Geospatial Statistical Analysis of the Comparative Associations of Poverty and Public Assistance Program Usage from U.S. Census Data in Core City, Suburban and Rural Areas in a Western New York County"

Background and purpose

Poverty policies are subject to political shifts by local and federal governments regarding how much support the poor should receive and how those supports should be funded. The Personal Responsibility and Work Opportunity Reconciliation Act (1996) changed federal funding from direct benefit support to discretionary state-wide block grants. This shifted much of the financial burden to state and local governments (Trattner, 2007), reducing safety net programs to limited and short term health and human services  (Waitzkin, 2005).

There is a continuing push to reduce federal safety net funding by $20 million over 10 years (Legislative background on SNAP funding,  2013) at the same time as enrollment has increased by over 70% since 2007 (Cancian, Han, & Noyes, 2014; Ferguson, Militana, & Brasher, n.d.; Gilbert, Nanda, & Paige, 2014). Despite prejudicial assumptions, little is known about profiles of recipients and recipient families as they compare to poverty levels (Mykerezi, Mills, & Melo, 2013). This paper presents a geospatial statistical analysis by poverty rate of usage of public assistance programs that enrollees can apply for online versus those that require travel to county offices.

Methods

This study performed a secondary analysis of aggregate data from 230 census tracts within a metro Western New York county using the 2014 U.S. Census Bureau American Community Survey 5-year estimates collected from the AmericanFactfinder website. Data were imported into SPSS for analysis and percentages in each tract were calculated for rates of households below the poverty level in the prior 12 months. Poverty rates were then matched with rates for households receiving public assistance income (typically require in person applications), and households receiving food stamps (includes options for online application). Furthermore, each tract was coded by ring position (core city, n=78; first ring suburbs, n=86; second ring suburbs, n=39; rural, n=27) to the county social service center where in-person public assistance applications were conducted. Using Poverty Rate as an independent variable, the relationship between food stamps and public assistance income rates were compared within the county overall and within clustered tracts relative to ring positions.  The census tract data were imported into ArcGIS to present usage rates by the regional rings visually.

Results

Poverty rates were highly correlated with both Food Stamps rate (r=.94) and public assistance rate (r=.78). However, public assistance rate (.034) was significantly lower than either the poverty rate (.163) or the Food Stamps rate (.180). Across the four rings, the poverty rate to Food Stamps enrollment correlation ranged between .76 and .91. However, the poverty rate to public assistance rate correlation was markedly lower for the rural ring (.18) than in the three inner rings (.51-.68).  City core and first ring suburbs had significantly higher rates of both poverty and Food Stamp use than the second and rural rings. All rings differed significantly from the city core in public assistance rates. 

Implications for Practice

This study demonstrates the importance of the collaboration between researchers and county social service agencies to assess potential barriers to enrollment within geographic locations and suggests possible remedies.