Abstract
Illegal street racing has become a pressing problem for cities and counties throughout the United States. In response to the problem, San Diego was the first city to pass a spectator ordinance, making attendance at an illegal street race an arrestable offense. Shortly after the spectator ordinance went into effect, a forfeiture ordinance was passed. This provides for forfeiture of vehicles used in illegal speed contests. A recent drop in illegal street‐racing casualties has been attributed to both ordinances, but other explanations are plausible, including an increase in media attention to the problem, a highly publicized prosecution for murder stemming from two illegal street racing fatalities, an increase in the number of sanctioned racing events, and enforcement of certain provisions in California’s Vehicle Code. Even in the face of these possible alternative explanations, several regression analyses revealed that the forfeiture ordinance had the most pronounced effect on street‐racing casualties.
Notes
1. The spectator ordinance provides that “Any individual who is knowingly present as a spectator, either on a public street or highway, or on private property open to the general public without the consent of the owner, operator, or agent thereof, at an illegal motor vehicle speed contest or exhibition of speed is guilty of a misdemeanor subject to a maximum of six months in jail and a fine of $1,000” (San Diego Municipal Code, 0–19118 N.S.). It also applies to people involved in the preparation of illegal speed contests and those who are present, though not necessarily observing, such events.
2. The forfeiture ordinance first provides that “a motor vehicle shall be declared a nuisance and forfeited subject to this division if … [i]t is used in violation of California Vehicle Code sections 23109(a) or (c); and … it is being driven by the registered owner of the vehicle, the registered owner is a passenger, the registered owner’s immediate family members is driving or riding in the car, or the driver or passenger lives at the same address as the registered owner.” Additionally, the offender who triggers the ordinance must have had one or more previous convictions under specific provisions of the Vehicle Code. Thus, some type of previous conviction is necessary for forfeiture to occur. Several other procedural protections are extended to those whose vehicles are targeted for forfeiture.
3. These data were provided to us by the DragNet team in the San Diego Police Department. The data were an improvement over ordinary traffic injury/fatality data because each casualty was validated by a DragNet team member. If, for example, a California Highway Patrol officer suspected that a crash on a San Diego freeway was caused by street racing, or even that street racing was involved, they would contact DragNet, and a representative would be sent to the crash site to determine whether the crash was in fact street‐racing‐related.
4. We explored other coding schemes, such as gradual, permanent effects, but the forfeiture and spectator coefficients did not achieve significance. Also, recall that the spectator and forfeiture ordinances went into effect on October 21, 2002 and April 21, 2003, respectively. Because our observations were weeks, and because the ordinances were not necessarily enacted at the start of one of those weeks, the first week containing a code of “one” for the ordinances was the week with the most days on which the ordinance was in effect. For example, the spectator ordinance variable was coded with a 1 starting on the week beginning on October 22, 2002. We did not feel comfortable coding the preceding week with a 1 because the ordinance was in effect for only one day during that week.
5. The main reason we did this was that autocorrelation functions did not clearly demonstrate whether autocorrelation was present. To err on the side of caution, though, we nevertheless assumed that it could have been and thus estimated several autoregressive Poisson regression models using Stata’s user‐written ‐arpois‐ command. Such models have been both developed and used extensively in the field of epidemiology (e.g., Katsouyanni et al., Citation1996; Schwartz et al., Citation1996). We discuss these analyses in the Results section.
6. The t tests were set up such that there were equal numbers of observations before and after the ordinances went into effect (84 observations for the spectator ordinance and 72 for the forfeiture ordinance). This approach restricted the number of observations in each test but was necessary to make the t tests valid.
7. ZINB models assume there are two latent groups‐one in the always‐zero group and another in the not‐always‐zero group. In the present context, this would mean weeks that will always have no street‐racing casualties and weeks that will always have a positive probability of having a street‐racing casualty. We report the ZINB results that assume there is a positive probability that there will be a street‐racing casualty in any given week. The reason is that these results are most comparable with the NBRM results in the first column of Table .
8. In some early analyses, we included arrests in models of street‐racing casualties. Arrests was never significant, so we excluded it. It was also excluded in such models because the standard errors were very high, even though arrests was not highly correlated with other covariates.
9. The convention in the literature is to argue that crime‐control interventions can have deterrent and incapacitative effects (Kessler & Levitt, Citation1999). However, since forfeiture targets property and does not require a criminal prosecution or conviction, or even confinement, the forfeiture ordinance could not have had an incapacitative effect. Had the ordinance been used during the period covered by our analysis, it could have had a “forfeiture effect.” But since it was not used, we are left to conclude that the ordinance had a deterrent effect.