On the Limits of Social Control: Structural Deterrence and the Policing of “Suppressible Crimes
Robert J. KaneJustice QuarterlyVol. 23 No. 22006pp. 186-213,
Kane examines the relationship between arrests and levels of suppressible crimes such as burglary and robbery. Kane’s research hypothesis is that arrest activity will predict a decrease in burglary and robbery rates. Arrests will have no impact on assault rates because there is no economic motivation with assaults, which makes them not suppressible. The author’s data are from the New York Police Department arrest data from 1989-1998. Kane uses a one-month lag between observation points to obtain the most accurate representation of arrest and crime trends during the study period. The author does note that he cannot distinguish between when police are using crackdowns and the general arrest practices. Kane measured neighborhood by police precincts, and only assault, robbery, burglary and murder were factored into the crime rate for each precinct.
Kane’s analysis measured arrest in two ways, and asserts that arrest is an indicator of police aggressiveness. The first way Kane measured arrest is a raw arrest number of the entire precinct for a particular month. The second way Kane (2006) measured arrest was the number of arrests per officer per month per precinct. More specifically, Kane “[summed] the number of monthly arrests within precincts for the crimes of homicide, robbery, and aggravated assault,then divided that total by the number of officers assigned to each precinct during that month” (p. 197).
Because Kane used the police precinct measurement there was a potential for spatial autocorrelation because crime in one precinct could possibly be influenced by crime in an adjoining or relatively close precinct. To control for spatial autocorrelation, Kane created a variable using the general population potential procedure, and this procedure estimates the possible effects of autocorrelation “while controlling for correlation error terms and the spatial effects variable” (p. 202). Kane also used a two stage least squares regression analysis.
Kane found that there were no direct linear effects of arrests, for both measurements of arrest, on robbery and burglary rates. He did find that the non-linear effects were significant for the per officer arrest measurement. This indicates that there is a saturation point with arrests. Once a certain level of arrests is met, arrests discontinue having an effect on crime rates. Consistent with his hypothesis, Kane found that there was no effect, linear or nonlinear,of arrests on assault rates. The author also found the same results with the raw arrest measurement as with the per officer arrest measurement. This study did find that his autocorrelation variable was significant in all models, indicating a need for the control. The author does note that while his findings are moderately promising because of the non-linear effect findings, Kane cannot be sure whether the effects were because of deterrence or incapacitation of offenders.
Daniel Lytle, Troy PayneUniversity of Cincinnati
