Threat to Whom? Conflict, Consensus, and Social Control

Abstract

A number of observers have questioned the efficacy of model specifications designed to falsify the conflict perspective of crime control. Exploiting the quasi-experimental qualities of Cincinnati's recent race-related riot, this study employs interrupted time series techniques to evaluate a component of conflict theory, the racial threat hypothesis. In brief, the autoregressive integrated moving average (ARIMA) transfer function models indicate that, upon controlling for the potentially confounding effects of robbery offenses, the April 2001 riot led to a lasting increase in the level of robbery arrests. The implications of this investigation for distinguishing between the predictions of conflict and consensus theories of crime control are discussed.

Notes

Sources: Homicide Trends in the U.S.: Trends by Race; Table 42: Personal crimes of violence: Percent distribution of single-offender victimizations, based on race of victims, by type of crime and perceived race of offender, 1996–2004 (Bureau of Justice Statistics).

¹Given the focus of prior research (Eitle et al. 2002; Liska and Chamlin 1984), one might question the decision to examine the influence of the Cincinnati riot on total, rather than race-specific, robbery arrests. Recall, however, that these studies employ estimates of the racial composition of criminal victimizations to quantify the level of interracial threat. Thus, depending on the racial characteristics of the offender–victim dyads, one can readily derive unique predictions from racial conflict theory with respect to race-specific arrests. For example, based on the conflict perspective, one would hypothesize that black-on-white crime would lead to an increase in black arrests (the racial threat hypothesis), but that black-on-black crime would lead to a decrease in black arrests (the benign neglect hypothesis).

The relationship between race-related riots, as a measure of interracial threat, and arrests are another matter. Based on racial conflict theory, one would postulate that a white majority would use its influence to pressure legal authorities to devote their limited resources to respond to crimes that are most threatening to them. Thus, for the reasons articulated in the body of this article, I predict that the Cincinnati riot would produce an increase in robbery arrests. However, in the absence of information concerning the racial composition of the actors involved in post-riot robberies, there is no reason to anticipate that the riot will vary across racial sub-categories of arrest.

Lg = Natural logarithm transformation.

θ₁ = First order moving average parameter.

θ₁₂ = First order seasonal moving average parameter.

α = Constant.

e_t = Error term.

ω_o = Zero-order input parameter of a transfer function.

ω₁ = Transfer function parameter estimate for the effect of the prewhitened robbery crime series on robbery arrests.

Q = Test statistic for the null hypothesis that the model residuals are distributed as white noise.

B = Backward shift operator.

I_t = Intervention series.

Z_t = Prewhitened robbery crime series.

∗∗=p < .01.

∗∗∗=p < .001.

²Monthly arrest data for index offenses, including robbery, are unavailable for 2005. Hence, the robbery arrest time series for Dayton and Columbus span the years 1996 through 2004 (n = 108).

³An anonymous reviewer suggests that I compare the results from the ARIMA analyses to findings that from alternative statistical techniques. I decline to do so because I am aware of no statistical procedure that can evaluate the impact of the 2001 Cincinnati riot on robbery arrests as efficiently as ARIMA models. To be sure, one could do a t-test to examine the change in the level of an outcome series before and after the occurrence of an intervention. However, this approach could lead one to conclude that there is a statistically significant change in the mean level of an outcome series when the impact of the intervention is minuscule. Suppose, for the sake of the discussion, that our outcome time series were decreasing for the entire period under investigation. Even though there were no change in the pattern of arrests after the intervention, a mean difference t-test would probably reveal a significant decline in the level of arrests subsequent to the intervention point. ARIMA modeling procedures, which take into account ongoing process of change (via the identification and estimation of the noise component of a transfer function model), would capture this pattern of decline and yield null findings for the effect of the Cincinnati riot.

Additional information

Notes on contributors

Mitchell B. Chamlin

MITCHELL B. CHAMLIN is Professor of criminal justice at the University of Cincinnati. He is also Editor for The Journal of Crime and Justice. Currently, he is studying the contextual effects of ongoing natural experiments on crime and crime control outcomes.