View this issue as a PDF: CrimHighlightsV14N1.pdf
This issue of Criminological Highlights addresses the following questions:
- Is there evidence concerning the effects on crime of having police in schools?
- Why do people confess to crimes they didn’t do even when they experience no physical threats?
- How can racial profiling of Blacks increase crime?
- What determines changes in US residents’ support for punitive responses to offenders?
- How are the negative effects of child abuse affected by the neighbourhood in which the child lives?
- Do ordinary citizens want harsh sentences to be imposed on offenders in cases they know well?
- How can the state reduce future offending by those it has wrongfully convicted?
- Are large influxes of foreign immigrants into a neighbourhood associated with increased crime?
There is no plausible empirical evidence to support the hypothesis that non-educational police involvement in schools reduces crime in schools.
Although students are typically safer from crime in schools than elsewhere in their communities and school crime in most locations does not seem to be increasing (see Criminological Highlights 2(2)#1, 4(4)#6), some crime does take place in schools. Not surprisingly, therefore, police in many countries have partnered with schools to reduce crime, especially violence, in schools. In the U.S., police presence in schools often increases after high profile school shootings take place. In addition, various special enforcement programs (e.g., searches) and laws (e.g., special penalties for the possession of weapons at or near schools) have been instituted.
However, “there has not been a systematic review of the evaluative evidence specific to the ‘policing schools’ area” (p. 82), though there have been some evaluations of programs in which the police teach a structured curriculum (e.g., on drug abuse resistance). This study examined the existing research in western countries on school-based, non-educational strategies in which the police have a dominant role. Thus the programs that were examined involved such things as “school resource officer” programs where officers are, in effect, assigned to particular schools and attempt to develop rapport with students, provide a positive role model for students, or address problems such as youth gangs.
A very thorough search of published and unpublished research literature was undertaken electronically and manually for papers that included at least one measure of crime or disorder (e.g., police reports, self-reports, school disciplinary records). Bibliographies of school research were scanned. Google searches using various search terms were undertaken. Experts in the area were contacted. For a study to be included in the analysis that eventually was carried out, it had to have an adequate research design. Most importantly, “simple pre-post designs (comparing a before and an after period), which are quite common in policing studies, were not included… [Also not included were] studies that compared results for a [single] school [in which an intervention took place] to all statewide or city schools” (p. 84). These studies do not adequately control for other factors occurring at the time, long term trends, etc. In the absence of an equivalent control school (or schools) one cannot know whether the change was caused by the intervention. What is interesting, however, is that many of the simple, but inadequate, pre-post designs showed “large and dramatic decreases in school crime or student misbehaviour” (p. 92). This is not surprising given that police resources are often assigned to a particular school because of an unusual outbreak of violence that subsequently recedes on its own.
Perhaps the most remarkable finding of this study is that only 11 studies were located that met the eligibility criteria. Most of these 11 were quite weak methodologically. But even among these 11, none of the studies involved a randomized experimental design. And most, if not all, of the 11 studies that were located for the purposes of this review did not have fully adequate designs that would allow one to draw strong inferences even if there had been positive results. In any case, none of these 11 studies showed favourable impacts of police presence in the schools.
Conclusion: It is surprising, given the amount of police resources being expended in school settings, that we do not know more about whether there has been a positive impact of increased police resources in schools. Furthermore, we know almost nothing about what might provide positive results. Clearly what is needed is a program of randomized assignment of schools to different ‘treatment’ conditions. Given that most communities have limited resources to assign to police in schools, and many cities have many schools in their public school systems, random assignment of schools to receive (or not receive) police programs could provide both a fair distribution of resources and an opportunity to determine whether a school-police program was effective.
Reference: Petrosino, Anthony, Sarah Guckenburg, and Trevor Fronius (2012). ‘Policing Schools’ Strategies: A Review of the Evaluation Evidence. Journal of MuliDisciplinary Evaluation, 8 (17), 80-101.
Ordinary university students can be induced to plead guilty to something they did not do. All that is necessary is to make it clear that if they confess they can avoid a severe penalty.
Imagine this situation. You are stopped by a security guard outside of a store for theft of a pair of gloves that you are carrying in one of your hands. You explain that you simply forgot to pay for them, but the guard doesn’t believe you. You are turned over to the police and they offer you a deal: admit your guilt and immediately contribute $400 to a charity of your choice, and you will not acquire a formal criminal record. Or you can go to court sometime in the next few months and face the possibility that a judge will not believe that you had no criminal intent and will find you guilty. You will then acquire a criminal record in addition to whatever other penalty might be imposed.
This study suggests that pleading guilty to crimes one did not do is completely understandable. Furthermore, it suggests that ordinary intelligent people (university students) will make false confessions under circumstances analogous to those in which ordinary people confess to crimes they are (falsely) accused of. The problem for innocent defendants who are faced with a choice of accepting a plea bargain is that the alternative – a much harsher outcome if they are found guilty – is too much of a risk. Innocent defendants are, therefore, at special risk when faced with plea bargains; the result is that many confess to acts they did not do. And courts have generally accepted the idea that when the confession is ‘voluntary’ – when it is not induced “by actual or threatened physical harm or by mental coercion overbearing the will of the defendant – the bargain [is] permitted” (p. 13).
Students in an American university signed up for what they thought was a problem solving experiment. They worked first with someone who they thought was another student (but in fact was an employee of the experimenter’s). Then they were asked to solve various problems on their own. For half of the participants, the ‘other’ person asked the real study participant for help. Almost all of the students gave help, in clear violation of the rules that had been laid down. For the other half of the participants, the other person did not ask for help; and no help was ever offered. When the experimenter returned, she accused the study participant of cheating, noting that there was a pattern of wrong answers that was very unlikely to have occurred by chance. Then, in the “harsh sentencing” condition, the study participant was told that he could admit his guilt and lose credit for being in the study, or, if he did not, the study participant would be taken before an academic review board that in the majority of cases found people guilty. They were told that, if found guilty, there would be a rather extensive penalty imposed (including notification to various university offices and a required ethics class the next academic term). In the “lenient sentence” condition, the choice was between losing credit, going before the academic review board and probably having a penalty that was described as being not so severe.
Not surprising most (89%) of the ‘guilty’ students accepted the plea deal. But 56% of the completely innocent students – who had done nothing wrong – were willing to falsely admit guilt in return for a certain, but reduced, punishment. It seems that when people are placed in real bargaining situations that affect their future lives and presented with information about the choices, they are highly risk averse. The innocent students were slightly, but not significantly, more likely to falsely plead guilty to cheating when they faced the alternative of being likely to have the harsher penalty imposed.
Conclusion: The study demonstrates that ‘ordinary’ people (university students) will plead guilty to offences they did not commit to avoid the possibility of harsher outcomes. The findings challenge the notion that “innocent defendants [are] not vulnerable to the powers of bargained justice” (p. 46). It is quite clear that courts are incorrect in placing “confidence in the ability of individuals to assert their right to trial in the face of grave choices” (p. 48).
Reference: Dervan, Lucien E., and Vanessa A. Edkins (2013). The Innocent Defendant’s Dilemma: An Innovative Empirical Study of Plea Bargaining’s Innocence Problem. Journal of Criminal Law & Criminology, 103(1) 1-48.
Racial profiling can be counterproductive. Not only does it reduce public confidence in the justice system, but by profiling Blacks, Whites learn that they aren’t going to get caught.
Racial profiling reduces both citizens’ assessments of the legitimacy of police actions and citizens’ general support of the police (Criminological Highlights 7(1)#4). In addition, it has been suggested (Harcourt 2007: Against Prediction) that when members of the majority group (Whites) become aware that another group (e.g., Blacks) are the focus of enforcement attention, they (the majority group) learn they can offend without being apprehended. If this is the case, racial profiling could undermine the deterrent value of law enforcement because a large portion of the population believes that nobody is watching them.
In this study, cheating by students was the focus. The prediction was that “when Blacks were profiled for cheating, White participants would feel greater impunity and be more likely to cheat than when either Whites were profiled or no profiling occurred” (p. 449).
Black and White university students participated in a study in which small groups of individuals were given very difficult anagrams to solve. In addition, there were sometimes two ‘confederates’ of the experimenter embedded in the group. Students were told that they would be monitored for cheating. Some of the students observed the experimenter move close to the group and stare directly at the confederates (who were Black for some of the groups and White for other groups). The experimenter then asked the two confederates “to move up front so I can see you better” (p. 349). They moved to the front of the room. In about one-third of the groups, nobody was ‘targeted’ by the experimenter. Students were given 15 minutes to do the task. Cheating was made possible by having the correct answers available to the students, though they were told not to look at the correct answers when solving the anagrams.
There were no significant differences in the rate of cheating of Blacks and Whites in the control condition when nobody was profiled. Similarly, when Whites were profiled (i.e., singled out for special scrutiny by the experimenter) White and Black students cheated at the same (low) rate. However, when Blacks were profiled, the average rate of cheating (number of anagrams on which they apparently cheated) by the White students was roughly three times the rate in any of the other conditions.
Black students did not cheat more when White students were ‘profiled’ by the experimenter. One possibility is that Blacks did not identify this as ‘racial’ profiling – they simply saw it as behaviour on the part of the experimenter toward these two ‘participants’ rather than ‘profiling’ of Whites rather than Blacks.
Police often justify profiling by calling it ‘targeted policing’ – targeting those who, according to their own arrest statistics, are more likely to be involved in certain kinds of illegal activities (e.g., carrying weapons or drugs). However, this justification ignores two important issues. In the first place, if one group (e.g., Blacks) are disproportionately stopped, it is hardly surprising to find that they are also disproportionately arrested – even if their actual rate of illegal behaviour is the same as that of other groups. The second justification for profiling is deterrence: that the profiled group will be deterred by the very real possibility of being caught. This latter justification ignores findings such as those in this study: by focusing on one group (e.g., Blacks), another larger group (Whites) may learn that they aren’t likely to be caught and, therefore, can break the law.
Conclusion: In this study, the behaviour that was the focus of concern (cheating on a test) increased in its overall rate of occurrence because of the profiling of Black students. More generally, it would appear that reducing the perceived likelihood of apprehension for a crime by focusing on one group to the virtual exclusion of other, can, in fact, increase overall offending in part because, by definition, there are more people who can cheat or offend with impunity than there are people who are deterred as a result of the profiling.
Reference: Hackney, Amy A. and Jack Glaser (2013). Reverse Deterrence in Racial Profiling: Increased Transgressions by Nonprofiled Whites. Law and Human Behavior, 37(5), 348-353.
The support in the general public for criminal justice policies that punish offenders is driven by important political factors such as the framing of crime by political leaders as the natural result of permissive policies.
Although we know that ordinary people are not necessarily as punitive as criminal justice policies would suggest (e.g., Criminological Highlights, 14(1)#6), there is little research that explains where support for punitive policies comes from and even less on whether these sentiments change in understandable ways over time.
The public’s support for punitive criminal justice policies (e.g., capital punishment, mandatory minimum sentences) – or “punitive sentiment” – appears to be (at least in the US) a national phenomenon. Changes in state incarceration rates take place in a manner that appears to support the idea that there are important national changes occurring. For example, between the early 1970s and 2010, incarceration rates in all 51 US jurisdictions increased by at least 100%.
Measuring punitive sentiment historically is not a straightforward task. This paper identifies 24 different survey indicators of ‘punitive sentiment’ that were administered nationally in the US a total of 242 times between 1951 and 2006. These include such issues as support for capital punishment, 3-strikes laws, and increasing the authority of, or spending on, the police. In general, over this 55-year period, these indexes of punitiveness moved in similar ways. When these measures are combined statistically and an index is created for each year, it turns out that there are large and predictable changes in this index of overall punitive sentiment.
The largest upturn in punitive sentiment occurred in the 1970s. Punitive sentiment remained high until about 1997 when it started to decline. The challenge, of course, is to determine what was driving punitive sentiment in the US. One of the measures used in this study was an index of the ‘net punitive tone’ of presidential statements on crime. Statements about the ‘coddling’ of criminals or ‘permissive’ judges are obviously ‘punitive’ in nature whereas concerns about “an endless, self-defeating cycle of imprisonment” (Lyndon Johnson) were coded as supporting social approaches to crime. Punitive statements about crime by presidents during this 55-year period were associated with increases in the overall punitive sentiment of the public. Increased concern about crime and increases in homicide rates were also associated with increases in punitive sentiment. However, the number of incarcerations (controlling for the amount of crime) did not have a relationship with punitive sentiment. Lack of public support for government welfare policies was also correlated with changes in punitive sentiment. “This finding suggests that the attack on social liberalism was a successful strategy for increasing public support for punitive crime control policies” (p. 349).
In the US, levels of punitive sentiment among Republicans were steady (and high) throughout this 55 year period. The increase beginning in the 1970s and the decrease beginning in the late 1990s held for independents and Democrats only. However, in general, for other sub-groups in the population (defined by age, sex, region, race and education) the overall patterns were similar.
Conclusion: It would appear that one can, with empirical support, talk about broad changes in the US population in support for punitive policies. However, “that the increase in support for punitive policies occurred at the same time as the public turned away from [social] solutions to poverty is no coincidence…. The ups and downs of punitive sentiment are driven by important political factors such as the construction of crime by political leaders. The framing of crime as a problem of a permissive system and increasing perceptions of racial integration increased public demand for punitive policies” (p. 357).
Reference: Ramirez, Mark D. (2013). Punitive Sentiment. Criminology, 51(2), 329-364.
The effect of experiencing child physical abuse on violence in adolescence is weakest in the most disadvantaged neighbourhoods.
It is reasonably well established that those who suffer physical abuse as children are more likely to be violent as adolescents and young adults. It is less well known, however, whether the effects of abuse are amplified or made weaker by the social circumstances in which a child lives.
Previous research has demonstrated that disadvantaged neighbourhoods have higher rates of child abuse and maltreatment. Child abuse may be considered, in these neighbourhoods, to be more typical or commonplace than in less disadvantaged neighbourhoods. The abuse, then, may be more ‘normalized’ which may lead to less severe impacts on the victim.
This study examined data from 343 neighbourhood ‘clusters’ in Chicago, collected in two waves (1994-1997 and 1997-2000). Residents of each of these neighbourhood clusters were assessed on levels of ‘concentrated disadvantage’ (percent below poverty line, receiving public assistance, unemployed), tolerance for deviance (e.g., wrongfulness of drinking, drug use, fighting among teenagers) and the tolerance for family violence.
The measure of ‘child physical abuse’ was the report of the parent of the use, during the previous year, of any of four forms of physical abuse against the youth (e.g., hitting the child with a fist, beating the child). The nature of the relationship between parent and child, the socioeconomic status of the family, and parental criminality were also measured. When the children were (on average) 11 and 14 years old, they were asked about their involvement in the previous year in various forms of violence (including robbery) and using or carrying a weapon.
Controlling for various characteristics of the children, child abuse increased youths’ subsequent violence at age 14. However, high levels of disadvantage in the youth’s neighbourhoods reduced the impact of child abuse on subsequent violence. In other words, “the relationship between child abuse and violence became weaker as neighbourhood disadvantage increased” (p. 235). Indeed, in the most disadvantaged neighbourhoods, there was essentially no impact of child abuse on subsequent violence.
In addition, children living in neighbourhoods characterized as being tolerant of deviance, and neighbourhoods that tolerated fighting among family members and friends, tended to exhibit higher levels of violence at age 14. These effects appeared to be independent of characteristics of the youth and the youth’s family (age, race, parental warmth, parental criminality, self-control, delinquency of the youth’s friends).
Conclusion: Although being abused as a child was associated with higher levels of violent behaviour as an adolescent, this effect was reduced considerably in disadvantaged neighbourhoods. This may be the result of the fact that “in disadvantaged neighbourhood, where violent behaviours are in general more abundant, violence is more likely to be seen as a somewhat common, legitimate, or necessary way of interacting with others, at least under some circumstances” (p. 239). Alternatively, in more advantaged neighbourhoods, “for youth who [generally] experience few risk factors, the effect of any one risk factor is more readily expressed and potentially more detrimental” (p. 241).
Reference: Wright, Emily M. and Abigail A. Fagan (2013). The Cycle of Violence in Context: Exploring the moderating Roles of Neighbourhood Disadvantage and Cultural Norms. Criminology, 51(2), 217-249.
Citizens with intimate knowledge of criminal cases do not want most offenders to receive sentences as harsh as those required by the United States Sentencing Commission guidelines.
It is sometimes thought that members of the public want harsh sentences. After all, public opinion studies suggest that most Americans (as well as Canadians) say that sentences are too lenient. This study, like one carried out in Australia (see Criminological Highlights 11(6)#2), tested this hypothesis in a straightforward way – by asking citizens who knew the cases as well as the judge knew them what they thought the sentence should be. This, then, is a study of jurors’ views of what ‘their’ offender should receive as a sentence.
The US Sentencing Guidelines, like most American guideline systems, provide a narrow range of (presumptive) sentences based largely on a standardized (but complex) measure of the seriousness of the offence for which the offender was found guilty and a mechanical calculation of the seriousness of the offender’s criminal record. Generally speaking, characteristics of offenders, including their reasons for engaging in the crime, are deemed irrelevant to the determination of the sentence. However, at the time that the guidelines were developed, no careful analysis was carried out to determine what sentences the community would want. This study took advantage of the fact that jurors can provide a readily accessible and knowledgeable sounding board for sentences. The only problem is that they are seldom asked their opinions about sentencing, and typically aren’t present when sentences are imposed.
The author of this study notes that “After more than 11 years as a federal district court judge, I have observed that jurors, almost without exception, suggest that the Federal Sentencing Guidelines sentencing ranges are too severe… [When meeting with jurors after guilty verdicts] jurors would [invariably] ask what sentence the defendant they had just found guilty was likely to receive… The jurors would nearly unanimously express surprise at the length of the likely sentence [required by the guidelines]” (p. 186).
This study collected systematic data on jurors’ recommended sentences in 22 criminal trials. 88% of the jurors’ recommendations were lower than the minimum allowable sentence under the guidelines. “On average, the Guidelines-recommended minimum sentence was more than twice the juror-recommended sentence” (p. 189). For example, in a serious drug trafficking case involving firearms and an accused with an extensive criminal history, the Guidelines required a prison sentence of 15 to 17.5 years. Jurors suggested a median sentence of 3 years. In another case, a drug trafficker received the mandatory minimum sentence of 20 years; the jurors recommended an average of 4.5 years. In this case, the defendant was Black, whereas 11 of the 12 jurors were white. The jurors were generally from very different backgrounds from the defendant.
The only exceptions to this pattern were white-collar crimes. In those cases, jurors’ average recommended sentences were longer than the Guidelines’ sentences.
Conclusion: It is clear that jurors in US federal cases believe, generally, that the federal sentencing guidelines are too harsh. Because jurors are likely to have detailed information about the offence and the offender and are likely to reflect the sentiments of the community at large, they may well be a very useful source of information about the appropriateness of sentences being imposed in the cases that they heard. When politicians in some countries (such as Canada) suggest that the public ‘demands’ that sentences should be harsher, they might want to think, first, about what people really want.
Reference: Gwin, (Judge) James S. (2010). Juror Sentiment on Just Punishment: Do the Federal Sentencing Guidelines Reflect Community Values? Harvard Law & Policy Review, 4, 173-200.
Each year people who have been found guilty of crimes and sent to prison are exonerated and released. Some receive compensation for their wrongful conviction and imprisonment. Compensation in excess of $500,000 appears to pay off in terms of reduced levels of offending after release.
Previous research has suggested that being imprisoned for crimes one has not committed has qualitatively different and much more serious psychological impacts on prisoners than one would expect from the literature on the effects of long-term imprisonment (Criminological Highlights 8(3)#6). But in addition, like all those released from prison, the wrongfully convicted face serious challenges in getting jobs and, more generally, learning the skills necessary for coping in a society that has changed since they were incarcerated.
In the US, the wrongfully convicted spend, on average, around 12.5 years in prison before being released. The question raised by this study is whether the level of compensation they receive for being wrongfully convicted and imprisoned relates in any way to their ability to stay out of trouble after being released.
Compensation in many US states is not automatic. Only about half of US states have statutes to compensate the wrongfully convicted, but even then, many states restrict eligibility to only certain types of cases (e.g., those exonerated with DNA evidence; those who did not confess to the original crime). Furthermore, in some states, the amount of compensation is limited by statute. New Hampshire, for example, limits those wrongfully convicted to $20,000 in compensation from the state no matter how long they spent in prison for something they did not do.
This study looked at the probability of offending (after release from prison) of 118 people who were wrongfully convicted in four states (Illinois, New York, Texas, and Florida) for which criminal history data was available and the amount of compensation was public. All four of these states have compensation statutes. In 40.5% of the cases, no compensation was given. In 30.6%, an amount ranging from $100 to $500,000 was given, and in 28.8% of the cases, an amount larger than $500,000 was given.
Overall, 38% of the sample had a post-exoneration offence. Those with high levels of compensation ($500 thousand or more) had a considerably lower rate of post-exoneration offending (18.2%) than those who received less compensation (50% reoffended) or no compensation (40% reoffended). The other factor that predicted reoffending, not surprisingly, was the number of convictions prior to the wrongful conviction (57% had at least one prior conviction).
Conclusion: “It may be that a certain minimum amount of money… is necessary for an exoneree to get his life on track… It [also] may be that exonerees who receive substantial compensation feel fairly treated by the system and thus offend less often…. The difficulty for exonerees is that they face the same obstacles as ordinary offenders released from prison” (p. 576). Those found not to have committed the offence for which they are imprisoned may be disadvantaged in prison more than ordinary offenders. This could easily occur because they are seen as not taking responsibility for their offences. Given the problems that they face – in part because of the length of time that they were imprisoned, but also because they were imprisoned for offences that they did not do – their requirements for support in trying to establish themselves are considerable. But, in addition, reasonable levels of compensation may symbolize for them the fact that the community that is responsible for their wrongful conviction is, in effect, taking responsibility for its errors.
Reference: Mandary, Even J., Amy Shlosberg, Valerie West, and Bennett Callaghan (2013). Compensation Statutes and Post-Exoneration Offending. Journal of Criminal Law and Criminology, 103(2), 553-583.
High concentrations of immigrants in Los Angeles, California, are associated with lower levels of crime.
There is a long tradition in the U.S. and in Canada of blaming immigrants for crime. Nevertheless, the data generally show that immigrants themselves tend to have relatively low levels of offending, despite their tendency (for economic reasons) to settle in disadvantaged neighbourhoods. Their children, on the other hand, may have higher rates of involvement in crime in part because of the disadvantages they experience and the neighbourhoods in which they live (see Criminological Highlights, 13(6)#6).
Recent research suggests that high concentrations of immigrants in a city are associated with lower rates of crime. It is not clear, however, whether this is an effect simply of the higher concentration in ‘high immigrant’ cities of people who are concerned that being arrested could lead to deportation. The question, then, is whether increased concentrations of immigrants “will lead to reductions in neighbourhood crime [after controlling for] the effects of structural deficits in poverty concentration, mobility, and age structure” (p. 197).
This study looks at changes in crime in Los Angeles census tracts from 2000 to 2005 as a function of the change in the concentration of foreign born (largely Hispanic/Latino residents) between 1990 and 2000, controlling for various factors (e.g., poverty, the concentration of female headed households). There was considerable variability across neighbourhoods in the change in the concentration of immigrants in this 10 year period.
The approach that is used predicts, from neighbourhood characteristics in 1990, what the concentration of immigrants would be in 2000. Essentially, then, in order to control for pre-existing differences in neighbourhoods, the study “compares neighbourhoods with similar expected probabilities of receiving immigrant settlement and compares the difference in crime changes between those that actually received this treatment [large concentrations of new immigrants] versus those that don’t” (page 200).
The results show that high immigrant concentrations in 1990 were associated with decreased crime in the first part of the next decade. In addition, “higher predicted concentration of immigrants are associated with greater than expected reductions in total reported criminal offences and violent crime…. Interpreting the estimates at the sample means for total and violent crimes suggests that a 19.1% increase in the concentration of immigrants in Los Angeles neighbourhoods reduces the average amount of total crime in a neighbourhood by 35.7% and violent crime by 40.9%” (p. 205). These effects hold “even after taking into account that immigrants are likely to [move to] neighbourhoods with [other immigrants]” (p. 209).
Conclusion: Crime can have “negative effects on city economies, causing urban flight and population decline for entire cities or in specific neighbourhoods.” For Los Angeles, the early 1990s “was a period of increasing migration of predominately Latin American immigrants to neighbourhoods of concentrated poverty that produced net reductions in serious crimes reported to the police”. In other words, the new immigrants appear to have been responsible for large reductions in crime, especially in areas of the city that “have been historically linked to crime and gang violence among native Latino populations” (p. 210).
Reference: MacDonald, John M., John R. Hipp, and Charlotte Gill (2013). The Effects of Immigrant Concentration on Changes in Neighbourhood Crime Rates. Journal of Quantitative Criminology, 29, 191-215.