A Better Way to Measure Homelessness

...167 days to go...

The King County Point-in-Time Count is officially underway! I’m happy to report I’ll be volunteering next week.

If you’re not familiar with PIT counts, they’re our primary tool for understanding the homeless population: for instance, how many people are experiencing homelessness, whether they have children, whether they’re sheltered or unsheltered, what percentage struggle with substance abuse and the like. The counts happen every other year in King County, and they’re required for any community to receive federal funding from the Department of Housing and Urban Development (HUD).

When I signed up, I assumed the count worked like it does in most cities: an army of volunteers fanning out across the county at night, clipboards in hand, literally trying to “find” as many homeless people as possible before dawn.

Turns out, Seattle does it completely differently.

A Statistical Innovation

King County is one of only two communities in the United States that uses a methodology called Respondent Driven Sampling (RDS). We’ve used it since 2022, and it’s actually much more accurate at counting hard-to-find populations than the traditional approach.

Think about the limitations of the traditional method. Volunteers walk designated routes for a few hours one night in January. If someone is sleeping in the woods, under a bridge, in a car parked on a side street, or anywhere else volunteers don’t happen to look, they’re missed. And this is a population that often doesn’t want to be found—many have suffered significant trauma and have good reasons to avoid strangers with clipboards. In rural areas especially, it’s remarkably easy for homeless people to remain invisible.

How Respondent Driven Sampling Works

Here’s the process:

Step 1: Volunteers give a “coupon” and bus ticket to a small sample of homeless individuals across the county.

Step 2: Coupon recipients can travel to a nearby “HUB” site using the bus ticket.

Step 3: Volunteers at the hub—like me—greet each person, scan their coupon, administer a 15-20 minute survey, and then give them three additional coupons to share with other homeless people in their network.

Step 4: Respondents receive compensation for participating: $20 for individuals, $40 for heads of families. They can earn an additional $5 for each of their three coupons that gets used by someone else.

Instead of volunteers trying to find homeless people, homeless people recruit each other. They know where their friends are sleeping. They know who’s hiding under which bridge. They have access to networks that are invisible to housed volunteers.

Why This Generates Better Data

RDS is more effective than traditional counts for two key reasons:

First, participation is voluntary. Homeless individuals choose whether to travel to a hub and take the survey. This matters enormously. When you’re not being counted by a stranger who woke you up at 2am and is asking invasive questions, you’re much more likely to provide complete and accurate data. The $20-40 incentive doesn’t hurt either—it shows respect for people’s time.

Second, the survey captures network data. Respondents are asked to name everyone in their homeless network, providing just the first two letters of their first name and first two letters of their last name. This protects confidentiality (you can’t identify someone from “JOHA”) but provides enough information to eliminate duplicates in the statistical modeling.

So if “JOHA” is identified by three different people, the University of Washington statisticians can identify that those referrals are probably the same person and adjust the count accordingly.

Still an Undercount (But Better)

Let me be clear: the PIT count will still yield an undercount of homelessness, even with RDS.

RDS is better at finding hard-to-reach populations than a traditional one-night count. But like any point-in-time methodology, it only captures people who are homeless on one specific day in January. It misses everyone who becomes homeless in March, or July, or October. The true count of people experiencing homelessness at any point during 2026 could be as much as 3x higher than the PIT count will show.

Still, RDS represents a significant improvement. And I’m genuinely excited to see it in action.

The Statistical Complexity

One downside of RDS: it requires significant statistical modeling work, which will be performed by the University of Washington.

While traditional visual counts are simple, RDS is complex. Statisticians must weight the referral networks to generate population estimates. This complexity requires peer review, so results take longer to publish. We probably won’t see initial findings until May.

What I Expect to See

I wish I could say I’m optimistic about the results.

Sadly, the count is almost certain to continue the upward trend from 2024, which showed a 26% increase from the 2022 baseline.

The factors driving homelessness haven’t gotten better. They’ve gotten worse:

Housing costs keep climbing while wages stagnate. The gap between what lower income people earn and what rent costs continues to widen, pushing more people into precarious housing situations.

Federal benefit disruptions. The government shutdown created chaos in programs like SNAP. More significantly, eligibility restrictions passed in the “Big Beautiful Bill” are only beginning to take effect. Those impacts probably won’t hit this count much, but they’ll definitely show up in 2028.

Emergency Rental Assistance ended. The pandemic-era ERA2 program ended in September 2025. Many local programs closed even earlier as funds were depleted. These were the prevention programs that kept people from becoming homeless in the first place. Without them, more people slip through the cracks.

So yes, I expect the numbers to be bad.

Nonetheless, you can’t solve a problem you can’t measure. So next week I’ll be at the Lake City Hub scanning coupons, asking questions, and hopefully making a few people feel seen and heard in the process.

I’ll report back on what I learn. And when the results are published in May, I’ll break down what they mean for Seattle’s approach to homelessness.