>There is a strong selection bias here: homeless people who are addicts or suffer from mental health issues are  significantly less likely to be on a computer writing interesting articles of their experience.

Good point. But shouldn't we see those kinds of articles in other countries as well? Maybe #4 (down-on-their-luck) might be proportionally smaller in the US but bigger in comparison to the same group in other countries?

One interesting thing also is #3, why are veterans homeless? PTSD?