ATTACHMENT THEORY offers a means for understanding how parent-child relationships affect a child's early psychological organization and subsequent development.1 Annually in the United States, between 1.6 and 2.8 million youth aged 12 to 18 years run away or are thrown out of their family home.2-5 While some youth are forced out of their home following arguments with parents about substance use or gender identity issues or both,2,6 the majority leave home to escape family conflict, maltreatment, and neglect.4,5,7
Attachment organization has been identified as an important link between maltreatment in early childhood and psychosocial problems in adolescence.8 For example, early childhood maltreatment and later substance abuse have been highly associated in studies of housed adolescents.9,10 This association suggests that early parent-child attachment organization (how thoughts and feelings about attachment are effectively organized) may also be linked to behavioral risks such as substance abuse in homeless youth. Research shows that rates of substance use are extremely high among homeless and street youth.11-13 In fact, female youth who are still street dependent at the age of 21 years report substance abuse as a primary reason they remain homeless.14
The Adult Attachment Interview (AAI) is considered the most useful instrument for assessing differences in attachment organization in healthy adolescent and adult populations15 (C. George et al, unpublished data, 1985/1996). However, although the AAI has been used to study clinically distressed adolescents,16-18 little empirical data exist about the use of this instrument in homeless youth populations. Consequently, this study had 2 aims: to explore the utility of the AAI to assess state of mind regarding attachment (attachment patterns) in homeless adolescents, and to explore patterns of relationships among attachment patterns, social connectedness, and substance use in this population.
BACKGROUND
The development and organization of attachment bonds in early childhood are principal determinants of mental health.19,20 On the basis of repeated interactions with the caregiver over the first year of life, infants form expectations about the availability and responsiveness of their caregiver. Ainsworth and colleagues developed a laboratory procedure to assess the quality of the infant-caregiver relationship.21 Infants who sought contact and comfort from their caregiver when distressed were considered secure, infants who avoided contact were classified as insecure-avoidant, and those who appeared ambivalent, both seeking and resisting comfort, were classified as resistant.21
The quality of infants' early attachment to their caregiver provides adolescents and adults with a set of expectations for how to relate to others and how to interpret others' actions.22,23 On the basis of early experiences with the caregiver, children form cognitive schemata or internal working models. These internal working models reflect individuals' understanding of their self, relationships, and the outside world.19,23,24
A secure infant-caregiver attachment relationship is carried forward into adolescence as a secure-autonomous state of mind regarding attachment, whereas an insecure attachment is carried forward as either a dismissing or a preoccupied state of mind regarding attachment.22 Specifically, 72% of the 50 children followed from infancy to adulthood received the same secure-insecure attachment classification at each assessment time.21 In the case of extreme trauma, such as child maltreatment, trauma, or loss, an unresolved-disorganized state of mind regarding attachment may be carried forward into adolescence.25
Research indicates that insecure states of mind regarding attachment, and particularly unresolved-disorganized states of mind, are risk factors for mental disorders, but not necessarily indicative of pathology.24,26 Although attachment is relatively stable from childhood through adolescence, it is likely to change if life circumstances improve, if effective support from others is available, or if stress is decreased.17,23,24 Thus, information about attachment organization may be useful for tailoring psychosocial interventions to specific states of mind in homeless youth.
One limitation of previous attachment research in adolescents has been the use of primarily self-report and proxy measures of attachment (eg, social connectedness, social support). Of 2 homeless youth studies found in a review of the attachment literature,27,28 both used only self-report measures, that is, the Attachment Styles Questionnaire or the Attachment Questionnaire. The Attachment Styles Questionnaire29 measures conscious opinions about the self and the environment, whereas the Attachment Questionnaire (R. Kesslar & K. E. Potthharst, unpublished data, 1983) assesses basic areas of family life and family interactions. De Haas and colleagues suggest that self-report measures yield little or no information about the state-of-mind dimension of attachment.30 Indeed, they suggest that this kind of information "may not become accessible unless a clinically-oriented interview such as the AAI is conducted."30(p482)
In the AAI, an individual's early attachment relationships are explored through a series of questions and probes that elicit an account of past experiences and the effect these have had on development and present functioning. On the basis of an effective and social-cognitive perspective, the AAI is designed to assess the present state of mind with respect to early attachment-related experiences.31 Thus, while reliable self-report measures of different aspects of attachment have been developed, the AAI remains the most reliable overall measurement of attachment organization.
METHODS
This pilot study had a descriptive, exploratory design. The target population was homeless youth who sought health and social services from a nonprofit street outreach program in an urban southwestern community. Survey data were collected from participants using the Social Connectedness scale32; the Childhood Trauma Questionnaire33; and the Simple Screening Instrument for alcohol or other drug abuse34; and qualitative data were collected using the AAI tool (C. George et al, unpublished data, 1985/1996). Responses to the self-report measures and questionnaires were entered into an SPSS data file and descriptive statistics were obtained. Transcripts of the AAIs were scored and coded by 2 independent raters blind to the others' assessments using the adult attachment rating and classification (AAI-RC) system, developed by M. Main et al (unpublished data, 1985/2002). On the basis of previous adult studies using the AAI 4-way classification system,31,35 interrater agreement of 80% or greater ([kappa] = 0.68 or above) was considered acceptable.
SAMPLE CHARACTERISTICS
Homeless adolescents aged between 16 and 23 years, who sought health or social services or both from the street outreach program, and who had been living independent of adult supervision for at least 2 months, were invited to participate in the study. Twenty-seven older (mean age 19.8 years) homeless adolescents (13 males and 14 females) enrolled in the study; 52% of the participants self-identified as street or runaway youth. Two participants (7%) self-identified as Hispanic or of Latin ethnicity; 16 (59%) as White; 2 (7%) as Black; and 7 (26%) as more than one race. Eighteen (69%) participants self-identified as heterosexual, while 8 (31%) self-identified as nonheterosexual. Of the nonheterosexual participants, 2 males and 4 females self-identified as bisexual, while 1 male and 1 female self-identified as questioning their sexual orientation.
STUDY PROCEDURES
Institutional review board approval was obtained including waiver of parental consent and waiver of documentation of consent. The institutional review board waived the need for parental consent, as obtaining consent might have compromised the personal safety of some participants, particularly of those who had left home because of parental neglect or abuse. Participants were recruited by direct invitation of the street outreach program staff during street outreach or when youth registered for services at the agency. Caseworkers referred eligible youth to the researcher who discussed the research procedures and obtained informed consent prior to data collection. To reduce attrition, and because the participants were homeless and highly transient, all survey instruments and the AAI were conducted at 1 session. Willingness to complete the survey questionnaires was taken as consent of the youth to participate in the study. Participants received food coupons worth $10.00 and a grocery store card worth $25.00 upon completion of the survey instruments and the 1-hour AAI.
Study participants were interviewed using the AAI instrument, a qualitative, semistructured interview tool consisting of 18 standardized questions, with standardized follow-up questions or "probes" designed to assess attachment pattern. Participants were asked to list 5 words describing their early childhood experiences with each parent, and then to describe specific episodes that reflected those words. Other questions focused on specific instances of rejection, separation, being upset or ill, trauma, and loss. Finally, participants were asked to provide instances of changes in relationships with parents and the current state of those relationships.
Interviews were audiotaped and transcribed verbatim following the unpublished protocol (C. George et al, unpublished data, 1985/1996), and the interview transcripts were coded and scored by 2 trained and reliable coders using the AAI-RC system (M. Main et al, unpublished data, 1985/2002). Both coders had previously completed a 2-week AAI coding workshop, and passed the 30-case reliability test provided by Main et al for AAI coders-in-training (unpublished data, 1985/2002).
On the basis of discourse analysis rather than content analysis, the AAI-RC system focuses on the process and overall coherence or incoherence of the interview narrative regarding attachment experiences.17 The AAI-RC system consists of 14 continuous 9-point scales: 9 scales rate overall states of mind, including inability to recall childhood, tendency to derogate attachment, passivity of thought processes, metacognitive processes, fear of loss, unresolved loss, unresolved trauma, coherency of mind, and coherency of the transcript; and 5 scales rate parents for experiences of loving, rejection, role reversal, pressure to achieve, and neglect; for states of mind respecting parents over idealizing, current anger, and derogation of parents. Validity and interrater reliability of the AAI have been well established.13,17,23
On the basis of the AAI-RC system, AAI transcripts were either classified as secure-autonomous (F) or placed into 1 of 3 insecure groups: insecure-dismissing (D), insecure-preoccupied (E), or insecure-cannot classify. The secure-autonomous state of mind is characterized by an ability to describe childhood relationship experiences in a clear, believable, and truthful manner regardless of whether these experiences were positive or negative. Adolescents were judged dismissing when they minimized negative relationship experiences with their parents, for example, idealizing an abusive parent, failing to remember early experiences, or derogating attachment figures and attachment feelings.
Adolescents were placed into a "cannot classify" group when they could not settle on a consistent strategy (secure, dismissing, or preoccupied) for describing their relationship with their parents during childhood, showing a mix of secure-autonomous, dismissing or preoccupied or both strategies. The category "cannot classify" represents a broader, more severe lack of attachment organization, and is rare in healthy populations.
Unresolved attachment ratings were made on the basis of instances of being unresolved with respect to any of 3 types of past traumatic experiences (sexual abuse, physical abuse, or loss) either alone or combined. This category is assigned on the basis of a brief lapse in the monitoring of reasoning or discourse while discussing a traumatic experience. Thus, a person can be dismissing for most of the interviewing but show a lapse while describing having been sexually abused, for example, falling silent for 60 seconds in midsentence, and then changing the topic as if the silence never occurred. Since these lapses are brief and can only be coded while the person is discussing trauma, all transcripts placed in the unresolved group are also assigned a secondary classification (secure, dismissing, preoccupied, or cannot classify) to reflect the overall state of mind with respect to attachment.
FINDINGS
This was a highly traumatized sample with 74% of participants reporting emotional abuse in childhood; 70% reporting physical abuse; and 55% reporting sexual abuse. Furthermore, 81.5% of participants interviewed reported moderate-to-high-risk alcohol or other drug (AOD) abuse. Although 22% of participants scored high on social connectedness, 78% scored low to average social connectedness. No statistically significant correlations were found for attachment pattern, social connectedness, and alcohol or other drug use/abuse. This may be because of the small sample size.
All AAI transcripts were scored and coded. Eight transcripts chosen at random were scored by both coders, blind to each other's assessments. Interrater agreement on the primary AAI code for this set of transcripts was 87.5%, a mean concordance above the acceptable level set for the study.
The distribution of primary attachment codes in this study is summarized in Table 1. No participant interviews were coded as having a secure (F) primary attachment pattern. Four participant interviews were assigned the insecure (D) attachment pattern; one was assigned the insecure (E) attachment pattern; 15 were assigned the unresolved-disorganized (U) due to trauma or loss code attachment pattern; and 5 were assigned cannot classify (CC) as primary attachment pattern. Regarding the "cannot classify" category, an additional 13 participant interviews were assigned CC as a secondary or tertiary code. In total, 18 of the 25 interviews (78%) were assigned CC as a primary or other (eg, secondary, tertiary) attachment organization strategy.
DISCUSSION
The findings of this study suggest that the Adult Attachment Interview tool has important utility in the assessment of attachment states of mind in homeless youth. Because of the high rates of child maltreatment reported in previous homeless youth samples,4,5,7,36 we anticipated obtaining low rates of secure attachment (F) and high rates of insecure attachment (D and E) in the sample. Thus, we were dismayed at the high rates of unresolved-disorganized pattern and cannot classify attachment patterns assigned. In fact, these rates were higher than those reported in previous samples of maltreated children and hospitalized psychiatric adolescents.18,37
Although the AAI unresolved-disorganized pattern is usually overrepresented in clinically distressed samples,17 low to moderate rates of securely attached youth have been reported in previous adolescent studies.18,38,39 Allen and Hauser, for example, reported rates of 50% for healthy high school-aged youth versus 6% for hospitalized psychiatric youth in their attachment research.38 It is noteworthy, then, that not one of the participants satisfied requirements for a secure attachment pattern. This clearly supports the need for much more research in the area of attachment and homeless youth.
Attachment theory clearly brings coherence to clinical descriptions of parenting in families of homeless youth, particularly, lack of parental responsiveness, availability and accessibility, all of which are essential for the development of secure attachment in relationships. Early experiences of rejection or loss often have lasting effects on later expectations, feelings, and relationships. Thus, underlying the youths' internal working models of parents as unresponsive and unavailable, are corresponding models of themselves as unlovable and incompetent.
Attachment theorists suggest that unresolved loss and trauma tends to affect the way social behavior is organized later in adulthood,40 including how individuals cope with extreme stress. Recent research has shown that adults placed in the cannot classify group, that is, those who do not have an organized way to cope with, understand, or reflect upon early experiences, are more likely to be diagnosed with a borderline personality disorder and engaged in violent crimes.40. The findings of this study suggest that homeless youth are more likely to use maladaptive coping behaviors, such as alcohol and other drugs when under stress.
RESEARCH IMPLICATIONS
The profound rates of childhood abuse (ie, nearly 3 quarters of the study population with a history of emotional abuse, almost as many with physical abuse, and over half with sexual abuse), coupled with the incredibly high substance abuse rates, make clear the need for more research examining the mediating and moderating factors between homelessness, abuse, and substance abuse in youth. Given the high rates of childhood trauma reported by the sample participants, the similar high substance abuse rate in this sample is not surprising. Alcohol use affects consciousness, and as a form of defensive exclusion, it may be used by homeless youth to deactivate negative thoughts and feelings about past experiences.24,41 On a positive note, the substance abuse field recognizes social connectedness as a protective factor, worthy of targeting and enhancing in interventions.42 For this reason, studies of resilient youth, such as the 22% who scored high on social connectedness in this study, are highly recommended.
PRACTICE IMPLICATIONS
The Adult Attachment Interview has clinical utility. Not only can it be used as a screening tool to identify adolescents at risk for more serious psychopathology but also can be used in treatment. Experiencing the AAI itself has been shown to help people reflect on their own actions, providing them with insight, and helping them to take responsibility for their actions.43-45 It may be particularly difficult for homeless adolescents to seek help; therefore, it is important to note that such adolescents can benefit from participating in the interview.
Moreover, homeless adolescents tend to feel alienated from mainstream society. The AAI could strengthen the bond between the practitioner administering the AAI and the adolescent. Jones45 describes how the AAI evoked emotion and vulnerability in her clients who otherwise seem very tough and uncaring. Practitioners and therapists using this instrument may be able to understand their patients differently, feel more empathic, and thus, more motivated to help them. Finally, the AAI can help the therapist identify adults, other than an abusive parent, who were alternative attachment figures that provided more adequate care for the patient. Successful treatment plans could involve helping adolescents draw on a model or internal representation of a more caring adult and thereby allow them to enter into more trusting and satisfying relationships.46
POLICY IMPLICATIONS
At present, society tends to deal with homeless youth as pariahs, seeing them more as "broken" than unique and resilient. Youth who wind up on the streets are, at times, the strength of their families, as demonstrated by their visions of a better life, conviction that they do not deserve maltreatment and abuse, and ability to pursue and achieve a life with resources, connections, and dreams. Many youth use substances to cope with the pain, oppression, challenges, and injustices they have encountered in their lives and worlds. While they continue to be "invisible" in public policies, they will not go away. Without programs and policies that embrace such youth and engage them in programs that allow them to utilize their strengths, these youth will populate other social systems which are painfully lacking in resources such as juvenile justice system, mental health system, and substance abuse treatment systems. This study suggests that a proportion of the youth have strong capacity for social connectedness and attachment. The larger proportion that does not have these abilities should be reached out to and embraced to form protective attachments supported by risk and resilience research.
REFERENCES