WHOSE CIVIC COMMUNITY? TESTING ALTERNATIVE HYPOTHESES OF THE RELATIONSHIP BETWEEN CIVIC COMMUNITY AND RACIAL INEQUALITY IN ARREST RATES

Abstract

This analysis extends prior research on racial inequality in arrest rates by testing opposing hypotheses derived from the civic community tradition in sociology. One interpretation of this perspective suggests that communities with a locally-oriented economic infrastructure, an abundance of civic institutions, and civically engaged citizens should exhibit lower racial disparity in arrest rates. On the other hand, recent writings exposing a potential “dark side” of social capital suggest that civic community may exacerbate such racial disparties. Using spatial regression models to analyze data from nearly 1,800 U.S. counties, the current analysis examines these competing hypotheses regarding the effects of civic community. Results suggest mixed support for both the traditional and “dark side” predictions. Implications for theory and empirical research are discussed.

Notes

¹It should be noted that these studies combine indicators of political participation with other indicators of civic activity. Hence, the independent impact of political activity on crime rates is somewhat unclear.

²We also conducted our analyses using two alternative approaches to measuring the racial disparity in arrests. One method involved taking the logarithm of the ratio of the black-to-white arrest rates. The second method involved computing the raw difference between the black and white arrest rates and then dividing by the total arrest rate. The general pattern of results and the conclusions suggested by each method are quite similar.

³We multiply the 2000 census population figures by 3 so that the population base is on the same scale as the three-year sum of arrest counts.

⁴Preliminary models indicated that a positive decelerating slope best approximates the relationship between the small manufacturing measure and the racial disparity in arrests, so we transform this variable to the natural logarithm scale.

⁵See http://enceladus.isr.umich.edu/race/racestart.asp

⁶A measure of drug crimes known to the police is not available in the FBI's Uniform Crime Report.

Note. Unstandardized coefficients reported; standard errors in parentheses.

*Coefficient ≥1.96 times associated standard error.

^a In model for violent arrests, this measure reflects rate of violent “index” crimes known to police; in models for property arrests, this measure reflects rate of property “index” crimes known to police; in models for drug arrests, this measure is the rate of (total) index crimes known to the police.

Note. Unstandardized coefficients reported; standard errors in parentheses.

*Coefficient ≥1.96 times associated standard error.

^a In model for violent arrests, this measure reflects rate of violent “index” crimes known to police; in models for property arrests, this measure reflects rate of property “index” crimes known to police; in models for drug arrests, this measure is the rate of (total) index crimes known to the police.

⁷Consistent with current best practices, we assessed the extent to which the sets of explanatory variables in our regression models show indications of high levels of multicollinearity. Generally, they do not. The largest variance inflation factor (VIF) in any model that we estimate was 3.19, with the average VIF never rising above 2.0.

Note. “n.s.” indicates effect is not significant at .05 level; “+” indicates a significant positive effect; “–” indicates a significant negative effect.

N = 1,799.