Point-Prevalence Bias

I was looking for current data on homelessness and mental illness, since it has been known since at least the 1980s that many homeless people are mentally ill--and this is likely a causal factor in their homelessness. While hunting, I found this article about something called "point-prevalence bias" in homelessness statistics. What in the heck is that?

If you measure the number and characteristics of the homeless at a particular point in time (say, the night of December 15), you will get everyone that was homeless that night. People that are chronically homeless--the person who has been homeless for several years--will be overrepresented compared to those who are homeless for only a few weeks. This is "point-prevalence bias."

If you are having some trouble understanding why this is bias, so am I. Yes, it means that the chronically homeless will be given more weight in the final results than measuring what they call "lifetime" homelessness--how many and what sorts of people have ever been homeless. In reading the paper, I smell something that sounds rather like, "We're having trouble getting enough sympathy for homeless people because the surveys show that more than half have histories of jail or prison and a big fraction of the rest have been in psychiatric hospitals, so let's emphasize the people who are short-term homeless, even though this is at any given time, a small fraction of the homeless population." It also looks like they are trying to create a huge population of people who have ever been homeless, even if only for a few days. It smells rather like Mitch Snyder's opt-repeated by journalists, "three and half million homeless--and growing" claim of the 1980s.

By the way, I was surprised at the large percentage of this "point" studies that show jail or prison lockup. I learned something very interesting about this from talking to my daughter, who is working on finding housing for homeless families who are at the Booth Family Shelter. Here in Boise, the vast majority of landlords--even landlords who are cooperating on helping find permanent housing for these homeless families--will not rent to people with felony convictions in the last ten years, or were evicted and owe lots of back rent.

I could understand the felony convictions stigma. While not every felony is something horrifying, trying to distinguish murder/robbery/kidnapping from turning back a car odometer/passing bad checks might be a bit of a struggle, and landlords don't want trouble.

The evictions and back rent stigma, however, surprised me. My daughter tells me that some people owe many months of back rent from evictions. This tells me that there are landlords in this area who are not heartless. It appears that Idaho is one of those states where five days after you fail to pay the rent, the landlord can evict you. So if you owe many months of back rent, it means that your previous landlord must have accepted excuses for not getting rent paid far beyond what the law requires.