Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/254374406

Reasons for Homelessness: An Empirical Typology

Article in Vulnerable Children and Youth Studies · March 2012


DOI: 10.1080/17450128.2011.643832

CITATIONS READS

9 770

4 authors, including:

Hillary J. Heinze Debbie Jozefowicz


University of Michigan-Flint University of Windsor
7 PUBLICATIONS 300 CITATIONS 25 PUBLICATIONS 758 CITATIONS

SEE PROFILE SEE PROFILE

Paul A Toro
Wayne State University
85 PUBLICATIONS 3,198 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Literature Review View project

Social Work and Allied Knowledge Syntheses View project

All content following this page was uploaded by Debbie Jozefowicz on 23 April 2015.

The user has requested enhancement of the downloaded file.


Vulnerable Children and Youth Studies
Vol. 7, No. 1, March 2012, 88–101

Reasons for homelessness: An empirical typology


Hillary J. Heinzea*, Debra M. Hernandez Jozefowiczb , Paul A. Toroc and Logan R. Bluea
a
Department of Psychology, University of Michigan-Flint, Flint, MI, USA; b School of Social Work,
University of Windsor, Windsor, ON, Canada; c Department of Psychology, Wayne State University,
Detroit, MI, USA
(Received 22 June 2011; final version received 18 November 2011)

We investigated patterns of situational characteristics and service use among youth


residing at metropolitan area homeless shelters. One hundred and three youth provided
ratings on a five-point scale comprised of 17 reasons for becoming homeless. Cluster
analysis yielded five distinct reason typologies: destitute, threatened, pregnant, resis-
tant, and partnered. Demographic, social, and treatment characteristics were analyzed
across groups, and a concurrent measure related to the initial clustering variable, a
qualitative assessment of reasons for seeking services, was used to validate typologies.
Distributions of demographic, social, and service characteristics lend tentative support
to the typologies and suggest areas of unmet need.
Keywords: homelessness; youth; reasons for homelessness; typologies; homeless
shelters

Introduction
Each year, hundreds of thousands of adolescents spend the night on the streets, in a shelter,
hotel, or motel, or in some other temporary or inadequate location without supervision of
an adult caretaker (Hammer, Finkelhor, & Sedlak, 2002). In addition to immediate dan-
gers faced by these youth, several studies have linked youth homelessness to a number
of predisposing risk characteristics, such as family substance abuse, conflict and mal-
treatment, school, attention and learning problems, and psychological disturbance (Cauce
et al., 2000; McCaskill, Toro, & Wolfe, 1998; Paradise & Cauce, 2002; Whitbeck, Johnson,
Hoyt, & Cauce, 2004; Wolfe, Toro, & McCaskill, 1999). Episodes of homelessness may
amplify risk, as these youth are more likely to engage in delinquent behaviors, associate
with deviant peers, and be victimized by others (Whitbeck & Hoyt, 1999).
Youth differ in terms of how they become, experience, and exit homelessness. Across
studies examining circumstances leading to homelessness, approximately one-third to over
one-half of youth report family problems, such as poor communication, conflict, strict
rules, and fighting, as a primary contributing factor (Ennett, Federman, Bailey, Ringwalt,
& Hubbard, 1999; Kufeldt, Durieux, Nimmo, & McDonald, 1992; Mitchell, 2003; Peled,
Spiro, & Dekel, 2002). A substantial number of youth indicate abuse or neglect as a pri-
mary reason for leaving home (5–29%; Ennett et al., 1999; Kufeldt et al., 1992; Mitchell,
2003; Peled et al., 2002). Other reasons endorsed by youth vary across study, including

*Corresponding author. Email:hheinze@umflint.edu

ISSN 1745-0128 print/ISSN 1745-0136 online


© 2012 Taylor & Francis
http://dx.doi.org/10.1080/17450128.2011.643832
http://www.tandfonline.com
Vulnerable Children and Youth Studies 89

need for independence (23%; Ennett et al., 1999), housing instability (10%; Ennett et al.,
1999), and maladaptive youth behavior (truancy 11–24% and/or substance abuse 8–24%;
Kudfeldt et al., 1992).
To better understand the needs and clarify effective intervention strategies associated
with youth homelessness, researchers have identified typologies of homeless youth dis-
tinguished by patterns of situational, background, and behavioral characteristics. Farrow,
Deisher, Brown, and Kulig (1992) describe four categories of homeless youth based on
reasons for leaving their housing: situational runaways, who tend to leave home following
a disagreement and return after a few days; (chronic) runaways, who leave home due to
serious problems, such as abuse, and stay away for long periods of time; throwaways, who
are abandoned or “thrown out” by parents; and system youth, who have little family contact
and are living in foster care or institutional facilities before becoming homeless. Although
perhaps useful in advancing our understanding of the diversity in onset and patterns of
homelessness, such, often impressionistic, typologies are limited in that they may misrep-
resent the nature and complexity of the problem (Hammer et al., 2002; Toro, Dworsky, &
Fowler, 2007). For example, youth may fall into more than one category (system youth
who were “thrown out”) and youth within categories may leave home for different rea-
sons (due to lack of resources or family conflict). Few differences have been found in
current family relationships, psychological symptomology, or rates of recent victimization
among youth who ran away, were forced to leave, or were removed (MacLean, Embry, &
Cauce, 1999).
Other researchers have taken an empirical approach, differentiating youth typologies
based on self-reported risk and protective characteristics. Zide and Cherry (1992) used dis-
criminate analyses of scores for psychosocial variables associated with youth homelessness
to distinguish four typologies, ranging from youth having strong family, school, and
community bonds (running to), to youth with marginal resources, high levels of victim-
ization, and low scores on all social bonds (forsaken). More recently, researchers have
proposed typologies based on meaningful relationships among study variables. To this
end, cluster analysis has been recommended to identify and describe groups of individ-
ual cases defined by similarities among multiple dimensions, so that groups can be linked
to developmental outcomes and intervention strategies (Henry, Tolan, & Gorman-Smith,
2005). A few studies have applied this approach in homeless youth research. Mallett,
Rosenthal, Myers, Milburn, and Rotheram-Borus (2004) clustered youth by behavioral
patterns, using responses for how youth typically spend their time. Four subgroups were
identified: partnered homeless youth who engage in both positive and negative activities
with romantic partners; socially engaged youth who engage in positive and negative activ-
ities with friends; connected/harm-avoidant youth who engage in positive activities and
are more connected to school and services; and transgressive youth who are engaged
in harmful and illegal activities and less likely to be connected to schools, work, and
families. Subsequent analyses by Milburn et al. (2009) included scores for both pro-
tective (e.g., school, family, positive friends, survival skills) and risk factors (emotional
distress, risky sexual behavior, substance use) to identify three subgroups of newly home-
less youth differentiated by degree of risk (protected, at risk, and risky). In a related
study, Bucher (2008) identified four treatment groups of homeless youth based on clus-
tering of several risk characteristics: minimal treatment (low risk across categories);
therapeutic housing–addiction focus (high rates of abuse, moderate suicide risk, high
drug abuse); therapeutic housing–behavior management focus (high rates of abuse, high
involvement in criminal activity, low drug use); and comprehensive treatment (high risk
across categories).
90 H.J. Heinze et al.

Taken together, cluster analytic studies have increased understanding of how back-
ground factors and behavioral patterns vary across homeless youth, suggesting distinctive
groups that differ in terms of risk and behavior patterns. What still remains unclear, how-
ever, is how youth may differ in terms of specific factors driving the homeless episode –
the situational characteristics or events leading them to become homeless in the first
place. Moreover, treatment needs and intervention strategies for identified clusters have
been inferred from distributions of youth demographic and social information rather than
assessed directly.
In this study, we used cluster analysis to identify typologies of youth based on their
ratings on a scale comprised of 17 reasons for becoming homeless. In addition to analyz-
ing demographic and social variables across groups, we compared clusters derived from
the quantitative reasons for homelessness scale using a concurrently measured qualitative
assessment related to the initial clustering variable, for which youth were asked to list
their primary reasons for seeking shelter services. Finally, we examined similarities and
differences across clusters in services and resources received through the shelter.

Method
Participants and procedure
Six shelters providing services to homeless youth and young adults in a metropolitan area
were included in this study. Shelters varied in terms of setting (suburban, urban), range
of services offered (emergency shelter services, counseling, transitional housing), size and
target age group.
All youth and young adults involved in programming at the time of data collection were
invited to participate. Staff were not asked to provide information for youth unavailable or
unwilling to participate in order to minimize (1) additional tracking, inquiry, or related
activities requested of staff and (2) any influence of research activities on youth–staff
relationships and engagement in programming. Written consent was obtained from youth
aged 18 or older and parents of youth under age 18. Written assent was obtained from youth
under age 18. For participating agencies serving youth under age 18, state law requires con-
sent from a legal guardian within 48 hours of admission in order for youth to remain at the
shelter; parents and guardians were provided with study information and consent forms
prior to scheduled data collection. Questionnaires were administered orally by a member
of the research team in group format; participants recorded their responses directly onto
the questionnaire. Youth were paid $10.00 for participating, and $100.00 was donated to
each participating shelter.

Measures
All measures summarized below were created for use in the current study.

Background information
Demographic information. Participants were asked to report their gender, age, and
ethnicity.
Family and school information. Participants were asked to report whether they were
currently involved in a romantic relationship and their number of children (including cur-
rent pregnancy). Additionally, participants were asked to report how well they get along
with their primary caregivers using a five-point scale, response choices ranging from 1
Vulnerable Children and Youth Studies 91

(not at all) to 5 (very well). Finally, participants were asked if they were currently in
school.

Risk behavior. Participants were asked whether they had been charged or convicted of an
offense in the past year and to rate the extent of gang involvement (four-point scale, ranging
from 1 = no involvement to 4 = very involved) and drug and alcohol use (four-point scale,
ranging from 1 = no use to 4 = frequent use).

Reasons for seeking services (qualitative)


Participants were asked to generate up to three main reasons why they were in the shel-
ter (What are the main reasons(s) why you are here/in (this program)?). This assessment
preceded the reasons for homelessness scale.

Reasons for homelessness scale (quantitative)


Participants were asked to read the following description and then proceed with subsequent
questions if they met criteria for being homeless:

The next few questions refer to homelessness. For these questions, being without a home
means living on the street, in a shelter, in motels or hotels, living on and off with friends or
relatives or some other temporary place because you could not afford to live elsewhere.

Participants were presented with a list of 17 reasons for homelessness and instructed to rate
the extent to which each was a reason for them leaving home/being without a home (see
Table 1). Participants rated each reason using a five-point scale (response choices ranged
from 1 = not at all a reason to 5 = very much a reason). This scale was created by one of
the authors (Jozefowicz) for use in the current study, informed by previous work on youth
homelessness (Toro, Wolfe, & McCaskill, 2005).

Table 1. Original reasons for homelessness scale items.


1. I got kicked out of my housing
2. I left an abusive situation
3. My place was unsafe
4. There was no room for me at the last place
5. I could not afford to live there
6. I did not get along with the people I was living with
7. I do not like the rules at my last place
8. My family is without stable housing
9. I do not like where I was staying
10. I like being on my own
11. My parent(s) expect me to be on my own
12. My parent(s) do not have the resources to help me
13. My parent(s)/family is not available to help me
14. I wanted to be with my partner
15. I needed to get away from my school/neighborhood/peers
16. I needed to be on my own
17. I got pregnant

Note: Participants were asked to rate how much each of the items was a reason for leaving home/being without a
home, using a five-point scale (1 = not at all a reason; 5 = very much a reason).
92 H.J. Heinze et al.

Service utilization
Participants were given a list of services and resources and asked to indicate whether
shelters had provided or referred each service or resource (People at (this program) have
connected me with the following services).

Data analysis
One hundred and three youth (35 boys and young men; 68 girls and young women) who
completed the reasons for homelessness scale were included in the study analyses. Thirty-
three percent of youth were between the ages of 10 and 16 years; 38% were between the
ages of 17 and 20 years; and 29% were between the ages of 20 and 24 years.1
In order to maximize parsimony, interpretability, and utility of typologies, the original
reasons for homeless scale items were reevaluated to identify specific and distinct reasons
for the homeless episode. Items that did not suggest a specific reason for becoming home-
less (I like being on my own) or items that could be endorsed as a result of different reasons
for homelessness were excluded from the analysis. For example, the item I didn’t like where
I was staying could be endorsed by participants for any number of reasons (substandard
conditions, difficulties with housemates, safety concerns). Conceptually similar items were
combined into indices. Principal component analysis with varimax rotation was conducted
to corroborate conceptually formed indices. The results yielded five factors with eigenval-
ues greater than 1.0. The items I got pregnant and I wanted to be with my partner were
analyzed as separate variables (see Table 2). Final indices and items used in the cluster
analyses are presented in Table 3.2
Youth were clustered based on standardized (z-score) ratings for the six reasons for
homelessness (four factors plus two separate items) using agglomerative hierarchical clus-
tering. Hierarchical clustering combines cases sequentially, based on distances between
scores, to ultimately identify a set of groups (clusters) defined by maximal similarity of
cases within clusters and maximal dissimilarity of cases between clusters. Agglomerative
hierarchical clustering begins with each case as a separate cluster; at each successive level

Table 2. Factor loadings of reasons for homelessness items used in cluster analysis.
Factor

Scale item I II III IV V

I left an abusive situation 0.03 0.89 0.13 0.06 0.07


My place was unsafe 0.16 0.84 −0.04 0.07 0.03
There was no room for me at my last place 0.72 0.20 0.40 0.01 0.07
I could not afford to live there 0.75 −0.07 0.13 −0.14 0.12
My family is without stable housing 0.76 0.20 −0.01 0.19 0.07
I did not like the rules at my last place −0.04 0.04 −0.16 0.84 0.07
I did not get along with the people I was staying with 0.10 0.10 0.31 0.79 −0.02
My parent(s) expect me to be on my own 0.36 −0.03 0.69 0.12 −0.03
My parents/family is not available to help me 0.03 0.09 0.84 −0.02 0.14
I got pregnant 0.33 0.08 −0.05 0.03 0.71
I wanted to be with my partner −0.05 0.02 0.16 0.03 0.85
Eigenvalues 1.93 1.59 1.52 1.40 1.29
Percentage of variance explained (%) 17.62 14.48 13.86 12.73 11.70

Notes: Factor analysis was conducted using principal component analysis with varimax rotation and Kaiser
normalization. Factor loadings greater than 0.40 are represented as bold values.
Vulnerable Children and Youth Studies 93

Table 3. Intercorrelations among reasons for homelessness used in cluster analysis (N = 103).
Reasons for homelessness 1 2 3 4 5 6

1. Lack of safety 0.25∗ 0.17 0.12 0.15 0.08


2. Lack of resources 0.11 0.40∗∗ 0.29∗∗ 0.14
3. Rules and conflict at the last place 0.15 0.05 0.07
4. Lack of family support 0.16 0.14
5. Got pregnant 0.30∗∗
6. Wanted to be with partner

Notes: Correlation coefficients are presented in rows or columns (reason in rows and columns numbered accord-
ingly). Final scale items were adapted from the original scale as follows (internal consistency estimates are given
in parentheses): 1. Lack of safety (index; α = 0.70): I left an abusive situation, My place was unsafe; 2. Lack
of resources (index; α = 0.71): There was no room for me at the last place, I could not afford to live there, My
family is without stable housing; 3. Rules and conflict at the last place (index; α = 0.54): I do not like the rules at
my last place, I did not get along with the people I was living with; 4. Lack of family support (index; α = 0.53):
My parent(s) expect me to be on my own, My parent(s)/family is not available to help me; 5. Got pregnant (single
item): I got pregnant; 6. Wanted to be with partner (single item): I wanted to be with my partner.
∗ p < 0.05, ∗∗ p < 0.01.

of clustering, clusters are merged until only one remains (Milligan & Cooper, 1987). In this
study, clusters were joined via the “furthest neighbor” linkage method, which joins groups
based on the furthest distance between members in order to maximize the distance between
groups. Visual inspection of the dendrogram and agglomeration schedule suggested a five-
cluster solution (a large change in the agglomeration coefficient at the five-cluster stage).
The findings suggest the following cluster groups: destitute (n = 32; 31%); threatened (n =
18; 17%); pregnant (n = 8; 8%); resistant (n = 34; 33%); and partnered (n = 11; 11%).
Cluster means of each of the six reasons for homelessness are presented in Figure 1.

Clusters (1–3) Clusters (4–5)

1. Destitute 4. Resistant
2. Threatened 5. Partnered
Sample means
3. Pregnant
5 5

4 4
Mean score
Mean score

3 3

2 2

1 1
ty

es

er
t

t p rt

ty

es

nt

er
t

t p rt
an
lic

lic
o
fe

rtn

o
rc

fe

rtn
ga
rc
pp
nf

eg

pp
nf
Sa

Sa

re
co

pa
so

co

pa
su

su
es
e

nd

ith

nd
ily

ith
sr

ily
Go

sr

Go
sa

sa

m
ck

w
ck
fa

fa
Be
le
La

Be
le
La
Ru

of

Ru

of
ck

ck
La

La

Reason for homelessness (items) Reason for homelessness (items)

Figure 1. Reasons for homelessness item means by cluster.


94 H.J. Heinze et al.

Results
χ 2 -Tests were conducted to assess demographic, family, and risk differences across clusters
(see Table 4).3

Destitute (31% of the sample)


Although members of the destitute group did not demonstrate a single defining reason for
their episode of homelessness (spike/s), they produced the highest mean scores for lack of
resources and lack of family support. Destitute youth were more likely to be male (47%)
and African American (84%), and they were less likely to be in romantic relationships
(26%).

Threatened (17% of the sample)


Members of the threatened group produced high ratings for lack of safety and/or abuse.
Threatened youth were mostly female (78%) and more likely to be involved in romantic

Table 4. Demographic, family, and risk characteristics by cluster (N = 103).


Destitute Threatened Pregnant Resistant Partnered
(n = 32) (n = 18) (n = 8) (n = 34) (n = 11) Total

n % n % n % n % n % n %

Gender (female)∗ 17 53 14 78 8 100 20 59 9 82 68 66


Race/ethnicity∗
African American 27 84 11 61 4 50 15 44 5 46 62 60
Caucasian 4 13 6 33 1 13 15 44 4 36 30 29
Other/missing 1 3 1 6 3 38 4 12 2 18 11 11
Age (years)
10–16 7 22 2 11 3 38 16 47 6 55 34 33
17–20 13 41 9 50 3 38 9 27 5 46 39 38
21–24 12 38 7 39 2 25 9 27 0 0 30 29
Social (n/% yes)
Relationshipa,∗∗ 8 26 13 72 3 38 16 47 10 91 50 49
Childrenb,∗∗ 7 25 6 33 7 100 4 12 3 27 27 28
Positive relationship 14 47 8 47 3 50 11 39 5 56 41 46
with caregiver (F)c
Positive relationship 11 55 3 21 5 100 15 58 2 29 36 50
with caregiver (M)∗
In school 19 59 12 67 2 25 25 74 9 82 67 65
Convicted of offense 7 22 0 0 0 0 7 21 2 20 16 16
Gang involvement 3 10 3 18 1 14 12 39 3 27 22 23
Drug and alcohol use 16 50 9 50 5 71 21 70 8 73 59 60

Notes: Valid percentages are presented (missing data are not included in calculation except for race/ethnicity).
The number of missing participants by (cluster) is as follows: race/ethnicity: 1 (5); romantic relationships: 1 (1);
children 4 (1), 1 (3), 1 (4); positive relationship with female caregiver: 2 (1), 1 (2), 2 (3), 6 (4), 2 (5); positive
relationship with male caregiver: 12 (1), 4 (2), 3 (3), 8 (4), 4 (5); convicted of offense: 1 (4), 1 (4); gang activity:
1 (1), 1 (2), 1 (3), 3 (4); drug and alcohol use: 1 (3), 4 (4).
a Participant is involved in a romantic relationship.
b Participant has children or is pregnant.
c Participant reports having a positive relationship with their primary female (F) and male (M) caregivers (4 or
5 on five-point scale).
∗ p < 0.05, ∗∗ p < 0.01.
Vulnerable Children and Youth Studies 95

relationships (72%). Of the 14 youth providing male caregiver relationship ratings, only
three reported having positive relationships.

Pregnant (8% of the sample)


Members of the pregnant group produced high ratings for pregnancy and relatively high
ratings for wanting to be with their partners. All pregnant youth were female and reported
having children (including first pregnancy), and they were more likely to be women of
color (87%). Of the five youth providing male caregiver relationship ratings, all indicated
having positive relationships.

Resistant (33% of the sample)


Members of the resistant group produced high ratings for problems with rules and con-
flict with others and relatively lower ratings for lack of family support, lack of resources,
pregnancy, and wanting to be with their partners. Resistant youth were more likely to be
Caucasian (44%) and less likely to have children (12%).

Partnered (11% of the sample)


Partnered group members produced the highest ratings for wanting to be with their part-
ners, as well as relatively higher ratings for rules and conflict and lower ratings for lack
of resources. Group members were mostly female (82%), and they were more likely to be
involved in a romantic relationship (91%). Of the seven youth providing male caregiver
relationship ratings, only two reported having positive relationships.

Reasons for seeking service by cluster


To assess the robustness of identified clusters, we compared cluster membership (based on
ratings on the reasons for homelessness scale) with youth-generated (qualitative) reasons
for seeking services at participating shelters. Participants generated a total of 237 responses
reflecting 44 identifiable reasons for seeking services, which then were recoded into
eight service-seeking reason categories (see Table 5). χ 2 -Test of frequencies of the four
most frequently endorsed service-seeking reason categories (lack of/access to resources,
wanting to make a positive change, emotional/behavioral problems, and family-related
reasons) and a composite of the four other reason categories revealed significant variation
in reasons for service seeking across clusters, χ 2 (16, N = 237) = 40.57, p = 0.001 (see
Table 6).
When asked to list their reasons for seeking services, destitute group members listed the
highest number of resource needs (49% of total responses), while members of the threat-
ened, pregnant, and partnered groups were less likely to list resource needs (21%, 20%, and
13% of total responses, respectively). Members of the threatened group were most likely
to indicate wanting to make a positive change in their lives (32% of total responses), while
resistant group members were least likely to indicate wanting to make a positive change
(10% of total responses). Although they provided high ratings for problems with rules and
conflict on the reasons for homelessness scale, resistant group members did not indicate
high frequencies of emotional/behavioral reasons for seeking services. Resistant group
members did, however, list more family-related reasons than other groups (24% of total
responses).
96 H.J. Heinze et al.

Table 5. Coding examples for youth-generated reasons for seeking services.


Category 1= lack of/access to resources Category 4 = family related
• No place to go/no place to stay • Family problems
• Access to resources (i.e., shelter, food) • Kicked out by parents
• Unemployed/lost job Category 5 = child related
Category 2 = positive life change • Needed help with my children
• Get life together/make a fresh start • Got pregnant
• Independence/be on my own Category 6 = positive qualities of agency
• Positive/new experience • Good reputation (heard about it from . . .)
Category 3 = emotional/behavior problems • It is a nice/safe place
• Exit prostitution Category 7 = abuse/left abusive situation
• Stop using drugs or alcohol Category 8 = unclear or vague reason
• Court order/probation related • Ran away
• Emotional problems

Notes: Participants were asked to generate up to three main reasons why they were in the shelter/program of
study. This assessment preceded the reasons for homelessness scale.

Table 6. Frequencies of self-reported reasons for seeking services by cluster (N = 103).


Destitute Threatened Pregnant Resistant Partnered
(n = 32) (n = 18) (n = 8) (n = 34) (n = 11) Total

Reason category n % n % n % n % n % n %

Lack of resources 38 49 9 21 4 20 24 33 3 13 78 33
Positive change 11 14 14 32 5 25 7 10 5 22 42 18
Emotional/behavioral 14 18 3 7 2 10 14 19 9 39 42 18
problems
Family-related reasons 8 10 8 18 4 20 17 24 2 9 39 16
Child-related reasons 1 1 2 5 2 10 1 1 0 0 6 3
Qualities of agency 3 4 2 5 1 5 2 3 0 0 8 3
Abuse reasons 0 0 1 2 0 0 1 1 0 0 2 1
Vague 3 4 5 11 2 10 6 8 4 17 20 8
Total reasons 78 33 44 19 20 8 72 30 23 10 237 100
generated

Notes: Each participant listed up to three reasons why they sought services at the agency of study.
χ 2 -Test of frequencies of the four most frequently endorsed reasons (lack of resources, positive change,
emotional/behavioral problems, and family-related reasons) and a composite of all other reasons evidenced
significant variation in reason frequency across clusters, χ 2 (16, N = 237) = 40.57, p = 0.001.

Shelter services received or referred by cluster


χ 2 -Tests were conducted to compare patterns of services received or referred through shel-
ters across clusters (see Table 7). Overall, participants were most likely to report receiving
individual counseling (75% of the sample) and least likely to report receiving cash assis-
tance, substance abuse counseling, child-related services (parenting training or child care),
or government assistance (such as food stamps or social security). Destitute group mem-
bers were most likely to report receiving cash assistance (53%), while threatened group
members were most likely to receive medical services (78%). Pregnant group members
reported relatively higher rates of child-related services (63%) and clothing (100%), but
relatively lower rates of cash assistance (13%). Resistant group members reported rela-
tively lower rates of medical services (32%) and cash assistance (18%). Finally, partnered
group members reported higher rates of family counseling (73%) relative to other groups.
Vulnerable Children and Youth Studies 97

Table 7. Frequencies of services received (or referred) by cluster (N = 103).


Destitute Threatened Pregnant Resistant Partnered
(n = 32) (n = 18) (n = 8) (n = 34) (n = 11) Total

Service received n % n % n % n % n % n %

Family counseling∗ 13 41 3 17 3 38 16 47 8 73 43 42
Individual counseling 24 75 14 78 4 50 26 77 9 82 77 75
Drug and alcohol 12 38 4 22 1 13 14 41 7 64 38 37
counseling
Educational support 17 53 11 61 4 50 23 68 5 46 60 58
Child-related servicesa 10 31 6 33 5 63 9 27 2 18 32 31
Medical servicesb,∗ 16 50 14 78 4 50 11 32 6 55 51 50
Job servicesc 17 53 11 61 4 50 14 41 4 36 50 49
Housing servicesd 23 72 12 67 4 50 18 53 7 64 64 62
Clothing∗ 16 50 14 78 8 100 22 65 5 46 65 63
Transportation 18 56 10 56 4 50 15 44 7 64 54 52
Government servicese 15 47 10 56 1 13 8 23 3 27 37 36
Money∗∗ 17 53 9 50 1 13 6 18 2 18 35 34

Notes: Cluster distributions of categorical variables tested with χ 2 -Tests.


a Child-related services include child care and/or parenting training.
b Medical services include doctor/medical, prenatal care, prescriptions, and/or Medicaid.
c Job services include help finding jobs and/or job training.
d Housing services include temporary and/or permanent housing.
e Government services include food stamps, welfare benefits, social security, and/or Section 8.
∗ p < 0.05, ∗∗ p < 0.01.

Discussion
In this study, we used cluster analysis to identify typologies of homeless youth based on
circumstances or events driving the homeless episode. The results suggest a five-cluster
solution, each group endorsing a different pattern of reasons for being homeless: desti-
tute, threatened, pregnant, resistant, and partnered. Examination of demographic and social
information across clusters highlights important subgroup differences. Threatened, preg-
nant, and partnered groups were comprised of predominately girls and young women, while
resistant and destitute groups were comprised of similar numbers of male and female par-
ticipants. Second, African American youth were over six times more likely to endorse
lack of resources or family support as their primary reason for becoming homeless (as
compared with Caucasian youth), while Caucasian and African American youth were
equally likely to indicate problems with rules or conflict as reasons for becoming home-
less. Such differences reflect disparities that exist between men and women and between
Caucasians and African Americans more generally (DeNavas-Walt, Proctor, & Smith,
2010; Northridge, 2010). Greater percentages of girls and young women and African
American youth reported being homeless due to external influences and events rather than
from interpersonal conflict and behavior difficulties (relative to their Caucasian and male
counterparts).
To assess the robustness of identified clusters, we compared cluster membership (based
on quantitative ratings for reasons for homelessness) with youth-generated (qualitative)
reasons for service seeking. Although these assessments measure slightly different con-
structs (youth may become homeless because of abuse or family conflict, yet seek services
for a place to stay), distributions of reasons for service seeking generally supported cluster
membership. The destitute group generated the highest percentage of reasons reflecting
98 H.J. Heinze et al.

resource seeking and the pregnant group generated the highest percentage of child-related
reasons relative to other groups. Wanting to make a positive change was highly endorsed
by members of the threatened group, while those in the resistant group were least likely to
indicate wanting to make a positive change. Additionally, the resistant group produced the
highest percentage of family-related reasons for seeking services relative to other groups.
Finally, we examined services received or referred through shelters across clusters. The
findings provide some evidence for need-based service distribution. Most members of the
destitute group reported receiving housing services through shelters, and most members of
the pregnant group reported receiving child care and/or parenting training. Distributions of
services highlight potential service–need discrepancies. A substantial percentage of youth
were not receiving services through shelters that facilitate transition to independent living,
such as assistance accessing government programs and job seeking. This is particularly
problematic for youth with fewer family supports, such as destitute group members. Fewer
than half of the resistant group members reported receiving family counseling through
shelters, although they were most likely to report family-related reasons for seeking ser-
vices. Such gaps can alert providers and community stakeholders to resource and service
needs that may often go unmet.

Limitations
To lessen intrusiveness of data collection, information was collected only for participants
available and interested in participating. It is possible that youth who did not partici-
pate differed from participants in significant ways. Youth unable or unwilling to secure
parental consent may have limited family support or may have been more likely to have
left abusive or neglectful homes than those who easily secured consent. Similarly, not all
participants responded to every item. Most notably, relationship ratings were missing for a
substantial minority of participants, many of whom reported not having a male or female
caregiver. Distributions of demographic, family, and risk information across clusters should
be interpreted accordingly.
Study scales and items were created for use in this research, and, therefore, psychomet-
ric support is limited. Further, for service utilization, services received in addition to those
provided or referred by shelters were not assessed. Distributions of qualitative reasons for
seeking service and of services received or referred provide preliminary support for the
legitimacy of the identified clusters. However, assessment of reasons for homelessness and
clustering conducted within different settings and populations is necessary to strengthen
these findings.
In this study, we used a clustering algorithm to maximize differences across groups
in order to emphasize circumstances and need patterns unique to each group. The result-
ing distribution of participants limits our ability to detect group differences; only large
social and treatment differences emerging across groups were statistically significant.
Moreover, the small sample sizes of some groups constrain generalization of findings
beyond study participants. Future studies may consider recruitment at specialized shelters
or locations. Recruitment at family or domestic violence shelters could increase participant
representation within the threatened or pregnant groups.
Finally, reason patterns and subgroup differences were assessed only in youth accessing
shelter services. Reasons for homelessness and associated risk and protective charac-
teristics may differ for youth who do not stay in shelters. For example, youth in legal
trouble, estranged from caregivers, or concerned that they will be sent back home or
to foster care may be wary of shelter staff and related services. Previous studies have
Vulnerable Children and Youth Studies 99

demonstrated variation in reasons for running away among shelter and street youth; most
notably, street youth report higher rates of strictness, physical abuse, neglect, substance
abuse, and truancy than youth recruited at shelters (Kufeldt et al., 1992; Toro et al.,
2007). Analyses of reasons endorsed by street youth could therefore highlight different
or additional typologies.

Conclusions
The findings of this study support and extend previous research (Bucher, 2008; Mallett
et al., 2004; Milburn et al., 2009) that propose typologies of youth based on risk and pro-
tective characteristics. Similarities emerge across identified groups, despite differences in
purpose, sampling, and procedures. This study and Mallet et al. (2004) both identified
a partnered group (11% and 16% of the sample, respectively), with these youth in both
studies reporting higher rates of behavior problems, relative to other groups. Transgressive
youth have characteristics in common with our resistant group, with youth in both studies
distinguished by rebellious behavior, although varying in severity and nature (Mallet et al.,
2004). Perhaps youth who leave home due to difficulty following the rules are at greatest
risk to engage in delinquent behavior while homeless, which may further sever ties with
positive support systems, such as schools and positive adult role models.
Groups identified by Bucher (2008) and Milburn et al. (2009) are distinguished by
varying degrees of risk and protective factors. Milburn et al. (2009) suggest that groups
identified for risk and protective characteristics should be examined in relation to rea-
sons for leaving, noting that youth in their highest risk group were also most likely to
report abuse as a reason for leaving home. In this study, we formed clusters based on rea-
sons for leaving, then examined risk and protective behaviors across clusters. Percentages
of some risk behaviors, namely, substance abuse, gang activity, and convictions, varied
across groups, although these differences were not significant. A few groups appeared
to have greater percentages of protective factors in place. Most pregnant and partnered
group members who provided relationship ratings reported having positive relationships
with male caregivers. Several members of the threatened group listed wanting to make
a positive change as a primary reason for seeking services. Largely, however, results of
the current study do not indicate large differences across groups in terms of maladaptive
behavior and positive supports, suggesting that youth who leave home for similar reasons
can vary considerably in terms of risk.
Identification of subgroups of youth coming to shelters with similar background and
situational circumstances can inform targeted acquisition and distribution of services and
resources. Several studies have documented risk characteristics and maladaptive trajec-
tories associated with youth homelessness, yet most homeless youth are not accessing
services necessary to sufficiently address varied and extensive medical and psychosocial
needs (Thompson, Bender, Windsor, Cook, & Williams, 2010; Toro et al., 2007). Provision
of targeted services by providers they trust may be one way to address the need–service gap
among homeless youth.

Acknowledgements
The research reported in this article was supported in part by a grant from the W.K. Kellogg
Foundation Seed Grant program administered by the Merrill-Palmer Institute for Child and Family
Development, Detroit, MI. We thank the youth, staff, and administration at the participating agencies
who provided time, insight, and resources to make this project possible. We also thank the faculty and
100 H.J. Heinze et al.

students in the Research Group on Homelessness and Poverty at Wayne State University involved in
project planning, data collection, and dissemination: Bart Miles, Patrick Fowler, Carolyn Tompsett,
Mason Haber, Lindsey Sander, Karen Hobden, and Nicole Ouellette. Finally, we thank Laura Klem
and Joe Kazemi for their suggestions for the manuscript.

Notes
1. The decision was made to retain older participants in order to accurately reflect characteristics
and needs of clients seeking services within the shelters of the present and many other cities.
2. The decision was made to retain I got pregnant and I wanted to be with my partner as separate
variables. Although these items are correlated, pregnant and nonpregnant youth with or without
a partner could vary widely in terms of support and service needs.
3. Some participants left demographic, family, or risk items blank. For most items, information was
missing for six or fewer participants (missing data are presented by cluster in Table 4). Notably,
13 and 31 participants did not provide ratings for their relationships with primary female and
male caregivers, respectively. Prior to relationship ratings, youth were asked to indicate who their
primary caregivers are (e.g., biological mother, stepfather). Of those youth missing relationship
ratings, 9 and 21 youth checked “none” for primary female and male caregivers, respectively.

References
Bucher, C.E.C. (2008). Toward a needs-based typology of homeless youth. Journal of Adolescent
Health, 42, 549–554.
Cauce, A.M., Paradise, M., Ginzler, J.A., Embry, L., Morgan, C.J., Lohr, Y., & Theofelis, J. (2000).
The characteristics and mental health of homeless adolescents: Age and gender differences.
Journal of Emotional and Behavioral Disorders, 8, 230–239.
DeNavas-Walt, C., Proctor, B.D., & Smith, J.C. (2010). Income, poverty and health insurance cov-
erage in the United States: 2009. (US Census Bureau, Current Population Reports, P60-238).
Washington, DC: US Government Printing Office. Retrieved from http://www.census.gov/prod/
2010pubs/p60-238.pdf
Ennett, S.T., Federman, E.B., Bailey, S.L., Ringwalt, C.L., & Hubbard, M.L. (1999). HIV-risk behav-
iors association with homelessness characteristics in youth. Journal of Adolescent Health, 25,
344–353.
Farrow, J.A., Deisher, R.W., Brown, R., & Kulig, J.W. (1992). Health and health needs of homeless
and runaway youth. Journal of Adolescent Health, 13, 717–726.
Hammer, H., Finkelhor, D., & Sedlak, M.J. (2002). Runaway/thrownaway children: National esti-
mates and characteristics. National incidence studies of missing, abducted, runaway, and
thrownaway children. (Government No. NCJ 196469). Washington, DC: US Department of
Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.
Henry, D.B., Tolan, P.H., & Gorman-Smith, D. (2005). Cluster analysis in family psychology
research. Journal of Family Psychology, 19, 121–132.
Kufeldt, K., Durieux, M., Nimmo, M., & McDonald, M. (1992). Providing shelter for street youth:
Are we reaching those in need? Child Abuse & Neglect, 16, 187–199.
MacLean, M.G., Embry, L.E., & Cauce, A.M. (1999). Homeless adolescents’ paths to separation
from family: Comparison of family characteristics, psychological adjustment, and victimization.
Journal of Community Psychology, 27, 179–187.
Mallett, S., Rosenthal, D., Myers, P., Milburn, N., & Rotheram-Borus, M. (2004). Practicing
homelessness: A typology approach to young people’s daily routines. Journal of Adolescence,
27, 337–349.
McCaskill, P.A., Toro, P.A., & Wolfe, S.M. (1998). Homeless and matched housed adolescents: A
comparative study of psychopathology. Journal of Clinical Child Psychology, 27, 306–319.
Milburn, N., Liang, L., Lee, S., Rotheram-Borus, M.J., Rosenthal, D., Mallett, S., . . . Lester, P.
(2009). Who is doing well? A typology of newly homeless adolescents. Journal of Community
Psychology, 37, 135–147.
Milligan, G.W., & Cooper, M.C. (1987). Methodology review: Clustering methods. Applied
Psychological Measurement, 11, 329–354.
Vulnerable Children and Youth Studies 101

Mitchell, F. (2003). ‘Can I come home?’ The experience of young runaways contacting the Message
Home helpline. Child and Family Social Work, 8, 3–11.
Northridge, M.E. (Ed.). (2010). Adult health. American Journal of Public Health, 100(S1), 1827.
Paradise, M., & Cauce, A.M. (2002). Home street home: The interpersonal dimensions of adolescent
homelessness. Analyses of Social Issues and Public Policy (ASAP), 2, 223–238.
Peled, E., Spiro, S., & Dekel, R. (2002). Where do they go from here? Destinations of youth exiting
a shelter. Children and Youth Services Review, 24, 269–285.
Thompson, S.J., Bender, K., Windsor, L., Cook, M.S., & Williams, T. (2010). Homeless youth:
Characteristics, contributing factors, and service options. Journal of Human Behavior in the
Social Environment, 20, 193–217.
Toro, P.A., Dworsky, A., & Fowler, P.J. (2007). Homeless youth in the United States: Recent research
findings and intervention approaches. Toward understanding homelessness: The 2007 national
symposium on homelessness research. Washington, DC: US Department of Health and Human
Services.
Toro, P.A., Wolfe, S.M., & McCaskill, P.A. (2005). A comparison of homeless and matched housed
adolescents on family environment variables. In L.B. Alexander & P. Solomon (Eds.), The
research process in the social services: Behind the scenes. Belmont, CA: Thompson.
Whitbeck, L.B., & Hoyt, D.R. (1999). Nowhere to grow: Homeless and runaway adolescents and
their families. Hawthorne, NY: Aldine de Gruyter.
Whitbeck, L.B., Johnson, K.D., Hoyt, D.R., & Cauce, A.M. (2004). Mental disorder and comorbidity
among runaway and homeless adolescents. Journal of Adolescent Health, 35, 132–140.
Wolfe, S.M., Toro, P.A., & McCaskill, P.A. (1999). A comparison of homeless and matched housed
adolescents on family environment variables. Journal of Research on Adolescence, 9, 53–66.
Zide, M.R., & Cherry, A.L. (1992). A typology of runaway youths: An empirically based definition.
Child and Adolescent Social Work Journal, 9, 155–168.

View publication stats

You might also like