Age differences among homeless individuals: adolescence through adulthood

Abstract

The present study examines differences between homeless adolescents, young adults, and older adults served by homeless shelters or food programs to inform service provision. Four homeless studies using the same sampling and measurement methods were pooled to permit comparisons across age groups. Results showed that homeless adolescents demonstrated greater resilience than younger and older adults. Adolescents reported the shortest duration of homelessness, lowest number of life stressors, fewest physical symptoms, largest social networks, and fewest clinically significant mental health problems. Adolescents also received fewer alcohol and drug abuse diagnoses than younger and older adults. Younger adults reported less time homeless and fewer physical symptoms than older adults, but more life stressors. Younger adults also endorsed higher levels of hostile and paranoid psychological symptoms. Implications for provision and policy are discussed.

Many different kinds of people become homeless, leading to a diverse homeless population. Agencies serving homeless individuals have responded to this diversity in a number of ways, primarily by targeting their services to a specific sub-group or by offering different programs of services to different sub-groups of the homeless population. With limited funds for services, it is important that agencies targeting a specific age group of homeless include services that are most salient to their target group. There is some evidence that youth are better served by agencies specifically designed for them, rather than by adult-focused agencies, and it is likely that closer attention to differences among age groups will only lead to improved services from any age-targeted agency (Fitzpatrick, 2000). In addition, efforts to prevent homelessness may also benefit from attention to these demographic differences.

Experiences of Homelessness

Existing research on differences between age groups within the homeless population is incomplete. Some studies compare homeless individuals with housed individuals (of the same approximate age; e.g. Shinn, Knickman, & Weitzman, 1991), but few studies compare across age groups within the homeless population. The limited research on the topic alludes to some important patterns of age differences. To begin with, some of the basic experience of homelessness may be influenced by the age of the individual such that older individuals tend to be homeless longer or experience more episodes of homelessness than younger homeless individuals (Calsyn & Roades, 1994; Hecht & Coyle, 2001).

Physical problems experienced by homeless individuals may also vary by age. While homeless adolescents do tend to have more physical health problems than housed adolescents, within the homeless population age is associated with increased physical health problems (Garibaldi et al., 2005; Moore, 2005). It is possible that this increase is simply due to the natural aging process, although it seems more likely that the stresses associated with homelessness may take a greater toll on older individuals. Certainly, repeated experiences of homelessness are associated with higher rates of health problems, and to the extent that older individuals may experience more chronic homelessness, they are at increased risk of health problems (Goering, Tolomiczenko, Sheldon, Boydell, & Wasylenki, 2002). Although some suggest that older and/or more street-experienced individuals may have a greater suspicion of health providers and, therefore, receive inadequate health care, others emphasize that this suspicion is likely more characteristic of adolescents. Some research has found that older adults actually report more health service use than younger adults (Garibaldi et al., 2005; Robertson & Toro, 1999). It is possible that while older adults are more likely to use health services, the effects of their age and longer history of homelessness may contribute to their developing more physical problems.

Mental health needs are often the focus of service providers, and for good reason. A recent survey of homeless individuals using services indicates that approximately 39% of all homeless people experience mental health problems, not including substance abuse (Burt et al., 2001). While research indicates that both homeless adults and adolescents experience more mental health problems than their housed counterparts, it is unclear whether there is an age difference among homeless individuals (Burt et al., 2001; Whitbeck & Hoyt, 1999). Some evidence suggests that among homeless adults, no differences in mental health symptoms emerge between older and younger adults (DeMallie et al., 1997; Hecht & Coyle, 2001).

Few, if any, studies exist that specifically contrast rates of substance abuse among homeless adolescents and adults, although most researchers agree that substance abuse is a problem worth noting among any subgroups of the homeless population (Burt et al., 2001; Moore, 2005). Some suggest that younger homeless adults are more likely to use drugs than older or elderly homeless adults, while older adults may be more likely to use alcohol, although other studies fail to find such differences (DeMallie et al., 1997; Hecht & Coyle, 2001). While it seems probable that adolescents, like younger homeless adults, may be more prone to drug problems than older adults, at this time research in this area is incomplete. More information on the prevalence of drug and alcohol problems among sub-groups of the homeless population may help service providers target resources more effectively.

Finally, age differences in social support might also influence the approaches taken by agencies serving homeless clients. Studies of both homeless adults and adolescents yield conflicting findings on social support. While some find that homeless individuals are likely to have smaller social networks than the housed population (Bassuk et al., 1997), others do not (Shinn, Knickman, & Weitzman, 1991). In particular, studies of homeless adolescents often find that adolescents maintain some ties with friends and family despite their homeless status (Johnson, Whitbeck, & Hoyt, 2005; Whitbeck & Hoyt, 1999). Length of time homeless is sometimes inversely related to social support (Eyrich, Pollio, & North, 2003), sometimes not (Unger et al., 1998). Homeless individuals of any age are more likely to report a history of physical and/or sexual abuse (Bassuk et al., 1997; Cauce et al., 2000; Whitbeck & Hoyt, 1999), although family conflict and abuse more often appears as a proximal cause of homelessness in the literature on homeless adolescents. None of these studies offer support for any clear-cut hypotheses about differences in social support experienced by homeless individuals of different ages.

The current study seeks to clarify questions about differences between homeless adolescents, young adults, and older adults served by homeless shelters or food programs. The experience of homelessness itself, including time spent homeless and recent stressors are contrasted, as are the personal characteristics of physical health, psychological symptoms, and substance abuse. Finally, differences in social support are examined through measures of current social networks. It was difficult to formulate strong hypotheses due to the scarcity of prior studies examining age differences and conflicting findings where studies did exist. We predicted that older adults would demonstrate longer duration of homelessness and more physical health problems than younger adults and adolescents. We expected that adolescents would more likely to demonstrate mental health problems due to evidence identifying family conflict and abuse as a common cause of adolescent homelessness, but that adults would report higher rates of substance abuse. We could not predict whether age differences in social network variables would emerge.

Methods

Sample

Four separate studies, conducted by the same research group in the same mid-sized Midwestern city over the course of 10 years (1992–2002), were pooled to obtain the largest possible sample that covered the broadest age range. All four studies used similar probability sampling methods to obtain broad samples of homeless individuals (see Toro, et al. (1999)). Two of the studies collected data from homeless adolescents (Studies 1 and 3), recruiting their samples primarily from shelters that targeted single homeless adolescents. The other two studies collected data from homeless adults (Studies 2 and 4), recruiting their samples from shelters and food programs that targeted adult homeless individuals, and homeless families. Studies 2 and 4 used only data from homeless adults, and in shelters where families with children were present did not collect data from youth under 18. Demographic characteristics of each age group are presented in Table 1.

Table 1.

Demographic characteristics of age groups

			Race
Group	Age range	N	Gender (%male)	Caucasian	African American	Other
Adolescent	13–17	363	32.5%	46.3%	43.8%	9.9%
Young adult	18–34	157	63.1%	15.9%	77.1%	7.0%
Older adult	35–78	310	78.1%	11.0%	86.1%	2.9%

Measures

Time homeless

A single item from the Housing, Income and Services Timeline was used to determine the duration of the current episode of homelessness in each of the 4 samples (Toro et al., 1995). Respondents indicated the length of time they had currently been homeless from 0=1–7 days to 6= more than 1 year. Adult respondents had an additional 3 options specifying time periods between 1 year and over 3 years, however these options were collapsed into the “more than 1 year” category.

Stressful Life Events

A sum of reported stressful life events was calculated using 52 items common across all four studies. All studies used variations of the Modified Life Events Interview (MLEI; (Lovell, 1984), a scale designed to determine the levels of stress experienced by homeless adults. Items tap recent stressful life events in five domains: social relationships, housing situations, employment, education/job training, and mental and physical health.

Physical health

A Physical Health Checklist was used to create a sum of reported physical health symptoms (Toro & Wall, 1991). Because the checklist was updated between studies, only items common across all four studies were included in the sum. The common items included a wide variety of chronic and acute symptoms, with a possible physical health symptom count ranging from 0–40.

Psychological symptoms

Psychological symptoms were assessed using the Brief Symptom Inventory (BSI; (Derogatis, 1993). Each symptom is rated on the amount of distress it has caused the respondent in the last week from 0 (not at all) to 4 (extremely). Nine subscales for clusters of symptoms are calculated. Each subscale score ranges from 0–4. Data for separate normative samples of non-patient adolescents and adults are provided to enable the conversion of raw subscale scores to T-scores. T-scores represent the score of an individual relative to the general population of the same age group and gender, on a scale from 1–100 with a mean of 50 and a standard deviation of 10. T-scores above 63 are considered clinically significant (Derogatis, 1993).

Substance abuse

Substance abuse was assessed using modules of the Diagnostic Interview Schedule for Children for Studies 1 and 2 (Fisher, Wicks, Shaffer, Piacentini, & Lapkin, 1992), and the Diagnostic Interview Schedule for Studies 3 and 4 (Robins et al., 1999). Each of these measures produces a diagnosis for alcohol abuse, alcohol dependence, other substance abuse, and other substance dependence (where “other substance” could refer to any illegal drug). Substance abuse items differ between the DISC and the DIS due to age-specific concerns. For example, the DISC includes more items assessing the onset and frequency of alcohol use as use in itself is considered problematic among children under 17, the DISC also includes wording that assesses impairment at work or at school. Due to low base rates of substance dependence among adolescents, only the abuse diagnoses were used. Diagnostic criteria for abuse are the same across the DISC and the DIS: at least one symptom in each of 2 or more problem areas, with problem areas including impairment of functioning at work, school or in relationships, recurrent use in hazardous situations, substance-related legal problems, and persistent use despite any of the aforementioned problems. For the purposes of the current study, those who in addition to meeting abuse diagnostic criteria also met criteria for dependence (by evidencing tolerance and/or withdrawal symptoms) were simply given the diagnosis of abuse. As a result, it was decided that diagnoses between adolescents and adults are comparable, even while using the more age-sensitive measures.

Social Network Interview

All studies used the same form of the Social Network Interview, which was designed to measure the size of the social support network of interviewees (Bates & Toro, 1999). Each study participant was asked to list the people who were important in their everyday lives and with whom they have had contact in the past six months. Then, for each named individual, the participant was asked to classify them as friend or family, and rate how frequently they had contact with this individual from 1=about daily to 5=about once a year or less. Four scales were created: total number of family members, total number of friends, average frequency of seeing family, average frequency of seeing friends. One discrepancy among studies emerged: for all studies except Study 3, if participants named more than 10 family members and 10 friends, they were asked, “if we stopped now, would we be missing anyone important”, and only then would additional network members be coded. Study 3 participants were allowed to volunteer an unlimited number of network members without interruption. As a result, when total number of family and total number of friends were calculated, they were each truncated at 10 members. The average frequency of contact scales were examined using correlated-samples t-tests to determine if this discrepancy in instructions would impact these ratings: it was found that Study 3 members who rated more than 10 family members gave higher frequency of contact ratings to the first 10 rated than to those rated later (t(51)=3.70, p<.001). No frequency differences were found for friends, and among the few respondents from the other studies who listed more than 10 network members in either category, no differences in frequency of contact were found between the first 10 and last 10. As the only difference in frequency found actually increased Type II error (see Results), average frequency of contact scores were calculated using ALL named members.

Procedure

For all four studies, researchers visited specific sites identified by key informants as having significant numbers of homeless people. All potential sites within the metropolitan area were sampled; that is for Studies 1 and 3 all sites serving single homeless adolescents were visited, and for Studies 2 and 4 all shelters and most food programs serving single homeless adults were visited. At each site, individuals were randomly chosen to participate in the full interview by a pre-determined order (e.g., first person in line one day, second person the next day). Brief sampling surveys were administered to determine if a potential participant was currently homeless before administering the full-length interview (especially important at soup kitchens where many poor, but not homeless, people could often be found).

Data from the four samples were pooled, and divided into three age groups: adolescent (age 13–17; N=363), young adult (18–34; N=157), and older adult (35–78; N=310). Chi-square analyses indicate that there were significant differences by age group in distributions of gender (χ²(2)=145.11, p<.01) and race (χ²(4)=146.29, p<.01). Demographic differences between age groups could be due to two factors: differences in sampling, or underlying differences in the population. Because this sample was recruited from agencies targeting adolescents vs. adults, it is possible that agencies serving adolescents are more welcoming for females than are agencies for adults; or that these agencies serving adolescents are located in neighborhoods with a greater Caucasian representation. However, some evidence suggests that at least among adults, older homeless individuals are more likely to be male, indicating that some of the demographic differences may have reflected actual population differences (DeMallie et al., 1997). Pairwise comparisons of the two adult groups indicated that there were significantly more males in the older group (χ²(1)=11.91, p<.01), and that there were more African Americans relative to two other race groups among the older adults (χ²(2)=7.07, p<.05). As all adults were recruited from the same locations, these differences indicate that sex and race differences among age groups might be due to actual differences among these subpopulations.

As the focus of the current paper is on age differences, and age to some extent appears to be correlated with gender and race, all analyses of age differences controlled for gender and race. Many outcome variables were significantly positively skewed, and were transformed by either taking the square root of each score (outcome variables: stress, physical health, number of friends), or by taking the natural logarithmic value for each score (all BSI subscales) to create normal distributions for each variable.

Results

Results of three multivariate analyses of variance (MANOVAs) examining age differences are presented in Tables 2 and 3. Older adults reported the longest duration of homelessness as well as the highest number of physical symptoms, followed by younger adults, followed by adolescents. Young adults reported the highest number of life stressors, followed by older adults, with adolescents reporting the lowest number of stressors. While older adults may experience longer episodes of homelessness, it appears that younger adults may experience greater chaos in their lives.

Table 2.

Age differences on homeless experience and social support

Mean
	Adolescent (A)	Young Adult (Y)	Older Adult (O)	Univariate F	Post-hoc
Homeless experiencea
Time homeless (0–6)	1.46	3.98	4.60	91.22 (1, 790)***	O>Y>A
Life stress	11.50	16.53	14.88	13.94 (1, 790)***	Y>O>A
Physical symptoms	6.43	8.56	10.00	13.76 (1, 790)***	O>Y>A
Social supportb
Number of family	5.24	3.60	3.22	2.49 (2, 602)	N/A
Number of friends	4.22	1.99	2.04	10.21 (2, 602)***	A>O, Y
Frequency of family contact	4.18	3.67	3.53	28.99 (2, 602)***	A>Y, O
Frequency of friends contact	4.52	4.08	4.10	17.13 (2, 602)***	A>O, Y

Note. Age differences control for gender and race. Means presented are for original, non-transformed scales. F values reflect variables transformed to correct skew.

^aMANOVA F=32.47 (6, 1578), p<.001

^bMANOVA F=13.98 (8, 1200), p<.001

^*p<.05,

^**p<.01,

^***p<.001

Table 3.

Age differences on mental health^a

Mean
	Adolescent (A)	Young Adult (Y)	Older Adult (O)	Univariate F	Post- hoc
Brief Symptom Inventory age and gender-normed T- scoresb
Depression	47.94	67.18	68.69	54.10 (2, 782)***	O, Y>A
Anxiety	51.64	62.77	62.34	17.44 (2, 782)***	O, Y>A
Somatization	50.95	54.60	58.62	10.75 (2, 782)***	O>Y>A
Obsessive-Compulsive	50.96	62.00	60.98	21.77 (2, 782)***	Y, O>A
Interpersonal Sensitivity	48.44	64.15	62.80	34.06 (2, 782)***	Y, O>A
Hostility	54.93	65.82	57.60	15.49 (2, 782)***	Y>A, O
Phobic Anxiety	51.13	63.85	63.68	19.08 (2, 782)***	O, Y>A
Paranoid Ideation	54.42	72.32	67.71	50.56 (2, 782)***	Y>O>A
Psychoticism	51.24	73.13	71.43	44.18 (2, 782)***	Y, O>A

^aAge differences controlling for gender and race

^bMANOVA F=10.40 (18, 1550), p<.001;

^*p<.05,

^**p<.01,

^***p<.001

Adolescents reported more friends, although not more family members in their social network, and reported greater frequency of contact with both friends and family members than either adult group. These results indicate that although homeless adolescents may experience more family dysfunction than their housed age-mates, when compared to homeless adults they are more likely to have regular contact with a wide network of friends and family.

By analyzing the BSI T-scores (rather than raw scores), we were able to examine whether homeless adolescents are more dramatically worse off compared to the general population of adolescents than homeless adults are when compared to the general population of adults. Adolescents scored significantly lower than both adult groups on all BSI subscales. These results indicate that homeless adolescents report rates of mental health problems that are closer to the general population of adolescents, compared with homeless adults who demonstrate a greater deviation from the general population of adults. In fact, all of the adolescent subscale means fell below the clinical cut-off of T=63, while all of the adult means fell at or above this cut-off.

Rates of alcohol abuse diagnosis were as follows: 10.9% of adolescents, 50.6% of young adults, 56.5% of older adults. Drug abuse rates were: 19.0% of adolescents, 47.4% of young adults, 56.3% of older adults. Logistic regression analysis was used to examine the effects of age group on diagnosis of drug abuse or alcohol abuse, independent of the effects of race and sex. Young adults and older adults were contrasted with adolescents as the reference group. Young adults demonstrated significantly higher rates of diagnosis of alcohol abuse (B=2.03, OR=7.63, p<.01) as well as drug abuse (B=1.35, OR=3.86, p<.01). Older adults also showed more diagnoses than adolescents, with higher rates of alcohol abuse (B=2.18, OR=8.67, p<.01) and drug abuse (B=1.64, OR=5.13, p<.01). Logistic regressions were re-analyzed with younger adults as the reference group to permit comparison between the two groups of adults; no significant differences were found between young and older adults on either diagnosis.

Discussion

Homeless adolescents, young homeless adults, and older homeless adults all face challenges that are not faced by housed individuals their own age. When comparing the different age groups of homeless individuals, it is important to bear in mind that within each age group there will always be individuals who have needs more typical of older or younger individuals, and so all agencies must be prepared to work with a variety of needs. However, our results indicate that there are some significant differences between homeless individuals of different ages. We suggest that these differences have ramifications for future research with homeless populations, and for service providers.

Different age groups may have fundamentally different experiences of homelessness due to differences in duration of homelessness. As seen in previous research, adolescents reported significantly shorter duration of the current homeless episode, indicating that shelters serving adults are more likely to have long-term clients, while youth-focused shelters should be prepared for a high rate of turnover in clientele (Calsyn & Roades, 1994; Hecht & Coyle, 2001). Adolescents may be less likely to remain homeless for long periods due to their different legal status; as minors they are required to be supervised by a parent or guardian and so may experience more pressure to return to their prior living situation or to seek out some other alternative to homelessness. In addition, social institutions, especially schools and extended family networks, may monitor the well-being of adolescents and be more likely to address housing instability. Fewer institutions may protect homeless adults. It is less clear why older adults report longer duration of homelessness than younger adults. Other studies have proposed that older adults may have fewer social resources than younger adults, but the current study failed to find differences in social network size (Calsyn & Roades, 1994; Hecht & Coyle, 2001). It is also possible that older homeless adults may face greater obstacles, including age discrimination, in obtaining employment or stable housing, or that older homeless adults have had more time to experience an accumulation of chronic disadvantages that impede their ability to exit homelessness.

Younger adults reported the highest rates of recent stressful life events, indicating that while their duration of homelessness tends to be shorter than older adults’, they may experience more chaos in their lives around the time of their homeless episode. During early adulthood, individuals have to navigate a number of challenges such as establishing independence, going to college and/or obtaining gainful employment, and beginning important romantic relationships. It appears that this age group also experiences higher rates of stressors within the homeless population.

Higher rates of physical health symptoms among older homeless adults are congruent with previous findings among homeless researchers as well as in the general literature. Older adults are at higher risk for many of the symptoms included on the physical health checklist, including hearing loss, joint pain or stiffness, high blood pressure and diabetes, to name a few. As Garibaldi and colleagues (2005) point out, older homeless adults are more prone to health problems regardless of their access to health services. However, service providers should be aware that this group is more likely to need assistance with physical health needs, and should be prepared to facilitate access to health services.

Mental health findings were surprising in that adolescents demonstrated less clinically severe problems across all measured subscales. This trend was unexpected in light of the literature emphasizing the high rates of abuse and family dysfunction among homeless adolescents, and the relatively greater attention to economic (rather than psychological) factors causing adult homelessness. Adolescents may demonstrate more resilience in the face of pathogenic stressors, or adults may be subject to more psychological stressors than previously recognized. Nonetheless, our results indicate that older adults have higher rates of somatization, which is consistent with the greater experiences of physical health problems among older adults. Younger adults reported higher rates of hostility, and greater paranoid ideation than both adolescents and older adults, suggesting that they also may be more challenging to work with in a shelter setting.

Diagnoses of alcohol and drug abuse were sensitive to age-based normative levels by using age-specific diagnostic measures. Despite this greater sensitivity in the adolescent measures, adults reported higher rates of diagnosis for both alcohol and drug abuse. Contrary to some prior findings with homeless adults, no differences in abuse were found between young and older adults (DeMallie et al., 1997; Hecht & Coyle, 2001). It is possible that the most elderly homeless individuals (over age 60) would differ from the rest of adult homeless, but unfortunately the current sample included too few elderly homeless individuals to further test this theory (probably because such “elderly homeless” are, in fact, rare). Rates of substance abuse diagnoses were somewhat lower among the adolescent population than has been reported in previous studies of homeless adolescents, which is likely due to the use of a formal diagnosis rather than simply the presence of any symptoms of drug or alcohol abuse. Adolescents who abuse drugs or alcohol are at high risk for continuing to have substance abuse problems in adulthood, and so even these relatively low rates of diagnosis are cause for concern. Finally, it is likely that, by using a sheltered population, our sample reported lower rates of substance abuse than would be found among homeless individuals living on the street, as most shelters do not permit alcohol or drug use or intoxicated individuals on their premises.

While both groups reported similar numbers of family members, adolescents reported having more friends and a greater frequency of contact with both family and friends. These results cohere with evidence that adolescents maintain social networks even when homeless, and suggest that social isolation may be more of a problem for homeless adults than adolescents (Johnson et al., 2005; Whitbeck & Hoyt, 1999). It should be noted that the current study was not able to examine characteristics of network members, and so results of social network analyses are incomplete. While adolescents clearly enjoy greater contact with social network members, it is not clear whether these members provide support or are a source of stress (Bao et al., 2000). It is possible that network members who are in frequent conflict or who engage in antisocial behaviors may be a source of stress for homeless adolescents, or may influence them in other negative ways. Service providers should be aware that they may need to provide homeless adults with more assistance in alleviating social isolation and making social contacts, but should be careful to recognize that homeless adolescents may also need help in managing the social contacts that they have.

Several strengths of the current study increase the utility of the findings. Adolescents and adults were sampled in the same way, providing a large representative sample of homeless individuals using services within the same metropolitan area. Most of the same measures were used, providing comparable data across all age groups. However, limitations also exist in the sample used. By sampling only those individuals who used services, findings cannot be expected to generalize to homeless individuals who live doubled up with friends or family, or to those individuals who avoid services and live solely on the streets. In addition, the cross-sectional nature of the study limits our understanding of risk factors, as does the limited knowledge of the history of respondents. Future research using longitudinal methods could allow a closer examination of the risk factors that lead to homelessness, or that lead to different outcomes among homeless individuals. In particular, future research with homeless adults should include more information about earlier experiences with homelessness, to determine the extent to which adolescent experiences with homelessness contribute to later adult outcomes.

Despite these limitations, the current findings can help shape the work both of future researchers and current service providers. Most agencies serving the homeless population do currently specialize in one age group or another. Recognizing that homeless individuals of different ages share many needs but also have some differing strengths and weaknesses can help service providers better allocate their resources to provide optimal service. For example, all homeless individuals need help finding a stable place to live, and any agency should at least be prepared to provide referrals for help with substance abuse, mental health, and physical health needs. However, a recognition that homeless adults are in some ways even more troubled than homeless adolescents can lead to greater sensitivity among agencies serving adults. While much work remains to be done in the field of homeless research, a better understanding of differences among sub-groups of the homeless population is an important step towards better services and preventive efforts.