, 11)
FR GA(3 ,2) GA(3 ,4) GA(3 ,6) GA(3 ,7) GA(3 ,9)
FR GA(4, I) GA(4,9)
FR GA(5,4) GA(5 ,8) GA(5,9)
ST 0.02 GA(l,1) GA(l ,2) GA(l ,3) GA(l ,4) GA(l ,5) GA(l ,6)
GA(I ,7) GA(I ,8) GA(l ,11)
ST O.Q3 GA(2,I) GA(2 ,2) GA(2,4) GA(2,7) GA(2,8) GA(2,9)
GA(2, 10) GA(2,11)
ST 0.02 GA(3 ,2) GA(3 ,4) GA(3 ,6) GA(3 ,7) GA(3,9)
ST 0.02 GA(4,9)
ST 0.02 GA(5,4) GA(5,8)
ST 0.1 GA(5 ,9)
ST0.1 GA(4,1)

FR PS(I , I) PS(2,2) PS(3 ,3) PS(4,4) PS(6,6)
VA 0.988 PS(5 ,5)
ST 0.5 PS(l , I) PS(2 ,2) PS(3 ,3)
ST 0.4 PS(4,4)
ST 0.2 PS(6,6)
VAO.OOOTE(l , I)
VA 0.000 TE(3,3)
VA 0.000 TE(2 ,2)
VA 0.0343 TE(5,5)
VA 0.0328 TE(4,4)
VA 0.0120 TD(I , I)
VA 0.0000 TD(2 ,2)
VA 0.2337 TD(3 ,3)
VA 0.0025 TD(4,4)
VA 0.2738 TD(5 ,5)
VA 0.0997 TD(6 ,6)
VA 0.0000 TD(7 ,7)
VA 0.0000 TD(8 ,8)
VA 0.0000 TD(9 ,9)
VA 0.0433 TD(I0 ,10)
VA 0.0380 TD(l l ,11)
EQ BE(l ,2,1) BE(2,2,1)
EQ BE(l ,2,3) BE(2,2,3)
EQ BE(l,3,1) BE(2,3, I)
EQ BE(l ,4,2) BE(2,4,2)
EQ GA(l ,1,2) GA(2 , l ,2)
EQ GA(l ,1,7) GA(2,1 ,7)
EQ GA(l ,1,8) GA(2, l,8)
EQ GA(l ,2,2) GA(2,2,2)
EQ GA( l ,2,7) GA(2,2,7)
EQ GA(l ,2,8) GA(2,2,8)
EQ GA( l,2 ,9) GA(2,2,9)
EQ GA(l ,3,2) GA(2,3,2)
EQ GA(l ,3,7) GA(2,3,7)
EQ GA(I ,5,8) GA(2,5,8)
EQ GA(l ,5,9) GA(2 ,5,9)
EQ PS(l ,1,1) PS(2,l , I)
EQ PS(l ,2,2) PS(2,2,2)
EQ PS(l ,3,3) PS(2 ,3,3)
EQ PS( 1,4,4) PS(2 ,4,4)
PD
OU MLALLAD=OFF

• variances and covariance's substituted from the Primary Group for the Supplemental Group

100