Resilience in Homeless Youth: The Key Role of Self-Esteem
Sean Kidd, PhD, CPRP
Golan Shahar, PhD
Ben-Gurion University and Yale University
This study examined the protective role of self-esteem, social involvement, and secure attachment among
homeless youths. These protective factors were examined as they ameliorate risks among 208 homeless
youths surveyed in New York City and Toronto. Both mental and physical health indicators were
employed in this study, including loneliness, feeling trapped, suicidal ideation, subjective health status,
and substance use. Self-esteem emerged as a key protective factor, predicting levels of loneliness, feeling
trapped, and suicide ideation, and buffering against the deleterious effect of fearful attachment on
loneliness. Findings highlight the role of the self-concept in risk and resilience among homeless youth.
Keywords: homeless youth, resilience, attachment, risk
The purpose of this study was to examine the interactions
between risk and resilience factors among homeless youths. The
size of the North American homeless youth population has steadily
and sharply increased in the past 20 years, with recent estimates
indicating a population size of over 1 million (Kidd & Scrimenti,
2004). Research in the area of youth homelessness has likewise
grown incrementally, focusing on the impacts of abusive and
chaotic family/home backgrounds as they relate to poor mental
health, substance use, and suicidality on the streets (Kidd, 2006).
Poor physical health (Ensign, 1998), victimization (Whitbeck,
Hoyt, & Yoder, 1999), and criminality (Baron, 2003) have also
received substantial attention. Within this body of literature, a
portrait of an extremely high-risk population has emerged, with
mortality rates up to 40 times that of the general youth population
(Shaw & Dorling, 1998), and suicide and drug overdose found as
leading causes of death (Roy et al., 2004).
While the majority of the homeless youth literature has focused
primarily upon risk, increasing attention has been given to coping
and resilience (Kidd & Carroll, 2007; Rew & Horner, 2003a).
Resilience can be understood as an ability to mobilize personal and
social resources to protect against risks (Rew & Horner, 2003b).
While in the homeless youth literature there have been numerous
examinations of risks and resilience independently, they have
seldom been considered in an integrated manner. The present study
examined interactions between risk and resilience domains to
identify key protective factors that might be targeted in interven-
tions for homeless youths.
Social Risk and Resilience
Several prestreet experiences have consistently emerged as be-
ing related to risk behaviors and mental health in the street context.
Greater frequency and intensity of abuse and neglect by caregivers
have been shown to be related to poor mental health and victim-
ization on the street (Kidd, 2006; Whitbeck, Hoyt, & Bao, 2000;
Yoder, 1999). The circumstances surrounding youths leaving
home have likewise been found to be an important consideration
with respect to risk. Throwaways, referring to youths who report
being forced to leave the home by a caregiver, have higher levels
of mental health difficulties than youths who report leaving home
of their own volition (Ringwalt, Greene, & Robertson, 1998).
Another important consideration for many youths is sexual orien-
tation. The narratives of gay, lesbian, and transgender youths are
replete with descriptions of abuse and being thrown out of the
home related to their coming out to their parents (Kidd & Kral,
2002). Such youths have been found to engage in more risky
survival strategies, are victimized more frequently, and have
poorer mental health relative to heterosexual youth (Whitbeck
et al., 2004). Finally, the present study considered school experi-
ences as having implications for later physical and mental health in
the street context. While a history of being bullied at school has yet
to be examined among homeless youths, it has been shown to be
related to poor mental health and suicide among the general
adolescent population (Mills, Guerin, Daly, Lynch, & Fitzpatrick,
An important factor that includes elements of both risk and
resilience is social involvement on the street. Varying as a function
of peer group composition, negative impacts of social involvement
have been found to include increased risk behavior and depression
with positive impacts including perceived emotional support, pro-
tection, and decreased risk behavior (Bao, Whitbeck, & Hoyt,
2000; Ennett, Bailey, & Federman, 1999; Hagan & McCarthy,
1997; Unger et al., 1998). Generally, affiliation with peers engaged
in drug use and other forms of criminality, thought to be charac-
terized by exploitative and coercive interactions, has a negative
impact (Bao et al., 2000; Rice, Milburn, Rotheram-Borus, Mallett,
& Rosenthal, 2005). In addition, having friends who have at-
tempted suicide has been found to be associated with poor mental
Sean Kidd, PhD, CPRP, Department of Psychiatry and Behavioural
Neurosciences, McMaster University; Golan Shahar, PhD, Department of
Psychology, Ben-Gurion University of the Negev, Israel, and Department
of Psychiatry, Yale University School of Medicine.
This project was funded by a grant from the Canadian Social Sciences
and Humanities Research Council.
For reprints and correspondence: Sean Kidd, PhD, CPRP, Centre for
Mountain Health Services–Mental Health Rehabilitation, 100 West 5th
Street Hamilton, ON L8N 3K7. E-mail: firstname.lastname@example.org
American Journal of Orthopsychiatry Copyright 2008 by the American Psychological Association
2008, Vol. 78, No. 2, 163–172 0002-9432/08/$12.00 DOI: 10.1037/0002-9418.104.22.168
health including suicidality (Yoder, Hoyt, & Whitbeck, 1998).
Although the mechanism of this latter finding has not been exam-
ined among homeless youths, it may be suggestive of the outcome
of having supports with significant psychological problems and
internalization of the idea of suicide as a means of ending distress
(Rew & Horner, 2003b). Finally, sex trade involvement has been
repeatedly associated with high-risk contexts and with poor out-
comes including suicidality (Greene, Ennett, & Ringwalt, 1999)
and victimization (Kidd & Kral, 2002). Accordingly, social in-
volvement in and of itself was anticipated to be an equivocal factor
in the present study, with the exception of peer suicidality and sex
trade involvement, which were expected to be linked to greater risk.
Personal Risk and Resilience
Attachment theory has been posited as a useful tool for under-
standing social relations and health among this group (Paradise &
Cauce, 2002). An extensive body of the general adolescent liter-
ature has highlighted the relationship between attachment style and
both physical and mental health. This includes the impact of
attachment style upon adolescents’ ability to make use of the
support of a social network (Larose & Bernier, 2001), ability to
engage in adaptive coping, and establishment and maintenance of
good self-esteem and self-efficacy (Howard & Medway, 2004;
Liable, Carlo, & Raffaelli, 2000; Liable, Carlo, & Roesch, 2004).
A lack of secure attachment, particularly involving negative self-
perception, has been linked to youth suicidality (de Jong, 1992).
Few studies have explored attachment styles among homeless
youths. Homeless youths have been found to have lower rates of
secure attachment than housed youths (Tavecchio, Thomeer, &
Meeus, 1999). Another study, employing an attachment theoretical
framework, found that social support and social connectedness had
a differential relationship with intention and self-efficacy in con-
dom use (only social support increased intention and efficacy)
(Taylor-Seehafer, Johnson, Rew, & Fouladi et al., 2007). Further-
more, these authors found that length of time on the street was
related to increased connectedness but not support. This is sug-
gestive, as posited by these authors, of the importance of peer
social networks in the development of secure models of social
attachment (Taylor-Seehafer et al., 2007). In the present study,
secure models of attachment were anticipated to be a source of
resilience and, for youths with stronger secure attachment patterns,
enhance their ability to gain support through their social involve-
ment. Conversely, as commented upon in the general adolescent
literature (de Jong, 1992), insecure attachment was expected to be
a risk with a negative impact on the potential benefits of social
Previous work has also highlighted the centrality of models of
self in attachment style (Bartholomew & Horowitz, 1991). The
formation of secure attachment with family members in early
development and, later, with peers facilitates adolescents’ devel-
oping and maintaining a model of the self as worthy and deserving
of love and friendship (Laible et al., 2004). For homeless youths,
many of whom have experienced abusive upbringings and/or dis-
ruptions in primary caregiver and who typically face multiple
forms of social stigma, self-esteem is often low and related to poor
mental health and risk behavior (Kidd, 2006). While low self-
esteem has been previously highlighted as a risk, little is known to
date how this variable interacts with social involvement and per-
sonality factors such as attachment among homeless youths. It was
anticipated in the present study that good self-esteem would serve
as a protective factor.
Mental and Physical Health Outcomes
The mental health variables examined in the present study were
loneliness, feeling trapped, and suicidal ideation. Loneliness was
examined due to its close association with poor social support and
connectedness and its’ relationship with suicidality (Rew, 2002;
Rokach, 2005). Loneliness is common among these youths who
often feel alienated from larger society, are typically estranged
from family, and for whom social support is often unreliable
(Kidd, 2004). The experience of being trapped was incorporated as
it has emerged in both exploratory (Kidd, 2004) and confirmatory
(Kidd, 2006) analyses as being a central component of emotional
distress among homeless youths. The term “trapped” refers to
youths feeling unable to engage effective coping in the face of
numerous personal (e.g., traumatic memories, addictions) contex-
tual (e.g., victimization by peers) and structural (e.g., multiple
Suicidal ideation was included as it has been found to be a
leading cause of death in this population (Roy et al., 2004).
Physical health is often seriously compromised among homeless
youths due to a range of conditions secondary to factors such as
poor diet, poor access to health care, drug abuse, and unsafe sexual
behaviors (Ensign, 1998). Accordingly, in an effort to evaluate
physical and mental health in an integrated manner, subjective
health status was included in this analysis. Finally, substance use
was examined as a problem behavior. With implications for both
physical and mental health, substance use is an important variable
for this group given the finding of overdose as a leading cause of
death (Roy et al., 2004).
The Present Study
In Table 1, we summarize the risks, resilience factors, and
mental and physical health outcomes that were examined in this
study. An integrated analysis of the interactions between risk and
resilience was employed in a sample of 208 youths dwelling in the
streets of New York City and Toronto. Based on the literature
reviewed above, the following hypotheses were stipulated:
Hypothesis 1: Each of the risk factors was expected to be
significantly related to poor mental health (i.e., loneliness,
feeling trapped, suicidal ideation), poor subjective health sta-
tus, and substance use.
Hypothesis 2: Each of the protective factors, namely, social
involvement, secure attachment, and self esteem, were ex-
pected to be related to better mental and physical health and
lower levels of substance use, over and above the risk factors.
Hypothesis 3: Each of the protective factors was expected to
buffer against the effects of each of the risk factors.
Participants were 208 youths (age range ⫽ 14–24 years, M ⫽
20.25, SD ⫽ 2.39, median ⫽ 20) who either had no fixed address
KIDD AND SHAHAR
or were living in a shelter at the time of the survey. Interviews
were conducted in agencies and on the streets of New York City
and Toronto. Street interviews were done in a range of locations
where homeless youths congregated and/or pan handled (e.g.,
sidewalks, public parks). Agency interviews included a youth
agency in New York providing an alternative high school and
health services and two agencies in Toronto that provided drop-in
services. One hundred youths were interviewed in New York City
(39 agency, 61 street) and 108 in Toronto (31 at each agency, 46
street). Participants were reimbursed with 20 dollars in restaurant
coupons, and 97% of those approached agreed to participate.
Participants began by participating in an open-ended interview on
the topic of resilience (findings reported in Kidd & Davidson,
2007) and then proceeded to independently complete the struc-
tured survey that was analyzed quantitatively. The data in this
study were derived entirely from the structured, quantitatively
analyzed material. All procedures used in this study were approved
by an institutional review board. For participants who were mi-
nors, their consent was accepted since they were self-reliant, not in
contact with or under the protection of their legal guardians, and
were regarded as emancipated.
One hundred twenty-two of the participants were male (59%),
84 were female (40%), and 2 were transgender (male to female)
(1%). Fifty-six percent of the participants were White, 12% Black,
12% Hispanic, 5% Native, 14% of mixed ethnicity, and the re-
mainder varied. The average age of the youths’ first experience of
homelessness was 15, with a mean education level of 10.6 years.
Most youths (57%) reported having been homeless for more than
2 years with the remainder reporting having been homeless for less
than 2 years. With respect to circumstances surrounding their
leaving home, 24% indicated that they were thrown out of their
homes (i.e., “throwaways”). Thirty-three percent reported contin-
uous homelessness, and 40% reported having had conventional
housing 25% of the time. The majority of participants resided in
street and/or squat locations (47%), with 26% “couch surfing”
(temporarily residing in the homes of others), and 14% living in
shelters. Most reported multiple income sources with pan handling
(45%), dealing drugs (23%), a job (23%), and sex trade involve-
ment (15%) appearing with the most frequency.
Brief and abbreviated measures were used for most of the
variables evaluated in this study. This was necessary as the study
involved a lengthy interview and a survey evaluating multiple
domains. The employment of full and/or lengthy measures in all
domains was not practical given the context of many interviews
(street environments with many youths coming and going) and the
nature of the population, the majority of whom have little tolerance
for extremely lengthy surveys and with whom more than one
contact with youths was seldom feasible. All scales used were
completed in a self-report format.
Maltreatment. Physical abuse by family members was mea-
sured by seven items derived from the Conflict Tactics Scale
(CTS; Straus & Gelles, 1990), with three items derived from
Whitbeck and Simons (1990) scale assessing contact and noncon-
tact sexual abuse. This set of items have been frequently used
among this population and have demonstrated good reliability (0.76 to
0.93) (Whitbeck et al., 1999; Whitbeck & Simons, 1990), and previ-
ous work has suggested good construct validity of the CTS among
adolescents (Cascardi, Avery-Leaf, O’Leary, & Slep, 1999).
Neglect was measured by an item previously developed for this
population that inquired whether the participant had ever felt
neglected by their parents/guardians specifying physical, emo-
tional, or a combination of physical and emotional neglect (Rus-
sell, 1998). Response categories for items evaluating abuse and
neglect ranged from 1 (never)to4(all of the time), with abuse and
neglect items found to have a reliability of 0.89.
Sexual orientation was measured using the Kinsey Scale (Kin-
sey, Pomeroy, & Martin, 1948). This widely used measure is
regarded as having stronger validity that those that dichotomize
sexual orientation (Sell, 1997). It is a seven-point scale with 1 ⫽
straight,4⫽ bisexual, and 7 ⫽ gay/lesbian. For the purpose of
simplification, we contrasted heterosexual (0) with gay/lesbian/
bisexual (1) using a binary dummy variable.
Bullying. The experience of having been bullied was assessed
through the binary categorical item: “bullied by other students”
(yes vs. no).
Suicidality of friends was assessed using a single item asking
participants to indicate the number of friends who had attempted or
completed suicide (range ⫽ 0 – 12; M ⫽ 3.27). For the sake of
simplification, a binary dummy variable was used contrasting
participants reporting at least one friend attempting/committing
suicide with those not reporting such an incidence.
Sex trade involvement was measured by a single item used
previously with this population (Greene et al., 1999) that inquired
if the youth “had ever had sex or done sexual acts with someone
to get money, food, drugs, a place to stay, or something else [they]
wanted?” with responses ranging from 1 (never)to4(11 or more
times). For the sake of simplification, a binary dummy variable
contrasting sex trade with its absence was used.
Social involvement was measured by an instrument derived from
previous qualitative work with this population (Kidd, 2003), was
Variables Included in the Analysis
Risks Protective factors Health indicators
Abuse and neglect in childhood Social involvement Loneliness
Throwaway status Secure attachment Feeling trapped
Sexual orientation Self-esteem Suicidal ideation
Bullied in school Subjective health status
Friends committing/Attempting suicide Substance use
Sex trade involvement
RESILIENCE IN HOMELESS YOUTH
indicated by the following responses regarding current social in-
volvement: “Keep to myself,” “1–2 close friends,” “3–5 close
friends,” and “More than 5 close friends.” The qualifier “close
friends” was used due to the fact that most youths have many
acquaintances on the streets and far fewer persons who might be
regarded as close friends (Kidd, 2003).
Attachment style was assessed using the Relationship Question-
naire (RQ) (Bartholomew & Horowitz, 1991). This measure con-
sists of 4 short paragraphs describing the 4 prototypical patterns of
attachment (secure, preoccupied, dismissing, fearful) that are rated
by respondents on a seven-point scale ranging from 1 ⫽ not at all
like me to 7 ⫽ very much like me. This measure has demonstrated
good convergent and discriminant validity (Bartholomew &
Shaver, 1998). Secure attachment, as a protective factor, was
indicated by high ratings on the paragraph indicating a sense of
emotional closeness with others, comfort in depending on others
and having others depend on oneself, and a lack of concern and
worry regarding being alone. Preoccupied attachment indicates an
anxious need to gain acceptance and validation of others. Dismiss-
ing attachment suggests avoidance and devaluing of closeness of
others and fearful attachment indicates avoidance of others to
reduce the pain of rejection. Means (and SDs) for responses were
2.85 (1.84) for secure attachment, 3.75 (2.01) for fearful attach-
ment, 2.82 (2.05) for preoccupied attachment, and 3.17 (1.95) for
Self-esteem was measured using five items from the Rosenberg
Self-Esteem Scale (1989) used previously among this population
(Adlaf, Zdanowicz, & Smart, 1996) (e.g., “I am satisfied with
myself”), employing a five-point scale with responses ranging
from strongly agree to strongly disagree (␣⫽.83).
Loneliness was evaluated using four items from the UCLA
Loneliness Scale (e.g., “I feel isolated from others”). These items
were found to have the highest loadings on the general loneliness
factor in a factor analytic study of the measure (Russell, 1996) and
participants responded using a four-point scale with items ranging
from never to always (␣⫽.87).
Trapped was a variable that emerged as being central to home-
less youths’ suicide risk in both qualitative (Kidd, 2004) and
quantitative analyses (Kidd, 2006). As measured here, Trapped
was comprised of factor scores derived from a factor analysis of
three scales—feeling trapped, helplessness, and hopelessness, all
measured on a four-point scale ranging from the response 1 ⫽
never to 4 ⫽ Always. Hopelessness was measured using the
participants’ total scores on three items from Beck’s Hopelessness
Scale (Beck, Weissman, Lester, & Trexler, 1974). These three
items were found to have the highest correlations with the general
hopelessness factor (Aish & Wasserman, 2001) and were found in
the present study to have a reliability of 0.79. Items assessing the
participants’ experience of being and feeling trapped was derived
from participants’ descriptions in the qualitative work (Kidd,
2004). These items were “I feel trapped,” “I feel like I don’t have
any real choices,” and “I feel like I don’t have anywhere else to
turn” (␣⫽.86). Two items (␣⫽.79), assessing the helpless
component of the trapped experience, had high loadings on the
helplessness general factor, and were derived from the brief form
of Lester’s Helplessness Scale (Lester, 2001). Total scores from
the hopelessness, trapped, and helplessness items were found to
load onto a single factor (Factor loadings: Hopelessness ⫽ .85,
Trapped ⫽ .89, Helplessness ⫽ .88; accounted for 76% of vari-
ance), with individual factor scores derived using the regression
method. Thus, the factor score, labeled “Trapped” was used for the
outcome variable in the analyses.
Suicidal ideation of the participants was derived from a total
score from a 4 item scale commonly used in studies of nonhome-
less youth (Lewinsohn, Rohde, & Seeley, 1996): “I thought about
killing myself,” “I thought about death/dying,” “I felt that every-
one would be better off if I was dead,” and “I thought about plans
to kill myself,” answered on a four-point scale (1 ⫽ never and 4 ⫽
all of the time), with an alpha coefficient of 0.87. This scale has
shown good reliability (0.80 – 0.89) in previous research with this
population (Yoder et al., 1998). This scale was logarithmically
transformed to adjust for skewness.
Subjective health status was measured with an item from the
Short Form General Health Survey (Stewart et al., 1989): “Com-
pared with others your own age how would you rate your health
overall?” (1 ⫽ poor through 4 ⫽ excellent)(M ⫽ 2.77). Responses
to this question have been shown to have good convergent validity
when compared with other objective and subjective health mea-
sures (Schultz et al., 1994).
Substance use was assessed by items that assessed how often
(1 ⫽ never;5⫽ every day) individuals used marijuana, cocaine,
tranquilizers, psychedelics and ecstasy, amphetamines, and heroin
or narcotics (␣⫽.69). This scale has been previously employed to
assess drug use among this population (e.g., Baron, 1999).
First, correlations and contingency tables were computed to
examine the associations among the study variables. These anal-
yses allowed us to test Hypothesis 1, according to which each of
the risk and resilience factors were expected be associated in
predictable ways with each of the outcome variables (loneliness,
feeling trapped, suicidal ideation, subjective health status, and
Next, five multiple hierarchical regression analyses were con-
ducted; one for each of the outcome variables. Block 1 in these
hierarchical regression analyses included risk and protective vari-
ables shown in correlational analyses to be associated with the
outcome variables. Accordingly, this block addressed Hypothesis
2–the hypothesis that the three protective factors (social involve-
ment, secure attachment, and self-esteem) would predict good
adaptation over and above the risk factors.
Block 2 included two-way interactions between each of the
protective factors and each of the pertinent risk factors and ad-
dressed Hypothesis 3–the hypothesis that each of the three protec-
tive factors would buffer against the adverse impact of the risk
Because as many as five outcome measures were examined,
thereby increasing the number of statistical tests and inflating Type
I error, we employed the Bonferroni correction method by dividing
the conventional level of statistical significance ( p ⫽ .05) by the
number of outcomes examined (5 outcomes) and considered only
effects reaching a statistical significance of p ⫽ .01. As well,
gender and age effects were explored and were not found to have
KIDD AND SHAHAR
a significant impact on regression analyses.
setting (United States vs. Canada; Agency vs. Street) did not
emerge as a notable consideration in the analysis with only a single
significant association found.
Associations Among the Study Variables
Correlations between the outcome variables (loneliness, feeling
trapped, suicidal ideation, subjective health status, and substance
use) are presented in Table 2. All correlations are of a small-to-
modest magnitude. The only exception was the correlation be-
tween loneliness and feeling trapped, (r ⫽ .65, p ⬍ .001), the
strength of which suggested that these two measures are indicators
of overarching emotional distress.
Next, correlations among the protective factors were computed.
Secure attachment was mildly associated with social involvement
(r ⫽ .33, p ⬍ .001) and with self-esteem (r ⫽ .20, p ⬍ .01). The
correlation between social involvement and self-esteem was small
and nonsignificant (r ⫽ .11, p ⫽ .09).
Then, associations between the risk factors were examined.
Focusing first on the continuous risk factors (i.e., physical and
sexual abuse, childhood neglect, fearful, preoccupied, and dismis-
sive attachment), we found that physical abuse was correlated with
sexual abuse (r ⫽ .37, p ⬍ .001), and with neglect (r ⫽ .62, p ⬍
.001). Neglect was associated with sexual abuse (r ⫽ .40, p ⬍
.001). Fearful attachment was associated with physical abuse,
sexual abuse, and neglect (all rs ⫽ .16, p ⬍ .05), as well as with
preoccupied and dismissive attachment (r ⫽ .16 and .17, p ⬍ .05,
respectively). Finally, preoccupied and dismissive attachment
were inversely correlated (r ⫽⫺.21, p ⬍ .01).
These continuous risk factors were then correlated with the
binary risk factors: Throwaway status, sexual orientation, exposure
to bullying, presence/absence of friends attempting/committing
suicide, and sex trade involvement. This was done using biserial
correlations. Throwaway status was associated with physical abuse
(r ⫽⫺.16, p ⬍ .05) and neglect (r ⫽⫺.15, p ⬍ .05). Sexual
orientation was associated with fearful and preoccupied attach-
ment (r ⫽ .18, p ⬍ .001; r ⫽ .15, p ⬍ .05, respectively; non-
straight orientation was related to elevation in both insecure at-
tachment styles). Being bullied was also associated with fearful
attachment in the positive direction (r ⫽ .19, p ⬍ .01). Sex trade
involvement was associated with sexual abuse (r ⫽ .25, p ⬍ .01;
presence of sex trade involvement was associated with reports of
childhood sexual abuse), and with fearful attachment (r ⫽ .16, p ⬍
.05; presence of sex trade involvement was associated with an
elevation in fearful attachment).
A series of contingency tables were then run to examine asso-
ciations among the five binary risk factors: throwaway status,
sexual orientation, exposure to bullying, the presence of friends
who attempted/committed suicide sex trade involvement. Of the 10
tests conducted, three statistically significant associations
emerged: between sex trade involvement and sexual orientation,
(1) ⫽ 29.05, p ⬍ .001 (gay and lesbian youths were more likely
to engage in the sex trade), exposure to bullying and sexual
(1) ⫽ 6.92, p ⬍ .05 (the differences between bullied
and nonbullied youths was particularly pronounced among hetero-
sexual participants), and between sexual orientation and the pres-
ence of friends who attempted/committed suicide,
(1) ⫽ 3.64,
p ⫽ .05 (the differences between the frequencies of youths who
had friend(s) who attempted/committed suicide was particularly
pronounced among gay and lesbian youths).
We then correlated risk factors with the resilience factors (see
Table 3). Social involvement was associated with the presence of
friend(s) who attempted/committed suicide (r ⫽ .28, p ⬍ .001),
with childhood neglect (r ⫽⫺.16, p ⬍ .05), and with fearful and
dismissive attachment (rs ⫽⫺.16, p ⬍ .05). Secure attachment
was associated with low levels of childhood sexual abuse (r ⫽
⫺.24, p ⫽ .001), childhood neglect (r ⫽⫺.24, p ⬍ .001), fearful
attachment (r ⫽⫺.25, p ⬍ .001), preoccupied attachment (r ⫽
.14, p ⬍ .05) and dismissive attachment (r ⫽⫺.20, p ⬍ .01).
Finally, self-esteem was inversely associated with: sex trade (r ⫽
⫺.14, p ⬍ .05), with the presence of friend(s) who attempted/
committed suicide (r ⫽ .18, p ⬍ .01), physical abuse (r ⫽⫺.17,
p ⬍ .05), sexual abuse (r ⫽⫺.24, p ⬍ .001), and childhood
neglect (r ⫽⫺.24, p ⫽ .001), fearful attachment (r ⫽⫺.21, p ⬍
.01) and preoccupied attachment (r ⫽⫺.19, p ⬍ .01).
Finally, in Table 4, correlations between risk/resilience vari-
ables and dependent variables are presented. Loneliness was
We did not have a priori hypotheses regarding age and gender. Nev
ertheless, we explored the associations between these demographic vari-
ables and the study variables. Age was associated with the choice to leave
home (r ⫽ .17, p ⬍ .05, older participants decided on their own to leave
home), having experienced bullying (r ⫽⫺.14, p ⬍ .05, older age was
associated with fewer participants experiencing bullying), and substance
use (r ⫽ .15, p ⬍ .05). Gender was associated with sexual orientation,
(1) ⫽ 9.33, p ⬍ .05 (men had the highest pro
portion of heterosexuals), and presence of a friend who attempted/
committed suicide, Maximum Likelihood
(1) ⫽ 5.80, p ⬍ .05 (in both
groups there were more incidents of friends attempting/committing suicide,
but the difference was stronger among women). As well, gender was
associated with sexual abuse, t(200) ⫽ 4.03, p ⬍ .001 (women higher than
men), fearful and secure attachment, t(200) ⫽ 2.06 and 2.24, p ⬍ .05
(women higher than men), self-esteem, t(200) ⫽⫺2.49, p ⬍ .05; (men
higher than women), and subjective health status t(200) ⫽⫺2.91, p ⬍ .01
(men higher than women). Because gender and age were associated with
two of our outcomes, namely, subjective health status and substance use,
respectively, we repeated the analyses while controlling for gender and
age. The pattern of results was identical to the one reported earlier.
While the study was initially designed to probe for differences between
national contexts (United States and Canada) and agency versus service
settings, these differences did not emerge as being significant across most
variables. Of the outcome variables, only substance use emerged as being
significantly different as a function of country (Canadian youth reported
more drug use), and no significant differences were noted as a function of
setting (agency vs. street).
Correlations Among the Study Outcomes
Feeling trapped .64
Subjective health ⫺.23
Substance use .06 .16
Suicidal ideation .38
p ⬍ .05.
p ⬍ .01.
p ⬍ .001.
RESILIENCE IN HOMELESS YOUTH
correlated with physical and sexual abuse (rs ⫽ .18 and .16, p ⬍
.05), neglect (r ⫽ .32, p ⬍ .01), throwaway status (r ⫽⫺.19, p ⬍
.05), sex trade involvement (r ⫽ .18, p ⬍ .05), fearful and
dismissive attachment (rs ⫽ .23 and .30, p ⬍ .01), secure attach-
ment (r ⫽⫺.17, p ⬍ .05), social involvement (r ⫽⫺.26, p ⬍ .01)
and self-esteem (r ⫽⫺.38, p ⬍ .001). Feeling trapped was
associated with physical and sexual abuse (r ⫽ .18, p ⬍ .05),
neglect (r ⫽ .32, p ⬍ .01), fearful and preoccupied attachment
(rs ⫽ .23 and .23, p ⬍ .01), and self-esteem (r ⫽⫺.44, p ⬍ .001).
Suicidal ideation was associated with physical and sexual abuse
(r ⫽ .27, p ⬍ .01 and .15, p ⬍ .05), neglect (r ⫽ .23, p ⬍ .01),
sexual orientation (r ⫽ .22, p ⬍ .01), being bullied (r ⫽ .16, p ⬍
.05), friends committing/attempting suicide (r ⫽ .24, p ⬍ .01),
fearful and preoccupied attachment (r ⫽ .23, p ⬍ .01 and r ⫽ .16,
p ⬍ .05), secured attachment (r ⫽⫺.15, p ⬍ .05), and self-esteem
(r ⫽⫺.44, p ⬍ .001). Subjective health status was associated with
physical and sexual abuse (rs ⫽⫺.16 and ⫺.26, p ⬍ .05), neglect
(r ⫽⫺.24, p ⬍ .01), friends committing/attempting suicide (r ⫽
⫺.27, p ⬍ .01), fearful attachment (r ⫽⫺.16, p ⬍ .05), secure
attachment (r ⫽ .23, p ⬍ .01), and self-esteem (r ⫽ .39, p ⬍ .001).
Finally, substance use was associated with friends committing/
attempting suicide (r ⫽ .36, p ⬍ .01), social involvement (r ⫽ .23,
p ⬍ .01) and self-esteem (r ⫽⫺.24, p ⬍ .01).
Results of the regression analyses are summarized in Table 5, in
which we present statistically significant coefficients. When lone-
liness was considered as an outcome, the following statistically
significant predictors were found in Block 1: childhood neglect
(␤⫽.22, p ⬍ .01, observed power: .80), self-esteem (␤⫽⫺.31,
p ⬍ .001, observed power: .99), dismissive attachment (␤⫽.26,
p ⬍ .001, observed power: .99). Block 1 accounted for 35% of the
variance of loneliness, F(10, 193) ⫽ 10.35, p ⬍ .001). In Block 2,
in which interactions between protective and risk factors were
examined, a statistically significant interaction between self-
esteem and fearful attachment was found (␤⫽⫺.16, p ⫽ .01,
observed power: .67). The regression model accounted for 41% of
the variance of loneliness (F(10, 172)⫽ 3.90, p ⬍ .001). The
two-way interaction was probed according to the recommenda-
tions put forth by Aiken and West (1991). The pattern depicted in
Figure 1 was identified. Consistent with the expected buffering
effect of self-esteem, fearful attachment was associated with lone-
liness when self-esteem was low, but not when it was high.
With feeling trapped as the dependent variable, the following
statistically significant predictors were found in Block 1: child-
hood neglect (␤⫽.28, p ⬍ .001, observed power: .92) and
self-esteem (␤⫽⫺.35, p ⬍ .001, observed power: .99). Block 1
accounted for 28% of the variance of feeling trapped, F(8, 195)⫽
9.72, p ⬍ .001. In Block 2, none of the statistically significant
interactions met our stringent level of statistical significance. This
block accounted for 35% of the variance of feeling trapped, (F(23,
1802)⫽ 4.13, p ⬍ .001.
When suicidal ideation was considered, statistically significant
predictors in Block 1 were physical abuse (␤⫽.19, p ⫽ .01,
observed power: .66), and self-esteem (␤⫽⫺.21, p ⬍ .001,
observed power: .99). Block 1 accounted for 29% of the variance
of suicidal ideation (F
⫽ 7.91, p ⬍ .001). In Block 2, no
statistically significant interaction was found. The regression
Correlations Between Risk and Resilience Variables
Physical abuse .02 ⫺.09 ⫺.16
Sexual abuse ⫺.09 ⫺.23
Throwaway status .04 ⫺.00 .01
Sexual orientation .03 .05 ⫺.13
Bullied in school .09 .06 ⫺.07
Sex trade involvement .03 .08 ⫺.14
Fearful attachment ⫺.13
Preoccupied attachment .01 .14
Dismissive attachment ⫺.16
p ⬍ .05.
p ⬍ .01.
p ⬍ .001.
Correlations Between the Risk/Resilience Factors and the Outcomes
Physical abuse .18
Sexual abuse .16
Throwaway status ⫺.19
⫺.12 ⫺.03 ⫺.00 .04
Sexual orientation .12 .04 .22
Bullied in school .14 .14 .16
Friend suicide .05 .07 .24
Sex trade involvement .18
.14 .08 ⫺.10 .06
Fearful Attachment .24
Preoccupied attachment .11 .22
Dismissing attachment .30
.09 .06 ⫺.05 .13
Secure attachment ⫺.17
Social involvement ⫺.26
⫺.08 ⫺.00 ⫺.01 .23
p ⬍ .05.
p ⬍ .01.
p ⬍ .001.
KIDD AND SHAHAR
model accounted for 36% of the variance of suicidal ideation,
F(317, 172)⫽ 3.14, p ⬍ .001.
When subjective health status was considered, the presence of
friends who attempted or committed suicide emerged as a statis-
tically significant predictor in Block 1 (␤⫽⫺.19, p ⬍ .001;
observed power: .83), and so was the case for self-esteem (␤⫽
.28, p ⬍ .001; observed power: .98). Block 1 accounted for 25%
of the variance of subjective health status, F(8, 195)⫽ 8.20, p ⬍
.001. In Block 2, no statistically significant interaction was found.
The regression model accounted for 29% of the variance of sub-
jective health status, F(23, 181)⫽ 3.44, p ⬍ .01.
Finally, when substance use was examined, the following sta-
tistically significant predictors were found in Block 1: presence of
friends who attempted or committed suicide (␤⫽.27, p ⬍ .001,
observed power: .97), social involvement (␤⫽.18, p ⫽ .01,
observed power: .72), and self-esteem (␤⫽.20, p ⫽ .001, ob-
served power: .87). Block 1 accounted for 18% of the variance of
substance use, F(4, 199)⫽ 11.57, p ⬍ .001. In Block 2, no
statistically significant interaction was found. The regression
model accounted for 19% of the variance of substance use, F(7,
196) ⫽ 6.72, p ⬍ .001.
The present study was conceived to extend the current literature
through an integrative analysis of well-established risks and less
thoroughly examined aspects of resilience as they relate to the
mental and perceived physical health of young homeless persons.
As expected, the majority of risk and resilience variables examined
in this study demonstrated significant associations with the mental
health (loneliness, feeling trapped, suicidal ideation), subjective
health status, and substance use behavior of the participants. Only
one finding, however, consistently emerged as confirming the
hypotheses that resilience would predict outcomes over and above
risks and would buffer the impact of risks on outcomes. Self-
esteem was the one resilience variable that was reliably associated
with all outcomes and served to buffer the impact of fearful
attachment on loneliness.
Previous work has highlighted the centrality of loneliness as a
key aspect of emotional distress among homeless youth. Loneli-
ness is the subjective experience underlying broken relations with
family, peers, and stigmatization by larger society (Kidd, 2004;
Rew, 2002; Rokach, 2005). In the present study, loneliness was
significantly accounted for by self-esteem, neglect by caregivers,
and dismissing attachment. This profile of findings is different
from the general youth population with respect to attachment.
Among the general adolescent population secure attachment is
strongly related to perceived loneliness (DiTommaso, Brannen-
McNulty, Ross, & Burgess, 2003). In addition, elevated self-
esteem was found to buffer the impact of fearful attachment on
loneliness. This effect would seem to be related to previous studies
that have demonstrated the role of self-esteem as a mediator
between attachment and psychological health among the general
adolescent population (Huntsinger & Luecken, 2004; Wilkinson,
The subjective experience of feeling/being trapped, similar to
the experience of helplessness, has been identified in past explor-
atory and confirmatory work as being the core element of emo-
tional distress experienced by homeless youth and a key precipi-
tant of suicide attempts (Kidd, 2004, 2006). In the present study,
feeling trapped was significantly related to self-esteem and neglect
by parents. Self-esteem, along with physical abuse, also emerged
as having a strong relationship with suicidal ideation. While the
importance of self-esteem to suicidality among homeless youth has
been highlighted previously (Kidd, 2004, 2006; Unger, Kipke,
Simon, & Montgomery et al., 1998; Yoder, 1999), its importance
relative to other variables would not appear to have been com-
Summary of the Generalized Linear Models Results
Physical abuse .19
Neglect .22 .28
Dismissive attachment .26
Social involvement ⫺.15 .18
Self-esteem ⫺.31 ⫺.35 ⫺.35 .38 ⫺.20
Friend suicide ⫺.19 .27
Fearful attachment ⫻ self-esteem .16
Note. Numbers are standardized regression coefficients.
Low FA High FA
Figure 1. Loneliness as a function of an interaction between fearful
attachment and self-esteem. Lon ⫽ loneliness; FA ⫽ fearful attachment;
SE ⫽ self-esteem.
RESILIENCE IN HOMELESS YOUTH
Self-esteem also emerged as central in its relationship with the
two other outcome variables examined in this study: subjective
health status and substance use. Subjective health status was sig-
nificantly related to better self-esteem and having fewer friends
who had attempted or completed suicide, two variables that were
also significantly related to substance use in addition to social
involvement. Though not examined previously in the literature, it
is possible that the suicidality of friends may be an indirect
indicator of peer networks in which there is a high degree of poor
mental health and/or traumatization and grief. It might, therefore,
be posited that having such a social network might lead to poor
self-care in regards to physical health with their being influenced
by greater and riskier drug use which has been associated with
passive suicidality (Mino, Bousquet, & Broers, 1999). The linkage
between social involvement and drug use is likely reflective of the
previously commented on association between peer involvement
and risk amplification (Bao et al., 2000).
Two factors may be involved in the importance of self-esteem
across the health indicators examined in this study. First, as noted
above, self-esteem may be more directly related to (i.e., a key
mediator) psychological health relative to other variables such as
attachment (Huntsinger & Lueken, 2004; Wilkinson, 2004). Sec-
ond, homeless youths regularly comment that, as a function of
unstable and unsafe social environments, greater emphasis is
placed on the self in terms of resilience. Self-resources commented
on by these youths include efficacy, intelligence, personal strength,
resourcefulness, and wariness around others (Kidd & Davidson,
2006). It would be expected, in this context, that when self-esteem
is not present to serve as a foundation of resilience, these young
people may be particularly vulnerable to negative mental and
physical health outcomes.
Certain elements of the street environment might, in part, ac-
count for the above-noted discrepancies between the present find-
ings regarding attachment and those typically found among the
general youth population. First, youths on the street are frequently
victimized by peers (Baron, 2003) and are stigmatized by larger
society (Kidd, 2004, 2007). This consideration, combined with
findings that in socially hostile environments youths develop
avoidant attachment styles with peers (Ireland & Power, 2004),
adds a dimension of complexity to how attachment style might
affect risk on the street. In certain street contexts (e.g., youths new
to the street, less familiar social settings), development of fearful
and/or dismissing attachment style may be adaptive. Youths who
approach these kinds of situations with a secure attachment model
may be at an increased risk of being victimized (i.e., may be more
easily lured into trusting perpetrators). In this scenario, and as the
findings of this study suggest, an ability to maintain good self-
esteem while remaining fearful in some contexts may be adaptive.
Indeed, many youths comment on a crucial aspect of “street
smarts” (often associated with resilience) as involving being very
careful about who can be trusted (Kidd & Davidson, 2007).
Most notable of the limitations of this study is our reliance on a
cross-sectional research design, which does not allow for a defin-
itive conclusion regarding the direction of the relationships be-
tween variables. Another limitation is the relatively narrow opera-
tionalization of the constructs examined which occurred out of the
necessity to keep this evaluation brief as a part of a survey
examining several other domains. More refined measures of at-
tachment (e.g., attachment with peers vs. attachments with main-
stream persons/adults), self-esteem (i.e., global vs. specific), social
network composition, health and substance use/dependence may
prove very fruitful in better delineating the active elements in this
spectrum of risks and protective factors. A particularly promising
avenue for future investigation is the roles of relations with family
members and nonstreet involved peers in the risk and resilience
trajectories. Many youths maintain relationships with nonstreet
involved friends and almost a third having been found to receive
some form of emotional and/or instrumental support from a family
member (Johnson, Whitbeck, & Hoyt, 2005). In addition, the use
of subjective health status as an indicator of physical health status
was not confirmed by objective health data and, as such, both
perceptual and reporting biases may have influenced the results.
Lastly, although this study did not indicate significant differences
across demographic variables (e.g., gender, nationality, agency vs.
street setting), the heterogeneity of this population cannot be
overlooked. It is also possible that the broad age range employed
in this study may have, in combination with the cross-sectional
design, glossed over the variables examined here as they evolve
over youths’ street careers. More refined measures of the con-
structs examined in this study as they relate to demographics will
no doubt prove extremely important in the development of focused
and effective interventions.
The key implication for intervention highlighted in this study
lies in the importance of self-esteem to risk and resilience for this
group. It is of note that the emphasis on self-esteem has also
emerged in the narratives of counselors in describing effective
practice with this population (Kidd, Miner, Walker, & Davidson,
2007). Accordingly, a focal aspect of treatment should be evalu-
ating the self-esteem of homeless youth clients and have as major
explicit and implicit goals of programming the need to improve
self-esteem. Approaches might include: (i) helping youths work
through past and present threats to self-esteem and reframe them as
being indicative of external problems and situational factors rather
than personal failures, (ii) emphasizing the positive implications
for valuing the self as they relate to successes and achievements,
and (iii) exploring how elements of empirically supported inter-
ventions (e.g., Cognitive Behavioral Therapy, Dialectical Behav-
ioral Therapy, Motivational Interviewing) might be adapted for
service delivery contexts for this group.
Adlaf, E. M., Zdanowicz, Y. M., & Smart, R. G. (1996). Alcohol and other
drug use among street-involved youths in Toronto. Addicted Research &
Theory, 4, 11–24.
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage.
Aish, A., & Wasserman, D. (2001). Does Beck’s Hopelessness Scale really
measure several components? Psychological Medicine, 31, 367–372.
Bao, W., Whitbeck, L. B., & Hoyt, D. R. (2000). Abuse, support, and
depression among homeless and runaway adolescents. Journal of Health
& Social Behavior, 41, 408 –420.
KIDD AND SHAHAR
Baron, S. W. (1999). Street youth and substance use: The role of back-
ground, subculture, and economic factors. Youth & Society, 31, 3–26.
Baron, S. W. (2003). Self-control, social consequences, and criminal
behavior: Street youth and the general theory of crime. Journal of
Research in Crime and Delinquency, 40, 403– 425.
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.
Bartholomew, K., & Shaver, P. M. (1998). Methods of assessing adult
attachment. In J. A. Simpson & W. S. Rholes (Eds.), Attachment theory
and close relationships (pp. 25– 45). New York: Guilford Press.
Beck, A. T., Weissman, A., Lester, D., & Trexler, L. (1974). The mea-
surement of pessimism: The hopelessness scale. Journal of Consulting
and Clinical Psychology, 42, 861– 865.
Cascardi, M., Avery-Leaf, S., O’Leary, K. D., & Slep, A. (1999). Factor
structure and convergent validity of the Conflict Tactics Scale in high
school students. Psychological Assessment, 11, 546 –555.
de Jong, M. L. (1992). Attachment, individuation, and risk of suicide in late
adolescence. Journal of Youth and Adolescence, 21, 357–373.
DiTommaso, E., Brannen-McNulty, C., Ross, L., & Burgess, M. (2003).
Attachment styles, social skills, and loneliness in young adults. Person-
ality and Individual Differences, 35, 303–312.
Ennett, S. T., Bailey, S. L., & Federman, E. B. (1999). Social network
characteristics associated with risky behaviors among runaway and
homeless youth. Journal of Health and Social Behavior, 40, 63–78.
Ensign, J. (1998). Health issues of homeless youth. Journal of Social
Distress and the Homeless, 7, 159 –174.
Greene, J. M., Ennett, S. T., & Ringwalt, C. L. (1999). Prevalence and
correlates of survival sex among runaway and homeless youth. American
Journal of Public Health, 89, 1406 –1409.
Hagan, J., & McCarthy, B. (1997). Mean streets: Youth crime and home-
lessness. Cambridge, United Kingdom: Cambridge University Press.
Howard, M. S., & Medway, F. J. (2004). Adolescent’s attachment and
coping with stress. Psychology in Schools, 41, 391– 402.
Huntsinger, E. T., & Luecken, L. J. (2004). Attachment relationships and
health behaviour: The mediational role of self-esteem. Psychology and
Health, 19, 515–526.
Ireland, J. L., & Power, C. L. (2004). Attachment, emotional loneliness,
and bullying behaviour: A study of adult and young offenders. Aggres-
sive Behavior, 30, 298 –312.
Johnson, K. D., Whitbeck, L. B., & Hoyt, D. R. (2005). Predictors of social
network composition among homeless and runaway adolescents. Jour-
nal of Adolescence, 28, 263–275.
Kidd, S. A. (2003). Street youth: Coping and interventions. Child and
Adolescent Social Work Journal, 20, 235–261.
Kidd, S. A. (2004). “The walls were closing in and we were trapped”: A
qualitative analysis of street youth suicide. Youth and Society, 36,
Kidd, S. A. (2006). Factors precipitating suicidality among homeless
youth: A quantitative follow-up. Youth & Society, 37, 393– 422.
Kidd, S. A. (2007). Youth homelessness and social stigma. Journal of
Youth and Adolescence, 36, 291–299.
Kidd, S. A., & Carroll, M. (2007). Coping and suicidality among homeless
youth. Journal of Adolescence, 30, 283–296.
Kidd, S. A., & Davidson, L. (2007). “You have to adapt because you have
no other choice”: The stories of strength and resilience of 208 homeless
youth in New York City and Toronto. Journal of Community Psychol-
ogy, 35, 219 –238.
Kidd, S. A., & Kral, M. J. (2002). Street youth suicide and prostitution: A
qualitative analysis. Adolescence, 37, 411– 430.
Kidd, S. A., Miner, S., Walker, D., & Davidson, L. (2007). Stories of
working with homeless youth: On being “mind-boggling.” Children and
Youth Services Review, 29, 16 –34.
Kidd, S. A., & Scrimenti, K. (2004). The New Haven homeless count:
Children and youth. Evaluation Review, 28, 325–341.
Kinsey, A. C., Pomeroy, W. B., & Martin, C. E. (1948). Sexual behavior
in the human male. Philadelphia: W. B. Saunders.
Laible, D. J., Carlo, G., & Raffaelli, M. (2000). The differential relations
of parent and peer attachment to adolescent adjustment. Journal of Youth
and Adolescence, 29, 45–59.
Laible, D. J., Carlo, G., & Roesch, S. C. (2004). Pathways to self-esteem
in late adolescence: The role of parent and peer attachment, empathy,
and social behaviours. Journal of Adolescence, 27, 703–716.
Larose, S., & Bernier, A. (2001). Social support processes: Mediators of
attachment state of mind and adjustment in late adolescence. Attachment
and Human Development, 3, 96 –120.
Lester, D. (2001). An inventory to measure helplessness, hopelessness, and
haplessness. Psychological Reports, 89, 495– 498.
Lewinsohn, P. M., Rohde, P., & Seeley, J. R. (1996). Adolescent suicidal
ideation and attempts: Prevalence, risk factors, and clinical implications.
Clinical Psychology: Science and Practice, 3, 25– 46.
Mills, C., Guerin, S., Daly, I., Lynch, F., & Fitzpatrick, C. (2005). The
relationship between bullying, depression, and suicidal thoughts/
behaviours in Irish adolescents. Irish Journal of Psychological Medi-
cine, 22, 112–116.
Mino, A., Bousquet, A., & Broers, B. (1999). Substance use and drug-
related death, suicidal ideation, and suicide: A review. Crisis, 20, 28–35.
Paradise, M., & Cauce, A. M. (2002). Home street home: The interpersonal
dimensions of adolescent homelessness. Analyses of Social Issues and
Public Policy, 2, 223–238.
Rew, L. (2002). Relationships of sexual abuse, connectedness, and lone-
liness to perceived well-being in homeless youth. Journal for Specialists
in Pediatric Nursing, 7, 51– 63.
Rew, L., & Horner, S. D. (2003a). Personal strengths of homeless adoles-
cents living in a high-risk environment. Advances in Nursing Science,
26, 90 –101.
Rew, L., & Horner, S. D. (2003b). Youth resilience framework for reduc-
ing health-risk behaviors in adolescents. Journal of Pediatric Nursing,
18, 379 –388.
Rice, E., Milburn, N. G., Rotheram-Borus, M. J., Mallett, S., & Rosenthal,
D. (2005). The Effects of peer group network properties on drug use
among homeless youth. American Behavioral Scientist, 48, 1102–1123.
Ringwalt, C. L., Greene, J. M., & Robertson, M. J. (1998). Familial
backgrounds and risk behaviors of youth with throwaway experiences.
Journal of Adolescence, 21, 421– 452.
Rokach, A. (2005). The causes of loneliness in homeless youth. Journal of
Psychology: Interdisciplinary and Applied, 139, 469 – 480.
Rosenberg, M. (1989). Society and the adolescent self-Image. Middletown,
CT: Wesleyan University Press.
Roy, E., Haley, N., Leclerc, P., Sochanski, B., Boudreau, J., & Boivin, J.
(2004). Mortality in a cohort of street youth in Montreal. Journal of the
American Medical Association, 292, 569 –574.
Russell, D. W. (1996). UCLA Loneliness Scale (Version 3): Reliability,
validity, and factor structure. Journal of Personality Assessment, 66,
20 – 40.
Schultz, R., Mittelmark, M., Kronmal, R., Polak, J., Hirsch, C. H., German,
P., et al., (1994). Predictors of perceived health status in elderly men and
women. Journal of Aging and Health, 6, 419 –447.
Sell, R. L. (1997). Defining and measuring sexual orientation: A review.
Archives of Sexual Behavior, 26, 643– 658.
Shaw, M., & Dorling, D. (1998). Mortality among street youth in the UK.
Lancet, 352, 743.
Stewart, A. L., Greenfield, S., Hays, R. D., Wells, K., Rogers, W. H.,
Barry, S. D., et al., (1989). Functional status and well being of patients
with chronic conditions: Results from a medical outcomes study. Jour-
nal of the American Medical Association, 262, 907–913.
RESILIENCE IN HOMELESS YOUTH
Straus, M., & Gelles, R. (1990). Physical violence in American families.
Brunswick, NJ: Transaction.
Tavecchio, L. W., Thomeer, M. A., & Meeus, W. (1999). Attachment,
social network and homelessness in young people. Social Behaviour and
Personality, 27, 247–262.
Taylor-Seehafer, M., Johnson, R., Rew, L., Fouladi, R. T., Land, L., &
Abel, E. (2007). Attachment and sexual health behaviors in homeless
youth. Journal for Specialists in Pediatric Nursing, 12, 37– 48.
Thompson, S. J., Sayfer, A. J., & Pollio, D. E. (2001). Differences and
predictors of family reunification among subgroups of runaway youths
using shelter services. Social Work Research, 25, 163–172.
Unger, J. B., Kipke, M. D., Simon, T. R., Johnson, C. J., Montgomery,
S. B., & Iverson, E. (1998). Stress, coping, and social support among
homeless youth. Journal of Adolescent Research, 13, 134 –157.
Whitbeck, L. B., Chen, X. J., Hoyt, D. R., Tyler, K. A., & Johnson, K. D.
(2004). Mental disorders, subsistence strategies, and victimization
among gay, lesbian and bisexual homeless youths and runaway adoles-
cents. Journal of Sex Research, 41, 329 –342.
Whitbeck, L. B., Hoyt, D. R., & Bao, W. (2000). Depressive Symptoms,
Depression, and Comorbidity among Homeless and Runaway Adoles-
cents. Child Development, 71, 721–732.
Whitbeck, L. B., Hoyt, D. R., & Yoder, K. A. (1999). A risk-amplification
model of victimization and depressive symptoms among runaway and
homeless adolescents. American Journal of Community Psychology, 27,
Whitbeck, L. B., & Simons, R. L. (1990). Life on the streets: The victim-
ization of runaway and homeless adolescents. Youth & Society, 22,
Wilkinson, R. B. (2004). The role of parental and peer attachment in the
psychological health and self-esteem of adolescents. Journal of Youth
and Adolescence, 33, 479 – 493.
Yoder, K. A. (1999). Comparing suicide attempters, suicide ideators, and
nonsuicidal homeless and runaway adolescents. Suicide & Life Threat-
ening Behavior, 29, 25–36.
Yoder, K. A., Hoyt, D. R., & Whitbeck, L. B. (1998). Suicidal behavior
among homeless and runaway adolescents. Journal of Youth and Ado-
lescence, 27, 753–771.
Received May 30, 2007
Revision received February 17, 2008
Accepted March 19, 2008 䡲
KIDD AND SHAHAR