Criminological Highlights Vol. 7, No. 5 – November 2005

View this issue as a PDF: PDF iconCrimHighlightsV7N5.pdf

This issue of Criminological Highlights addresses the following questions:

  1. Why don’t the police stop crime?
  2. Were the large scale crime prevention programs in New York, Boston, and Richmond successful?
  3. Are wrongful convictions a large problem in criminal courts?
  4. Why did New Zealand increase its level of imprisonment?
  5. Can lay judges make reliable judgements about pretrial release of criminal defendants?
  6. Can judges predict recidivism?
  7. Do boot camps reduce reoffending?
  8. How might judges help the media understand complex judicial decisions?

Why don’t the police stop crime?  Largely because they are not well placed to do so.

The public often looks to the police to stop almost all types of crimes – from household burglaries, vandalism, and impaired driving to domestic violence, gun crimes, gang violence, and pornography (though they typically are not seen as responsible for preventing crimes involving senior officers of large corporations such as Hollinger, Inc.)  If this is the way in which they are seen, why do we have so much crime? A number of different answers can be offered.

First, their main role in dealing with crime has to do with apprehension of offenders and aiding in their prosecution.  This has little effect on crime rates.  Though some crime is avoided through the apprehension and imprisonment of serious offenders, the impact of these activities on overall crime rates is limited (see  Criminological Highlights, 3(1)#1).  The presence of the police generally may have a deterrent effect. This is sometimes evident in property crime sprees when police go on strike or when, for other reasons, they are not available for apprehending offenders. However, their overall impact on crime in normal circumstances is clearly limited, and it is probably limited to certain types of offences.  

Second, “there is increasing recognition that policing is not just the business of … police forces,” and that “problem oriented strategies involve cooperation in crime reduction with a wider range of departments and agencies” (p. 5).   Long term trends in crime rates (e.g., the decrease in reported rates of serious violence that have occurred in the U.S. and to some extent in Canada in the past 10-15 years) appear to have little to do with the police.  Though the police sometimes take credit for crime drops (e.g., New York City – see Criminological Highlights, 1(4)#5 and this issue, #2), the evidence often is otherwise.  

Third, it appears that many traditional police tactics are not very effective. Much crime is not detectable by the police and is unaffected by traditional unsystematic police patrols. This activity consumes large portions of police budgets yet it appears to have little overall impact.  This is not surprising: crimes are rare events and it is unlikely either that a police officer would be in a position to intervene or apprehend offenders at the scene of a crime.  Increasing patrol density does not appear to have an impact on crime just as decreased “response times” seldom affect crime levels.  The theory that through dealing aggressively with minor incivilities, crime can be reduced, though popular, appears to be without empirical foundation (see Criminological Highlights, 5(1)#6; 3(3)#1). Police organizations have a range of legitimate priorities and concerns that may, at times, not be consistent with strict enforcement of the law.

Finally, even if certain techniques could be identified that were effective, the police organization is one that can easily resist change. In part change in police organizations is difficult to accomplish because “discretion increases as one moves down the hierarchy” (p. 15) in police organizations.  Evidence-based policies are therefore, more difficult to implement than they would be in other organizations.  

Conclusion.  To say that the police are not an important force in preventing crime is not a criticism of police organizations.  “[Police] need to be alert to the dangers of concentrating single-mindedly on traditional approaches to crime reduction.  Doing so not only has inherent dangers, but it can also divert attention from other tasks and objectives of policing” (p. 19).  One might suggest, therefore, that those responsible for policies related to policing should examine carefully how police resources can best be allocated to accomplish the various responsibilities allocated to the police.  Such an approach might lead to a different, and more effective, allocation of scarce resources. 

Reference: Dixon, David (2005) Why Don’t the Police Stop Crime? Australian and New Zealand Journal of Criminology, 38(1), 4-24.

Back to top

Widely publicized police interventions in three American cities show more consistency in their claims than in their effects in reducing homicide rates. 

During the 1990s, homicide rates were dropping in many American cities.  It was inevitable, therefore, that there would be a “chorus of self-congratulation” from politicians and police chiefs who could claim to have changed some part of the criminal justice system prior to or during the drop in reported crime.  Hence aggressive policing, youth curfews, targeting career criminals, adding more police officers, and policies that encouraged community policing were all used to explain local crime drops. These explanations ignored the fact that the programs were typically implemented locally, but the “crime drop” was widespread. 

This study looked at homicide rates in three cities with highly publicized crime reduction programs: Boston (Operation Ceasefire), New York (Comstat) and Richmond, Virginia (Project Exile).  Using data from 95 large U.S. cities, the basic design involved examining the change in homicide rates in each of these three cities to see if the change could be attributed to the program implemented in that city, given the pre-existing downward trends across the country as well as the known determinants of homicide rates (e.g., resource deprivation).  

Boston’s Operation Ceasefire focused on communication with gang youth, telling them in face-to-face meetings that firearm possession would not be tolerated, and that a tough approach toward youth gangs would be followed as long as the problem existed. Those apparently responsible for violence were also told that “all available levers would be pulled to ensure swift and tough punishment of violators” (p. 423). New York’s Comstat focused on being intolerant of minor crimes and disorder and aggressively restoring order, as well as making the police managers “responsible” for crime patterns in their districts.  Richmond’s Project Exile used a traditional deterrence and/or incapacitation logic, focusing on harsher penalties for violence or drug crimes in which firearms were used. Extensive use was made of advertising the criminal justice consequences of illegal firearm possession and use. 

All three cities, like U.S. cities on the whole, showed decreases in homicide rates.  At the beginning of the interventions, Boston’s homicide rate was about 18 (per 100,000 in the population), New York’s was about 20, and Richmond’s was about 70. [In contrast, homicide rates in Canada’s 9 largest cities have averaged between 1.25 and 2.86 in the last decade.]   In Boston, the drop in youth firearms homicides was insignificant once existing trends in other cities and other known contributors to homicide were taken into account. In New York there was no evidence of an effect of the police intervention program on homicides overall, or on firearms homicides in particular.  In Richmond there was a significant decline in firearms homicides when other known determinants of homicide were taken into account, but not when looking at homicide rates in isolation from other factors.  

Conclusion. One of the difficulties with all evaluations of single-city programs such as these is that the programs themselves are multi-faceted, and the manner in which they are implemented and the cities themselves vary considerably.  In addition, different evaluations of these same programs have arrived at a range of different findings.  The variation in findings is not surprising, given that there is no unambiguously “best” or broadly accepted model for evaluating programs such as these.  Indeed, part of the problem may be that homicide rates themselves vary dramatically and the effect of interventions may be specific to local conditions, including local homicide rates.  Richmond’s homicide rate varied from a low of about 36 per hundred thousand in the population (in 2001) to a high of 80 (in 1994) – rates that are dramatically higher than the average U.S. city. Large Canadian cities show much less year-to-year variation.  A conservative conclusion, therefore, might be that one cannot be confident that any of these highly publicized programs would have a significant impact on homicides (or gun homicides) in cities in which they might be implemented.  

Reference: Rosenfeld, Richard, Robert Fornango, and Eric Baumer.  (2005) Did Ceasefire, Comstat, and Exile reduce homicide?  Criminology and Public Policy, 4(3), 419-450.

Back to top

In the 14 years ending in 2003, 340 people convicted of serious crimes in the U.S. were found to be innocent. This number almost certainly underestimates the number of wrongful convictions. Nevertheless, these cases help identify factors associated with false convictions.

It is difficult for a defendant, once convicted, to prove conclusively that he or she is innocent of the crime.  The use of DNA evidence, beginning in 1989, provided one mechanism for proving innocence.  Other convicted people have been exonerated on the basis of new evidence which demonstrated that the defendant had no role at all in the offence.  This paper uses a very strict definition of innocence. For example, cases in which there was strong evidence of factual innocence, but “unexplained physical evidence of the defendant’s guilt” are excluded.  In 263 of the 340 cases, the charges were dismissed by a court after new evidence, such as DNA, emerged. Interestingly, however, in some cases – including rape cases in which the convicted defendant was exonerated by DNA – prosecutors maintained their beliefs in the defendant’s guilt. In one rape case, prosecutors maintained that the original defendant was guilty even after DNA evidence (commissioned over the objections of the state) implicated a previously convicted serial rapist. 

Most of the exonerations were for murder (60%) or for rape/sexual assault (36%).  About a third of the murder exonerations involved people sentenced to death.  The pattern of exonerations almost certainly reflects the amount of effort expended in examining potentially false convictions rather than the pattern of actual false convictions. It also reflects the types of evidence that are available in different types of criminal cases.  Robbery trials are considerably more likely to occur than rape/sexual assault trials.  Eyewitness misidentifications (which for the most part occur only in the case of strangers) are, therefore, more likely in robberies than in rapes.  In the 121 exonerations in rape cases, almost all (88%) were originally convicted on the basis of mis-identification.   DNA was used to clear most of the falsely convicted rape defendants, but DNA evidence is for the most part irrelevant in robberies. If mis-identifications in cases of robbery are no less prevalent than in rape cases, one would expect very large numbers of false convictions for robbery. But DNA evidence is almost always absent in robbery cases and there may be less effort expended investigating the convictions of those convicted of robbery than there is in convictions for rape. One would suspect, therefore, that there are hundreds, if not thousands, of falsely convicted robbery defendants.  

Wrongful convictions are disproportionately likely to be identified in cases of capital murder either because more effort is spent in examining these cases and/or because there are more false convictions in these serious cases. The latter explanation could occur because of the enormous pressures to solve heinous crimes.  “Considering the huge discrepancies between the exoneration rates for death sentences, for other murder convictions, and for criminal convictions generally…. [it seems likely that] we are both much more likely to convict innocent defendants of murder – and especially capital murder – than of other crimes, and a large number of false convictions in non-capital cases are never discovered because nobody ever seriously investigates the possibility of error” (p. 533).  In addition, of course, for the most part only those with long sentences or facing the death penalty ever get a chance to prove their innocence: the average time from conviction to exoneration is more than 11 years (p. 535).  The reasons for false convictions vary.  In many instances the central problem was a mistake. For example, it is known that white witnesses make many errors in the eyewitness identification of black suspects. In addition, false confessions are not uncommon (see Criminological Highlights 7(4)#7). But “in at least 60 of the 340 exonerations, the defendant was deliberately falsely accused at trial….” (p.543). 

Conclusion. It is clear that there are large numbers of wrongful convictions in U.S. trials. Canada has also had a number of highly celebrated wrongful convictions. It would appear from this paper that wrongful convictions are likely to be found wherever effort is spent in finding them.  Hence, apparently low numbers of cases of convicted persons who later are completely exonerated may only reflect the lack of resources available to investigate these cases.

Reference: Gross, Samuel R., Kristen Jacoby, Daniel J. Matheson, Nicholas Montgomery, and Sujata Patil (2005).  Exonerations in the United States, 1989 through 2003.  Journal of Criminal Law and Criminology, 95(2), 523-560.

Back to top

Increased imprisonment in New Zealand in recent years has more to do with “penal politics” than with crime.

New Zealand’s imprisonment rate in 2004 was about 174 per hundred thousand residents, second only to the U.S. among OECD countries. Most of the increase occurred in the previous 15 years – a period during which crime, if anything, decreased. In 1980, the imprisonment rate (per hundred thousand residents) was 88; by 1990 it had risen to 117.  (In comparison, Canada’s imprisonment rate, which has been fairly stable since the early 1960s, was 103 per hundred thousand residents in 2002/3.)

In 1999, a “Citizens Initiated Referendum” obtained 92% support for the view that there should be “minimum sentences and hard labour for all serious offences” (p. 305).   The referendum results became a standard against which subsequent sentencing legislation could be compared.  For example, legislation in 2002 increased some penalties, “exhort[ed judges] to make more use of maximum penalties” (p. 305), restricted parole opportunities for violent offenders, made community risk the sole factor to be considered in deciding parole, and allowed victims to attend and/or provide written statements for parole hearings.  Prior to 2002, law-and-order politics had been associated mainly with attempts by the police to generate support for their organizations. By 2002, all political parties except the Green Party had formed a consensus supporting punitive approaches: Crime was seen as a serious, central problem to be responded to with tough measures. Academics and a Governor General who had suggested that prisons wouldn’t solve the crime problem were denounced as being “anti-populist.” It appears that there were four “distinctive factors, beginning in the mid-1980s… ultimately coalescing and converging in the late 1990s” (p. 307) that account for the change in New Zealand’s crime policy.

First, economic problems in the 1970s and 1980s combined with dramatic changes in government social policies led to a “widespread decline in trust of politicians… [and] dissatisfaction with the democratic processes” (p. 308).  Dramatic changes were made in the manner in which governments were elected and it became easier for referendums to be held. 

Second, at a time when New Zealand society was changing (racially, economically, socially, and politically) “three incidents of mass murder between 1990 and 1992 allowed concerns about the general direction of New Zealand society to surface” (p. 311).  Reported crime was increasing, and even when reported crime rates stabilized in the mid-1990s, the public appeared to continue to believe that crime was increasing.  As in Canada, crime in one’s own neighbourhood was not seen as being as much of a problem as was crime elsewhere.  Nevertheless, the public perceived crime as being out of control and the justice system as being too lenient on those who were sentenced.

Third, groups representing victims of crime became more important as a result of a highly publicized brutal attack leading, ultimately, to the 1999 referendum referred to above. Harsh punishment was the focus of these groups.

Finally, the decline in the importance of government, academic, and judicial experts coincided with an increased acceptance of “personal experience, common sense, and anecdote rather than social science research” as the basis of policy.  The families of victims were accorded “expert” status by the media. Social science findings were seen as less persuasive than personal views or accounts. 

Conclusion: “What seems to have been particularly important in the New Zealand context was the disenchantment with the existing democratic process…” (p. 318).   Though the focus of much of the move toward increasing punitiveness was on violent crime, policies dealing with other types of crimes were more moderate.  Hence, the overall impact was less than it might have been had the punitive provisions been applied to a broader range of offences.  For example, though it became more difficult for violent offenders to be paroled, it became less difficult for others.  Consequently, even in this context, the punitive effects of these changes were somewhat muted.

Reference: Pratt, John and Marie Clark. (2005) Penal populism in New Zealand.  Punishment and Society, 7(3), 303-322.  

Back to top

Judges’ decisions about bail are not reliable:  when deciding on the pretrial detention or release of identical cases, different judges arrive at different decisions. 

Disparity of judicial decisions has largely been examined in the context of sentencing. In Canada, for example, there are studies demonstrating that judges faced with written descriptions of cases vary dramatically in their recommended sentences (for both adult and young offenders).  In England, where this study was carried out, lay judges (equivalent in background to Justices of the Peace in Canada) decide most cases involving questions of pretrial release. The accused is generally supposed to be released unless it is believed that he or she will not appear in court as required, will offend while on bail, or will interfere with the administration of justice. 

In this study, 61 lay judges from 47 different adult courts were presented with written descriptions of 27 cases. These cases varied according to the gender, race, age, and criminal history (convictions and bail record) of the accused person; the seriousness of the offence; the relationship of the accused to the victim; strength of community ties; and strength of the prosecution’s case. The judges rated the risk that the offender would abscond (i.e., not appear in court as required), would offend on bail, or would obstruct justice.  Finally, they indicated their overall decision whether to detain or release them pending trial. 

Decision models – essentially the judge’s “theory” of the factors that were related to each of the outcome variables – were constructed.  Judges used different factors in arriving at their assessments of the cases. For example, when deciding on whether the accused would appear in court, 60% of the judges took account of the accused person’s ties to the community, 49% used previous criminal history, and 20% used the seriousness of offence in arriving at their decision. In attempting to determine whether the accused would offend if released on bail, previous criminal history was important for 85% of the judges. In assessing whether the accused would obstruct justice if released, 61% used offence in making this assessment, and 22% used the relationship to the victim and criminal history.  

There were three possible decisions for each of the 27 cases: unconditional release, conditional release, and remand in custody.  For each case, the modal (most common) decision was used as the standard.  Between 8% and 59% of the judges disagreed with the modal decision, with an average disagreement of 28%.  On the individual ratings of the likelihood of absconding, offending on bail, and obstructing justice, “there was less variability among judges on those cases where the mean risk posed by the defendant was judged as relatively low” (p. 377).  Generally speaking, as one would expect, “judges’ bail risk judgments were predictive of their subsequent bail decisions, and for the majority of judges the decision was driven by only one of the three risk judgements” (p. 381). 

Conclusion: The results suggest that not only is there disagreement on what is perceived to be the appropriate outcome of the bail hearing among lay judges, but these same judges disagree on how the judgement should be arrived at.  However, certain things were predictable: greater disagreement was observed in cases that were judged to pose a greater risk of absconding, offending or obstructing justice while on bail.  The reduction of disparity might be addressed through the use of “well defined and structured guidelines” (p. 282) that “more precisely specify the factors judges should use when making their bail risk judgements, how each factor should be weighted, and how risk judgements should inform bail decisions” (p. 383).

Reference: Dhami, Mandeep K. (2005) From Discretion to Disagreement: Explaining Disparities in Judges’ Pretrial Decisions. Behavioural Sciences and the Law, 23, 367-386.

Back to top

Judges are not good predictors of recidivism. 

Fifty years ago, psychologist Paul Meehl noted that when there exists a statistical predictor of some behaviour (e.g., recidivism), individual “clinical” judgments of this same behaviour are typically less effective than the statistical prediction.  In addition, the usual finding is that clinical “adjustments” of a statistically based prediction decrease the accuracy of the prediction.

Under the U.S. federal guidelines, a sentence is determined by an “offence score” (based broadly on various characteristics of the offence) and a “criminal history score” (based largely on the offender’s previous convictions).  Though not explicitly developed as an actuarial prediction of recidivism, judges are allowed to depart from the guideline-dictated sentence if they believe that the criminal history score is not an accurate predictor of the offender’s likelihood of committing crimes in the future. This study examined 102 cases in which judges departed from the guideline-designated sentence (50 upward and 52 downward departures). Two measures of recidivism were employed: arrests (including supervised release violations) and reconvictions.  

Approximately half of the sample re-offended in the ten years following sentencing.  As it turned out, neither the standard criminal history score nor the judge-adjusted criminal history score did a very good job of predicting recidivism.  A potentially more sensitive analysis looked at reoffending within six years after release.  Again, neither the “standard” criminal history score nor the judge-adjusted score performed significantly above chance in predicting recidivism.  The most important finding was that in both analyses the “pre-departure” criminal history score outperformed the judge-adjusted model.  In other words, judges’ clinical judgment about the offender they were sentencing made the prediction of recidivism worse than it already was.

Conclusion: Judges, it would seem, “by using their clinical or intuitive judgement to depart from…the Guidelines, did not improve on predictions of recidivism and appear to have worsened them” (p. 746).  One of the purposes of the guidelines was to limit the ability of judges to create non-uniform sentences based on factors not relevant to sentencing.  Clearly departures from the guidelines that are based on judges’ intuitions about reoffending increase, rather than decrease, disparity in sentencing. 

Reference: Krauss, Daniel A. (2004). Adjusting Risk of Recidivism: Do Judicial Departures Worsen or Improve Recidivism Prediction Under the Federal Sentencing Guidelines?  Behavioural Sciences and the Law, 22, 731-750. 

Back to top

A randomized experiment demonstrates, once again, that boot camps do not reduce recidivism.

Previous research (see, for example, Criminological Highlights, 2(4)#1, 3(4)#3, 4(1)#1) has not found any persuasive evidence that boot camps reduce recidivism.  This is not a very surprising finding, given that there is little in the boot camp model that relates to our understanding of the causes of criminal behaviour. Nevertheless, the boot camp model appears to have maintained its attractiveness.

This study, carried out in California, has a methodological advantage over most other evaluations: it consisted of an experiment in which youths were assigned at random either to a boot camp or to a standard custodial institution and aftercare.  The boot camp program was designed to save money and to reduce recidivism. It targeted the California Youth Authority’s least serious male offenders (mostly property offenders). The two camp sites had almost twice the number of staff as a standard facility.  “Boot camp sites generated lively, lengthy daily schedules of physical training, military drill and ceremony exercises, school classes, group counselling sessions, substance abuse treatment groups…” (p. 314). There was evidence, as in other such settings, that youths in these boot camps felt less fear of being attacked by other youths and were generally enthusiastic about the military milieu, the physical training, and the various treatment programs.  Youths in the boot camp spent less time in custody and more on parole.  While on parole, the boot camp youths received more face-to-face contacts and more drug tests. 

Youths who dropped out of the boot camp (more than a quarter of those assigned to it) were appropriately maintained as “boot camp” youths in the study. All youths (boot camp and the youths assigned to traditional institutions) were then followed for an average of 7.5 years (range: 2 to 9 years).  Arrests for charges other than probation violations were recorded.  Sixteen different recidivism comparisons were examined: four “offence types” (all, serious, violent, property) and four time periods (one, two, and three years, as well as all available data). In addition, “time to first arrest” was examined for the two groups. 

Almost all comparisons showed no difference between the boot camp youths and the controls.  The one exception was for the two year follow-up period.  But overall (and for all other time periods) there were no positive effects of the boot camp experience.  In any case, both immediately (year one) and in the long term, boot camp youths were just as likely to reoffend as were youths sent to ordinary custodial facilities.

Conclusion. The lack of positive impacts of a boot camp experience is not surprising.  While “many [boot camp] staff were good role models and clearly cared about their cadets, the program itself was not specifically designed to incorporate any of [the elements of] effective [correctional] treatment” (p. 328).  In addition, for political reasons, there was no serious attempt to build in effective treatment. Though “continuously refined in an ad hoc but often creative manner, the [boot camp] was fundamentally a militarized quick fix and its aftercare a hastily designed and unevenly implemented… service… [The program] did not focus much on individual needs or provide much by way of treatment services” (p. 328-9). Thus it is hardly surprising that the boot camp experience had no overall impact.

Reference:  Bottcher, Jean and Michael E. Ezell (2005). Examining the Effectiveness of Boot Camps: A Randomized Experiment with a Long-term Follow up.  Journal of Research in Crime and Delinquency, 42(3), 309-332.

Back to top

How might judges explain to representatives of the mass media why specific decisions were made in criminal cases?  Judges in the Netherlands use a “press judge” – a judge whose responsibility it is to act as a spokesperson for the court. 

One hardly needs research to discover that the public does not have a very deep understanding of the manner in which decisions are made in court, nor does one need to do a content analysis to conclude that mass media stories seldom do an adequate job of describing the complexities of criminal (or other) cases.  How might these related problems be addressed?  The Dutch have institutionalized the position of a “press judge” – a fully qualified judge whose role it is to discuss individual cases and the law with the mass media.

Media criticism of judges has been described by an Australian judge as being “a universal phenomenon” (p. 452).  The suggestion has been made that “judges should shoulder part of the blame for inaccurate media reporting [of judicial proceedings] if they fail to actively involve themselves in the way in which public information about the courts is disseminated” (p. 453). The typical approach to addressing this issue is to encourage largely abstract public legal education about the law.

In 1974, courts in the Netherlands first appointed press judges, but they did not take an active public role until the late 1980s. In the late 1990s, communication advisors to support press judges were recruited.  These communications advisors “have a predominantly supportive role; stepping into the limelight is a monopoly reserved for press judges” (p. 471). 

In their discussions with the media, press judges typically took what might be called an “orientation role” – in which they attempted to give meaning or direction to “raw information.” In a given judgement, then, their role would be to help the journalist frame a story in a manner which was consistent with the court judge’s judgement.  Hence the judge must “resort to a calculation exercise allowing him to determine what form of presentation gives… the best chance of getting the message across in all its integrity” (p. 464).  Judges tried to draw a line between explaining and commenting.  Given their role as “translators” of judgements from the court to the media, it is not surprising that in some instances press judges “suggest to a presiding judge improvements to the text of a judgement so that it would be easier to explain to the media” (p. 466).

Conclusion.  “The institutionalization of the press judge as a function which deserves recognition through a partial exemption from ordinary judicial responsibilities is an indication that the Dutch judiciary is acknowledging the importance of embracing a wider audience…. Addressing a media audience is seen as an almost natural extension of judicial communication, despite the obvious struggle of some press judges to conquer the obstacle of ruthless media editing” (p. 471).  The institution of the press judge is not seen as an attempt of the court to become part of popular culture. Instead it is seen as a move “from an isolationist position to [one in which judges take] a much more outward looking yet nevertheless controlling stance” (p. 471-2).

Reference: Gies, Lieve. (2005) The Empire Strikes Back: Press Judges and Communication Advisers in Dutch Courts.  Journal of Law and Society, 32(3), 450-72. 

Back to top

Publication Type

Volume Number


Issue Number