Touched by Homelessness: An Examination of Hospitality for the Down and Out

Abstract

Objectives. This study investigated patterns of “doubled-up” homelessness using an indirect measure based on host households.

Methods. In random household telephone surveys conducted in Alabama between 1990 and 2000 and nationally in 1997, respondents indicated whether any individual had stayed with them during the past year because that person was homeless.

Results. The percentage of Alabama households providing shelter during the past year declined from 16.2% in 1990 to 7.1% in 2000. The national rate for providing shelter in 1997 was 18.0%.

Conclusions. Many households provide shelter to people to prevent them from being literally homeless. As the economy has expanded, these rates have declined in Alabama.

Recent studies by Link and his colleagues1 demonstrate that the prevalence of homelessness in the United States is far greater than had been previously imagined. In addition to direct effects on homeless people themselves, however, homelessness may indirectly affect the lives of many others, particularly the people who provide temporary shelter for those who are near homelessness or who provide respite from the streets for those who are literally homeless. Rossi2 reported that a typical homeless family in Chicago spent approximately 4 years “doubled up” with family and friends before becoming literally homeless, and other studies have shown that between 46% and 82% of homeless families lived in doubled-up situations immediately before becoming literally homeless.3–6 Some studies7,8 suggest that people move back and forth between states of literal and doubled-up homelessness.

There is limited information about the prevalence of doubled-up living arrangements, however, with different studies estimating the ratio of people living in doubled-up situations to those literally homeless to be as low as 1:19 and as high as 20:1.10 Census statistics from 1990 place the number of single-parent families living in someone else's home at 2 million. McCallum et al.11 reported that more than 16% of respondents in a 1990 telephone survey of Alabama residents (n = 481) indicated that during the past year, an individual had stayed with them temporarily because that person was homeless. The current study sought to expand these results by examining data on doubled-up homelessness in Alabama between 1990 and 2000 and for the nation in 1997.

METHODS

We replicated the methodology used by McCallum et al.11 to produce 4 additional sets of data regarding doubled-up housing, 3 from random household telephone surveys of Alabama residents conducted between 1993 and 2000 and the other from a national random household telephone survey conducted in 1997.

In 1993 (n = 451), 1997 (n = 450), and 2000 (n = 504), the University of Alabama's Capstone Poll asked respondents from randomdigit-dialed samples of households in Alabama whether any individual had stayed with them temporarily during the past year because that person was homeless. As a follow-up, respondents who answered affirmatively were asked to describe several characteristics of the guests and their stay.

The national data were collected during a random-digit-dialed telephone survey (n = 1021) conducted by the Survey Research Center at the University of Maryland in 1997. This survey involved the same questions as the Alabama survey.

Data for all samples were weighted by sex, race, age, and education to adjust sample distributions to match population totals. Population totals were based on 1990 census reports for the Alabama samples and on the 1996 Current Population Survey for the national sample.12

RESULTS

The percentages of Alabama households that reported taking in someone who was homeless during the past year showed a linear decline over time, from 16.2% (95% confidence interval [CI] = 12.9%, 19.5%) in 1990 to 12.6% (95% CI = 9.5%, 15.7%) in 1993, 10.3% (95% CI = 7.5%, 13.1%) in 1997, and 7.1% (95% CI = 4.9%, 9.3%) in 2000. Differences between 1990 and 1997 (z = 2.65), between 1990 and 2000 (z = 4.35), and between 1993 and 2000 (z = 2.75) were statistically significant (P < .01 for all comparisons). This trend mirrored the economic improvements occurring over the same period. Somewhat surprisingly, 18.0% (95% CI = 15.6%, 20.4%) of respondents in the 1997 national survey reported taking in someone who was homeless—a rate nearly twice as high as the 1997 Alabama rate.

Table 1 ▶ reports characteristics of those persons who stayed doubled up from the national sample of survey respondents; the small numbers of households taking in guests in the Alabama samples prevented development of comparable estimates. In the vast majority of households in the national sample (73.2%), guests were adults without children. However, among a substantial segment of the national sample (21.7%), guests were families with children.

TABLE 1.

—Characteristics of People in Doubled-Up Arrangements: National Sample, 1997

	Sample
	No.	%	95% Confidence Interval
Family status
Adults	135	73.2	66.8, 79.6
Children	9	5.1	1.9, 8.3
Families	40	21.7	15.7, 27.7
Relationship
Related to respondent	81	44.0	36.8, 51.2
Related to someone else	21	11.5	6.9, 16.1
Close friend	54	29.4	22.8, 36.0
Acquaintance	13	6.9	3.2, 10.6
Stranger	10	5.6	2.3, 8.9
Other	5	2.7	0.4, 5.0
Length of stay
≤1 week	27	14.5	9.4, 19.6
>1 week to 1 month	41	22.4	16.3, 28.5
>1 month to 6 months	64	35.0	28.1, 41.9
>6 months to 1 year	27	14.6	9.5, 19.7
>1 year	25	13.4	8.5, 18.3

In addition, most of the people staying doubled up were either related to (55.5%) or a close friend of (29.4%) someone in the household where they stayed. Finally, the period of stay among those living doubled up was relatively long, with 35% of guests staying between 1 and 6 months. Neither very short nor very long stays were common; only 14.5% of guests stayed for a week or less, and 13.4% stayed for longer than a year.

Table 2 ▶ reports national differences in hospitality rates between households with annual incomes of less than $30 000 and those with annual incomes of more than $30 000. Lower-income households were more likely to provide shelter than higher-income households (χ²₁ = 13.19, P < .001). Lower-income households were also more likely than higher-income households to offer shelter to families with children (χ²₁ = 15.77, P < .001) and to offer shelter to friends and acquaintances (χ²₂ = 7.75, P < .05). There was no relationship between household income and length of stay. As before, similar comparisons were not made for the Alabama samples because of the small numbers of households that took in guests.

TABLE 2.

—Doubled-Up Shelter, by Income of Respondent Household: National Sample, 1997

	Income ≤ $30 000			Income > $30 000
	No.	%	95% Confidence Interval	No.	%	95% Confidence Interval
Provided shelter	69	26.1	20.8, 31.4	106	16.1	13.3, 18.9
Entire family	30	43.6	31.7, 53.5	17	16.7	9.5, 23.9
Relationship
Related to someone in household	31	44.4	32.6, 56.2	67	63.8	54.5, 73.1
Close friend	23	32.9	21.8, 44.0	26	24.6	16.4, 32.8
Acquaintance/stranger/other	15	22.6	12.7, 32.5	13	11.6	5.4, 17.8
Length of stay
≤1 month	23	32.9	21.6, 44.2	40	37.8	28.4, 47.2
>1 month to 6 months	24	35.6	24.0, 47.2	37	35.3	26.2, 44.4
>6 months	21	31.5	20.3, 42.7	28	26.9	18.3, 35.5

DISCUSSION

Large numbers of people across the nation have hosted homeless individuals or families; extrapolating our results to the national population, in 1997 approximately 18 million households provided temporary accommodations for people who would otherwise have been homeless. Lower-income households were more likely than higher-income households to host those in need of shelter. While the majority of guests were single adults, 26.8% of the host households in the national sample provided shelter to children (either alone or in families). Lower-income households were more likely than higher-income households to take in families with children.

Because the economy has expanded over the past decade, we might expect to see the prevalence of doubled-up homelessness decrease and fewer people indirectly affected by it; this is exactly what we observed in the Alabama data, with rates of doubled-up housing decreasing from 16.2% in 1990 to 7.1% in 2000. Thus, a booming economy might reduce the number of people turning to doubled-up housing.

Cross-sectional time-series data such as those reported here allow indirect investigation of how doubled-up housing rates change as welfare reform runs its course and families lose their eligibility for Temporary Assistance for Needy Families, or as economic declines and expansions occur. Thus, it is useful to continue collecting these data to assess the impact of systemic changes on risk of near homelessness and as a potential leading indicator of alterations in the prevalence of literal homelessness in our society.