So far, this chapter has explored homeless assistance programs and services at the national level and through the broad geographical designations of central cities, suburban areas, and rural areas. NSHAPC also has the capacity to examine variations in programs and program contacts within the survey's 76 primary sampling areas (called "sampling areas" hereafter).⁴ Program contact information can be segmented to reveal the proportion of services within a sampling area that are shelter/housing program contacts, food program contacts, health program contacts, and other program contacts. Further, program contact information can be used to calculate a rate of program contacts per 10,000 clients in poverty, which is a good measure for comparing the level of service availability across sampling areas of very different sizes (e.g., a city with more than a million people and a rural area of a few thousand people).⁵

Source: Weighted NSHAPC data representing programs operating during "an average week in February 1996." Includes all programs, of every type.

Figure 4.12

Distribution of Program Contacts within Sampling Areas

All other things being equal, one might expect the sampling areas with the largest population to provide the most homeless assistance program contacts. To examine this expectation, figure 4.13a arrays each of this study's 76 primary sampling areas from left to right according to the size of its population (largest on the left, smallest on the right).⁶ Each bar shows the total estimated number of program contacts on an average day in February 1996,⁷ combining program contacts of all types.

From this figure one can see that the expected relationship of more program contacts in the larger sampling areas is generally true but there are exceptions. The average estimated number of program contacts per sampling area is about 17,600 on an average day in February 1996. But the estimated numbers of program contacts range from a high of about 186,000 to a low of nothing (for two sampling areas that had no programs of any kind) (table 4.2). And variation exists even at the highest end. For example, providers in the largest sampling area estimate only about two-thirds the number of program contacts (about 123,000) as do providers in the next largest sampling area (about 186,000).

To examine how much population size accounts for the differences observed in figure 4.13a, the estimated number of program contacts per 10,000 people is employed. Figure 4.13b shows this rate for each of the 76 sampling areas, arrayed in the same order as figure 4.13a. The average estimated rate of program contacts per 10,000 population in a sampling area is 122. The use of a common denominator (10,000 people) reduces the differences among sampling areas quite a bit. Now one can see that some of the smaller sampling areas in the middle and toward the right of the figure appear to provide more units of homeless assistance services per capita than do some of the largest sampling areas.

Yet another way to look at these data is to ask whether the variability in service levels can be accounted for by the size of a sampling area's population in poverty, and not just by the total number of people in the sampling area. There is some reason to expect that services should be related to poverty, and the number of poor people in a sampling area is the best measure of need that is available for all 76 sampling areas. Some sampling areas could have a lot of people but not very many poor people, while some smaller sampling areas might actually have more poor people than some larger areas. Therefore a second rate was constructed for each sampling area—its rate of program contacts per 10,000 poor people. Figure 4.13c shows the results.

The average estimated rate of program contacts per 10,000 poor people in a sampling area is 1,437. The rate of contacts per 10,000 poor people reduces even further the level of variability in service provision among the largest sampling areas at the left of the graph. The variability in the middle of the graph (medium- and small-sized metropolitan areas) appears to have increased in relation to that in figure 4.13b. The mostly rural areas to the right of figures 4.13b and 4.13c appear to have the greatest variability whichever rate is used.

From the three graphs in figure 4.13 one can draw the conclusion that the biggest sampling areas, which comprise the nation's biggest cities, do not always provide the most services on a per capita basis, even though they obviously provide very large numbers of services. One can also conclude that a great deal of intercommunity variability remains in the provision of homeless services, even after controlling for levels of population and poverty. This level of variability probably stems from important differences in philosophies, policies, resources, and experience among communities.

Distribution of Services within Sampling Areas by Program Type

The next issue to be examined is how the total estimated number of program contacts within each sampling area are distributed among the four major program types of shelter/housing, food, health, and other. The results, shown in figure 4.14, reveal great variation in the proportion of service contacts in sampling areas within shelter/housing, food, health, and other program types.

Figure 4.14a-d shows four panels, one each for shelter/housing, food, health, and other program contacts. The average proportion of program contacts reported by shelter/housing programs is 24 percent, by food programs is 49 percent, by health programs is 5 percent, and by other programs is 19 percent (table 4.2 gives means, highs, lows, and standard deviations). In comparing parts (a) through (d) of figure 4.14, one can see the predominance of food program contacts and the relative paucity of health program contacts. Food program contacts comprise at least 40 percent of all program contacts in most sampling areas (only 17 of the 74 areas with any services have less than 40 percent of their program contacts at food programs, and one-third have more than 60 percent of program contacts at food programs). In contrast, only five sampling areas have as much as 20 percent of program contacts occurring at health programs, and most have less than 10 percent in the health area.

Figure 4.13

The greatest variability occurs in smaller metropolitan areas and rural areas, which are the most likely to have either much more of a concentration in a particular type of service than is true nationally, or much less of a concentration. Some of these sampling areas have all or virtually all of their program contacts in housing programs, others have all or almost all their contacts in "other" programs (such as outreach, drop-in, or housing/financial assistance programs), and a few have a significant share in health programs.

Figure 4.14

Distribution within Sampling Areas of Contacts with Different Types of Shelter/Housing Programs

Shelter/housing program distributions in sampling areas reflect very different decisions about where to invest homeless housing resources. This analysis uses a rate of shelter/housing program contacts per 10,000 poor people. Figure 4.15 provides this information, first for all shelter/housing program types (figure 4.15a), and then separately for each type of shelter and housing program (emergency shelter—figure 4.15b; transitional housing—figure 4.15c; permanent housing for the formerly homeless—figure 4.15d; and vouchers for temporary shelter—figure 4.15e).

Figure 4.15

The estimated national rate of program contacts with all types of shelter and housing programs for homeless people is 195/10,000 poor people. In addition to the two sampling areas with no programs of any kind, one additional sampling area has no shelter/housing program contacts at all.

Emergency shelter contacts per 10,000 poor people in the study's primary sampling areas average 81/10,000. Six sampling areas offer 150 or more shelter/housing contacts per 10,000 poor people, while nine sampling areas offer 20 or fewer emergency shelter contacts per 10,000 poor people including four that do not offer any.

Variability is even greater among primary sampling areas for rates of transitional housing, permanent housing, and voucher distribution. Transitional housing contacts within sampling areas have a national average of 49/10,000 poor people, with eight sampling areas offering more than 100 transitional housing contacts per 10,000 poor people and 23 offering 20 or fewer, including 13 that offer none.

Permanent housing contacts within sampling areas have a national average of 40/10,000 poor people, with seven sampling areas offering more than 100 permanent housing contacts per 10,000 poor people and 42 offering 20 or fewer, including 20 that offer none.

Voucher distribution contacts within sampling areas have a national average of 26/10,000 poor people, with four sampling areas offering more than 100 voucher program contacts per 10,000 poor people and 50 offering 20 or fewer, including nine that offer none.

Previous

Contents

Next