Dissertations & Theses (Open Access)

Date of Award

Fall 12-2018

Degree Name

Doctor of Philosophy (PhD)

Advisor(s)

Steven Kelder, Phd

Second Advisor

Jennifer Gonzalez, Phd

Third Advisor

Casey Durand, Phd

Abstract

Homelessness is a significant barrier to the quality of an individual’s health. Federal housing interventions attempt to eliminate homelessness by providing varying degrees of subsidization, but available units do not currently keep pace with the scale of the problem. Coordinated Assessment is a process requirement put in place by HUD to organize and prioritize the waiting list for housing, based on community priorities. Several tools measuring health and social vulnerability, including the Vulnerability Index-Service Prioritization Decision Assistance Tool (VI-SPDAT), have become popular options for prioritizing housing resources. Such assessments currently in use are not thoroughly researched and the effects of this new paradigm of prioritization based on vulnerability are not well understood.

Confirmatory factor analysis applied three theoretical models to the data and the model with best goodness-of-fit characteristics was further improved using modification indices (RMSEA=0.036; CFI=0.904; SRMR=0.035). Multiple group testing across demographics and exposure to homelessness consistently demonstrated weak invariance. Comparison of the self-reported items measuring health conditions and healthcare utilization to abstracted hospital electronic medical records and data from a Health Information Exchange demonstrated superior data quality from the HIE. Generally, self-reported items tended to show higher specificity and low sensitivity for diagnostic records, although this effect varied between conditions.

Responses to several items on the VI-SPDAT were found to differ significantly by demographic groups or levels of exposure to homelessness. Demographics, homelessness exposure, and multiple individual items were shown to be associated with overall score on the measure in multivariate, negative binomial models. Finally, overall score on the VI-SPDAT was found to have no association with selection for housing placement and housing exit type. However, several items were found to be either positively or negatively associated with probability of housing entry or of negative housing exits.

There are several findings from this research which have immediate relevance and application to the practice of Coordinated Assessment. The racial, ethnic, and gender disparities in overall score and individual items are potential sources of concern. The weak invariance of the factor model may give partial explanation for these. The lack of association between assessment score intended to prioritize housing and selection for housing entry is another point of concern which needs to be clarified. Recommendations for changes to practice and policy are described and specific needs for further research to follow this is explored.

Download

Share

COinS