Friday, 14 January 2005 - 10:00 AM

This presentation is part of: Child Welfare Services

Using Exploratory Spatial Data Analysis to Examine Neighborhood Structure and Foster Care Entry

Purpose: Child welfare practitioners are increasingly focusing on the role of neighborhood quality of life in their goals to improve permanency outcomes for abused and neglect children. However, most research has focused on social structural factors associated with child maltreatment and has largely ignored spatial context within which these processes operate. To the extent that geography is taken into account, neighborhood incidence rates and descriptive GIS (Geographic Information System) maps displaying rates or locations of child maltreatment may help target services to high-risk areas. These tools are useful for preliminary assessment but they are limited in two respects: 1) they are subject to misinterpretation due to the spatial nature of the data (lack of independence among the units of analysis) and the underlying population at risk (rare events like foster care entry create unstable rates across neighborhoods), and 2) they do not reveal information about neighborhood-level mechanisms or risk factors associated with child maltreatment.

Method: This study uses a cross-section of first entries to foster care from a California county from 2000 through 2002 to perform exploratory spatial data analysis (ESDA) of the relationship between neighborhood structure and foster care entry. First, variance instability of foster care entry rates is examined across zip codes, census tracts and block groups in search of spatial outliers and autocorrelation that can lead to faulty inference in non-spatial descriptive analysis, particularly when using administrative units as proxies for neighborhoods. Next, the following question is addressed using Monte Carlo simulation univariate and bivariate tests for spatial autocorrelation: what is the impact of foster care entry in one location on entry in nearby locations?

Findings: While a standard quartile incidence map displays several neighborhoods at elevated risk, one area emerges as a spatial outlier after adjusting for areas with small populations at risk and spatial autocorrelation. Preliminary results at the zip code level offer evidence that neighborhoods with high rates of foster care entry are significantly clustered both globally (Moran’s I=.32, p<.01) and locally. At the zip code level, all five measures of impoverishment in neighborhoods are related to rates of foster care entry in nearby neighborhoods, pointing to a possible diffusion process.

Implications: For child welfare service planning and research, ESDA descriptive and graphical tools can be used to detect spatial outliers, patterns of spatial association and clusters that non-spatial data analysis and traditional maps are not designed to reveal. These tools can generate hypotheses about complex spatial processes that can be tested with confirmatory spatial data analysis. Specifically, this study offers evidence that certain neighborhood structural characteristics and their geographic locations are related to foster care entry risk. Child welfare planners and practitioners can use ESDA maps that adjust for variance instability and non-independence among neighborhood units to more accurately target preventive services and foster parent recruitment in the areas at highest risk. In addition, although the underlying mechanics of ESDA are complex, the resulting maps and plots are visually appealing and easily understood by practitioners and others who may not have statistical backgrounds.

See more of Child Welfare Services
See more of Oral and Poster

See more of Celebrating a Decade of SSWR (January 13 - 16, 2005)