Criminological Highlights Vol. 7, No. 6 – February 2006

View this issue as a PDF: PDF iconCrimHighlightsV7N6.pdf

This issue of Criminological Highlights addresses the following questions:

  1. Can police crackdowns on illegal firearms on the street reduce gun crimes?
  2. How can information about the structure of gangs guide police actions?
  3. Does police misconduct increase crime?
  4. Why doesn’t variation in the severity of sentences affect crime levels?
  5. Does reporting domestic violence to the police deter those who have offended?
  6. Are girls getting more violent?
  7. Do police crackdowns on driving offences save lives?
  8. What are the risks of carrying out routine crime audits?

Though not all police crackdowns on gun violence are effective, some seem to be able to suppress gun violence, at least temporarily.  

“Project Safe Neighbourhoods” brought over a billion dollars to cities in the United States to reduce crime, especially  gun violence. The political debate about guns in the U.S. has focused largely on attempts to restrict access to guns by “high risk” individuals (e.g., youths or those with criminal records).  More recently, however, “targeted enforcement” strategies have been used to try to reduce gun violence.  It appears that such strategies  are sometimes effective, but often are not. 

It would seem that there are a few general principles that apply in this area: (a) efforts to reduce gun misuse are likely to be more cost effective than broad efforts to reduce availability, and (b) “for a given level of law enforcement spending, we may achieve a greater deterrent effect by increasing the certainty rather than the severity of punishment” (p. 680; See also Criminological Highlights, 6(2)#1). This paper, therefore, suggests that “enforcement activities could be made more effective by prioritizing… targeted police patrols that seek to deter high risk people from carrying guns illegally” (p. 681). 

A number of strategies that were part of Project Safe Neighbourhoods almost certainly were ineffective or have large costs relative to their value.  These include providing gun locks (to prevent thefts), school prevention programs, better controls on gun purchases, and better tracing of guns involved in crime. Two programs that are sometimes seen as being effective – those in Richmond, Virginia and in Boston (see Criminological Highlights, 7(5)#2) – are shown to be largely ineffective: “Homicide rates in Richmond were trending downward even before the launch of Project Exile” (p. 693-4) just as the decreases in Boston were no larger than decreases in other parts of the state.  Simply put, the flaw in simple claims of success with respect to many American “violence prevention” programs in the 1990s is that the programs were started during a period of broad decline in violence rates in many American cities. Said differently, crime rates went up in the latter part of the 1980s and then began declining before programs such as those in Richmond and Boston were implemented.  Those cities that showed the largest increases in the 1980s showed the largest decreases in the 1990s.  The gun homicide rate in Boston, for example, had decreased by about half  prior to the beginning of the program. It then continued to decrease for about three more years. 

One potentially promising strategy that appears to have been evaluated carefully was used in Pittsburgh. In response to concerns about guns being illegally carried in public places, concentrations of police on the street were increased dramatically (20%-50%) in high risk areas, during high crime periods (specified days and times). The police officers involved in this show of force did not have to respond to normal calls for service. Their focus, instead, was on traffic stops and “stop-and-talk” activities with pedestrians who appeared to have a high “risk” for carrying guns. The analysis involved comparisons of intensively policed areas with control areas, pre- and post-implementation during the targeted times and the “regular patrol density” times. 

Using “assault related gunshot injuries” and reports of “shots fired” as the measures of success, there appeared to be larger decreases in the densely-patrolled areas during the times when there were many police present.  Furthermore, perhaps because of intensive officer training, focused activities, and community involvement, the decrease in gun violence was apparently accomplished without aggravating community-police relationships. The concern, obviously, is that if targeted patrols of this sort were employed in a city, they could be seen giving the police a license to target certain racial (or other) groups. 

Conclusion.  It would appear that targeted increases in police patrols can suppress gun violence at least during the time that the police are present, and that with proper procedures, it is possible to do this without creating strained relationships between the police and the community.   To the extent that the focus can be narrow (i.e., on people and locations likely to have a high rate of carrying illegal guns), and to the extent that there is “extensive officer training and… [involvement of the] community in project design and implementation”  (p. 682) the overall impact can be positive. Nevertheless, it would appear that the effectiveness of such strategies is likely to be limited to those times and locations in which the concentration of police is high.

Reference: Ludwig, Jens (2005).  Better Gun Enforcement, Less Crime. Criminology and Public Policy, 4 (4), 677-716.

Back to top

Systematic analyses of police information about street gangs can be used to break up gangs.

Though it is often difficult to know how to define a gang or to determine whether a particular individual should be considered to be a gang member (see Criminological Highlights, 4(1)#8), there is no doubt that in many cities street gangs are a social problem.  

One of the more popular police approaches to gangs that is often cited, without convincing evidence, as having been effective in reducing homicide rates, is Boston’s Operation Ceasefire (see Criminological Highlights, 7(5)#2 and #1, this issue). Its focus on collective accountability of group members  assumes that gangs have a known hierarchy, are cohesively organized, and are capable of controlling themselves or their members. In practice, this may not always be so.  A cohesive or “tight” gang structure might be disrupted by removing key individuals.  On the other hand, one researcher warns that “treating chaotic groups as cohesive groups may… create a self-determining prophecy. In fact, a law enforcement focus on gangs, through arrest, patrol, surveillance and other mechanisms, can be a powerful external source of cohesion” (p. 610). 

This paper suggests that police should make use of network analysis  -- a focus on the “dynamic interaction among people or groups rather than on the attributes of such individuals or groups” (p. 611).  The importance of knowing the gang structure can be illustrated by a simple comparison.  If a gang is organized hierarchically, such that orders and control come from above, then those at the top of the structure might well be the appropriate people for the police to focus their efforts on.  On the other hand, consider an organization that is not hierarchical, but rather has a few key members who serve, informally, as communication links among non-hierarchically organized members. Searching for “leaders” may accomplish little compared to focusing on individuals who serve this “central” role. 

In this study, police in Newark, New Jersey were able, collectively, to gather information from “interviews” with members of street gangs. Putting together the information that was available about the links between individual gang members made it clear that, in Newark, gangs were not tightly organized. However, an analysis of gang structure revealed that there were certain individuals who served as the connection points between other individuals or groups of individuals.  In other words, if these “connecting” individuals were to disappear, there would be no linkages among various sub-groups or individuals.  For example, in one gang, one individual served as the only connection between two large groups of gang members.  These subgroups, themselves may have internal cohesion, but the “gang” as a whole did not.  In other words, by using information gathered by police, it was possible to understand the importance of specific individuals to the overall structure of the gang.  Descriptions of the “general characteristics” of a gang (e.g., hierarchically structured vs. loosely organized) does not reveal important characteristics of the networks that exist among gang members. By systematically analyzing “knowledge about particular individuals and their known associates, one has the capacity to gather some interesting and powerful information…. Social patterns in relationships can be easily missed or overlooked” (p. 623). 

Conclusion. Police intervention with gangs can be effective, but it also can have paradoxical effects. It would appear that the gangs in this one city were not structured such that a collective-responsibility based approach by police could possibly reduce gang activity. Indeed, such a police strategy might have created a more cohesive and organized gang structure (p. 624). Focusing attention, on the other hand, on individual gang members who occupy key locations within the social structure may be effective in destroying the gang structure.  Obviously, “the utility of this analytic technique for interventions is, at this stage, hypothetical” (p. 628). Nevertheless interventions based on an empirically based “network analysis” would appear to have a higher likelihood of success than interventions based on hypothetical gang structures that may not exist. 

Reference: McGloin, Jean Marie (2005) Policy and Intervention Considerations of a Network Analysis of Street Gangs.  Criminology and Public Policy, 4 (3), 607-636.

Back to top

Police misconduct in highly disadvantaged neighbourhoods can lead to increases in violent crime.

“Conflict between the police and public in structurally disadvantaged neighbourhoods may undermine police legitimacy … If members of disadvantaged communities perceive mistreatment and marginalization by the police, then they may rely on informal methods to redress conflict rather than seek police assistance.  Such a response to compromised police legitimacy may lead to increases in violence… as some residents cease their cooperation with formal legal authorities” (p. 470). 

In this study, data from 74 local police precincts in New York City for the 22 year period from 1975 through 1996 were examined.  An index of structural disadvantage was created by combining data on the proportion of female headed households with children, the percent of black residents, the proportion of households receiving public assistance, the unemployment rate, and the proportion of residents with low educational achievement.  Police misconduct was operationalized as the number of officers compulsorily separated from the department due to misconduct including the number of officers allowed to resign under “questionable circumstances” (e.g., while under suspension or after having been charged).  The dependent measure was the violent crime rate. 

The results are quite straightforward.  Precincts were divided into low, high, and extreme (structural) disadvantage.  Within high and extreme disadvantage precincts, the level of police misconduct predicted the violent crime rate. The effect of police misconduct was higher in the extremely disadvantaged communities. There was no impact of police misconduct on violent crime rates in precincts characterized by low structural disadvantage. 

Conclusion.  The results of this study suggest that police misconduct can lead to increases in crime in the most disadvantaged neighbourhoods.  The findings are consistent with the view that formal institutions, as well as informal institutions, can be important determinants of the crime rate in certain neighbourhoods. “In [the poorest] communities, residents may feel the most marginalized and socially dislocated and they may respond the most adversely to (real or apparent) violations of procedural justice norms by the police, who represent the most visible agents of official social control” …   These findings suggest the importance of police departments meeting procedural justice expectations, specifically in extremely disadvantaged communities” (p. 492). 

Reference: Kane, Robert J. (2005)  Compromised Police Legitimacy as a Predictor of Violent Crime in Structurally Disadvantaged Communities.  Criminology, 43 (2), 469-498. 

Back to top

Harsh sentences don’t deter crime in part because there is no relationship between the sentences that are handed down and people’s knowledge of those sentences.

“The deterrence doctrine asserts that some people will refrain from some acts because they perceive a risk of punishment” (p. 623).  Deterrence-based policies assume that there will be some link between actual punishment and perceptions of punishment. When politicians raise penalties, they assume that people will know about the changes and will act accordingly.  The levels of certainty, severity, and swiftness of punishment can only affect behaviour if they are known. 

This study took place in 54 of the largest counties in the United States. Two sets of measures were collected: criminal justice processing measures, and estimates from members of the general public of these same measures. Data for four offence types were examined: robbery, homicide, aggravated assault, and burglary. For each of the four offence types, “actual punishment” measures in the respondent’s county were obtained on certainty (arrest per 100 offences known to the police and adults convicted per 100 adults arrested), severity (adults sentenced to prison per 100 adults convicted, average maximum sentence imposed), and swiftness of punishment (average days between arrest and sentencing).  Survey respondents in each county were asked to estimate each of these. A number of “control variables” were also measured including the actual crime rate in the county, victimization experience, TV news viewing, the respondent’s view of the importance of the crime problem, as well as various demographic characteristics. 

In general, respondents to the survey underestimated the proportion of offenders who received prison sentences, but were reasonably accurate on the length of the average sentence.  However, the averages tell us nothing about whether individuals living in more punitive locations also perceive their locations to be more punitive. There were no consistent findings that could support deterrence notions.  There were four offences, and five measures of punishment, leading to 20 tests of the deterrence notion. Two were significant and positive (supporting deterrence) – one on certainty and one on swiftness (none on severity). One relationship was in the opposite direction from that predicted by deterrence theory (on severity of punishment).   Since these findings exclude the 15-20% of respondents who were not willing to venture a guess about punishment levels in their area, it would appear that the authors, if anything,  biased the findings “in favour of finding a high correspondence between reality and perception, yet [they] still found virtually no association” (p. 651). 

Conclusion.  In general, it would appear that these findings, derived from a sophisticated survey covering 54 different counties across the United States, support previous findings showing that people are largely ignorant of punishment levels in their communities.  Not surprisingly, then, changes in actual penalties being handed down are not accompanied either by changes in the proportion of citizens who think that sentences are too lenient or by changes in offending rates.  The deterrent effect sentences may have, then, is largely independent of actual severity of these sentences (See also, Criminological Highlights, 6(2)#1). Said differently, people cannot be deterred by severe sentences if they don’t know about them.  “These findings suggest that conventional efforts to increase general deterrent effects beyond their current level are so unpromising that policy makers should consider more productive alternatives beyond merely increasing punishment levels” (p. 655). Indeed, on the basis of these findings, decreases in punishment levels would be just as effective as increases in deterring people from offending.

Reference: Kleck, Gary, Brion Sever, Spencer Li, and Mac Gertz. (2005).  The Missing Link in General Deterrence Research.  Criminology, 43 (3), 623-659.

Back to top

Reporting an incident of domestic assault to the police reduces the likelihood that the victim will again be assaulted.

The effect of various police actions on domestic violence has been a focus of research for the past 20 years. Although initial studies suggested that the arrest of offenders deterred future assaults, later more methodologically sophisticated studies suggested that this conclusion was over-stated. Some studies show no effect of arrest on offending; others show different effects for different subgroups of the population. From a policy perspective, of course, the concern is not only that calling the police, or arresting the offender, might not lead to favourable outcomes, it is also that such actions could conceivably lead to retaliation by the offender.

This paper took advantage of the fact that the National Crime Victimization Survey in the United States interviews people six times during a 3-year period. People who reported being victims of a domestic assault in the beginning of this three year period were the focus of the study. They can be seen as being a representative sample of domestic violence victims during the 1990s. Victims were asked if the police were informed of the assault (by the victim or by someone else) and if the offender had been arrested.  Subsequent waves of the survey, then, were used to determine if the offender re-victimized the domestic assault victim as a function of whether the initial victimization was reported to the police (and if the offender was arrested).   Various “control” factors were included (e.g., the use of a weapon, whether the victim was injured, drinking/drug use, age, race, education of the victim, whether the victim was living in poverty). 

The likelihood of a repeat victimization was dramatically higher for those victims who indicated that the incident did not come to the attention of the police.  In other words, there was an important protective effect of making the police aware that the victim had been assaulted.  Victims were divided into two groups: those who reported that the offender had committed similar offences prior to the incident that was recorded in the first victimization survey and those who reported that this was a first offence.  In those cases in which the victim reported on the victimization survey that it was the first offence, the probability of re-offending was about 10% for incidents that were not was not reported to the police, but only 5% for incidents that had been reported to the police.  

The re-offending rate was, not surprisingly, considerably higher for those offenders with a prior history of offending. Once again, however, the re-offending rate was considerably higher (35%) in those cases that were not reported to the police than in those cases that were reported (22%).   There was no statistically significant effect, however, of arrest. In other words, the evidence appears to suggest that it is the mere fact that the police find out about the assault that is important in deterring the offender.  Furthermore, for those cases that were reported to the police, it did not matter whether the victim or someone else reported them to the police. There was, therefore, no evidence that offenders would be particularly likely to retaliate against victims who, themselves, reported the assault to the police. 

Conclusion. It would appear that bringing a domestic assault to the attention of the police can have protective effects for the victim.  Moreover, “the effects of reporting and arrest on re-offending are similar for serious and less serious offenders” (p. 579).  Furthermore, the effect of reporting to the police appears to be the same for couples of various socioeconomic backgrounds.  Similarly, the lack of an effect on re-offending of the police decision concerning the arrest of the offender does not depend on the socioeconomic or marital status of the couple.  “It may be that a visit from the police changes the offenders’ attitudes toward their behaviour.  Offenders may redefine their behaviour as a criminal act. It may also change their perception of the costs of further violence” (p. 581-2).  Clearly, however, encouraging the reporting of domestic assaults to the police can reduce re-victimization.

Reference: Felson, Richard B., Jeffrey M. Ackerman, and Catherine Gallagher (2005).  Police Intervention and the Repeat of Domestic Assault.  Criminology, 43 (3), 563-588.

Back to top

Girls are not getting more violent. 

There appears to be a belief – among some academics and in the popular press – that girls are becoming more violent than they were at some unspecified point in the past.  To a large extent, this belief is based on reports of high profile violent offences involving girls.  The “scientific” evidence of an apparent increase in violence by girls comes from data suggesting that youth court cases involving girls now constitute a higher proportion of the cases coming to court than they did 10-15 years ago in the United States (and in Canada). 

The data on violence show important gender differences.  The “gender gap” (the difference in the rates for boys and girls) is larger for the more serious forms violence than it is for minor instances of  violence. Hence a policy that increases the number of “low end” cases being recorded is almost certainly going to lead to a larger proportional increases in total violence for girls than for boys. In addition, if, in the past, girls – more than boys –  tended to be screened out of the youth justice system if they were apprehended for minor instances of violence, then a “gender neutral” screening process would naturally lead to an apparent increase in “girl crime.” 

This study looks at three types of data for the period 1980-2003: Police arrests of young people,  victimization reports in surveys, and adolescent self-reports of offending.  The findings suggest that a convergence of the rates of male and female youth violence (i.e., increasing similarity in the rates of violence for girls and boys) is only evident in the police data for overall violence and for assaults.  It is interesting to note that the convergence appears to be larger for simple assault than for aggravated assault.  There is no trend (i.e., no evidence of converging or diverging rates) in the police data for homicide or robbery. It is, therefore, assaults – offences for which there is likely to be more opportunity for police and others not to charge a youth – that are creating the apparent increased convergence of rates. 

What is notable about the curves derived from police recorded data on violence is that much of the recent convergence of the rates for girls and boys is due to relatively recent decreases in the rates for boys rather than increases in the rates for girls. In other words, girls’ rates tend to be fairly constant. Boys’ rates have, in recent years, decreased. Hence girls are described as being (relatively) more violent when the real cause of the change is boys: they are becoming relatively less violent. (There are similar findings from Canadian data.)  The trends for self-reported violence, however, show something quite different. There is “very little overall change both in girls’ assault levels and in the Violent Crime Index and, most notably, essentially no change in the gender gap or female-to-male percentage of violent offending.”  Said differently, girls are not getting more violent.

Conclusion. The findings – that the only evidence of “convergence” of violence rates of girls and boys comes from police-recorded data –  suggest that the perception that girls are becoming more violent is a result of changes in the manner in which youths (girls, but equally importantly, boys) are being dealt with by the system.  The high visibility of police (and court) statistics and the low visibility of much of youth violence may partially explain the perception of increasing female youth violence. Furthermore, “there is no shortage of intuitively appealing speculations for explaining the apparent rise in girls’ violence…” (p. 390).  “Particularly important… is the media[’s]… eagerness to both create and spread conceptions of purported shifts in girls’ violence” (p. 391). 

Reference: Steffensmeier, Darrell, Jennifer Schwartz, Hua Zhong, and Jeff Ackerman (2005) An Assessment of Recent Trends in Girls’ Violence Using Diverse Longitudinal Sources: Is the Gender Gap Closing? Criminology, 43 (2), 355-405.

Back to top

Police crackdowns on bad driving can prevent serious traffic accidents.

In 2001, Canada’s traffic accident death rate was 8.9 per hundred thousand residents – about five times the country’s homicide rate for that year. Various enforcement programs have been used to try to control bad driving.  Most of these programs have the goal of increasing the perceived risk of apprehension.  Although there have been thousands of local police programs designed to reduce the number of serious traffic accidents, only a tiny proportion of these have received adequate evaluations.   

This paper examines programs designed to crack down on problematic driving. Thirty-three high quality evaluations were found.  These involved studies of the effects of random breath testing or “sobriety check points” to reduce impaired driving (15 studies), the use of speed (5 studies) and red-light  cameras (4 studies), studies evaluating a combination of these two “camera” deterrents (4 studies), random “road watch” by police looking for traffic law violators (1 study), and mixed programs (4 studies) using one or more approach.  

The theory behind all of these programs is the same. Rather than focusing on what to do with law violators once they are caught, these programs attempt to affect the behaviour of those unapprehended individuals who might be sensitive to the threat of punishment. If completely effective, then, nobody would be apprehended and punished, but everyone would perceive a high likelihood of apprehension if they were to offend.  Given this underlying theory, it is not surprising that most programs –  23 of the 33 (70%) –  were implemented along with media campaigns.  There was, however, some variation on this dimension.  In 13 of 15 (87%) drinking-driving programs there were media campaigns, compared to only 7 of the 13 (54%) speeding or red-light camera programs

The results were overwhelmingly positive: 30 of the 33 programs showed reductions on such measures as fatal accidents or traffic accidents resulting in serious injuries.  The three studies that did not show a reduction had one thing in common: they were “automated” systems – red light or speed cameras.   It is possible that these “failures” reflect bad judgement concerning the locations in which the cameras were placed or the failure of the program to communicate the increased threat of apprehension that the cameras provided to law violators.

In some cases there was evidence that the effect of the crackdown generalized to locations not covered by the program itself.  Though there was some variability in the size of the effect of these programs (e.g., the reduction in the percent of accidents with injuries ranged from 23% to 31%), there appeared to be more variability in the size of the effect for the “automated” red light or speed cameras. 

Conclusion.  It is clear that police practices that increase the perceived likelihood of apprehension for traffic violations can reduce the number of serious traffic accidents.  It appears that police crackdowns for impaired driving as well as crackdowns of speeding and running red lights can be equally effective.

Reference: Blais, Etienne and Benoit Dupont (2005) Assessing the Capability of Intensive Police Programmes to Prevent Severe Road Accidents. British Journal of Criminology, 45, 914-937.

Back to top

The “crime audit” – an attempt to evaluate how well local governments and police are doing in responding to crime – can have a negative impact on the manner in which a community responds to crime. 

In England, communities must conduct “crime audits” every three years. The audit process involves gathering information about crime from local agencies (e.g., police, probation, hospitals), analysis of this information (e.g., locating problem areas), and disseminating it to the community.  This paper suggests that audits should not be seen as being “neutral” descriptions or analyses of social problems. 

In the first place, crime audits begin with quantitative data that are readily available (e.g., police data) rather than any other information (e.g., citizen experiences of disorder) that are not so available. Moreover, they often focus on only a subset of crimes.  The purpose of the “crime audit” is, in part, to allow local governments to focus resources in effective ways. In addition the audit is supposed to help identify sources of information that are needed. 

One problem is that “What comes to matter [in a crime audit]… is the collection of measurable auditable data, rather than any particular concern with what is being measured” (p. 171). Furthermore, though masses of data may be gathered, the ability of any local government to use such data to guide policy (and ultimately to improve communities) is limited.  There is also a serious problem sometimes referred to as colonization which occurs when “organizational action comes to be guided by the values of the audit” rather than the values underlying the original problem.

In a study of the impact of the audit process on a group of organizations involved in the treatment of alcohol and drug problems within marginalized communities, it was found that the nature of the data required for the audit was seen as “changing the nature of the agency and the workers’ understanding of their tasks” (p. 175).  The data requirements, for example, tended to conflict with the reality of working with drug users which is “often a ‘stepwise’ process [because] many of the clients… feel that they have previously been ‘failed’ by the state agencies” (p. 176).  As one worker put it, “If you ask too many questions too early the clients run off…. If we change too much in order to be able to supply information for the audit, then we’ll have to re-think what we’re about” (p 176).  Because of funding requirements, however, “most [workers] were resigned to making changes in order to fit into the audit process” (p. 177).

Conclusion.  The difficulty with audits is easy to describe. There is a “need of the  audit [process] to have working practices that are auditable [or easily quantifiable] in order that the data being fed into the audit ‘fit’ other types of data collected by a variety of agencies.”  Furthermore, “the need to be part of the audit process and to be included in local crime reduction partnerships has placed pressure on… agency managers to at least consider their priorities when thinking about work and what constitutes good practice” (p. 177). There is nothing wrong with changing priorities or practice.   What appears to be disturbing in cases such as these is not that agencies are changing what they are doing. It is that they are changing what they are doing in order to be “auditable” rather than in order to become more effective.

Reference: Gilling, Daniel and Adrian Barton.  Dangers Lurking in the Deep: The Transformative Potential of the Crime Audit.  Criminal Justice, 5 (2), 163-180.

Back to top

Publication Type

Volume Number


Issue Number