The Relationship between Attachment and Externalizing Behaviors in

advertisement
Attachment and Externalizing Behaviors
The Relationship Between Attachment and Externalizing Behaviors in
Normative versus High-Risk Samples
Emilie A. Paczkowski
Distinguished Majors Program
University of Virginia
Advisor: N. Dickon Reppucci
Second Reader: Joseph P. Allen
Running head: ATTACHMENT AND EXTERNALIZING BEHAVIORS
1
Attachment and Externalizing Behaviors
2
Abstract
This study examined how gender and sample type affect associations between attachment and
externalizing behaviors. Normative participants (87 male and 98 female; mean age 14.25) and
high-risk participants (166 male and 105 female; mean age 14.34) completed the Adolescent
Attachment Interview and Family Attachment Interview, respectively. The Youth Self-Report
(YSR) measured externalizing behaviors. Insecure attachment styles and externalizing behaviors
were more common among high-risk participants, but no gender differences were found on these
measures. Secure participants exhibited lower externalizing behaviors overall and within both
males and females. Confirmatory factor analytic techniques provided support for measurement
invariance across normative and high-risk samples on YSR externalizing subscales. While no
relationship was found between attachment security and externalizing behavior in the high risk
sample of youth, structural modeling techniques indicated the presence of this relationship within
the normative sample.
Attachment and Externalizing Behaviors
3
The Relationship Between Attachment and Externalizing Behaviors in
Normative versus High-Risk Samples
Externalizing behavior problems are the single most common reason for which young
children are referred for psychological treatment (Richman, 1985). Severe externalizing
behaviors, such as aggression and attention problems that arise in early childhood are likely to
endure into later childhood and adolescence (Campbell, 1995). Additionally, many youths
become involved in some type of delinquent externalizing behavior over the course of
adolescence (Moffitt, 1993) at great cost to the individuals involved, as well as to the
community. Although the precise etiology of these behaviors is unknown, it is certain that these
problems do not develop in a vacuum. Rather, child and adolescent development is influenced
by multiple contexts, one of the most salient of which is the family. Since its inception,
attachment theory (Bowlby, 1969/1982) has served as a useful lens through which to examine
the influence of family factors on development in childhood and, more recently, adolescence.
Though previous research has investigated relationships between externalizing behaviors
and attachment style, few studies have examined the way these relationships may function
differently across groups. In particular, researchers have not grappled with the question of
whether the relationship between attachment style and the development of externalizing
behaviors changes form after a certain threshold. Specifically, the question of whether the
relationship between attachment style and externalizing behaviors differs across normative and
high-risk groups of adolescents has not been addressed. In addition, there is a growing body of
literature suggesting that findings based on normative versus clinical/forensic studies of girls’
externalizing behaviors are inconsistent. The goal of the present study, therefore, is to examine
the nature and the function of the relationship between attachment styles and levels of
Attachment and Externalizing Behaviors
4
externalizing behaviors across both group and gender. This objective will be accomplished
through secondary data analysis. The value of this approach lies in the ability to integrate
existing databases to fill an essential void in attachment research. With the advantage of having
access to both normative and high-risk samples of adolescent males and females, the present
study will build on previous research by testing for invariant relationships across samples and
gender. This research will also extend attachment literature that has dealt extensively with
children but has only recently begun to examine these issues in adolescence.
The review of the literature begins with an overview of attachment theory, including a
discussion of the assessment of attachment in adolescence as it relates to the classification
scheme used in the present study. Externalizing behavior is then addressed with a focus on the
measurement of the behaviors of interest, aggression, delinquency, and attention problems.
Next, research examining each of these constructs from the perspective of attachment theory is
reviewed. Differences in attachment style and levels of externalizing behaviors in males versus
females and normative versus high-risk samples are then explored with regard to how these
differences will drive the current research questions.
Attachment Theory
Bowlby’s (1969/1982) influential work in attachment theory was derived largely from
ethological studies of animal behavior. He claimed that attachment is an instinctual system that
operates to maintain proximity to the mother, contributing to the survival of the individual or the
species. Attachment behavior encompasses the actions a child takes to maintain proximity to the
attachment figure. Stressful situations that constitute a threat to the individual activate the
attachment system and elicit attachment behavior (Rice, 1990). Cicchetti, Cummings,
Greenberg, and Marvin (1990) elaborated on these aspects of attachment, defining its three
Attachment and Externalizing Behaviors
5
“essential features:” 1) its function, protecting children from danger, 2) its outcome, regulating
proximity to the attachment figure, and 3) its set goal, establishing a state of security.
Ainsworth, Blehar, Waters, and Wall (1978) described the quality or security of attachment
relationships as dependent on the attachment figure’s responsiveness. Children whose caretakers
respond reliably and appropriately to their attachment behavior are generally securely attached,
whereas those whose caretakers are unresponsive or unreliably responsive are generally
insecurely attached. In addition to appropriate response in times of distress, parents of securely
attached children have also been found to facilitate their children’s independent exploration
(Ainsworth, 1989).
Ainsworth et al. (1978) developed a classification system for attachment in infancy that
categorized children as avoidant, secure, or ambivalent. An avoidant child is characterized by a
lack of exploration before separation from the attachment figure and the tendency to ignore the
mother when they are reunited. Children who are classified as securely attached willingly
explore when under minimal stress and seek contact when they experience distress. It is believed
that the secure child is able to derive comfort from this contact and is then able to return to play.
An ambivalent child engages in little exploration and seeks closeness to the attachment figure
when experiencing minimal stress prior to separation. Upon reunion the child remains unsettled
and will seek and resist contact with the attachment figure.
Attachment relations are believed to persist even when attachment figures are not present.
Ainsworth (1969) and Bowlby (1969/1982) have both noted this durability of attachment
relations and have theorized that it results from the formation of what Ainsworth (1969) referred
to as “intra-organismic structures,” and what Bowlby (1969/1982) referred to as “internal
working models.” Bowlby contended that a child internalizes representations of the attachment
Attachment and Externalizing Behaviors
6
figure and his/her relationship with the attachment figure. The cognitive structures derived from
these representations are hypothesized to allow individuals to understand and anticipate what
occurs in the world around them and to assimilate information relevant to themselves and their
relationships with others.
Defining and measuring attachment beyond infancy. Ainsworth and Bowlby contended
that the internal working models developed in infancy maintain themselves by biasing perception
and cognition and by influencing how the child shapes his/her own interpersonal environment
(Bowlby, 1969/1982, 1973). Bowlby suggested that because of the endurance of working
models, attachment behavior in adolescence and adulthood is a direct extension of childhood
attachment behavior. The evaluation of attachment beyond childhood through instruments such
as the Adult Attachment Interview (AAI), involves assessing the internal working models
established earlier in life (Main, Kaplan, & Cassidy, 1985). The AAI does not seek to elicit
objective memories of past attachment related events; rather it endeavors to infer the individual’s
strategies for regulating the attachment system through the analysis of his/her narrative of
childhood attachment experiences (Van Ijzendoorn, 1995). The interview assesses the security
of the individual’s attachment by examining the coherence of his/her depiction of attachment
experiences and how well he/she is able to incorporate specific memories into a broader
understanding of the parent-child relationship. Unlike assessments of attachment in infancy, this
interview does not evaluate the current security of attachments and is not relationship specific
(Main et al., 1985).
Main and colleagues (1985) identified three major patterns for classifying adult
attachment. Within this scheme, individuals who are able to discuss past attachment experiences
coherently and integrate these experiences into their representations of themselves in
Attachment and Externalizing Behaviors
7
relationships are classified as securely attached. These individuals also value attachment
relationships. Dismissing attachment is marked by incoherent discourse regarding attachment
experiences. These individuals may idealize attachment figures, claim they are unable to recall
attachment experiences, or dismiss the impact of non-supportive experiences. Individuals
classified as dismissing may further belittle the need for attachment or attachment figures.
Individuals who are identified as having a preoccupied attachment style often exhibit anger in
discussing attachment relationships, have trouble separating past and present relationships, and
waiver between positive and negative appraisals of attachment experiences. Though they are
given different names, the classifications of insecure adult attachment described above are
analogous to the aforementioned insecure attachment classifications for infants. In other words,
dismissing attachment in adulthood is the counterpart of avoidant attachment in infancy (Maio,
Finchman, & Lycett, 2000), and preoccupied attachment is the counterpart of ambivalent
attachment (Allen et al., 2002).
Bartholomew (1990) expanded upon these three categories of adult attachment, defining
four attachment classifications based on two dimensions of internal working models. These two
dimensions include a self-model, characterized by the self-worth and anxiety experienced in
attachment relationships, and an other-model, characterized by the tendency to seek out or avoid
support. In this system, secure attachment involves having positive self- and other-models,
which enables one to develop intimate relationships while maintaining autonomy. Those with a
preoccupied pattern of attachment have a negative self-model and a positive other-model,
leading to the anxious pursuit of intimacy and reassurance. Dismissing attachment consists of a
positive self-model and a negative other-model, which results in high self-esteem, coupled with a
desire to preserves one’s independence in relationships. The fourth category in this system,
Attachment and Externalizing Behaviors
8
fearful attachment, involves negative self- and other-models. Those with a fearful attachment
pattern are believed to avoid intimacy because they fear loss.
Externalizing Behaviors
The behavioral and emotional problems that past research has examined in relationship to
attachment fall into two major categories, known as “externalizing” and “internalizing” behavior.
Aggressive and delinquent behavior and, in some cases, attention problems and hyperactivity fall
into the externalizing domain, whereas anxiety, depression, somatic complaints, and withdrawal
are classified within the internalizing grouping (Achenbach, 1985). The present research will
concentrate specifically on the relationship between the attachment patterns and externalizing
behavior, which previous studies have shown to be significantly higher in insecurely versus
securely attached children (Speltz, Greenberg, & Deklyen, 1990).
Why is it important to study externalizing behaviors? Externalizing behavior is of
particular interest for a number of reasons. This type of disruptive behavior is the most common
referral problem for preschool children brought to child psychiatry clinics (Richman, 1985).
Externalizing behavior that is present at age three or four years is also likely to persist into
elementary school and early adolescence (probability around 50%) (Campbell, 1995). Moffitt
(1993) recognized this stability specifically in serious antisocial behavior over the lifespan but
also noted the temporary increase in the number of people involved in serious acts of
delinquency in adolescence. These features of externalizing behavior make it a particularly
interesting construct to investigate in adolescence, especially as it relates to attachment, a
construct rooted in childhood and theorized to influence behavior over the lifespan. Both
externalizing behavior and attachment boast strong developmental theories and empirical
research to support the nature of their growth from childhood to adolescence, making a
Attachment and Externalizing Behaviors
9
compelling case for a line of inquiry, like that undertaken in this study, aimed at developing a
more comprehensive understanding of the interplay between the two constructs.
Measuring externalizing behavior. As noted above, externalizing behavior is an umbrella
term, encompassing a variety of problem behaviors that have been defined both by clinical
diagnoses and empirically based problem behavior syndromes. Two of the most widely used and
validated measures of externalizing behavior are the Child Behavior Checklist (CBCL) and the
Youth Self Report (YSR) (Achenbach, 1987). The CBCL is a measure that allows parents to
report on the competencies and behavioral and emotional problems of 4-18 year olds. The YSR
is a self-report measure designed for children with a mental age of at least 10 years that allows
adolescents to report on their own competencies and problems. The items used on the two
measures are identical to a large degree and yield scores for specific scales, including Attention
Problems, Delinquent Behavior, and Aggressive Behavior. The DSM diagnoses of attention
deficit hyperactivity disorder (ADHD), conduct disorder (CD), and oppositional defiant disorder
(ODD) have also been viewed as externalizing behavior problems (Hinshaw, 1992). Gould,
Bird, and Jaramillo, (1993) investigated the convergence between the behavior problem
syndromes derived from the CBCL and YSR and psychiatric diagnoses based on the DSM-III.
The authors found a strong linear relationship between scores on the Attention Problems scale
and ADD, F (1, 225) = 55.1, p < .001. Scores on the Delinquent scale were linearly related to a
diagnosis of either conduct disorder or oppositional disorder, F (1,306) = 86.5, p < .001. A
strong linear relationship was also found between a diagnosis of either conduct disorder or
oppositional disorder and scores on the Aggressive scale, F (1, 306) = 142.7, p < .001.
Attachment and Externalizing Behaviors 10
The Relationship Between Attachment and Externalizing Behaviors
Research has shown that parental warmth and attachment may reduce the negative effects
of stress and promote adaptive functioning in children (Garmezy, 1983). Conversely, insecure
attachment has been identified as a risk factor that interacts with other factors within the family
and the child to increase the likelihood of childhood behavior problems (Greenberg & Speltz,
1988). Arguably, insecure attachment may lead to deviant behavior when children whose
parents are not responsive and supportive develop models of attachment characterized by anger
and hostility. These children are also likely to believe that the people in their lives will not meet
their needs (Loeber & Dishon, 1983). Toth and Cicchetti (1996) proposed a similar idea,
claiming that a “maladaptive pathway” may link early insecure attachment to the development of
negative models of relationship figures in later childhood. In addition to developing negative
internal working models of attachment relationships, insecure children may learn to over or
under regulate their affect and behavior in reaction to caretakers who selectively respond to their
emotional needs (Sroufe, 1983). Other research indicates that close and affectionate
relationships between children and caretakers facilitate children’s internalization of rules of
conduct and increase the likelihood that children will feel committed to the welfare of others.
(Kochanska, Tjebkes, & Forman, 1998). The maladaptive externalizing behaviors that may
ultimately result from insecure attachments to parents are the focus of the current study.
The relationship between aggression and attachment. Past researchers have explored the
aforementioned externalizing behaviors, including aggressive behavior, delinquent behavior, and
attention problems, within the framework of attachment theory. Studies of aggression in
children have built on Bowlby’s (1969/1982, 1973) idea that the maladaptive views of the self
and others that result from insecure parent-child relationships put a child at risk for aggression.
Attachment and Externalizing Behaviors 11
Main and Goldwyn (1984) found that insecurely attached infants acted more aggressively toward
their mothers than did securely attached infants. This greater aggressiveness has been shown to
carry over into childhood among children with an avoidant attachment classification from lowincome samples (Renken, Egeland, Marvinney, Mangelsdorf, & Sroufe, 1989). Studies of
middle-income samples, however, failed to find the same association (Fagot & Kavanagh, 1990;
Lewis, Feiring, McGuffog, & Jaskir, 1984) with the exception of Teti and Ablard’s (1989)
sibling study, which found securely attached children to be less aggressive and more compliant
than insecure children. Aggression and attachment have also been examined among physically
abused, neglected, and nonabused/nonneglected children (Finzi, Ram, Har-Evan, Shnit, &
Weizman, 2001). This research found that abused children were significantly more likely to
have an avoidant attachment style and higher aggression scores than the other two groups,
whereas neglected children were significantly more likely to display ambivalent attachment and
lower aggression than abused children. Nonabused/nonneglected children were characterized by
a secure attachment style (68.6%) and low aggression.
Aggression and noncompliance in childhood have also been linked to antisocial behavior
in adolescence (Loeber & Dishon, 1983), which researchers have only more recently begun to
examine in relation to attachment. Many studies of adolescents’ attachment to parents, however,
have focused on current relationship quality rather than attempting to assess the adolescent’s
models of attachment relationships formed in childhood. For example, Simons, Paternite, and
Shore (2001) found that adolescents’ perceived quality of mother-adolescent attachment was
negatively correlated to self-reported aggression. Other research has used attachment theory to
investigate how attachment to parents carries over specifically into other intimate relationships.
For instance, insecurely attached adolescents have been found to be more likely to be engaged in
Attachment and Externalizing Behaviors 12
reciprocally aggressive dating relationships than are securely attached adolescents (Bookwala &
Zdaniuk, 1998). In terms of more global measures of aggression and other externalizing
behaviors, the child or adolescent’s internal working model of attachment relationships is
thought to be the link between poor parenting in childhood and problem behavior in adolescence
(Allen, Aber, & Leadbeater, 1990). The present research hopes to elucidate the nature of the link
between attachment representations formed in early childhood and general aggressive behavior.
The role of attachment in explaining delinquency. The study of delinquency as it relates
to attachment is rooted directly in Bowlby’s (1944) original conception of attachment theory,
which was developed in part to explain the personality of juvenile thieves. Bowlby posited that
these juveniles had developed internal working models of others as unworthy of trust, empathy,
and concern, leading to a callous interpersonal style. In Hirschi’s (1969) criminological theory
of delinquent behavior, attachment was described as an affective relationship that facilitates the
internalization of norms. Those with insecure attachments may lack the social bonds that would
cause them to identify with the social order.
More recent work has produced evidence to support these claims that attachment is
related to delinquency. For example, Arsenio, Shea, and Sacks (2000) found that juvenile
offenders were more likely to be insecurely attached than their peers, while Allen, Moore,
Kuperminc, and Bell (1998) found that a combination of adolescent-reported self-worth,
adolescent-reported attachment to mother, and mother-reported maternal control predicted
mother-reported delinquent behavior. It was posited that delinquency may be a form of rebellion
against attachment figures’ norms and controls among insecure-dismissing adolescents (Allen,
Moore, & Kuperminc, 1997). Among insecure-preoccupied adolescents, delinquency may act as
a dysfunctional form of attachment behavior, increasing the intensity of interaction with
Attachment and Externalizing Behaviors 13
attachment figures (Allen et al., 1998). In this scenario, the increasing autonomy that
characterizes parent-child interactions in adolescence is hypothesized to be particularly
threatening to preoccupied adolescents (Allen et al., 2002). This combination of increasing
autonomy and the pre-existing vulnerability of preoccupied attachment may help to explain the
great increase in delinquency that occurs in adolescence (Moffitt, 1993). The present research
will attempt to replicate and further clarify the link between delinquency and specific attachment
styles.
Exploring the relationship between attention problems and attachment. Though less
research has been devoted to exploring the relationship between attention problems and
attachment, a few studies have attempted to link these two constructs. Ladnier and Massanari
(2000) pointed out that behaviors characteristic of avoidant children are similar to symptoms of
ADHD, but failed to produce evidence that ADHD occurs with greater frequency among
children with an avoidant attachment style. Alternatively, another recent study found that the
type of attachment insecurity that was present in those children diagnosed with ADHD was
consistent with an ambivalent attachment style (Clarke, Ungerer, Chahoud, Johnson, & Stiefel,
2002). Smith (1994) posited a phenomenon he referred to as dis-attachment that results from a
failure of the mother to bond with a child who is very demanding and difficult to comfort.
Almost all of the children Smith studied who experienced this dis-attachment also exhibited
symptoms of ADHD. Rather than proposing that the lack of secure attachment causes ADHD,
however, he claimed that an underlying deficit in the neurotransmitter, serotonin, is responsible
for the symptoms of ADHD as well as those behaviors that prohibit proper parent-child bonding.
In a similar vein, Ladnier and Massanari (2000) described ADHD as a consequence of
“attachment deficits” that result from “bonding breaks.” A bonding break is an event that
Attachment and Externalizing Behaviors 14
prevents a child from forming a secure attachment to a caretaker. The trauma of failed
attachment interferes with neurological development, resulting in attachment deficits, or
shortcomings in the child’s emotional development. These deficiencies in emotional
development are reflected in the emotional and behavioral symptoms of ADHD.
The direction of this past research points to a connection between attention problems and
attachment style, but the work done thus far has failed to operationalize attachment in a
consistent and meaningful way and has looked only at attention problems as they are expressed
in those diagnosed with ADHD, ignoring those who may have subclinical levels of these
problems. Research examining attention problems as they exists on a continuum from the
perspective of attachment theory may be revealing of how attachment style relates to attention
difficulties in those who fall both above and below the clinical cutoff for a diagnosis of ADHD.
Does Gender Matter in the Examination of Attachment and Externalizing Behaviors?
Although researchers have drawn comparisons between male and females in the study of
attachment, these studies have focused almost exclusively on children. The present study seeks
to extend this research into adolescence. According to Bowlby’s (1969/1982) conception of
attachment, males and females are equally likely to be securely attached, yet he noted that
females were more likely to exhibit ambivalent attachment, whereas males were more likely to
exhibit avoidant attachment. Current research will examine whether this trend continues into
adolescence, with females more likely to exhibit preoccupied attachment and males more likely
to exhibit dismissing attachment.
With respect to externalizing behavior, previous research has consistently found males to
display higher levels of this behavior than females. An investigation of problem behavior in
childhood found that boys were more often placed in the clinical cut-off group for problem
Attachment and Externalizing Behaviors 15
behavior than were girls (Moss, Rousseau, Parent, St-Laurent, & Saintonge, 1998). Psychiatric
diagnoses of behavior disorders also differ between males and females. Boys outnumber girls
with a ratio of 3:1 for ADHD and with a ratio of 2:1 to 3:1 for CD (Robins, 1991). In a study
employing the CBCL, girls also received lower scores in delinquent and aggressive behavior
(Gjone & Stevenson, 1997). An exception to these findings is the fact that females who are
found in clinical/adjudicated populations often display a greater number of co-morbid problems
than males in these settings (Moretti & Odgers, 2002).
Still, few studies have examined these constructs in a way that allows for a direct
comparison of the relationship between attachment style and externalizing behavior in males and
females. Moreover, the research that exists has yielded conflicting findings. For example, some
evidence has pointed to a stronger relationship between problem behavior and insecure
attachment in boys than in girls (Lewis et al., 1984), showing that 40% of insecure males
compared to 6% of secure males scored above the 90th percentile on the CBCL Problem Total
score, while no significant effects of attachment security were found for girls. This same study,
however, found that boys classified as insecurely attached had more internalizing problems than
securely attached boys, whereas girls classified as insecurely attached had more externalizing
problems. Furthermore, another study found that girls who were categorized as insecureavoidant were rated as more difficult than securely attached girls and boys (Fagot & Kavanagh,
1990).
Unique Relationship in High-Risk Samples
Studies of attachment and externalizing behavior have also explored these constructs
within both normative and high-risk populations. The research presented thus far has included
work done with samples drawn from both of these populations. Clinical/high-risk samples
Attachment and Externalizing Behaviors 16
deserve special attention, however, as the prevalence of attachment insecurity and externalizing
behavior as well as the relationship between these two constructs in these groups may differ from
normative samples. Scharfe (2002) found that attachment could be reliably measured in a
clinical sample of adolescents. Nevertheless, the distribution of children and adolescents across
the attachment classifications varies between high-risk and normative samples. Among
preschool aged children, Speltz, Greenberg, and Deklyen (1990) found that only 20% of clinic
children exhibited secure attachments, whereas 72% of the comparison group was securely
attached. An investigation of attachment in an adolescent clinic sample revealed that 94% of the
sample was insecurely attached (Scharfe, 2002). The present study will seek to replicate these
findings.
Externalizing behavior has also been the focus of many studies comparing high-risk and
normative samples. Ramos-Marcuse & Arsenio (2001) found that these behaviors were greater
among clinic-referred children than among non-clinic-referred children. Moreover, comorbidity
rates of the externalizing behaviors currently under investigation may be higher in a clinical
sample than in a normative sample due to “Berkson’s bias” (Berkson, 1946). Berkson’s bias
refers to the higher probability that those with multiple disorders will be referred for mental
health services.
Despite a body of research that compares normative and high-risk samples with regard to
attachment style and externalizing behavior, past research has yet to explore how the relationship
between these constructs may differ between normative and high-risk samples. Although direct
comparisons between these two samples cannot be drawn from the research to date, studies of
high-risk samples of children have found a relationship between insecure attachment and
externalizing behavior (Renken et al., 1989) that studies of normative samples have failed to find
Attachment and Externalizing Behaviors 17
(Fagot & Kavanagh, 1990; Lewis et al., 1984). With the advantage of access to both a normative
and high-risk sample, the present research will improve upon these studies by testing for
invariant relationships across samples and exploring these questions among adolescents.
The research reviewed above has shown that adolescents’ attachment styles reflect the
relationships forged with their parents in early childhood (Bowlby, 1969/1982). Through the
adolescent’s internal working models of attachment relationships, poor parenting in childhood
may continue to influence problem behavior in adolescence (Allen, Aber, & Leadbeater, 1990).
Past research has established that certain groups, like males and females (Bowlby, 1969/1982)
and normative and high-risk populations (Speltz, Greenberg, and Deklyen, 1990; Scharfe, 2002),
vary in the prevalence of specific attachment styles. Additionally, males (Moss, Rousseau,
Parent, St-Laurent, & Saintonge, 1998; Gjone & Stevenson, 1997) and those drawn from highrisk groups (Ramos-Marcuse & Arsenio, 2001) have been found to exhibit higher levels of
externalizing behaviors. Despite these bodies of research dealing separately with attachment and
externalizing behaviors in these groups, previous studies have yet to examine how the
relationship between these constructs may differ between males and females or normative and
high-risk samples. The present study will explore how the association between attachment style
and externalizing behaviors varies by gender and sample (normative versus high-risk).
Hypotheses
1. A greater proportion of the normative sample will be securely attached.
2. Males will display a dismissing attachment style more often than females, while females
will display a preoccupied attachment style more often than males.
3. High-risk participants will exhibit higher levels of attention problems, aggressive
behavior, and delinquent behavior than normative participants.
4. Males will exhibit higher levels of attention problems, aggressive behavior, and
delinquent behavior than females.
Attachment and Externalizing Behaviors 18
5. Insecure attachment will predict higher levels of each type of externalizing behavior.
6. Attachment insecurity will be a better predictor of externalizing behavior among females
than among males.
7. Attachment insecurity will be a better predictor of externalizing behavior in a high-risk
sample than in a normative sample.
To test these hypotheses, the present research drew samples from larger studies of
normative and high-risk adolescents. Attachment and externalizing behaviors were examined
independently and in relation to one another within each sample. Externalizing behaviors were
assessed as three constructs, attention problems, aggressive behavior, and delinquent behavior.
In addition, samples were split by gender to allow for comparisons of males and females. By
these means, the current research sought to determine how gender and population from which
one is drawn (normative versus high-risk), may affect how attachment style relates to
externalizing behaviors.
Methods
Participants
Normative sample. This sample was drawn from a larger longitudinal investigation of
adolescent social development in familial and peer contexts. Participants included 186
adolescents (87 male and 98 female). The mean age of the adolescents was 14.25 years (SD =
0.82), with a range from 12 to 16 years. The racial/ethnic background of the sample was 60.8%
Caucasian, 38.7% minority, with 0.5% unreported. The sample was primarily (82.4%) of upper
middle and upper socioeconomic status.
Adolescents were recruited from a public middle school drawing from suburban and
urban populations in the Mid Atlantic United States. Recruitment consisted of an initial mailing
to all parents in the school along with follow-up contact efforts at school lunches. Participants
Attachment and Externalizing Behaviors 19
provided informed assent before each interview session, and parents provided informed consent.
All interviews took place in private offices within a university academic building.
High-risk sample. This sample was drawn from a larger longitudinal study of attachment
and antisocial behavior among high-risk adolescents. Participants included 271 adolescents (166
male and 105 female). The mean age of the adolescents was 14.34 years (SD = 1.49), with a
range from 10 to 18 years. The racial/ethnic background of the sample was 48% Caucasian, 14%
minority, with 38% unreported. Though 83.4% of the sample failed to report their
socioeconomic status, 88.9% of those who did report on SES were of lower and lower middle
status.
Adolescents were recruited from a mental health center that serves a largely urban
population in Western Canada. Participants were recruited upon entry to the mental health
center, either on an inpatient or outpatient basis. Active parental consent was obtained. All
interviews took place in private offices within the mental health center.
As these were convenience samples, they were not matched a priori with respect to
demographic characteristics. As such, the results must be interpreted with this limitation in mind.
Measures
Adolescent Attachment Interview (AAI) (Carlson, 1989). Adolescents in the normative
sample were interviewed with the AAI, a modified version of the Adult Attachment Interview
(Main, Kaplan, & Cassidy, 1985) that maintains the same basic interview format. This semistructured interview was designed to investigate adult’s attachment representations by probing
for descriptions of early attachment relationships and specific memories that support and
contradict these descriptions. For example, participants were asked to list five words describing
their early childhood relationships with their parents and then to describe specific episodes that
Attachment and Externalizing Behaviors 20
reflected those words. Other questions focused on instances of distress, loss, separation, trauma,
and rejection. Lastly, the interviewer asked the participant to give descriptions of changes in
relationships with parents and the current state of those relationships. Scoring focused on the
individual’s state of mind regarding attachment by examining the accessibility of early
experiences to memory and the coherence of the participant’s narrative. Based on this interview,
adolescents were classified as secure, preoccupied, dismissing, or unclassifiable.
Family Attachment Interview (FAI) (Bartholomew & Horowitz, 1991). Adolescents in
the high-risk sample were interviewed with the FAI. This interview was designed to assess
experiences and feelings in family relationships and the coherence of accounts of those
relationships. Participants were asked to describe their family history and their feelings about
the importance of family relationships. With regard to relationships with their caregivers,
participants were asked to describe their reactions to instances of separation or loss, feelings in
the relationship, and changes since childhood. Participants were classified as secure,
preoccupied, dismissing, fearful, or even split.
The various attachment classifications derived from the AAI and the FAI were combined
to create four common categories of attachment style for use in the present study. This course of
action is justifiable based on research that has found a 78% correspondence between the three
main categories of attachment in the AAI (secure, preoccupied, and dismissing) and the
corresponding categories for the FAI (Bartholomew & Shaver, 1998). Those who were
unclassifiable based on the AAI were combined with those who were labeled even split based on
the FAI, as these categories have roughly the same meaning in their respective coding systems.
Those who were classified as fearful based on the FAI were labeled dismissing for the purposes
Attachment and Externalizing Behaviors 21
of this study, as both of these orientations to attachment drive avoidant behavior (Bartholomew,
Henderson, & Dutton, 2001).
Youth Self Report (YSR) (Achenbach, 1987). The YSR is a self-report measure designed
for children with a mental age of at least 10 years that allows adolescents to report on their own
competencies and problems. The original form of the YSR includes 112 items divided into 9
scales: Withdrawn, Somatic Complaints, Anxious/Depressed, Social Problems, Thought
Problems, Attention Problems, Delinquent Behavior, and Aggressive Behavior. Participants
were asked to rate how well a number of descriptions of symptomatic behaviors applied to them
within the past 6 months, on a scale of 0 = not true, 1 = somewhat or sometimes true, and 2 =
very often or often true. Both the normative and high-risk samples were given shortened
versions of the YSR. The present study employed only those items from the YSR that were
drawn from the Attention Problems, Delinquent Behavior, and Aggressive Behavior scales and
were administered to both samples.
Results
Analyses I: Prevalence of Attachment Styles by Sample and Gender
As illustrated in Table 1, the normative sample had a greater proportion of securely
attached participants, while the high-risk sample had a greater percentage of participants in each
of the other attachment classifications, even split, preoccupied, and dismissing. While no gender
differences were found in even split or secure attachment, a greater percentage of females were
classified as preoccupied, whereas a greater percentage of males were classified as dismissing.
Additional analyses yielded the same pattern of gender differences within the normative and
high-risk samples (see Table 2).
Attachment and Externalizing Behaviors 22
_____________________
Insert Tables 1 & 2 here
______________________
Analyses II: Levels of Externalizing Behavior
Univariate analyses of variance, reported in Table 3, examined whether levels of
externalizing behavior differed significantly between sample or gender groupings. The high-risk
sample exhibited greater attention problems, aggression, and delinquent behavior than the
normative sample. While no significant gender differences were found in attention problems,
aggressive behavior, or delinquent behavior in the combined sample or in the normative sample
alone (see Table 4), high-risk females exhibited significantly greater attention problems than did
their high-risk male counterparts.
______________________
Insert Tables 3 & 4 here
______________________
Analyses III: The Relationship Between Attachment and Externalizing Behaviors
Additional univariate analyses of variance were performed to investigate possible
associations between the various attachment styles and externalizing behaviors. These analyses
were carried out with the entire sample and with each of the gender and sample groups
separately. Results for the combined sample (see Table 5) indicated that secure attachment was
associated with fewer attention problems and less aggressive and delinquent behavior.
Preoccupied attachment predicted greater attention problems and aggressive and delinquent
behavior. Dismissing attachment was associated only with greater delinquent behavior.
___________________
Insert Table 5 here
____________________
Attachment and Externalizing Behaviors 23
When the entire sample was split by gender, secure attachment was associated with fewer
attention problems and less aggressive and delinquent behavior in both males and females (see
Table 6). Preoccupied attachment was associated with greater attention problems and aggressive
and delinquent behavior in females, but predicted only greater attention problems and aggressive
behavior in males. Dismissing attachment was not associated with any difference in
externalizing behaviors except for greater attention problems in females.
______________________
Insert Table 6 here
______________________
With the exception of greater aggressive behavior among preoccupied normative
participants (F (1, 136) = 4.67, p = .032), the associations between attachment styles and
externalizing behaviors did not exist in the normative and high-risk samples when they were
analyzed separately. Males and females were also analyzed separately within each sample,
revealing greater aggressive behavior among normative females with a preoccupied attachment
style (F (1, 69) = 5.43, p = .023). This finding constituted the only significant association
between attachment style and externalizing behavior within the four sub-samples defined by
gender and sample type (see Table 7). In order to examine the nature of these relationships more
closely, a series of advanced statistical analyses were performed. Specifically the goals of these
analysis were to; (1) ensure that the same latent construct of aggression was being measured
across groups, and (2) test for the possibility that the decrease in statistical significance was due
the reduced sample size created by within group analyses.
______________________
Insert Table 7 here
______________________
Attachment and Externalizing Behaviors 24
Analyses IV: Structural Modeling
The first set of advanced analyses involved fitting a series of multigroup factor analytic
models and testing for invariance across the high risk and normative samples. This procedure
was used to test for measurement equivalence across the two groups. Although the same measure
of externalizing behavior was used (Youth Self Report), the possibility existed that that the
underlying construct that they were tapping into was qualitatively different (McArdle, 1996;
McArdle & Nesselroade, 1994). In order to determine that aggression had been measured
equivalently across the samples the factor loading patterns from observed indicators of
aggression to the latent constructs of aggression had to be equal across the samples. The items
used as indicators included I destroy other people's things, ‘destroy,’ I disobey at school,
‘school,’ I get in many fights, ‘fights,’ I have a hot temper, ‘temper,’ I threaten to hurt people,
‘threaten.’
The key question guiding this analysis was whether:
1. the factor structure of aggression was invariant across normative and high-risk samples.
Previous research using similar measures has assumed that the structure of aggression is
equivalent across gender and has simply summed the items to form subscales.
In order to determine the best model for the data, a series of confirmatory models were
fit. Prior to testing for invariance across the normative and high-risk samples, a baseline model
was established for each group separately. Alternative models were then fit separately. The fit of
the model that constrained both the variances and loadings to be equal was acceptable (χ2 = 35.2,
df = 14, RMSEA = .06). Specifically, there was no significant improvement in fit when
constraints were lifted.
______________________
Insert Figures 1 & 2 here
______________________
Attachment and Externalizing Behaviors 25
The second set of advanced analyses also involved fitting a series of multigroup
confirmatory factor analytic models. Here, the question was whether information could be gained
by using the true score of aggression within a factor model as the dependent variable in a
structural framework. This “true” score was free of measurement error and allowed for a more
precise estimation of the association between the independent and dependent variables.
The key questions guiding these analyses were whether:
2. the relationships between gender, attachment and aggression were better understood within
a structural modeling framework, and whether
3. the relationship between gender, attachment, and aggression differed significantly between
high-risk and normative adolescents.
In order to determine the best model for the data, a series of confirmatory models were fit.
Prior to testing for invariance across the normative and high-risk samples, a baseline model was
established for each group separately. Alternative models were then fit and evaluated against the
baseline model. The first model was the most restrictive and constrained all of the parameter
estimates and variances to be equivalent across the sample. While this model demonstrated an
acceptable level of fit (χ2 = 63.3. df = 33, RMSEA = .04), the fit improved significantly when the
regressions from ‘gender’ and ‘secure’ to ‘aggression’ were allowed to vary.
______________________
Insert Figures 3 & 4 here
______________________
Discussion
The present study sought to develop a greater understanding of how gender and the
nature of the sample (high-risk versus normative) may affect the relationships between
participants’ specific attachment styles and self-reported externalizing behaviors. This study has
yielded several significant differences between the groups in question and affirms the need for
Attachment and Externalizing Behaviors 26
research comparing normative and high-risk samples with respect to attachment style and
externalizing behavior.
The Role of Gender and Sample Type in the Prevalence of Specific Attachment Styles
It was expected that attachment security would be greater in the normative sample.
Findings supported this hypothesized difference in the frequency of various attachment styles
based on sample type, revealing that secure attachment was more common among normative
participants, while each of the other attachment styles, even split, preoccupied, and dismissing,
were found with greater frequency in the high-risk sample. These findings are consistent with
the findings of previous researchers (Speltz et al., 1990; Scharfe, 2002).
Hypotheses pertaining to gender predicted that males would be more likely to display
dismissing attachment than females, while females would be more likely to display preoccupied
attachment than males. Females were in fact more likely to be classified as preoccupied, while
males were more likely to be classified as dismissing. These results are in keeping with
Bowlby’s (1969) findings that female infants were more likely to exhibit ambivalent attachment,
while males were more likely to exhibit avoidant attachment. This finding provides evidence of
stability in gender differences in attachment style over the lifespan. The replication of this
gender difference also reaffirms the need to study the impact of gender on the relationship
between attachment style and its various correlates.
Is Aggression Qualitatively Different Across High-risk and Normative Youth?
Advanced statistical analyses were employed to determine whether the items used to
measure the externalizing behaviors were assessing the same three constructs in both the
normative and high-risk samples. The factor analytic techniques used to make this determination
could be employed only with the items measuring aggressive behavior, as the number of items
Attachment and Externalizing Behaviors 27
measuring delinquent behavior and attention problems were insufficient to perform the analyses.
Findings established that the construct of aggressive behavior that the selected YSR items were
attempting to measure was not qualitatively different across samples, despite the quantitative
differences in levels of aggressive behavior found in the two samples. This result holds promise
for the current and future research, in that the same measures can be used to measure this
construct in normative and high-risk samples, allowing for more straightforward comparisons of
these groups.
When was Gender of Importance in Predicting Externalizing Behavior?
As expected, high-risk youth had higher rates of externalizing behavior, however, there
were no significant gender differences found in any of the externalizing behaviors within the
combined sample. This finding is intriguing given that past research has consistently found
higher rates of externalizing behaviors in males (Moss, Rousseau, Parent, St-Laurent, &
Saintonge, 1998; Robins, 1991; Gjone & Stevenson, 1997). A possible explanation for the lack
of significant gender differences is the use of self-report measures. Such measures may yield
higher scores for externalizing behaviors among females, as they are free from the biases of
clinicians and other reporters who may be reluctant to give externalizing diagnoses or high
ratings on externalizing behaviors to females as these behaviors violate gender stereotypes.
Although no gender differences were found in the sample overall, separate analyses of
the high-risk and normative samples revealed gender differences within these groups. Univariate
analyses of variance found that high-risk females had higher levels of attention problems than
high-risk males, which could be explained by the idea that females who are found in a
clinical/adjudicated population often have more co-morbid problems than males in these settings
(Moretti & Odgers, 2002). Confirmatory factor analytic models revealed relationships that had
Attachment and Externalizing Behaviors 28
previously gone undetected; namely, that males within normative populations were more likely
to exhibit aggressive behavior. This finding could be a result of selection bias, since females may
need to have levels of externalizing behavior comparable to males’ to become part of the highrisk sample.
The Relationship Between Attachment and Externalizing Behaviors
Insecure attachment was hypothesized to predict higher levels of aggressive behavior,
delinquent behavior, and attention problems. When the sample was analyzed as a whole,
findings generally supported this hypothesis, with secure attachment associated with lower levels
of each behavior. These results are in line with previous research that has linked secure
attachment to lower levels of aggressive behavior (Main and Goldwin, 1984; Teti and Ablard’s,
1989; Simons, Paternite, & Shore, 2001; Bookwala & Zdaniuk, 1998), delinquent behavior
(Allen, Moore, Kuperminc, and Bell, 1998; Allen, Moore, & Kuperminc, 1997; Allen et al.,
2002), and attention problems (Ladnier and Massanari, 2000; Clarke, Ungerer, Chahoud,
Johnson, & Stiefel, 2002; Smith, 1994).
The role of preoccupied attachment style. Preoccupied attachment was associated with
higher levels of each externalizing behavior. This finding conforms to past research that has
found relationships between attention problems and preoccupied attachment (Clarke et al., 2002),
as well as between delinquent behavior and preoccupied attachment (Allen, Moore, Kuperminc,
and Bell, 1998; Allen, Moore, & Kuperminc, 1997; Allen et al., 2002). Conversely, the specific
association found between preoccupied attachment and higher levels of aggressive behavior,
coupled with a lack of association between dismissing attachment and aggressive behavior, was
inconsistent with past findings (Renken, Egeland, Marvinney, Mangelsdorf, & Sroufe, 1989;
Finzi, Ram, Har-Evan, Shnit, & Weizman, 2001) linking aggression exclusively to avoidant
Attachment and Externalizing Behaviors 29
attachment. One potential explanation for these contradictory findings is the difference in the
ages of the samples being compared. Previous research has been done mainly with infants and
children, while the present study examined adolescents. The preeminence of preoccupied
attachment as a predictor of aggression among adolescents may be related to Allen and
colleagues’ (2002) idea that increasing autonomy in adolescence may be especially threatening
to preoccupied adolescents, causing an increase in their behavior problems at this age.
The implications of dismissing attachment style. Also in keeping with hypotheses and
past research (Allen, Moore, Kuperminc, and Bell, 1998; Allen, Moore, & Kuperminc, 1997;
Allen et al., 2002), a relationship was found between dismissing attachment higher levels of
delinquent behavior.
The Role of Attachment in Predicting Externalizing Behavior Among Males versus Females
Attachment insecurity was expected to be a better predictor of externalizing behavior
among females than among males. Partial support was found for this hypothesis, as preoccupied
and dismissing attachment styles were related to higher levels of a greater number of
externalizing behaviors in females than in males. Past research suggesting a stronger
relationship between externalizing behavior and insecure attachment in females than in males
(Lewis, Feiring, McGuffog, & Jaskir, 1984; Fagot & Kavanaugh, 1990) was echoed by the
discovery that insecure-preoccupied attachment predicted higher levels of all externalizing
behaviors for females, but failed to predict higher delinquent behavior in males. Additionally,
insecure-dismissing attachment failed to predict higher levels of any of the externalizing
behaviors for males, but did predict greater attention problems in females. One possible
interpretation of this finding draws on evidence from past research (Bowlby, 1969/1982), as well
as from the present study, that points to a lower proportion of females exhibiting dismissing
Attachment and Externalizing Behaviors 30
attachment. Studies have also found that females are less likely to be diagnosed with ADHD
(Robins, 1991). Thus, the association between dismissing attachment and attention problems
may be related to a third variable, gender role deviation.
Contrary to the hypothesis, however, the reverse relationship of secure attachment to
lower levels of externalizing behavior was not more prevalent among females, as this
relationship existed in both males and females when the sexes were analyzed separately. Beyond
providing further evidence that females’ externalizing behavior is better predicted by insecure
attachment than is males’ externalizing behavior, these findings indicate that externalizing
behaviors are associated with different types of insecure attachment in males versus females.
The Role of Attachment in Predicting Externalizing Behavior Among High-Risk Youth
Attachment insecurity was expected to be a better predictor of externalizing behavior in a
high-risk sample than in a normative sample. Univariate analyses of variance yielded no support
for this hypothesis, revealing that, despite a finding of greater aggressive behavior among
preoccupied normative participants, no associations were found between attachment style and
externalizing behavior when the normative and high-risk samples were examined separately.
Advanced analyses, however, found that, in the normative sample, aggressive behavior is lower
among those with a secure attachment style. No such relationship was found in the high-risk
sample. These findings are also contrary to the hypothesis and directly oppose past work that
has found a relationship between insecure attachment and externalizing behavior in a high-risk
sample (Renken et al., 1989), while studies of normative samples have failed to find such a
relationship (Fagot & Kavanagh, 1990; Lewis et al., 1984). A possible explanation for this
reversal of past findings is the distribution of participants in each sample over the attachment
classifications. Although the majority of normative participants in the present study were
Attachment and Externalizing Behaviors 31
classified as secure, this group showed greater variability in attachment classification than the
high-risk group, which was overwhelmingly insecurely attached (84.1%).
The current research replicated many of the findings of past studies, including the higher
levels of insecure attachment and externalizing behavior found in high-risk samples as well as
the relationship between secure attachment and lower levels of externalizing behavior found in
the combined sample. The failure to find a relationship between attachment style and
externalizing behavior in the high-risk sample, however, contradicted hypotheses and precluded
the possibility of comparing how that relationship might function differently in the normative
versus the high-risk sample. Still, these results should be considered a first step toward research
that allows such comparisons and should be viewed in light of the methodological weaknesses
inherent in any secondary data analysis.
Limitations
In the present study, data was brought together from two large-scale studies. Although
some of the same measurement instruments were employed in these studies, they were conducted
independently and for different purposes. Thus, the samples were not matched for demographic
characteristics, like socioeconomic status, living environment (small versus large urban area), or
ethnicity. Demographic dissimilarities between the samples present a problem, as they represent
other variables on which the samples differ. These other variables cannot be ignored as possible
explanations for any differences found between the two groups.
Other limitations of the present study center on the measurements of attachment
employed in each study. Different attachment interviews were used in the two samples and were
combined to create common categories of attachment style in which to classify each participant.
In doing so, benefits of each of the original measures of attachment may have been lost. The
Attachment and Externalizing Behaviors 32
Adolescent Attachment Interview yielded a continuous scale of attachment style that had to be
collapsed into categories, eliminating all variability that existed within each category of
attachment, which could have allowed for the analysis of relationships between levels of security
and insecurity and levels of externalizing behavior. The Family Attachment Interview included a
third type of attachment insecurity, fearful attachment. In combining the measures, this category
had to be merged with dismissing attachment. Although both fearful and dismissing attachment
are related to avoidant attachment behaviors, the fearful category is a well-validated
classification that does not overlap perfectly with any other insecure attachment style. Given
that the two insecure attachment styles were differentially associated with the three externalizing
behaviors in males versus females, the loss of this fourth category may have curtailed the ability
to find more specific relationships between externalizing behavior and different types of insecure
attachment in males and females.
Additionally, attachment interviews were not given to every individual in the samples.
ANOVAs were run to determine whether there was a significant difference in levels of
externalizing behaviors among those in each sample who were given attachment interviews
versus those who were not. These analyses revealed that high-risk participants who were given
the interview had higher levels of delinquent behavior (F (1, 245) = 4.90, p = .028) and that,
although the finding was not statistically significant, there was a trend showing normative
participants who were not given the attachment interview to be higher in aggressive behavior (F
(1, 171) = 2.91, p < .10).
Another limitation was the available measure of externalizing behaviors. These
constructs were assessed via items from the Youth Self report. Unfortunately, each study from
which data was drawn employed different shortened from of the YSR; therefore, only those
Attachment and Externalizing Behaviors 33
items that overlapped between the two studies could be used in the present study, leaving only
three markers of attention problems, three markers of delinquent behavior, and five markers of
aggressive behavior. Adding each participant’s scores on the items representing each of the
three externalizing behaviors created a score indicating the participant’s level of each behavior.
The small number of items available allowed for only a small possible range of scores defining
the levels of each externalizing behavior, reducing variability in those scores. Having only three
items to define attention problems and delinquent behavior also precluded the use of factor
analytic techniques to assess the quality of these items as measures of the intended constructs.
Implications, Recommendations, and Questions Raised by this Study
This study is a meaningful step toward research that is truly able to compare high-risk
and normative samples with regard to the relationship between attachment and externalizing
behavior. Future research should be designed with such comparisons in mind, bringing together
samples from each type of population, matched for various demographic confounds, and
assessed with the same measures of attachment and externalizing behavior.
Future research could also improve upon the current study by taking advantage of the
aforementioned benefits of the different measures of attachment and by adding measures of
externalizing behavior, like the CBCL and TRF, that would take into account the perspectives of
multiple reporters. Employing measures of internalizing behavior would also enhance future
research, as these behaviors may be less confounded with high-risk/clinical status and could,
therefore, yield greater variability within each type of sample.
Although the present findings partially support the idea that attachment is related to
externalizing behavior, the inability to draw causal conclusions from the available data raises a
question of causality that is worthy of further exploration. The externalizing behaviors measured
Attachment and Externalizing Behaviors 34
in this study may arise at different points in the life span but could also have precursors in
infancy that make the child more difficult to deal with. Attachment theory contends that
parenting behavior shapes the development of internal working models in infancy that are
expressed as secure or insecure attachment styles (Bowlby, 1969/1982). Insecure attachment has
been identified as a risk factor that interacts with other factors within the family and the child to
increase the likelihood of childhood behavior problems (Greenberg & Speltz, 1988). Further
research is necessary, however to determine whether parenting behavior works through
attachment to cause behavior problems or if early precursors of behavior problems alter
parenting behavior, resulting in different parent-child interactions that are later expressed as
various attachment styles.
The finding that stands out most in this investigation, however, is that secure attachment
is associated with lower levels of externalizing behavior in the normative sample yet not in the
high-risk sample. While normative adolescents’ internal models of attachment relationships are
significantly related to the likelihood that they will engage in externalizing behaviors, these
attachment representations do not predict behavior among high-risk adolescents. This finding
may indicate that, though high-risk adolescents are less likely to be securely attached, attachment
insecurity is not the key to understanding their high levels of externalizing behavior. These
results may point to a more far-reaching dissimilarity between normative and high-risk groups.
As such, work aimed at developing interventions to reduce problem behaviors in adolescence
should avoid generalizing from normative to high-risk samples and should focus treatments on
the unique needs of the targeted group. Though the present research is a first step in discovering
how these groups differ, far more work is needed to determine the various factors that influence
behavior among high-risk adolescents.
Attachment and Externalizing Behaviors 35
References
Achenbach, T.M. (1985). Assessment and taxonomy of child and adolescent psychopathology.
Newbury Park, CA: Sage Publications.
Achenbach, T.M. & Edelbrock, C. (1987). Manual for the Youth Self-Report and Profile.
Burlington, VT: University of Vermont, Department of Psychiatry.
Ainsworth, M.D.S., Blehar, M.C., Waters, E., & Wall, S. (1978). Patterns of attachment: A
psychological study of the strange situation. Hillsdale, NJ: Lawrence Erlbaum.
Ainsworth, M.D.S., & Wittig, B.A. (1969). Attachment and the exploratory behavior of oneyear-olds in a strange situation. In B.M. Foss (Ed.),, Determinants of infant behavior (Vol.
4, pp. 113-136). London:Methuen.
Allen, J.P., Aber, J.L., & Leadbeater, B.J. (1990). Adolescent problem behaviors: The influence
of attachment and autonomy. Psychiatry Clinics of North America, 13(3), 455-467.
Allen, J.P., Marsh, P., McFarland, C., McElhaney, K.B., Land, D., Jodl, K., & Peck, S. (2002).
Attachment and autonomy as predictors of the development of social skills and
delinquency during midadolescence. Journal of Consulting & Clinical Psychology,
70(1), 56-66.
Allen, J. P., Moore, C. M., & Kuperminc, G. P. (1997). Developmental approaches to
understanding adolescent deviance. In S. S. Luthar & J. A. Burack (Eds.), Developmental
psychopathology: Perspectives on adjustment, risk, and disorder (pp. 548–567). New
York: Cambridge University Press.
Allen, J. P., Moore, C., Kuperminc, G., & Bell, K. (1998). Attachment and adolescent
psychosocial functioning. Child Development, 69, 1406–1419.
Arsenio, W., Shea, T., & Sacks, B. (2000). Juvenile offenders’ and comparison adolescents’
Attachment and Externalizing Behaviors 36
conceptions of the emotional consequences of victimization: Relations with attachment
and empathy. Manuscript submitted for publication.
Bartholomew, K. (1990). Avoidance of intimacy: An attachment perspective. Journal of Social
and Personal Relationships, 7, 147-178.
Bartholomew, K., Henderson, A., & Dutton, D. (2001). Insecure attachment and abusive
intimate relationships. In C. Clulow (Ed.), Adult attachment and couple psychotherapy:
The ‘secure base’ in practice and research (pp. 43-61). London: Brunner-Routledge.
Bartholomew, K., & Horowitz, L. (1991). Attachment styles among young adults: A test
of a four-category model. Journal of Personality and Social Psychology, 61, 226-244.
Berkson, J. (1946). Limitations of the application of fourfold table analysis to hospital data.
Biometrics Bulletin, 2, 47-53.
Bookwala, J., & Zdaniuk, B. (1998). Adult attachment styles and aggressive behavior within
dating relationships. Journal of Social and Personal relationships, 15, 175-190.
Bowlby, J. (1944). Forty-four juvenile thieves: Their characters and home life. International
Journal of Psychoanalysis, 25, 19-52.
Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation. New York: Basic Books.
Bowlby, J. (1982). Attachment and loss: Vol. 2. Attachment. New York: Basic Books.
(Original work published 1969)
Campbell, S.B. (1995). Behavior problems in preschool children: A review of recent research.
Journal of Child Psychology and Psychiatry, 36, 113-149.
Carlson, E.A. (1989). Individual differences in quality of representational organization of
attachment of high risk adolescent mothers. Unpublished doctoral dissertation, Columbia
University, New York, New York.
Attachment and Externalizing Behaviors 37
Cichetti, D., Cummings, E.M., Greenberg, M.T., & Marvin, R.S. (1990). An organizational
perspective on attachment beyond infancy: Implications for theory, measurement, and
research. In M. Greenberg, D. Cicchetti, & E. Cummings (Eds.), Attachment in the
preschool years: Theory, research and intervention (pp. 3-49). Chicago: University of
Chicago Press.
Clarke, L., Ungerer, J., Chahoud, K., Johnson, S., & Stiefel, I. (2002). Attention deficit
hyperactivity disorder is associated with attachment insecurity. Clinical Child
Psychology & Psychiatry, 7(2), 179-198.
Fagot, B.I., & Kavanagh, K. (1990). The prediction of antisocial behavior from avoidant
attachment classifications. Child Development, 61, 864-873.
Finzi, R., Ram, A., Har-Even, D., Shnit, D., & Weizman, A. (2001). Attachment styles and
aggression in physically abused and neglected children. Journal of Youth and
Adolescence, 30(6), 769-786.
Garmezy, N. (1983). Stressors of childhood. In N. Garmezy & M. Rutter (Eds.), Stress, coping
and development in children (pp. 43-84). New York: McGraw-Hill.
Gjone, H., & Stevenson, J. (1997). A longitudinal twin study of temperament and behavior
problems: Common genetic or environmental influences? Journal of the American
Academy of Child & Adolescent Psychiatry, 36(10), 1448-1456.
Gould, M.S., Bird, H., Jaramillo, B.S. (1993). Correspondence between statistically derived
behavior problem syndromes and child psychiatric diagnoses in a community sample.
Journal of Abnormal Child Psychology, 21(3), 287-313.
Greenberg, M.T., & Speltz, M.L. (1988). Contributions of attachment theory to the
understanding of conduct problems during the preschool years. In J. Belsky & T.
Attachment and Externalizing Behaviors 38
Nezworski (Eds.), Clinical implications of attachment (pp. 177-218). Hillsdale, NJ:
Erlbaum.
Hinshaw, S. P. (1992). Academic underachievement, attention deficits, and aggression:
Comorbidity and implications for intervention. Journal of Consulting & Clinical
Psychology, 60(6), 893-903.
Hirschi, T. (1969). The causes of delinquency. Berkeley, CA: University of California Press.
Kochanska, G., Tjebkes, T.L., & Forman, D.R. (1998). Children’s emerging regulation of
conduct: Restraint, compliance, and internalization from infancy to the second year.
Child Development, 69(5), 1378-1389.
Ladnier, R.D., & Massanari, A.E. (2000). Treating ADHD as attachment deficit disorder. In
T.M. Levy (Ed.), Handbook of attachment interventions (pp. 27-65). San Diego:
Academic Press, Inc.
Lewis, M., Feiring, C., McGuffog, C., & Jaskir, J. (1984). Predicting psychopathology in sixyear-olds from early social relations. Child Development, 55, 123-136.
Loeber, R., & Dishion, T. (1983). Early predictors of male delinquency: A review.
Psychological Bulletin, 93, 68-99.
Main, M. (1991) Metacognitive knowledge, metacognitive monitoring, and singular (coherent)
vs. multiple (incoherent) model of attachment: Findings and directions for future
research. In C.M. Parkes, J. Stevenson-Hinde, et al. (Eds.), Attachment across the life
cycle. (pp. 127-159).
Main, M., & Goldwyn, R. (1984). Predicting rejection of her infant from mother’s
representation of her own experience: Implications for the abused-abusing
intergenerational cycle. Child Abuse and Neglect, 8, 203-217.
Attachment and Externalizing Behaviors 39
Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, childhood, and adulthood: A
move to the level of representation. In I. Bretherton & E. Waters (Eds.), Growing points
in attachment theory and research. Monographs of the Society for Research in Child
Development, 50, 66-104.
Maio, G.R., Fincham, F.D., & Lycett, E.J. Attitudinal ambivalence toward parents and
attachment style. Personality & Social Psychology Bulletin, 26(12), 1451-1464.
McArdle, J. J. (1996). Current directions in structural factor analysis. Current Directions in
Psychological Science, 5, 11-18.
McArdle, J. J. & Nesselroade, J.R. (1994). Structuring data to study development and change.
In, S.H. Cohen & H.W. Reese (Eds.), Life-Span developmental psychology:
methodological innovation (pp. 223-267). Hillsdale, N.J.: Erlbaum.
Moffitt, T.E. (1993). Adolescence-limited and life-course-persistent antisocial behavior: A
developmental taxonomy. Psychological Review, 100(4), 674-701.
Moretti, M., & Odgers, C. (2002). Aggressive and violent girls: Prevalence, profiles, and
contributing factors. In R.R. Corrado, R. Roesch, S.D. Hart, & J.K. Gierowski (Eds.),,
Multi-problem violent youth: A foundation for comparative research on needs,
interventions, and outcomes (pp. 116-129). Washington, DC: ISO Press.
Moss, E., Rousseau, D., Parent, S., St-Laurent, D., & Saintonge, J. (1998). Correlates of
attachment at school age: Maternal reported stress, mother-child interaction, and behavior
problems. Child Development, 69(5), 1390-1405.
Ramos-Marcuse, F., & Arsenio, W.F. (2001). Young children's emotionally-charged moral
narratives: Relations with attachment and behavior problems. Early Education &
Development, 12(2), 165-184.
Attachment and Externalizing Behaviors 40
Renekn, B., Egeland, B., Marvinney, D., Mangelsdorf, S., & Sroufe, L.A. (1989). Early
childhood antecedents of aggression and passive-withdrwal in early elementary school.
Journal of Personality, 57, 257-281.
Rice, K.G. (1990). Attachment in adolescence: A narrative and meta-analytic review. Journal
of Youth and Adolescnce, 19(5), 511-538.
Richman, E.A. (1985). Disorders of preschool children. In M. Ruttner & L. Hersov (Eds.),
Child and adolescent psychiatry: Modern approaches. Boston: Blackwell Scientific
Publications.
Robins, L.N. (1991). Conduct disorder. Journal of Child Psychology and Psychiatry and Allied
Disciplines, 32, 193-212.
Scharfe, E. (2002). Reliability and validity of an interview assessment of attachment
representations in a clinical sample of adolescents. Journal of Adolescent Research, 17
(5), 532-551.
Simons, K.J., Paternite, C.E., & Shore, C. (2001). Quality of parent/adolescent attachment and
aggression in young adolescents. Journal of Early adolescence, 21(2), 182-203.
Smith, C. A. (1994). Dis-attachment. Australian & New Zealand Journal of Psychiatry, 28(4),
691-693.
Speltz, M.L., Greenberg, M.T., & Deklyen, M. (1990). Attachment in preschoolers with
disruptive behavior: A comparison of clinic-referred and nonproblem children.”
Development & Psychopathology, 2(1), 31-46.
Sroufe, L.A. (1983). Infant-caregiver attachment and patterns of adaptation in preschool: The
roots of maladaptation and competence. In M. Perlmutter (Ed.), Minnesota symposia on
child psychology (Vol. 16, pp. 41-81). Hillsdale, NJ: Erlbaum.
Attachment and Externalizing Behaviors 41
Teti, D.M., & Ablard, K.E. (1989). Security of attachment and infant-sibling relationships: A
laboratory study. Child Development, 60(6), 1519-1528.
Toth, S.L., & Cicchetti, D. (1996). Patterns of relatedness, depressive symptomology, and
perceived competence in maltreated children. Journal of Consulting and Clinical
Psychology, 64, 32-41.
Van Ijzendoorn, M.H. (1995). Adult attachment representations, parental responsiveness, and
infant attachment: A meta-analysis on the predictive validity of the Adult Attachment
Interview. Psychological Bulletin, 117, 387-403.
Attachment and Externalizing Behaviors 42
Tables & Figures
Table 1. Prevalence of Attachment Styles by Sample and Gender
%
High-risk Normative
(N = 170) (N = 143)
χ2
df
p
Secure
6.5
58.0
98.31 1
.000
Preoccupied
24.7
7.7
15.98 1
.000
Dismissing
59.4
32.9
21.96 1
.000
Even split
9.4
1.4
9.20
1
.002
Male
(N = 175)
Female
(N = 138)
Secure
25.7
35.5
3.52
1
.061
Preoccupied
9.7
26.1
14.71 1
.000
Dismissing
57.1
34.8
15.48 1
.000
Even split
7.4
3.6
2.06
.151
1
Attachment and Externalizing Behaviors 43
Table 2. Prevalence of Attachment Styles by Gender within Samples
%
High-Risk
Male
(N = 105)
Female
(N = 65)
Secure
7.6
4.6
.44 1
.333
Preoccupied
15.2
40.0
.00 1
.000
Dismissing
66.7
47.7
.01 1
.011
Even split
10.5
7.7
.55 1
.376
.22 1
.144
Secure
Normative
Male
Female
(N = 70)
(N = 73)
52.9
63.0
χ2
df
p
Preoccupied
1.4
13.7
.01 1
.005
Dismissing
42.9
23.3
.01 1
.010
Even split
2.9
0
.15 1
.238
Table 3. Levels of Externalizing Behavior by Sample and Gender
M
High-risk
Normative
F
p
Attention Problems t
3.2
1.6
110.02
.000
Aggressive Behavior t t
3.9
1.0
225.44
.000
Delinquent Behavior t
2.3
0.7
171.51
.000
Male
Female
Attention Problems t
2.5
2.6
0.59
.442
Aggressive Behavior t t
2.9
2.4
3.81
.052
Delinquent Behavior t
1.7
1.5
3.28
.071
t Scores
are on a 0-6 scale
t t Scores
are on a 0-10 scale
Attachment and Externalizing Behaviors 44
Table 4. Levels of Externalizing Behavior by Gender within Samples
M
Normative males
Normative females
F
p
Attention Problems t
1.5
1.7
1.21
.273
Aggressive Behavior t t
1.2
0.9
1.84
.176
Delinquent Behavior t
0.7
0.6
0.82
.366
High-risk males
High-risk females
Attention Problems t
3.0
3.4
3.90
.049
Aggressive Behavior t t
3.9
3.9
0.00
.947
Delinquent Behavior t
2.3
2.3
0.00
.997
t Scores
are on a 0-6 scale
t t Scores
are on a 0-10 scale
Attachment and Externalizing Behaviors 45
Table 5. Levels of Externalizing Behavior by Attachment Style
M
Secure
Non-secure
F
p
Attention Problems t
1.7
2.7
26.54
.000
Aggressive Behavior t t
1.3
3.1
41.80
.000
Delinquent Behavior t
0.8
1.8
33.58
.000
Preoccupied
Non-preoccupied
Attention Problems t
3.1
2.4
12.37
.001
Aggressive Behavior t t
3.8
2.5
19.97
.000
Delinquent Behavior t
2.3
1.5
11.01
.001
Dismissing
Non-dismissing
Attention Problems t
2.5
2.5
1.27
.260
Aggressive Behavior t t
2.7
2.7
1.75
.187
Delinquent Behavior t
1.8
1.5
4.95
.027
t Scores
are on a 0-6 scale
t t Scores
are on a 0-10 scale
Attachment and Externalizing Behaviors 46
Table 6. Levels of Externalizing Behavior by Attachment Style in Males and Females
Males
Females
M
M
Secure
Non-secure
F
p
Secure
Non-secure
F
p
Attention
Problems t
1.8
2.5
6.03
.015
1.6
3.1
28.65
.000
Aggressive
Behavior t t
1.6
2.9
11.35
.001
1.0
3.2
34.32
.000
Delinquent
Behavior t
1.1
1.9
10.02
.002
0.6
2.0
23.42
.000
Preoccupied
Non-preoccupied
Preoccupied
Non-preoccupied
Attention
Problems t
3.1
2.2
4.25
.041
3.2
2.4
6.25
.014
Aggressive
Behavior t t
4.2
2.4
9.28
.003
3.6
2.0
12.96
.000
Delinquent
Behavior t
2.3
1.7
2.80
.096
2.3
1.2
10.61
.001
Dismissing
Non-dismissing
Dismissing
Non-dismissing
Attention
Problems t
2.3
2.3
0.01
.924
3.0
2.4
4.89
.029
Aggressive
Behavior t t
2.6
2.5
0.19
.664
2.8
2.2
1.79
.183
Delinquent
Behavior t
1.9
1.5
1.96
.163
1.8
1.4
2.10
.150
t Scores
are on a 0-6 scale
t t Scores
are on a 0-10 scale
Attachment and Externalizing Behaviors 47
Table 7. Levels of Externalizing Behavior by Attachment Style within Sub-Samples Defined by
Gender and Sample
M
High-risk
Normative
Male
Female
Male
Female
Secure
Non-secure
Secure
Non-secure
Secure
Non-secure
Secure
Non-secure
Attention
Problems t
3.0
2.9
3.0
3.5
1.5
1.3
1.6
2.2
Aggressive
Behavior t t
4.1
3.6
4.0
4.2
1.1
0.9
0.8
1.1
Delinquent
Behavior t
2.3
2.4
1.7
2.6
0.9
0.6
0.6
0.7
Preoccupied
Preoccupied
Preoccupied
0.0
Nonpreoccupied
1.5
Preoccupied
3.5
Nonpreoccupied
3.5
2.5
Nonpreoccupied
1.7
Attention
Problems t
3.3
Nonpreoccupied
2.8
Aggressive
Behavior t t
4.3
3.6
4.5
4.0
2.0
1.0
1.6
0.8
Delinquent
Behavior t
2.5
2.4
2.8
2.4
0.0
0.7
0.9
0.6
Dismissing
Dismissing
Dismissing
1.4
Nondismissing
1.4
Dismissing
3.6
Nondismissing
3.4
1.9
Nondismissing
1.7
Attention
Problems t
2.7
Nondismissing
3.2
Aggressive
Behavior t t
3.4
4.2
3.9
4.4
1.0
1.0
0.8
0.9
Delinquent
Behavior t
2.4
2.3
2.5
2.6
0.6
0.8
0.5
0.6
t Scores
are on a 0-6 scale
t t Scores
are on a 0-10 scale
Attachment and Externalizing Behaviors 48
Figure 1. Factor Model (Normative Sample, n = 186)
Figure 2. Factor Model (High Risk Sample, n =271)
Attachment and Externalizing Behaviors 49
Figure 3. SEM Model (Normative Sample)
Gender
Figure 4. SEM Model (High Risk Sample)
Gender
Attachment and Externalizing Behaviors 50
Download