Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Understanding the Personal Networks of People Experiencing Homelessness in King County, WA with aggregate Relational Data

Zack Almquist, Ihsan Kahveci, Owen Kajfasz, Janelle Rothfolk, Amy Hagopian

Abstract

The social networks of people experiencing homelessness are an understudied but vital aspect of their lives, offering access to information, support, and safety. In 2023, the U.S. Department of Housing and Urban Development reported 653,100 people experiencing homelessness on any given night -- a 23% rise since 2022, though likely an undercount. This paper examines a unique three-year dataset (2022-2024) of survey responses from over 3,000 unhoused individuals in King County, WA, collected via network-based sampling methods to estimate the unsheltered population. Our study analyzes the networks of the unsheltered population, focusing on acquaintance, close friendship, kinship, and peer referral networks. Findings reveal a decline in social connectivity over time. The average number of acquaintances dropped from 80 in 2023 to 40 in 2024. Close friendship levels remained stable at 2.5, but given the growth in the homeless population, this suggests decreased network connectivity. Kinship networks expanded, indicating that more family members of unhoused individuals are also experiencing homelessness. These trends suggest increasing social disconnection, possibly driven by displacement and a rise in newly homeless individuals. The growing isolation may reduce opportunities for information sharing and mutual support. However, the increased reliance on family networks highlights the shifting dynamics of social support within this community. This research underscores the need for policies fostering social connections and community building, such as reducing displacement and providing spaces for congregation, to counter the growing anomie among unhoused populations.

Understanding the Personal Networks of People Experiencing Homelessness in King County, WA with aggregate Relational Data

Abstract

The social networks of people experiencing homelessness are an understudied but vital aspect of their lives, offering access to information, support, and safety. In 2023, the U.S. Department of Housing and Urban Development reported 653,100 people experiencing homelessness on any given night -- a 23% rise since 2022, though likely an undercount. This paper examines a unique three-year dataset (2022-2024) of survey responses from over 3,000 unhoused individuals in King County, WA, collected via network-based sampling methods to estimate the unsheltered population. Our study analyzes the networks of the unsheltered population, focusing on acquaintance, close friendship, kinship, and peer referral networks. Findings reveal a decline in social connectivity over time. The average number of acquaintances dropped from 80 in 2023 to 40 in 2024. Close friendship levels remained stable at 2.5, but given the growth in the homeless population, this suggests decreased network connectivity. Kinship networks expanded, indicating that more family members of unhoused individuals are also experiencing homelessness. These trends suggest increasing social disconnection, possibly driven by displacement and a rise in newly homeless individuals. The growing isolation may reduce opportunities for information sharing and mutual support. However, the increased reliance on family networks highlights the shifting dynamics of social support within this community. This research underscores the need for policies fostering social connections and community building, such as reducing displacement and providing spaces for congregation, to counter the growing anomie among unhoused populations.

Paper Structure

This paper contains 49 sections, 4 equations, 17 figures, 15 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Study participants
  4. Study design
  5. Network Instruments
  6. Statistical modeling approach
  7. Count data models with GLM structure
  8. Zero-Inflated Poisson (ZIP) Model
  9. Zero-Inflated Negative Binomial (ZINB) Model
  10. Model selection
  11. Model fitting and selection for count data models for degree
  12. Degree distribution model for the peer referral
  13. Results
  14. Network degree of acquaintanceship, close friends and kinship
  15. Mean degree and degree distribution
  16. ...and 34 more sections

Figures (17)

  • Figure 1: Illustration of the longest chain in 2024 with the degree distribution and geospatial location of the interview (labeled as RDS hub location).
  • Figure 2: King County, WA plotted with the 17 RDS Hub locations from the 2024 PIT survey. Urban areas are colored in red, and rural areas are colored in grey. King County, WA, which has around 2.3 Million people, contains Seattle, WA, the largest city at around 750,000 people and is the region's economic hub.
  • Figure 3: Degree distribution for kinship, list-based network, and peer referral network for 2022, 2023, and 2024. Note that 0s in close friendship in 2023 and 2024 are artificially inflated by including skips; based on the aggregate data, we would not expect more than 9-10 percent of the sample to be zeros and those only in the "seeded" individuals.
  • Figure 4: Degree distribution of the kinship network for 2022 to 2023 with density plot (red).
  • Figure 5: Degree distribution of the close friendship network for 2023 and 2024. In 2023, we top-coded the instrument at 15 (subjects were not made aware of this fact); in 2024, we top-coded at 20.
  • ...and 12 more figures