Research That Matters (January 17 - 20, 2008) |
Method: This study utilizes a pilot survey of homeless individuals participating in Project Homeless Connect in December 2006. The pilot study attempted to develop a scale to measure the causes of homelessness. In total, one hundred fifty five homeless adults were randomly surveyed in seven cities across Los Angeles County. The self-perception scale entailed fifteen questions utilizing a five-point Likert scale. All fifteen questions have been created for six latent variables, work, housing, substance abuse, personal relation, childhood, and services, covering aspects related both childhood and adulthood. For testing measurement model, Amos 5.0 was utilized to conduct confirmatory factor analysis.
Result: Confirmatory factor analysis for self-perception of the causes of homelessness using Amos statistical software suggest that measurement model was a moderately good fit [NFI = .955; NNFI = .954; CFI = .964; RMSEA = .05]. Substance abuse including alcohol and drugs correlated with eviction, divorce, domestic violence or partner abuse, a conflict with family or friends. The model specified the correlation between structural factors and individual aspects.
Implication: The causes of homelessness are not simply one-way path. Complicated paths over measured variables and latent variables improve our understanding of the phenomenon of homelessness. Further investigation in structural model of the causes of homeless should be conducted. Not only shelter programs but also treatment for the homeless who have problems of substance abuse and mental illness increase their chance to cycle out of homelessness.