Examining how structural characteristics and the physical environment simultaneously impact crime in neighborhoods: Using a semi-parametric strategy

Abstract

This study examines the associations between various social and physical environmental characteristics and their interrelated influence on neighborhood crime. Using Kernel Regularized Least Squares (KRLS), we estimate the marginal effects of each independent variable at each datapoint by providing pointwise estimates of partial derivatives. Then we regress the derivative values for each independent variable on each other variable in the model to examine whether these derivative estimates (marginal effects) vary by other variables in the model. We found that the effects of the physical environment on different types of crime in neighborhoods vary by different levels of social structural characteristics. We simultaneously assess how the two different types of neighborhood environments can work together in a semiparametric way, theoretically integrate both social disorganization and criminal opportunity perspectives, and thus provide a more comprehensive as well as nuanced explanation of neighborhood crime.