Emergency homeless shelter use in the Dublin region 2012–2016: Utilizing a cluster analysis of administrative data

City:

Dublin, Ireland

Organization:

·      School of Natural and Built Environment, Queen's University Belfast

·      Local Government Management Agency, Ireland

·      School of Architecture, Planning and Environmental Policy, University College Dublin

Project Start Date:

2017 (Inferred)

Project End Date:

October 2018 (Paper submitted)

Reference:

Waldron, R., O’Donoghue-Hynes, B., & Redmond, D. (2019). Emergency homeless shelter use in the Dublin region 2012–2016: Utilizing a cluster analysis of administrative data. Cities, 94(October 2018), 143–152. https://doi.org/10.1016/j.cities.2019.06.008.

Problem:

Ireland has experienced a dramatic rise in homelessness since the 2008 economic recession, with an increase of 130% in demand for emergency homeless accommodation from July, 2014 to December, 2016 alone. Dublin accounts for over 70% of the total homeless population of the country. Understanding the patterns of emergency accommodation use among the city’s homeless population will help Dublin’s municipal government create meaningful strategies that allocate limited resources to the actual needs of those experiencing homelessness.

Technical Solution:

·      K-Means Cluster Analysis was conducted within SPSS to create three unique clusters of homeless clients using the Z scores for the ‘total nights stayed” and ‘total homeless episodes” variables, applying Kuhn & Culhane’s (1998) cluster typology

·      Chi-Square tests examined whether statistically significant relationships exist between the homeless clusters and the demographic and patterns of use variables. Relationships were considered significant at an alpha level of 0.05

·      Cramer’s V tests were conducted to measure the strength of association between variables within clusters

Datasets Used:

• Dataset: Pathway Accommodation and Support System (PASS), Dublin Region Homeless Executive, January 2012 to December 2016, inclusive

Outcome:

Alignment between Dublin cluster patterns and those from similar studies in cities in U.S., Canada, and Denmark.

K-Means Clusters:

·      Transitional:

o   78% of clients; 35% of total nights

o   72% total nights stayed: < 100 days; 75% of total homeless episodes: = 1

o   64% number of emergency accommodations providers used: 1-2

o   60% Irish

·      Episodic:

o   10% of clients; 15% of total nights

o   56% total nights stayed: 100 > 500; 53% total homeless episodes: 3 < 5

o   40% number of emergency accommodations providers used: 11-20

o   65% Young adults; 33% Middle age

o   77% Irish; 4% non-EEA

·      Chronic:

o   12% of clients; 50% of total nights

o   K-Means51% total nights stayed: 500 >1000, 11% > 1000

o   17% number of emergency accommodations providers used: 11-20

o   58% Young adults; 39% Middle age

o   83% Irish; 6% non-EEA

Significant variables within clusters, by Chi-Square and/or Cramer’s V values:

·      Homeless Nights

o   X² = 789.057, p = 0.000

o   Cramer’s V = 0.546

·      Homeless Episodes

o   X² = 8359.200, p = 0.000

o   Cramer’s V = 0.546

·      Accommodation providers

o   Cramer’s V = 0.416

·      Age

o   X² = 504.479, p = 0.000

·      Country of origin

o   X² = 451.949, p = 0.000

Issues that arose:

·      The dataset only captured cases of reported homelessness where an individual has sought emergency accommodation support, and thereby logged into the PASS system. Therefore, the study does not capture the ‘hidden homeless’ e.g. individuals couch surfing or sleeping outside

·      Until 2014, the PASS system only logged individual adults contacting homeless services, and did not distinguish if they were accompanied by children or not, so difficult to identify the number of families experiencing homelessness from 2012 to 2014.

·      Difficult to link data from PASS to other datasets relating to health or social protection systems due to privacy issues, so study was limited in demographic data, particularly around reasons for homelessness e.g. physical or mental health issues

Status:

Terminated.

Entered by:

October 30, 2020: Malini Pandya, malini.pandya@utoronto.ca