Program Use by Families in the Project-Based

문서에서 Family Options Study (페이지 126-135)

Chapter 8. Impacts of Project-Based Transitional Housing

8.2 Program Use by Families in the Project-Based

(PBTH) Versus Usual Care (UC) Comparison

Each impact comparison in the study may be thought of as a distinct experiment or test, and this chapter addresses only the comparison between PBTH and UC, without reference to the families who were randomized to the permanent housing subsidy (SUB) or community-based rapid re-housing (CBRR) interventions. In total, 707 families took part in the test of PBTH versus UC. These families all had the opportunity to be assigned to PBTH or UC at the point of random assignment and were assigned to one of these two interventions—368 families to PBTH and 339 families to UC.110 Seventy-nine percent of these 707 families (294 PBTH families and 262 UC families) responded to the 18-month followup survey, and thus a total of 556 families are included in the PBTH-versus-UC impact comparison reported in this chapter. This section describes the extent to which the 294 PBTH families used transitional housing and other types of homeless and housing assistance during the followup period. Parallel in-formation is presented for the 262 UC families. The data on program use do not distinguish between subtypes of transi-tional housing and include transition-in-place assistance, so this section uses the abbreviation “TH” rather than PBTH to describe the broader category of assistance.

Exhibit 8-5. Types of Program Rules in PBTH

Types of Program Rules Percent of Families Referred to PBTH Programs With These Types of Rules (N = 293)a

Weekday curfew 52

Weekend curfew 10

Limit on daytime visitors 9

Limit on overnight visitors 42

Compliance with mandatory service requirements 87

PBTH = project-based transitional housing.

a Program data on client ratios were not collected from 17 programs that collectively had 83 PBTH family referrals.

Note: Percentages are unweighted.

Sources: Program data; random assignment records

109 The percentages of study emergency shelters with these types of rules are shown in Gubits et al. (2013), Exhibit 2-19. In general, study emergency shelters were more likely to have these types of rules than study PBTH programs. Most UC families had weekday curfews (93 percent), limits on daytime visitors (70 percent), and limits on overnight visitors (96 percent) in the study shelters from which they enrolled in the study.

110 In the entire study, 746 families were randomly assigned to UC, but only 339 of these families had PBTH available to them when they were randomized. Therefore, only these 339 UC families are part of the PBTH-versus-UC comparison sample. All 368 families randomly assigned to PBTH during the course of the study had UC available to them, so all are part of the PBTH-versus-UC comparison sample.

Exhibit 8-6 shows the use of seven types of homeless and housing programs by these families. The first column shows the percentage of families assigned to PBTH who ever used each program type during the followup period.111 The third row (shaded in the exhibit) shows the use of some type of TH by the families assigned to PBTH; 53.6 percent of fam-ilies in PBTH received TH assistance at some point during the followup period—meaning that they either followed up on the referral and moved into the PBTH facility or entered another TH program.112, 113

The second column shows the percentage of families assigned to UC who ever used each program type during the followup period.114 The shaded row of the second column shows that 29.1 percent of the UC families received TH assistance during the followup period, despite not being given priority access to PBTH. Emergency shelter staff were requested by the study to not refer families assigned to UC to one of the active interventions. Nevertheless, as shown in the exhibit,

families were able to learn about transitional housing pro - grams in their communities, and these programs had program slots available at some point during the followup period.

Rows 1, 2, and 4 through 7 of the exhibit show participation in other types of homeless and housing programs. Twenty-five percent of these UC families (not shown in the exhibit) found their way to SUB or other permanent housing programs during the followup period, presumably through the regular process of coming off waiting lists and leasing units, and 12 percent found their way to rapid re-housing assistance, despite the lack of preferential access to those programs through the study. The use of programs other than TH is generally higher for the UC group than for the PBTH group, presumably because the UC group did not have the PBTH intervention easily available and so turned to other types of programs. The eighth row shows the percentages of families in the PBTH and UC groups who used none of the seven

Exhibit 8-6. PBTH Versus UC: Program Use Since Random Assignment

Type of Housing Assistance

Percent Ever Used From RA to 18-Month

Followup Surveya

Number of Months Used From RA to 18-Month Followup Survey, if Ever Used Type of Housing Assistance

Percent Used in Month of Followup

Survey Response

PBTH UC PBTH UC PBTH UC

Mean Median Mean Median

Subsidy (SUB)b 5.7 6.9 9.4 8.5 12.0 9.5 4.6 5.7

Rapid re-housing (CBRR) 10.1 12.2 7.6 5.5 6.4 4.5 2.2 2.1

Transitional housingc 53.6 29.1 11.5 11.5 8.4 6.0 21.7 9.8

Permanent supportive housing 6.4 7.8 9.2 8.5 9.1 7.5 3.9 5.6

Public housing 4.7 4.9 9.8 10.5 10.6 11.5 4.1 4.2

Project-based vouchers/Section 8 projects 4.4 7.0 10.8 12.5 15.2 15.5 3.4 6.3

Emergency shelterd 82.9 88.4 3.3 2.0 4.2 2.3 4.7 9.5

No use of homeless or housing programse 20.4 30.4 19.0 18.0 19.6 19.0 56.7 57.4

N 294 262 294 262

PBTH = project-based transitional housing. UC = usual care.

RA = random assignment.

a Percentage of families who ever used a type of assistance program during the period from the month of RA to the month of the 18-month followup survey response (median period duration: 21 months). Percentages do not add to 100 because some families used more than one program type during the followup period.

b Subsidy assistance is housing choice vouchers plus site-specific programs offered to families assigned to the permanent housing subsidy (SUB) group in Bridgeport, Connecticut and Honolulu, Hawaii.

c Includes use of project-based, scattered-site, and transition-in-place transitional housing. The inclusion of transition-in-place assistance makes this category broader than the study-defined PBTH intervention.

d All families were in emergency shelter at random assignment. Percentages less than 100 percent for ever used emergency shelter are due to missing data on shelter use.

e No use of homeless or housing programs (ever used) indicates no use of the seven program types in this table during any of the followup period and no use of emergency shelter after the first 6 months after RA. No use in the month of the followup survey response indicates no use of any of these seven program types.

Notes: Percentages are regression adjusted, controlling for site and randomization ratio. Percentages, means, and medians are weighted for survey nonresponse to represent full comparison sample.

Source: Family Options Study Program Usage Data

111 The followup period is from the calendar month of random assignment through the calendar month of response to the 18-month followup survey. Therefore, the length of the followup period differs across families. This period lasts for a median of 21 calendar months for the full sample.

112 All percentages, means, and medians in the exhibit are weighted to adjust for survey nonresponse and hence as best possible represent the full experimental sample of 707 families. The findings here on program use are thus in line with similarly weighted impact estimates provided later in the chapter.

113 The unweighted number of PBTH families who used TH during the followup period is 159 families. Of these, 92 families were confirmed by enrollment verification to have used the program to which they were referred by the study. It is not known how many of the other 67 families received PBTH or some other form of TH.

114 The percentages in the first six rows of these columns are not mutually exclusive because some families use more than one program type during the followup period.

Short-Term Impacts of Housing and Services Interventions for Homeless Families 96 Compared With Usual Care (UC)

types of programs during the followup period, nor used emergency shelter from the seventh month after random assignment onward. About 20 percent of PBTH families and 30 percent of UC families fall into this group.

The mean and median number of months of use for each program type are also shown in the exhibit (third and fourth columns for PBTH families, fifth and sixth columns for UC families) for only those families who ever used a given program type.115 As one might expect, given that TH was readily avail - able to PBTH families, the number of months of TH use is higher for the families who had priority access to PBTH (median of 12 months) than for the 29.1 percent of UC families who found their way to TH (median of 6 months) at some point during the followup period.

The last two columns of Exhibit 8-6 show the percentages of PBTH and UC families who received each type of program in the calendar month of the followup survey response. Although the study team expects most outcomes in the report to be influenced by assistance received during the entire followup period, some outcomes will be particularly strongly influ-enced by assistance received at the time of followup survey response. The shaded row of the seventh column shows that the TH assistance had ended for about 60 percent of the PBTH families who ever received it and for 66 percent of the UC families who ever received it.116 About one-fifth (22 percent) of PBTH families were still in TH in the month of followup survey response compared with one-tenth of UC families. The majority of both PBTH and UC families (57 percent each) were not using a homeless or housing program at the time they responded to the followup survey.

Thus, substantial differences are expected not in outcomes of PBTH and UC families in areas that reflect the families’

current experience, but only in those that reflect a lasting influence of families having received the intensive case management and services provided by PBTH.

As Exhibit 8-6 makes clear, the PBTH families used a range of programs in addition to the program to which they were referred by the study, which is consistent with the design of the study. Families were not required to use the intervention

to which they were assigned and also were not forbidden from using other programs that were available to them in their community. The intent of the study was to maximize use of the assigned active intervention (in this case, maximize use of the PBTH intervention by the PBTH families) and to create the largest possible contrast between the program mixes of different assignment groups (in this case, PBTH versus UC). As shown in the exhibit, the use of PBTH was quite different for the PBTH and UC groups. The contrast in usage of TH—53.6 percent for PBTH families and 29.1 percent for UC families—is sizable, although smaller than the analogous contrast in either the SUB-versus-UC or CBRR-versus-UC comparisons.

As is conventional in random assignment analyses, our goal is to estimate the intention-to-treat (ITT) impact—that is, the difference in impact by the program to which families were assigned, regardless of whether they actually used that program (or some other program). This goal is consistent with the policy option of making a treatment available to a family, but without the ability to force a family to use that treatment.

Because not all families randomly assigned to PBTH used TH, and some families assigned to UC did use TH, the true ITT impact is likely smaller than it would have been had the gap in TH usage been wider (assuming that TH truly has a nonzero impact on families who use it). In particular, the difference in the use of TH by the PBTH and UC groups is narrow enough, given the relatively small sample size available for analysis, that the study may have failed to detect as statistically significant one or more ITT impacts large enough to be of policy importance.

Additional detail about the use of TH assistance by PBTH families is shown in Exhibit 8-7. The exhibit shows that about one-fifth (21 percent) of PBTH families who used TH did so for 18 or more months.

The remainder of the chapter reports estimated impacts in the various outcome domains that—if statistically significant—

can be causally attributed to the offer of project-based transi - tional housing to the families randomly assigned to PBTH.

115 Hence, 0 values are not factored into the means, nor do they pull downward the medians of the various distributions.

116 Because 21.7 percent of PBTH families were still using TH in the followup survey month, and 53.6 percent had used TH at some point during the followup period, it can be calculated that 1 – (21.7/53.6) = 59.5 percent of PBTH families who used TH at some point had stopped using it by the survey month. A similar calculation, 1 – (9.8/29.1), yields 66.3 percent for UC families.

Exhibit 8-7. Number of Months of Transitional Housing Receipt During Followup Period by PBTH Families Who Ever Used TH

PBTH = project-based transitional housing. TH = transitional housing.

a Percentages are weighted for survey nonresponse to represent all families in comparison sample.

Note: N = 159.

Source: Family Options Study Program Usage Data

8.3 Impacts on Housing Stability in the Project-Based Transitional Housing (PBTH) Versus Usual Care (UC) Comparison

Proponents of PBTH emphasize that most families who become homeless have barriers that make it hard for them to secure and maintain housing, as Chapter 3 addresses.

Thus, housing subsidies alone may be insufficient to ensure housing stability and other desirable outcomes. Family needs may arise from poverty, health, disability, or other problems that led to homelessness to begin with or from the disruptive effects of homelessness on parents and children. Proponents of PBTH believe that by addressing these barriers and needs in a supervised residential setting, PBTH lays the best founda - tion for ongoing stability. The PBTH-versus-UC comparison offers evidence on whether this approach is effective in the

short term in improving family outcomes. It is important to keep in mind that the followup period covered in this report is not quite long enough to observe the effect of PBTH after program completion. About one-fifth of PBTH families (and one-tenth of UC families) were still in TH in the month of followup response.

Exhibit 8-8 shows the impacts of the offer of PBTH on home - lessness, housing independence, residential moves, and hous ing quality.117 The first row of the exhibit shows the impact on the confirmatory outcome of the study (homeless or doubled up in the past 6 months, or stay in emergency shelter in the past 12 months). The impact estimate indicates that PBTH causes a 7.7-percentage-point reduction on the confirmatory outcome (from about 52 to 44 percent). This estimate is statistically significant at the .10 level before the adjustment for multiple comparisons but is not statistically significant after adjustment.118

117 The homeless outcomes in this study diverge from the homeless definition final rule in that they do not include stays in transitional housing in their definitions of being homeless. Additional impacts on the use of transitional housing during the followup period are provided in Appendix E.

118 The study estimates impacts on this confirmatory outcome for each of the six paired comparisons and four pooled comparisons. Seven of these estimates have been prespecified as “confirmatory tests.” A multiple comparison procedure is performed to compute adjusted p-values for these tests to reduce the possibility of chance findings of statistical significance. The details of this procedure are provided in Appendix C.

25

20

15

10

5

Number of months of transitional housing receipt 0

0.25–0.75 1.00–1.75 2.00–2.75 3.00–3.75 4.00–4.75 5.00–5.75 6.00–6.75 7.00–7.75 8.00–8.75 9.00–9.75

10.00–10.7511.00–11.7512.00–12.7513.00–13.7514.00–14.7515.00–15.7516.00–16.7517.00–17.7518.00 or more

Percent of PBTH families who ever used THa

Short-Term Impacts of Housing and Services Interventions for Homeless Families 98 Compared With Usual Care (UC)

Exhibit 8-8. PBTH Versus UC: Impacts on Housing Stability

Outcome PBTH UC ITT Impact Effect

Sizea

N Mean (SD) N Mean (SD) Impact (SE)

Homeless or doubled up during the followup period At least 1 night homelessb or doubled up in past 6 months or

in shelter in past 12 months (%) [Confirmatory]c 294 44.3 (49.9) 262 52.0 (50.1) – 7.7*b (4.4) – 0.14 At least 1 night homelessb or doubled up in past 6 months (%) 294 37.2 (48.8) 262 41.8 (49.1) – 4.6 (4.3) – 0.08 At least 1 night homelessb in past 6 months (%) 294 18.0 (39.1) 262 25.3 (43.0) – 7.3* (3.7) – 0.15 At least 1 night doubled up in past 6 months (%) 294 32.3 (47.5) 262 32.4 (46.3) – 0.1 (4.1) 0.00 Any stay in emergency shelter in months 7 to 18 after RA (%) 294 18.9 (38.8) 262 27.1 (44.9) – 8.2** (3.6) – 0.16 Number of days homelessb or doubled up in past 6 months 293 45.3 (71.0) 262 54.1 (75.1) – 8.7 (6.7) – 0.10

Number of days homelessb in past 6 months 291 15.5 (41.8) 261 23.2 (50.5) – 7.7 (4.7) – 0.14

Number of days doubled up in past 6 months 294 36.5 (65.5) 262 36.4 (64.4) 0.1 (5.8) 0.00

Housing independence

Living in own house or apartment at followup (%) 252 57.5 (49.6) 262 61.2 (49.0) – 3.6 (4.7) – 0.06 Living in own house or apartment with no housing assistance (%) 252 40.0 (49.3) 262 39.9 (48.1) 0.0 (4.6) 0.00 Living in own house or apartment with housing assistance (%) 252 17.6 (36.6) 262 21.3 (43.0) – 3.7 (3.7) – 0.08 Number of places lived

Number of places lived in past 6 months 293 1.7 (1.1) 261 1.8 (1.2) – 0.1 (0.1) – 0.07

Housing quality

Persons per room 290 1.7 (1.2) 250 1.9 (1.2) – 0.2 (0.1) – 0.11

Housing quality is poor or fair (%) 292 37.1 (48.7) 259 35.4 (47.3) 1.7 (4.4) 0.03

PBTH = project-based transitional housing. UC = usual care.

ITT = intention-to-treat. RA = random assignment. SD = standard deviation. SE = standard error.

*/**/*** Impact estimate is significantly different from 0 at the .10, .05, and .01 levels, respectively, using a two-tailed t-test.

a Effect size column shows standardized effect sizes, which were calculated by dividing impact by standard deviation for the entire UC group.

b The definition of homeless in this report includes stays in emergency shelters and places not meant for human habitation. It excludes transitional housing. Additional impacts on the use of transitional housing in the followup period are provided in Appendix E.

c After adjustment for multiple comparisons, the impact on the confirmatory outcome is not statistically significant at the .10 level for the PBTH-versus-UC comparison.

Notes: Impact estimates and outcome means are regression adjusted for baseline characteristics and are weighted to adjust for survey nonresponse. See Chapter 5 and Appendix B for outcome definitions.

Sources: Family Options Study 18-month followup survey; Program Usage Data

Two other homelessness outcomes show effects of PBTH relative to UC: (1) a 7.3-percentage-point reduction in expe - rience of homelessness in the past 6 months, and (2) an 8.2-percentage-point reduction in use of emergency shelter in months 7 to 18 after random assignment. The evidence of reductions in homelessness from two different data sources (the followup survey and the Program Usage Data, the latter of which reflects primarily Homeless Management Information System [HMIS] records) strengthens the finding that PBTH reduced homeless stays in shelters or places not meant for human habitation, even if not by an amount large enough to remain statistically significant after the multiple compari-sons adjustment. PBTH appears to have no effect relative to UC on the experience of being doubled up or the time spent doubled up in the 6 months before the survey response.

Exhibit 8-9 shows the month-by-month impacts on the pro-portion of families with at least 1 night in emergency shelter

during the month. This exhibit illustrates that a somewhat lower proportion of the PBTH group is in emergency shelter during the followup period as compared with the UC group.119 The last six rows of Exhibit 8-8 show that, relative to UC, PBTH has no effect on the proportion of families living in their own house or apartment (with or without assistance), no effect on number of moves, and no effect on housing quality.

It may be that these relatively modest effects of priority access to PBTH on housing stability result in part from the relatively low use of transitional housing among the PBTH assignment

It may be that these relatively modest effects of priority access to PBTH on housing stability result in part from the relatively low use of transitional housing among the PBTH assignment

문서에서 Family Options Study (페이지 126-135)