HPRI Fellow Summer Seminar 4:

Data, Homeless Count, and Demographic Survey Methods

Amanda Landrian Gonzalez, PhD

Scot Hickey, MA

Stephanie Kwack, BA

​

Seminar Roadmap

• Introductions
• Icebreaker Activity: Who Counts as Homeless?
• The Homeless Count: What is it? Why do we do it? How do we do it?
• Group Activity & Discussion #1: Trend Interpretation
• The Demographic Survey: A Closer Look
• Group Activity & Discussion #2: Survey Question Review

​

2

INTRODUCTIONS

3

ICE BREAKER ACTIVITY: WHO COUNTS AS HOMELESS?

• Instructions:
• We will read aloud 5 different individual “profiles” describing where a person is currently living/sleeping.
• For each profile, stand in the designated section of the room if you think the individual would be considered sheltered, unsheltered, or not homeless.

4

PROFILE #1:

​

A PERSON SLEEPING IN AN ABANDONED BUILDING

​

5

PROFILE #2:

​

A PERSON LIVING AND SLEEPING IN THEIR VEHICLE (CAR, VAN, OR RV)

​

6

PROFILE #3:

​

A PERSON WHO IS CURRENTLY SLEEPING ON A FRIEND’S COUCH

7

PROFILE #4:

​

A PERSON LIVING AND SLEEPING IN A MOTEL

8

PROFILE #5:

​

A PERSON WHO IS CURRENTLY INCARCERATED

9

THE HOMELESS COUNT

WHAT IS IT?

WHY DO WE DO IT?

HOW DO WE DO IT?

10

WHAT IS THE HOMELESS COUNT?

• The Homeless Count, or Point-in-Time (PIT) Count, is a federally required, one-night census of people experiencing homelessness across the United States
• Conducted during the last 10 days of January every year
• Consists of two components:
1. Sheltered count: People living in emergency shelters, transitional housing, safe havens
2. Unsheltered count: People living in places not meant for human habitation (e.g., on the street, vehicles, tents)

11

WHAT IS THE HOMELESS COUNT?

• Required for every Continuum of Care (CoC): the regional bodies that receive federal homelessness funding
• CoCs are required to provide basic descriptive data about their unhoused population (e.g., age, race/ethnicity, gender, veteran status, disability and health status, homelessness history)

​

12

THE GREATER LA HOMELESS COUNT

• Conducted by the Los Angeles Homeless Services Authority, a joint city-county agency
• Covers the LA CoC: nearly all of Los Angeles County except Glendale, Pasadena, and Long Beach
• Among largest homeless counts in the country—both in geographic coverage and size of the population counted
• USC partners with LAHSA, UCLA, and UW on methodology and analysis

​

13

WHY DO WE CONDUCT THE HOMELESS COUNT?

1. No count = no federal homelessness funding
2. Funding allocation
3. Resource planning
4. Program accountability
5. Filling critical data gaps
6. Public awareness and community engagement

14

Trends in National Homelessness

33% increase

15

Trends in Los Angeles CoC

16

2025 Los Angeles CoC

Overall Estimate: 67,777 (+/- 1,286)

HOW DO WE CONDUCT THE COUNT?

17

Street Count (Visual Tally)

HOW DO WE COUNT UNSHELTERED ADULTS?

Demographic Surveys

18

Unsheltered Adult Count:

Street Count

• Around 8,000 Volunteers

​

• Visual Tally of:
• Persons experiencing homelessness (individuals and families)
• Cars, vans, RVs, tents, makeshift shelters (i.e., CVRTM)

​

• Full enumeration
• Complete coverage of 2,310 census tracts conducted over a period of 3 nights

19

2025 Homeless Count

Street Count (Visual Tally) Results

What are the characteristics of these people?

How many people are living in CVRTMs?

20

Street Count (Visual Tally)

HOW DO WE COUNT UNSHELTERED ADULTS?

Demographic Surveys

21

Unsheltered Adult Count:

Demographic Survey

Critical for achieving 2 specific goals:

1. Providing estimates on the average number of people found in CVRTMs (i.e., the multiplier)
2. Describing the demographic and health characteristics of the unsheltered population (e.g., age, gender, race, veteran status)

​

​

22

• Trained survey teams deployed to selected census tracts (N=576)
• CTs covered over 3-month survey period (Dec-Mar)
• $20 incentive provided
• Total surveys collected: 6,158
• Eligible unsheltered adult surveys: 4,396

Unsheltered Adult Count:

Demographic Surveys

23

How do we estimate how many people are sleeping in observed CVRTM?

Demographic Survey questions ask:

• Where slept most of the last 30 days
• Where slept last night
• Household size
• Whether slept in a CVRTM in the last 6 months, and if so, with how many people

Street Count (Visual Tally)

• Counts of cars, vans, RVs, tents, and make-shift shelters (CVRTM) observed by volunteers

Estimated number of people in cars, vans, RVs, tents, and make-shift shelters

X

24

2025 Homeless Count:

Estimating People Based on Dwelling Counts

CVRTM

Multiplier =

1.75

​

43,073

25

2025 Los Angeles CoC

Overall Estimate: 67,777 (+/- 1,286)

HOW DO WE CONDUCT THE COUNT?

26

WHAT ABOUT UNSHELTERED YOUTH?

• Unsheltered youth (under 25) = 2% of LA CoC unhoused population and 3% of all unsheltered individuals
• Can’t be reliably counted via adult visual street tally method—youth are often hidden, highly mobile, and disengaged from services
• 2025 method used separate sampling strategy and youth enumeration survey, with volunteers deployed to sampled census tracts to attempt to locate and survey unsheltered youth
• Method was resource-intensive, depended heavily on youth-serving provider capacity
• Providers warned of continued undercounts

27

WHAT ABOUT UNSHELTERED YOUTH?

• 2026 change: LA CoC introduced Respondent-Driven Sampling (RDS) to estimate unsheltered youth
• Peer-referral method designed for hidden, hard-to-reach populations
• Surveyed youth recruit other unsheltered youth in their social networks to participate who then recruit other youth and so on…
• Pros and cons of a big methodological change like this?

28

GROUP ACTIVITY & DISCUSSION #1:

TREND INTERPRETATION

INSTRUCTIONS:

• Split into groups of 2-3
• Review the chart included in your handout
• Work together to:
1. Write a brief summary of the data and observed trends
2. Brainstorm potential hypothesis for trends observed

​

29

1. Write a brief data summary. Write 2-4 sentences summarizing what these trends show. For example, what's rising, falling, or roughly flat? When did the biggest change(s) happen?
2. Brainstorm potential reasons or explanations for the trends. What could be driving what you're seeing? Think about policy changes, economic conditions, and/or potential changes in who is becoming homeless.

30

• Homelessness is a system with inflows and outflows. When the counts drop, it can mean we're housing more people, fewer people are entering homelessness, or because we’re just seeing them less. The count alone can’t tell you which.
• The homeless count is one important window into homelessness but not the whole picture.

KEY TAKEAWAYS

31

5 MINUTE BREAK

32

THE DEMOGRAPHIC SURVEY:

A CLOSER LOOK

33

INFORMED CONSENT

• Covers purpose of the survey; types of questions included; how the data is to be used and by whom; that participation is voluntary; and the incentive offered.
• Consent language is reviewed every year for potential improvements (e.g., clarity/accessibility, trauma-informed language)
• For 2026 Demographic Survey, we refined language around:
• Information not being shared with law enforcement, including LAPD and ICE
• Data not being used to track or identify participants
• How data are analyzed (e.g., at a population level)
• Goal is to use the consenting process to build trust so participants feel they can engage safely and honestly

34

HUD-REQUIRED VS. LOCALLY-ADDED QUESTIONS

• Every CoCs survey must cover certain items to comply with HUD reporting standards/requirements
• Age, race/ethnicity, gender, veteran status, disability status/health conditions, chronic homelessness, domestic violence experience, and household composition
• LA CoC asks additional questions that have significant policy or research value
• Sexual orientation, foster care history, incarceration history, health insurance status, reasons for homelessness, location where last housed, number of pets

35

SURVEY QUESTION EXAMPLES OVER TIME

• Gender and race/ethnicity have changed substantially in recent years

2023

2025

36

SURVEY QUESTION EXAMPLES OVER TIME

2023

37

SURVEY QUESTION EXAMPLES OVER TIME

2025

38

WHAT THIS MEANS FOR ANALYSIS AND TREND INTERPRETATION

• “The world may not have changed even if the numbers did.”
• Survey results should be interpreted within the context of the questions (e.g., how they’re worded, how they’ve changed)
• Comparability vs. accuracy trade-off
• Old questions might be “wrong” but offer comparability across years
• New questions might be “right” but break the trend line

​

39

TRAUMA-INFORMED PRACTICES

• The demographic survey asks about experiences that are painful—housing loss, disability, violence, incarceration.
• Trauma-informed practices ensure we respect participants’ experiences and right to privacy while collecting honest, usable data
• Every question includes “Don’t know” or “Declined” as valid response options
• Participants are never forced to disclose or fabricate answers
• Sensitive question blocks are preceded by screening item(s)

​

40

GROUP ACTIVITY & DISCUSSION #2:

SURVEY QUESTION DESIGN

INSTRUCTIONS:

1. There will be 3 stations, each corresponding to a potential homeless count demographic survey question
2. Split up into 3 groups with one group at each station
3. Spend ~3 minutes at each station reviewing/critiquing the survey item with the USC data team/staff member lead

​

41

DRAFT HOMELESS COUNT DEMOGRAPHIC SURVEY ITEMS

1. How long have you been homeless?

____ months

​

2. How many people are in your family?

____ adults ____ children

​

3. Do you use drugs? ____ Yes ____ No

42

• Informed consent isn't just a formality; it's where trust and good data begin.
• How you ask a question shapes what answer you get and what your data can (or cannot) tell you later
• Survey items should always consider the research question(s) you want to answer
• When you read published statistics, consider the question that produced that number to understand potential limitation/potential biases

KEY TAKEAWAYS

43

CLOSING REMARKS

44

Thank You! Questions?

THANK YOU |

Please feel free to reach out anytime:

45