View this issue as a PDF: CrimHighlightsV19N5.pdf
This issue of Criminological Highlights addresses the following questions:
- When police speak to Black and White citizens, how do they communicate different messages even if the words are the same?
- Can you predict how youths will be processed by the criminal justice system by looking at their faces?
- Are school security cameras worth the investment?
- Why did crime drop in some cities when COVID-19 hit?
- How is it better to have been born in 1995 than in 1980?
- What might explain the substantial drop in youth crime in some western countries in this century?
- How does the death rate from opioid overdoses underestimate the true measure of opioid-related deaths?
- What should you look for when you hear that a diversion program for youths “worked”?
When the police stop Black drivers, the tone of the police officer’s voice in addressing the driver is judged by independent observers as being more negative than when they stop White drivers.
Proactive police stops of Black drivers is clearly an issue in many countries. There is substantial evidence that Blacks are more likely to be stopped and subjected to intrusive questioning than are Whites. This paper looks at the manner in which Black and White drivers are addressed by the police who stop them, by examining the tone of their speech, not its content.
The pitch, rhythm, and intonation of a person’s voice “is essential to parse the social meaning of verbal communication, such as the difference between a polite utterance and a sarcastic one” (p. 2). It is common to draw inferences about a speaker or a speaker’s attitude from even brief exposure to vocal tone. This paper examines “the extent to which an officer’s tone communicates respect, comfort, and friendliness toward Black and White men” (p. 3) who were subjected to traffic stops.
The research used the audio portion of body camera footage from traffic stops in a medium sized US city. Typically, “a police officer greets the driver, gives a brief explanation for the stop, and requests their license and registration before returning to their patrol car” (p. 3). These initial utterances were normally about 12 seconds long. Using recordings of the police officer’s initial statements to the driver, the researchers then removed certain frequencies of the officer’s speech. In this way, the content of the utterances was masked while such characteristics as the tone, rhythm, and quality of the voice was maintained. To test that the content had been removed, a professional transcriber was given the recordings and was only able to recognize 6% of the words – and then typically only words like “you”, “is” or “there”.
A number of separate replications of a basic experiment were carried out. In the first, for example, 84 university students each listened to 50 content-masked utterances, half involving stops of Black drivers and half involving stops of White drivers. In the second study, drivers at a Department of Motor Vehicles office were recruited. Participants were asked to provide a “gut impression” of the officer’s tone on dimensions such as whether the officer sounded tense or at ease and the officer’s friendliness and respectfulness toward the driver. Those listening to the clips were unaware of whether the person stopped and being spoken to was Black or White.
The participants who listened to the clips perceived officers’ tone as being more positive when the officer was known to be speaking to White drivers than when they were speaking to Black drivers even when controlling for the driver’s age, sex, whether a ticket was issued, and whether a search took place. “Participants were more likely to categorize officer speech toward Black drivers… as talking down” (p.5). The results of other studies were consistent with these findings, suggesting that the tone of voice of the officer can affect citizens’ trust in police.
Conclusion: It is important to realize that these aspects of the police officer’s tone would not be captured by a transcript of the interaction between the police officer and the citizen. The findings demonstrate, however, that the tone that police officers use differs when they are speaking to Black vs. White citizens. At the same time, citizens almost certainly respond, in part, to the manner in which police officers treat them. This study makes it clear that it is not just what is said that is important, but how it is said. The racial disparities in the officer’s tone of voice in speaking to Black and White citizens “can shape citizens’ trust in police and alter their interpretations of subsequent encounters” (p. 12).
Reference: Camp, Nicholas P., Rob Voigt, Dan Jurafsky, & Jennifer L. Eberhardt (2021). The Thin Blue Waveform: Racial Disparities in Officer Prosody Undermine Institutional Trust in Police. Journal of Personality and Social Psychology: Attitudes and Social Cognition (online).
Justice isn’t blind: Youths whose faces are judged to be more ‘dominant’ or less ‘trustworthy’ are treated more punitively by the youth justice system.
People often use “appearance-based features to infer information about a person’s inner character and traits” (p. 788). This paper examines whether the facial characteristics of youths affects whether they are formally or informally processed by the youth justice systems in 3 US states.
There is evidence from previous research that someone’s “looks” can affect the likelihood that they will be selected in a police lineup. In addition, “adult offenders are more likely to receive harsh… sentences when they appear to be less trustworthy, even when the defendant’s trustworthiness is not relevant to the case” (p. 790).
This study looks at the importance of first-time offending youths’ facial characteristics in affecting the manner in which they are processed by the youth justice system. It examines the outcome for male youths apprehended for the first time whose crime was judged to be a low-level misdemeanor offence such as vandalism, theft, and possession of marijuana. For these offences, it was plausible to expect that the youth could either be processed formally in court or dealt with informally. Photographs of the youth were obtained at the time that the youth first appeared at court intake.
Each of the photographs of 95 Latino or White youths were rated by 24 undergraduates on a number of dimensions. These dimensions were combined into two broad descriptions: dominance (the sum of ratings of the faces as dominant, aggressive, angry, mean and threatening) and trustworthiness (the sum of the ratings of the faces as trustworthy, caring, happy, likeable, and intelligent). There was very high agreement across the raters in the ratings of the accused youths. Furthermore, the two dimensions – dominance and trustworthiness – had high internal consistency across the dimensions. Because these two rated judgements were, themselves highly correlated (r=-.82), it was subsequently decided to combine them into a single measure of “negativity”.
Youths whose faces were independently rated as being more “negative” (more dominant and less trustworthy) were the most likely to be processed formally by the youth court rather than being diverted to a more informal outcome. There was also an indication of a separate effect: Youths with skin tone rated as darker were more likely to be formally processed. Clearly these effects had nothing to do with the youths’ criminal history or their offences since the sample of youths whose pictures were used had been charged with very similar offences and none had been to court before.
Conclusion: Given that people “rapidly form impressions of individuals based on their faces” (p. 800) and given that these impressions can be independent of relevant background factors, it is not surprising that youths whose faces are described as being less trustworthy and more dominating would be treated more harshly by youth justice systems. The challenge for youth justice systems, then, is to figure out how important decisions about youths – such as whether it is necessary to take the youth to court – can be made in a manner that is “blind” to what the youth looks like. Better, more structured, approaches for determining this decision clearly need to be developed.
Reference: Chen, Jacqueline M, Adam D. Fine, Jasmine B. Norman, Paul J. Frick, and Elizabeth Cauffman (2021). Out of the Picture: Latinx and White Male Youths’ Facial Features Predict Their Juvenile Justice System Processing Outcomes. Crime & Delinquency, 67(6-7), 787-807.
It would be nice to think that a simple policy change or some new piece of technology can reduce crime. However, this often is not the case. A previous study (Criminological Highlights 18(6)#7) for example, demonstrates that bringing more police “school resource officers” into schools doesn’t appear to affect public safety, but does increase the number of youths excluded from schools.
This paper looks at another attractive-sounding technology for schools: security cameras. Although the initial cost of this new technology may be small, the cost of monitoring the cameras is substantial. The theory, however, is not just that students will be deterred from committing crime by the presence of cameras, but also that offenders can more easily be apprehended. This longitudinal study of 850 schools examines data on various forms of crime that was collected at two points in time. In 150 of these schools, security cameras were installed between the two data collection waves. In the other 700 schools, there was no change in the use of cameras: they either had them before the first data collection point or they never got them. The hypothesis was that the implementation of security cameras would reduce school crime.
School administrators provided the data on not only the prevalence of various forms of crime, but also on the use of other security measures. They also indicated the frequency of what might be called “social disturbances”: racial tensions, bullying, gang activity, disorder in the classroom. Finally, various demographic measures (school size, an urban-rural indicator, crime in the neighbourhood, poverty level of the students, etc.) were collected for each school.
The results are simple to describe: The implementation of security cameras had no effect on serious or non-serious violent crimes, property crimes, drug crimes or crimes related to weapons. Similarly, the implementation of security cameras had no impact on the number of these types of crime reported to the police, nor on the frequency of social disturbances. Not surprisingly, then, the implementation of security cameras also had no effect on suspensions of students.
The analyses were also carried out comparing the schools that never had security cameras to those that implemented them between the first and second wave of data collection. The results were unchanged: Those that implemented security cameras apparently reaped no measurable benefits from them.
Conclusion: “The implementation of security cameras was unrelated to the recorded crime at the school, the rate of school crime reported to police, the frequency of noncriminal social disturbances, or the rate of schools’ use of exclusionary punishment” (p.39). Given these findings, it would appear that those advocating for the implementation of security cameras in schools should shoulder the burden of demonstrating, rather than simply asserting, that they will be effective.
Reference: Fisher, Benjamin W., Ethan M. Higgins and Emily M. Homer (2021). School Crime and Punishment and the Implementation of Security Cameras Findings from a National Longitudinal Study. Justice Quarterly, 38(1), 22-46.
The decrease in crime that corresponds with the onset of COVID-19 in many cities appears to have been caused, in part, by the stay-at-home restrictions that were imposed.
Many cities and countries experienced a reduction in police-reported crime in the latter part of 2020. Statistics Canada reports, for example, that in January-March 2020, there were 4% more crime incidents reported to the police than in the previous year. But in the final 9 months of that year, crime decreased such that, overall, there was a 10% reduction in reported crime. This paper attempts to understand why that reduction (in Canada and elsewhere) took place.
In this study, changes in police-reported crime rates in 27 cities (in 23 countries) across the Americas, Europe, the Middle East, and Asia were examined in the context of the variability that took place in stay-at-home restrictions across locations. The restrictions imposed in the various cities ranged from voluntary recommendations to avoid public spaces to almost complete lockdowns of everything but essential activities (often involving substantial financial penalties for non-compliance).
Early evidence about the effects of the COVID-19 pandemic on crime suggested that these effects were not universal across countries or crime categories. This study looked at 6 crime categories: assault, burglary, robbery, theft, vehicle theft, and homicide. Crime rates were examined for a period starting 30 days before local restrictions were imposed and ending 150 days after. In some cities, stay-at-home restrictions did not come into effect on a specific date; rather they were implemented over a longer time period. In these cases, local researchers provided an estimate as to when the restrictions might be considered to have been largely implemented.
The analyses that were used – interrupted time series analyses – took into account the season and covered the period starting from 1 January 2018 (or in some cases 2019). The main analyses examined crime trends starting 30 days before stay-at-home restrictions were implemented and ending 5 months after implementation.
Generally speaking, “stay-at-home restrictions are associated with declines in all types of crime, with the exception of homicide” but, “over time, the mean trend begins to return to pre-treatment [pre-restriction] levels of crime” (p. 870). The overall decline was about 37%, but this should be interpreted cautiously since there were substantial differences across cities and crime categories.
In a separate analysis, the stringency of the stay-at-home restrictions was estimated (on a four-point scale). The results suggest that cities with the most stringent stay-at-home restrictions showed the largest decline in four categories of crime: burglary, robbery, theft and vehicle theft, but not homicide. The effect of the restrictions on assault was significant only when data from one outlier (Barcelona) was removed from the analysis.
Conclusion: In the early days of the COVID-19 pandemic, police recorded crime – most notably property-related crime – decreased in many cities. The size of that decrease appears to be related to the timing and strength of the implementation of stay-at-home policies. However, these declines were short-lived, with the maximum drop occurring 2-5 weeks after the implementation followed by a gradual return to previous levels.
Reference: Nivette, Amy E. and 35 others (2021). A Global Analysis of the Impact of COVID-19 Stay-at-Home Restrictions on Crime. Nature Human Behaviour 5, 868-877.
The likelihood and timing of arrest over a youth’s first couple of decades of life depend, in large part, on when they were born, not on just individual characteristics of the youth, family background, economic status, or neighbourhoods. Those born in 1995 are much less likely to get arrested in their late teenage years than those born earlier.
People born at a particular time share a sociohistorical environment that differs from those born at other times. This paper examines differences in arrest likelihood for youths born between 1980 and 1995. Data from the Project on Human Development in Chicago Neighbourhoods allowed these comparisons to be made.
It is well established that the likelihood of arrest for a youth increases up until late adolescence/early adulthood (depending on the specific offence) and then drops off. The exact height and shape of this curve has typically been attributed to the impact of individual characteristics or characteristics of the family or neighbourhood.
The sample of youths in this study were broadly representative of children and adolescents in the mid-1990s. Three waves of data were collected in the first 5 years of the study and then again ten years later. Criminal history data (from the Illinois state police) were collected in 2015, meaning that arrest data were available for everyone up to about age 20 or older. Extensive information on demographic factors, behavioural problems, family structure and troubles, socio-economic factors, and neighbourhood issues were collected.
The findings show that when a youth was born makes a big difference. By the time youths reached their late teens, those born earlier were much more likely than those born in the 1990s to have been arrested. However, in various ways, youths born in Chicago in the 1990s were systematically advantaged compared to those born earlier. Their parents were less likely to be arrested and more likely to have a university degree. Compared to those born in the 1980s, those born later also were less likely to have been exposed to poverty, violence and concentrated incarceration.
Nevertheless, when these factors were controlled for, the pattern was the same: Those born later were still less likely to be arrested in their late teens than were those born in the 1980s. The data for the early years – up to about age 15 -- were very similar. But after age 15, the arrest rate for those born in 1995 dropped off whereas the arrest rate for those born earlier continued to rise until about age 19-20.
This pattern of different arrest rates in the late teens held for those who were advantaged and disadvantaged and for those whose measures of self-control were both high and low. When broken down by the crime for which people were arrested, the pattern held for drug, violence, and property arrests.
Conclusion: The pattern of arrests – dropping off earlier for those born in the 1990s than for those born in the 1980s – was not driven by differences in individual dispositions, demographic, economic, family factors or neighbourhood environments. “Rather, the substantial cohort differentials in arrest in late adolescence and the course of desistance in adulthood arise from the distinct sociohistorical environments through which each cohort aged” (p. 1169). From the perspective of getting arrested, compared to those born in the 1980s, those born in the 1990s were simply more fortunate to have grown up in an era when arrests were less likely.
Reference: Neil, Roland and Robert J. Sampson (2021). The Birth Lottery of History: Arrest over the Life Course of Multiple Cohorts Coming of Age, 1995-2018. American Journal of Sociology, 126(5), 1127-1178.
Youth crime in Sweden has been decreasing in the past two decades just as it has in some other western countries. The most important factors that appear to be accounting for the decline are that youths have changed their daily routine activities and are less likely to be drinking large amounts of alcohol.
Crime rates, including crimes attributed to youths, have decreased substantially in many western countries since the early 1990s. Although this observation has been made many times, there is relatively little high-quality research that has attempted to explain the drop.
There are many possible explanations for this drop in youth crime. An earlier American paper attributed the drop in that country primarily to reductions in unstructured socializing and use of alcohol (Criminological Highlights 19(3)#8). This paper examines the determinants of youth crime in Sweden, a country in which “crime trends… have closely mirrored those of many other countries” (p. 359).
The data for this study come from 8 sweeps of a nationally representative survey of about 48 thousand Grade 9 students (roughly 15 years old) between 1999 and 2017. The dependent variable – the frequency of self-reported delinquency – was derived from 19 questions on criminal offending where students indicated how often they had done each of these acts in the previous 12 months. Three measures were calculated: serious property offending, violent offending, and minor offending.
The explanatory variables included measures of parental monitoring, attitudes toward crime, the frequency with which the youth engaged in unstructured activities, the amount of evening time spent with friends, and alcohol intoxication.
The results show that all types of offending dropped substantially between 1999 and 2017. There were rather small changes in some of the explanatory variables: there was a small reduction in approval of crime, and small increase in parental monitoring. The largest changes in the variables associated with self-report offending were in the amount of unstructured activities that youths reported, the amount of evening time with friends, and alcohol intoxication, all of which decreased substantially.
Conclusion: Sweden, like many other countries, experienced a reduction in youth crime. Multivariate models confirmed that the declining trend in delinquency “can be almost completely explained by changes in social bonds, attitudes toward crime and routine activities” (p.371) as well as heavy alcohol consumption. These variables account for about 95% of the between-year variability in delinquency. These findings are important not only because they provide an explanation for a drop in youth crime but because they locate the “cause” of youth crime in the nature of the society in which youths live rather than in the youth justice system that responds to youth crime.
Svensson, Robert and Dietrich Oberwittler (2021). Changing Routine Activities and the Decline of Youth Crime: A Repeated Cross-Sectional Analysis of Self-Reported Delinquency in Sweden. Criminology, 59, 351-386.
Opioids overdoses often result in death. But opioid use is also associated with another form of premature death: homicides.
It is estimated that in 2017, there were 47,600 deaths due to opioid overdoses in the US, more than twice the number of homicide deaths. This paper examines the hypothesis that opioid use is directly related to homicide victimization by examining the relationship between local race-specific homicide rates and race-specific opioid use in these same locations.
A fair amount is known about opioid use in the US in large part because the use of opioids is quite common. “Big Pharma” (e.g., Purdue Pharma) successfully promoted opioids in the United States to populations of people who previously were not likely to become addicted to drugs. As it turns out, opioid addition is associated disproportionately with the non-Hispanic white population in the US. When addicted users could no longer get drugs through medical prescriptions, they typically moved to illicit markets, leading to competition among suppliers and drug-related violence. An important fact is that the opioid epidemic is distributed across age groups, community size, and geographic locations rather than being concentrated in, for example, poor inner-city neighbourhoods. Hence its social effects would not be expected to be as evident as they might be if the drug use were concentrated in certain types of neighbourhoods.
From a research perspective, the fact that the risk of death from overdoses is so high with opioids means that the rate of opioid-related deaths can be used as an indicator of the extent of opioid use in a particular county. This paper, then, is able to examine the hypothesis that “homicides increase as a function of growth in demand for opioids and the concomitant increase in transactions in the illicit opioid market” (p. 552). Data were collected from 1,421 counties (and clusters of low population counties) in the US. The main dependent variable was the rate of race-specific homicide victimizations in each of these locations. This was examined as a function of race-specific drug-related deaths (which, it was determined, were largely from opioids).
Various other characteristics of the counties were controlled for. These included alcohol-related mortality rates, economic factors, residential mobility, age, race, sex, geographic region, and firearm suicide rate (often used as an indicator of firearms availability). Because high homicide rates are also associated with high concentrations of white evangelical Protestants, this was controlled for as well. In all, over 20 factors were controlled for.
The results are quite straightforward. Looking (separately) at the predictors of white and black homicide victimization rates, opioid use predicted the rate of homicide victimizations above and beyond these other control factors. The factors other than the rate of opioid use that predicted white and black homicide rates varied a bit but were largely similar. In terms of the focus of this paper, however, there was an important consistency: “Homicide rates among both blacks and whites are higher in counties with higher black and white opioid death rates, controlling for other homicide covariates” (p. 569).
Conclusion: The focus of much of the concern about opioid use is on the number of deaths of those who overdose on these drugs. They – and their families and other loved ones -- are clearly victims of those who promoted opioid use. This paper, however, identifies a broader set of victims: those who happen to live in communities with high opioid use. They suffer from the impact of living in communities with high homicide rates.
Reference: Rosenfeld, Richard, Joel Wallman, and Randolph Roth (2021). The Opioid Epidemic and Homicide in the United States. Journal of Research in Crime and Delinquency 58(5), 545-590.
A school-based diversion program illustrates two important issues for evaluating such programs. First, although the program may have resulted in fewer arrests, the youths who were diverted to the program were disproportionately the less serious offenders. Second, although simple comparisons suggest that diversion reduced reoffending, more appropriate analyses demonstrated that this was likely a function of selection by the police of low-risk youths for diversion.
A traditional problem with programs designed to divert youths from court processing is that those referred to these programs often include large numbers of youths who, otherwise, would not have been charged in the absence of any alternative.
This paper examines a city-wide diversion program in Philadelphia, Pennsylvania, that was initiated by the Philadelphia police when it was discovered that they were charging youths in school at a rate that was “three to 25 times higher than the rate of most other Pennsylvania school districts” (p.167). Police referred youths to the program if they had committed a low-level offence and had not previously been arrested. A comparison sample was established of youths from the year prior to the establishment of the diversion program who would have been eligible for the diversion program if it had existed. These youths had, instead, been arrested.
One serious problem in understanding the degree to which youths were actually diverted is that there are indications that these schools were experiencing a decrease in the number of incidents that might have led to arrest. Hence charge rates would have decreased without the program. Nevertheless, it appears that the diversion program did attract some youths who would otherwise have been arrested.
However, the data describing the youths who were diverted illustrates a common phenomenon: those who were actually diverted appeared to be the least serious cases of those who were eligible for diversion. The diverted youths were significantly younger and considerably more likely to be diverted for marijuana possession and non-firearm weapon possession than a comparison group of eligible youths from the year before the program was created.
Comparing the diverted youths and the earlier group of all “eligible” youths, then, was not terribly helpful. Indeed, when re-arrests within 2 years of initial arrest (for the comparison group) or within two years of referral to the program (for the diverted youths) were examined, it was found that the diverted youths were less likely to be arrested. This might be seen – incorrectly – as supporting the conclusion that the program reduced offending. However, when a more appropriate “matched” sample was created of youths from the year prior to the establishment of the program who had similar backgrounds to the diverted group but instead had been arrested, there were no significant differences in re-arrests between the diverted youths and the matched control group. The program may have kept some of the youths from being charged, but it didn’t change their offending behaviour.
Conclusion: The data in this paper can be used to illustrate some relatively common points about diversion programs. First, those actually diverted tend to be at the less serious end of the dimension of those who are eligible for diversion. Second, simple “before-after” comparisons of recidivism rates of those who might have been ‘eligible” for diversion with those who were actually diverted may show lower recidivism rates for those diverted than those not diverted. However, this may be because of pre-existing differences between the two groups. In this study, when a more sophisticated analysis using “matched” groups was carried out, there were no differences in re-arrests between those diverted and those not. This is not to say that there may not be advantages to diversion over arrest. But we shouldn’t expect that programs for youths who have committed very minor offences will have much of an impact on them.
Reference: Goldstein, Naomi E.S. and 9 others. Preventing School-Based Arrest and Recidivism Through Prearrest Diversion. Law and Human Behavior, 45(2), 165-178.