Abstract: Applying Propensity Score Matching to Assess If Social Integration Predicts Neighborhood Perceptions (Society for Social Work and Research 23rd Annual Conference - Ending Gender Based, Family and Community Violence)

310P Applying Propensity Score Matching to Assess If Social Integration Predicts Neighborhood Perceptions

Friday, January 18, 2019
Continental Parlors 1-3, Ballroom Level (Hilton San Francisco)
* noted as presenting author
Jason T. Carbone, MSW, Doctoral Student, Saint Louis University, Saint Louis, MO
Background:  Objective neighborhood conditions have long been associated with individuals’ health and well-being outcomes. More recently, a growing body of literature has begun to explore the impact of neighborhood perceptions on health outcomes including mental health outcomes.  Social integration, which represents the degree to which individuals feel connected to society, is a factor that could potential influence neighborhood perceptions and by extension have important implications for health and well-being outcomes.  Using longitudinal data, this study tests the hypothesis that increased social integration will results in more positive neighborhood perceptions.

 Method:  Two waves of the Midlife in the United States (MIDUS) study are utilized in this study.  Social integration, based on summed scores of three Likert-style questions, was dichotomized (low and high social integration) and only respondents whose social integration scores remained constant across both waves were used in the analysis (n=800).  The MatchIt package in R was used to create propensity scores based on a wide range of socio-demographic variables, including age, sex, marital status, number of years they have lived in their current neighborhood, perceptions of neighborhood cleanliness, perceptions of physical neighborhood conditions, frequency of contact with neighbors, and frequency of conversations or get-togethers with neighbors.  Individuals in the low social integration group were then matched to those in the high social integration group via the nearest neighbor method with a caliper of 0.10.  A leave-one-out sensitivity analysis was completed to assess the degree of confounding introduced by each variable when removed from the matching model and the order of magnitude of other potential confounders.  The matched social integration data, along with all covariates included in the propensity score calculation, were included in a linear regression model to predict neighborhood perception.

 Results:  Controlling for covariates, high social integration results in a more than seven-percent increase in neighborhood perception (beta=0.22, p<0.001).  Baseline neighborhood perception was also a large predictor of neighborhood perception at the second time-point (beta=0.43, p<0.001).  Other significant covariates associated with lower neighborhood perceptions include being female (beta=-0.05, p<0.05) and being a divorcee (beta=-0.09, p<0.001).  Overall, the model accounts for more than one third of the variance in neighborhood perception at time two (R2=0.34).  Results of the sensitivity analysis suggest that additional confounding is likely to be minimal.

Conclusion and Implications:  As a whole, these findings have important implications for individuals working to facilitate positive community change.  While a growing body of literature has demonstrated that neighborhood perceptions are associated with health and well-being, research to date has failed to identify factors that influence perceptions of neighborhoods.  Although this study found that social integration is associated with neighborhood perception, high social integration has a relatively small impact on neighborhood perceptions.  While social integration is important, researchers need to further explore variables that impact neighborhood perceptions.  For community practitioners, strengthening social integration should be a part of their work, but other factors that may have a greater influence on neighborhood perceptions should be a higher priority.