Tipping the Scales of Justice

Abstract

Rose and Clear propose that neighborhood incarceration, after a tipping-point threshold, can disrupt informal social-control mechanisms in neighborhoods producing more crime and violence. They hypothesize that the incarceration-crime relationship at a neighborhood level is curvilinear. Using suggestions from Hannon and Knapp, the authors assess curvilinearity in the incarceration-crime relationship by comparing results across three different estimation techniques (ordinary least squares, heteroscedasticity consistent covariance matrix [HCCM] for small samples, called HC3, negative binomial). Data from 95 Portland, Oregon, neighborhoods are used for the study. The results are generally consistent and supportive of Rose and Clear’s curvilinearity hypothesis of neighborhood incarceration but only for explanations of violent crime, not property crime. Moderate levels of neighborhood incarceration are related to significant violent crime rate decreases, and high levels of neighborhood incarceration are significantly related to violent crime rate increases. However, models were sensitive to estimation technique and outlying observations. Policy considerations related to concentrated neighborhood incarceration are discussed.