Summary: In response to a congressional request, GAO examined the methodologies of current population estimates of the number of homeless chronically mentally ill persons to: (1) determine the soundness of current estimates; and (2) identify the best available methods for getting sound estimates.
GAO found that: (1) although there were no sound national estimates of the number of homeless persons who were chronically mentally ill, there were some relatively sound local estimates on both homeless persons and the chronically mentally ill; (2) while the estimates would not apply nationally, the methods used were applicable on a larger scale; (3) many of the estimates had methodological flaws associated with a high degree of uncertainty or bias that could lead to underestimates or overestimates; (4) of the 10 studies that GAO rated technically sound, 9 were based on surveys or censes and one was based on utilization data; (5) the rates of homelessness ranged between 6 and 95 per 10,000 in the study communities and were related to the quality of methodology; (6) analysis of a methodology judged to be of a higher quality lowered the range by almost half; (7) the proportion of homeless persons identified as mentally ill was between one-fifth and one-third, in the judgment of service providers, and between one-sixth and one-half using standardized instruments; and (8) because the definition of homelessness varies and homelessness itself could contribute to behavior and appearances that suggest mental illness, counting homeless mentally ill persons is never entirely precise.
