Internet Explorer 11 is not supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

How Michigan Got Better at Counting Homeless Veterans

Homeless veterans are notoriously difficult to count. Michigan found a way to test the accuracy of its numbers and deepen the state’s understanding of veteran homelessness today.

A homeless veteran in Portland.
FlickrCC/Erik Kurtz
For the past two decades, Barbara Ritter has published annual counts of homeless veterans with some skepticism. As the project director for Michigan's Homeless Management Information System (HMIS), Ritter knew the numbers were likely off; homeless veterans often choose not to identify themselves as veterans. But last year, Ritter used a new projection formula perfected by academic researchers in 2011 to test the magnitude of her undercount. Based on her second year of findings -- she thinks she's got better numbers.

Michigan’s most recent report, released in early October, shows that the total number of homeless veterans counted throughout 2012 increased from 4,073 in 2011 to 4,243 in 2012. That increase reflected not only an improvement in counting homeless veterans, but also an actual increase in the population because more veterans are returning from Afghanistan and Iraq. When Ritter compared her counts with the projections, she found that the newest data represented 95 percent of the expected number of homeless men and 94 percent of homeless women.

What Ritter is doing in Michigan may serve as a model for other states that want to know how accurately they’re counting a sub-population of the homeless that is inherently difficult to track. Ritter sat down with Governing to explain the problem of counting homeless veterans, how Michigan is trying to circumvent it, and what the state’s most recent count tells her about the state of veteran homelessness. The transcript below has been condensed and edited for clarity.

Read more health and human services stories here.

Can you explain why there was room for improvement in the counting? What’s hard about counting homeless veterans?

I always believed historically we had an undercount. A lot of veterans won’t admit they’re veterans. Prior to now, when they came asking for community services, being a veteran was a liability. All things being equal, when services are full, if a provider gets somebody who can be served at Veterans Affairs (VA), and that would open up that slot for somebody who doesn’t have that qualification, the provider’s going to send them to the VA. If you’re a vet and you know that, it’s just easier not to self-identify because then you don’t have to have that discussion.

We could see it in our data because we’d have somebody identified as a veteran and then if we looked at three or more [interactions] with other social service agencies, they’d be saying “no” half the time and “yes” half the time. So, [veterans] would look at what they were trying to get from the particular provider and if they were worried that being a veteran would decrease their chances of fitting into the eligibility criteria, then they would obviously not disclose that.

A lot of veterans have extremely strong feelings about the VA. My father was an example of that. He was hurt in World War II. He had a complete pension. The family never knew about it until he was in end-of-life care and then realized that he was fully covered and it was because he did not want to be referred to the VA hospitals.

Can you explain how you achieved improvements in counting?

The biggest thing that happened was the VASH [Veterans Affairs Supportive Housing] vouchers. Anything that was an overlay of health care or behavioral health care -- veterans may have thought that admitting they were vets would result in a referral rather than a service. But once [HUD] added the VASH housing vouchers [in 2008], then there was something concrete that [homeless veterans] wanted and being a vet got them in the door rather than excluded them.

We also followed up with training about how to ask a question. The HUD data standards have changed this year. That’s probably helpful. Prior to the new standards, we used to ask, “are you a veteran?” With the new standards, you actually ask a more specific question, which is, “have you ever been on active duty in the military?”

Could you give an example of how someone could identify as a veteran, but not qualify for benefits?

Yes. The National Guard. Unless they’ve been activated, they don’t qualify for a lot of stuff. Or the Army Reserve. There are a lot of people in the Reserve who may not have actually served in active duty. It’s a very complex picture on eligibility.

So, what’s the methodology you’re using now to verify that the counts are more accurate?

Dennis Culhane [a professor of social policy at the University of Pennsylvania] did a study [in 2011] and tried to figure out how many veterans we should see, cross referencing HMIS data with data from the American Community Survey. You could project how many men, women, singles and families should be in your homeless population. It allowed me to back out and say, I know how many overall homeless we have, and then how many of that overall homeless count we would expect to be veterans and what the real veterans count is. I was always hesitant to publish very much without knowing that I was somewhere in the ballpark of having who I should have in the system on the veterans side.

What did you do previously? You published it but just didn’t feel good about it?

We had one line. There’s this big demographics table that we do every year. I published it because everybody always asks, but I didn’t talk about it in great depth. We certainly didn’t do an entire chapter the way we’re doing now. When I talked about it in writing, I always said I believe this is an undercount. I had that caveat. But I think we’re getting close. We’ve got two years in a row where we’ve got close to what the study would estimate we should be. So I’m more comfortable with what we see in the data.

Which brings me to my next question. What are you seeing in the new and improved data?

There are two patterns going on in homeless data in general. When I looked at my data back in the 90s, I never had seniors. Now 17 percent of our singles are over 55. We have an aging population and a bunch of them are vets. And we also have a cohort of young people that is growing -- not as much in my data but around the country -- we see people who have not been able to get on the economic escalator at all. And then [in Michigan] we do have the under-30 veterans coming back into shelters and a high, high proportion of those were people with disabilities. If you’ve ever worked with PTSD [Post Traumatic Stress Disorder] or TBI [Traumatic Brain Injury] folks, having a family hang together with that is extremely difficult. So, it may be that the trajectory of homelessness is sped up for those young people coming back with a disability.

What happens to folks normally is that they don’t go homeless just because they lose their housing. They double up with family until they’re basically not invited to the next Thanksgiving dinner. It takes some time for them to work through their support systems and then they end up in the shelter. What I’m concerned about -- I don’t have a ton of data on this, but I think we need to watch it very carefully -- is that if we have veterans coming back in with significant illnesses, such as TBI or PTSD, they’re going to roll through their families a lot faster, and they may hit the shelter system faster. We did have a surprising number of veterans under 30. I think we have to be prepared to address that. Otherwise they could just age in the homeless system the way many Vietnam veterans did, and that’s a concern.

We look at exit-into-housing as one of our outcome measures. One of the normal features of exit-into-housing is that the younger you are the more likely you are to go back into housing, and that’s because you still have those support systems. They’re not completely fragmented. Your family might be willing to take you back. The older you are in homelessness, the less likely your support systems are in play. What I saw in the veterans data was actually the inverse of that. The youngest were the least likely to exit into housing. And I don’t know that that’s a significant difference because I don’t know if I have a big enough “n” to really draw causal conclusions, but it’s a trend we’re going to monitor.

Five figures from Michigan’s most recent HMIS data

  • Forty-five percent of homeless veterans in families had a documented disability, according to homeless service providers. For the civilian homeless in families, it was 19 percent.
  • Among homeless veterans with a documented disability, the broad category of mental health ranked first at 59 percent.
  • Homeless veterans were mostly male (89 percent).
  • The rate of advanced education was higher among homeless veterans (43 percent) than civilian veterans (27 percent).
  • For homeless families with veterans heading the household, the rate of homeless declined by 25 percent between 2011 and 2012.
J.B. Wogan is a Governing staff writer.
Special Projects