Since homelessness is mostly transient in general, I'm skeptical of the claim to effectiveness, since it's likely that a large portion of a sample would move on within a year anyway.