There’s been a lot of hype around one of the statistics released by the Human Services Agency in the 2007 Homeless Count Report: 31% of homeless people in San Francisco moved here after becoming homeless. We’re going to assume for most of this post that that number is valid, but let’s take a look at the method for just one moment: 292 homeless people at St. Anthony’s Dining Hall, Bayview Resource Center, Mission Neighborhood Center, Larkin Street Drop-In Center, and North Beach Citizens Drop In Center were interviewed after the Count, and asked, “Where were you most recently living at the time you became homeless?” Their options were: “San Francsico,” “Other County in CA,” and “Out of CA.” 182 answered that they’d been made homeless in San Francisco; 44 in other California counties; 46 outside of the state; and 20 declined to respond.
A sample of 292 is just over three quarters (76%) the sample size needed for a 95% confidence level with an unknown population size. It’s about 80% the sample size needed for a population of 6,377. So, to begin with, we don’t have a large enough sample to make a proper statistical assessment for even the known homeless population of San Francisco, let alone for the larger unknown population.
But are these even the right people? We don’t know, actually, and neither does anyone else. While HSA did a good job in spreading their sampling sites out geographically (though it should be noted that there were no sample sites near Golden Gate Park or the Sunset or Richmond), not all homeless people access these sorts of services. A principle of good statistical research is that population samples must be random. This sample very clearly omits all homeless people who do not take advantage of services, and underrepresents those who use services less.
The point, here, isn’t that this data is wrong—it’s that we don’t know how trustworthy it is.
But let’s say that population is representative—that the sample size hasn’t adversely affected the results, and that homeless people who use service agencies are no more nor less likely to be local than are those who don’t: 30.9% is a plausible figure. It’s also a very low one, if we think we’re handling homelessness properly. 62.3% of homeless San Franciscans were made homeless while residing in regular domiciles in this city. In other words, there are significant conditions within the city of San Francisco that make people homeless.
HSA and the Mayor’s Office are claiming that homeless immigrants explain homelessness’ increase. For this explanation to be valid, the number (not just the proportion) of homeless people coming from out of town would have had to have increased. Unfortunately, this data has not been collected in previous counts, which means that this claim is based on nothing but… well, nothing.
Lastly, San Francisco has evicted 1,800 homeless people from the city through the Homeward Bound program since the last count. The City claims that these people are being sent back to family or friends who can help ensure that they will become housed. The fact of the matter is that the City has done minimal tracking, and actually has no idea whether or not Greyhounded homeless people are finding housing. History indicates that this is probably not the case: What mental health advocates call “Greyhound therapy” has been a very common solution to homelessness in the United States. Recently, towns in Humboldt County and the South Bay have had to ask San Francisco to stop sending them homeless people. So 30.9% of homeless people in San Francisco were homeless when they got here? We shipped 28.8% of 2005’s homeless population out of town. What additional percentage left of their own accord?
To re-cap: There’s not much reason to believe that this 30.9% figure is accurate, but if it isaccurate, all it shows is that we bring in just about as many people as we drive out.