James Bonta, Ph.D.

R. Karl Hanson, Ph.D.

Corrections Research
Solicitor General Canada

30 May 1994

GAUGING THE RISK FOR
VIOLENCE: MEASUREMENT,
IMPACT AND STRATEGIES
FOR CHANGE

1994-09

The opinions expressed in this paper are those of the authors and do not necessarily represent the Ministry of the Solicitor General Canada. 

This document is available in French. Ce rapport est disponible en français.

Supply and Services Canada
Cat. No. JS4-1/1994-9E
ISBN: 0-662-22383-7

ACKNOWLEDGEMENT

We would like to thank Robert Cormier, Cathy Gainer, Alison MacPhail, Larry Motiuk and Richard Zubrycki for comments on an earlier draft.

SUMMARY 

This paper examines violent crime in Canada, how it is measured and how it has changed over the past few years. Our conclusion is that violent crime is a serious social issue but that it has not substantially increased in recent years. Victimization surveys from the United States, and the few conducted in Canada, suggest that victimization rates have, if anything, been decreasing over the past decade. Much of the public however, believe the rates of serious crimes to be increasing.

If the statistical facts about violence are incongruent with the public's fear of crime, then we are led to ask what some of the factors are that account for this discrepancy. There are a number of possible explanations: statistical reporting methods, media portrayals of violence, and demographic and social factors.

One conclusion is unequivocal. Providing good, objective data on violent crime will be insufficient for alleviating public fears surrounding violence. Although we must continue to improve our statistical reporting of crime, we also need to address the factors that lead to actual incidents of violence in our society. Improving offender rehabilitation programs and early prevention programs are two suggestions for a constructive social agenda.

Gauging the Risk for Violence: Measurement,
Impact and Strategies for Change 

Violence is an issue that touches all Canadians: 307,491 violent crimes were reported to the police in 1992 (Canadian Centre for Crime Statistics, CCJS, 1992) and there were many more victims of crime who never entered official statistics. To say, however, that violence has increased to the point where it is almost out of control, as suggested by some media accounts, distorts our understanding of violence and misdirects society's efforts in dealing with violent behaviour. In this paper, we review methods for measuring violent crime and how these measures provide different views of the incidence of violence. From this evaluation of the statistical facts surrounding violence, we explore explanations as to why the fear of violence is disproportionate to the actual risk of violence.

Our general conclusions are that the actual rate of violent crime has not increased substantially and that the perception of a rising tide of violence can be traced to several factors, including the methods used to "count" violent crime, how this information is conveyed to the public, and to a variety of social and psychological factors. Finally, we end the paper with a presentation of a general framework for understanding the fear of violence, and possibilities for diminishing the threats posed by violent acts. We hope that an improved understanding of these complex issues will direct us towards a constructive agenda that promotes the development of a society increasingly free from violent victimization. 

Defining Violent Criminal Behaviour

Not all violent behaviour is considered criminal. Certain sports (boxing) and medical treatments (surgery), for example, can be characterized as "violent" but they are not subject, under ordinary circumstances, to criminal justice interventions. The violent acts that attract public concern are generally those defined as criminal by the Criminal Code (e.g., murder, assault, sexual assault). These officially recognized violent crimes are the basis of our national statistics and a major focus of police and correctional policies.

The Criminal Code, however, is constantly debated and frequently revised based on changes in public attitudes towards certain behaviours. Prior to 1983, for example, it was not a crime in Canada for a man to rape his wife. Similarly, there is variability in whether specific acts of aggression are interpreted as fitting into existing Criminal Code categories. Wife assault, for example, has only in the past decade begun to attract attention as a serious crime (Burris & Jaffe, 1983) even though such acts have been officially defined as "violent crime" for many years.

There are also crimes that are neither clearly violent or clearly non-violent. For example, is the possession of a silencer for a gun a violent crime? Presently, it is counted as a non-violent offense although some may argue otherwise. Variations in definition are important in so far as they affect what may or may not be included in the official statistics. The official crime statistics, in turn, can have significant effects on public perceptions and criminal justice policy. 

The Measurement of Crime 

There are three major methods for measuring crime. The first method (i.e., official reporting) summarizes crimes that are reported to police. The second method uses victimization surveys and the third depends upon self-reports. These different methodologies produce different results and herein lies the difficulty in assessing the actual incidence of violent crime. 

Official Reporting: The Uniform Crime Reports

Uniform Crime Reports (UCR) is the major method in North America for summarizing crime reported to the police. Efforts to summarize crime statistics require addressing the variability in definitions of crimes across jurisdictions and across time. As well, the quality of the summary is based on the consistency with which the basic information is collected and reported. In order to address these concerns, the United States first adopted the UCR system in 1930 (McCaghy & Cernkovich, 1987). Seven types of crimes ("Index Offenses") were identified for monitoring. An eighth index offence, arson, was added in 1978. These index offenses do not represent all crimes, but identify the most frequent and serious (homicide, rape, robbery, assault, burglary, larceny and auto theft). In 1985, almost 16,000 agencies with policing responsibilities for 97% of the U.S. population reported index crime data to the FBI (McCaghy & Cernkovich, 1987).

Canada's uniform crime reporting system began in 1962 as a joint project of Statistics Canada and the Canadian Association of Chiefs of Police (CCJS, 1988). Unlike the American UCR, which collects information on only eight index offenses, the Canadian UCR includes almost all criminal code offenses as well as federal and provincial statutes. In 1981 the Canadian Centre for Justice Statistics (CCJS) formally assumed responsibility for the program and continues to be the repository of UCR data.

Because the UCR summarizes only officially reported crimes, it is important to understand how a crime comes to be recorded by the police. The police can either directly observe a crime, or the crime can be reported to them. Direct police activity (e.g., undercover operations, arriving at a crime scene) accounts for only 13% of officially recorded crimes in the United States (Reiss 1971), and although we have no direct evidence, it is likely that the figure would be similar in Canada. Consequently, the vast majority of offenses come to police attention from the reports of victims.

The process of crime reporting may then proceed as follows:

1. A citizen contacts the police to report what the citizen thinks was a crime committed.

2. The police record the most serious offence (the rules for this are explained below).

3. Police investigate the citizen report to determine whether or not the offence actually occurred or if the offence is "unfounded".

4. If further police investigation leads to a suspect being identified or arrested, the crime is said to be "cleared".

There are other police investigations not triggered by citizen calls or direct observation but these represent the minority of charges.

The police agencies collect the above information and submit it to the CCJS for national collation and reporting. In 1990, a remarkable 99% of urban police agencies submitted such reports.

There are two characteristics of the UCR system that can possibly influence the estimated rates of violent crime: a) there are different methods for counting violent and non-violent crimes, and b) there is a Most Serious Offence (MSO) rule, which may exclude non-violent crimes. Violent crimes (see Appendix A) are counted according to the number of victims (e.g., if one person assaults five people, five violent offenses are recorded). In contrast, only one non-violent crime is recorded per incident. When faced with an incident involving many non-violent offenses, the specific non-violent offense recorded is the offense considered the most serious (the MSO rule). Usually this is the offense with the maximum allowable penalty under law. For example, if someone is stopped by the police for possession of a stolen auto and refusing a breathalyser (one incident) only possession of a stolen auto is recorded. Robbery is an exception to the above rules: it is counted as an incident but categorized as a violent crime - one incident regardless of the number of victims.

The Most Serious Offense rule also gives precedence to violent offenses. That is, when violent and non-violent offenses occur together in the same incident, it is the violent offense that is recorded regardless of the maximum allowable penalty for the non-violent offense. In cases where the maximum penalties for two or more non-violent offenses are the same, the police have discretion in recording what they consider to be the most serious.

The recording methods for the UCR have the inevitable consequence that non-violent crimes are under-represented. Single incidents with multiple offenses that include a violent crime will be counted as violent; the non-violent offenses will not be recorded. Consequently, the ratio of non-violent crimes to violent crimes is actually larger than UCR statistics would suggest. This method of reporting also makes it difficult to use the UCR to address other issues such as the rate of violence committed in conjunction with non-violent offenses (Silverman & Teevan, 1986).

Police discretion and public policy (e.g., mandatory charging and prosecution) can also influence the UCR. This can operate at a number of points, as when considering "unfounded" complaints and deciding which offence is the most "serious" among multiple offenses. Some critics feel that police discretion in recording offenses is so capricious that the UCR has been described as "crime reported to and observed by police which they see fit to officially record" (McCaghy & Cernkovich, 1987: 44). McCaghy and Cernkovich's position is an extreme view considering that there is too little research measuring the possible impact of police discretion.

The major source of variability in the UCR data, however, appears to be found in the citizens' decision to report a crime to the police. In order to report a crime, the citizen must first define the act as a criminal offense, and secondly decide that reporting the crime to the police would be useful. Based on information from victimization surveys (to be discussed later), it is clear that there is considerable variability in the reporting rates for different offenses and types of victims. For example, most cases of motor vehicle theft are reported to police whereas only a fraction of sexual offenses against children appear in official crime statistics. In addition, members of disempowered groups (natives, poor, women, minorities) may fail to report crimes due to fears that the police will not take their cases seriously or that the police cannot protect them.

In 1988, the UCR was revised in an effort to take into account some of its shortcomings as well as collect more detailed information on the accused, the victims and the incidents. The revised UCR includes questions on the location of the crime (e.g., a gas station), use of alcohol or drugs by both the accused and the victim, level of injury, value of drugs seized, etc. In 1992, 51 police agencies were using the revised UCR and as the program becomes more widely implemented more detailed information regarding the context of crime will be available. Although there are some significant changes in the revised UCR, scoring rules allow the conversion of data between the two systems to ensure historical continuity.

Crime Trends as Measured by the UCR

With this introduction to the UCR we can now turn our attention to crime statistics as reported by the police. The most important measure is the offence rate. Absolute numbers fail to take into account growths in the general population. Thus, increases in absolute numbers may simply reflect increases in the number of people who are available to engage in the behaviour. Rates take the total population into account and are usually reported as per 100,000 people.

Table 1 summarizes 1982-1992 data from various Juristats (official publication of the CCJS) on some characteristics of violent crime and their rates. Keeping in mind the limitations of the UCR, a number of general observations can be made. First and foremost, violence is a serious social problem. In 1992, there were 307,491 violent offenses reported to the police. The evidence from victimization surveys and studies of self-reported crime (to be discussed later) suggest that even the high rate observed in the UCR substantially underestimates the rate of violent victimization in Canada. However, within a larger context, physical injuries as the result of traffic, occupational and household accidents are considerably more likely. For example, injuries stemming from alcohol use and requiring hospital attention were estimated to be between 100,000 to 240,000 for 1984 (Adrian, 1989).

Table 1. Violent Crime in Canada 

Homicides account for 0.3 to 0.4% of violent crime.
Violent crimes account for 8 to 10% of all criminal code offenses.
Note: F = female victim. 

The second general observation that we can draw from Table 1 is that the homicide rate (number of homicides per 100,000 population) has remained relatively stable over the 10 year period. As well, the victim-accused relationship and gender of the victim in homicides has remained essentially unaltered. Although males are at the greatest risk for being a victim of a homicide, women were victims in approximately 35% of homicides. Further, 75-80% of all offenders were known to their victims. The use of firearms has also shown little fluctuation; if we add UCR data from as far back as 1975, the use of firearms has actually decreased among both homicides and robberies (CCJS, August 1991). Taken together, this information suggests that:

1) the most extreme form of violence in society (i.e., homicide), as measured officially, has not increased.

2) males are the most likely victims of a homicide.

3) the highest risk of being a victim of homicide comes from acquaintances and close friends and not from strangers.

4) the use of firearms remained unchanged.

Despite showing little change in homicides in this country, the UCR statistics show a dramatic rise in the rate of violent crimes, from 685 in 1982 to 1,122 in 1992. Since homicides accounted for approximately 0.2% of all violent offenses in 1992, this increase must be accounted by non-fatal violent crimes.

Sexual offenses, in particular, have shown a dramatic rise during the past decade. As shown in Figure 1, the rate of reported sexual offenses increased only slightly form 1962 to 1982, after which it increased from 60 per 100,000 in 1983 to 140 in 1992 (CCJS). Part of the increase in the rate of reported sexual offenses can be traced to the implementation of new sexual assault legislation in 1983. This legislation, along with changes intended to improve the treatment of sexual assault victims, replaced previous offenses of rape, attempted rape and indecent assault with the new offenses of sexual assault, sexual assault with a weapon and aggravated sexual assault. Even within the category of sexual assault, there was marked variation in the rates. Almost all of the increase in sexual offenses after 1983 was due to increases in reports of Level 1 (absence of physical injury or use of weapon) or the less serious form of sexual assault. The rate for the most serious sexual offense categories (with a weapon, aggravated) remained unchanged (Roberts, 1990).

Although there are no national statistics on the rate at which spousal assaults are reported to police, there is some indication that there has been an increase in the reporting of these offenses. Manitoba probation service, one of the few agencies whose records distinguish between spouse abuse and other assaults, had only 22 identified cases of spousal assault in 1986 representing 1.4% of their adult caseload (Bonta, Parkinson, Pang, & Barkwell, 1994).

By 1992, the number of spousal assault cases had increased to 825 (27% of their caseload). This staggering increase may be due, in part, to the creation of a Family violence court. That is, criminal justice policy and processing may be a significant factor in official measures of crime.

To summarize, the increases in officially recorded violence between 1982 and 1992 may be affected by policy driven changes in reporting practices, legislative changes and criminal justice processing. The most serious forms of violent crimes have shown little change. After accounting for these factors, however, is there still a possibility that the actual rate of violent crime (not just crime reported to the police) has substantially increased within the last decade? The answer to this question takes us to a consideration of victimization surveys and self-reported criminal activity. 

Victimization Surveys 

Victimization surveys ask citizens directly if they have been victims of a criminal offence, and often ask further questions about the effects of the victimization. Victimization surveys have the advantage of detecting crimes that are unreported to the police, but such surveys cannot be expected to detect all offenses (Evans & Himelfarb, 1987). Murder rates obviously cannot be estimated from victimization surveys, and such surveys have difficulty addressing "victimless" crimes such as prostitution, gambling and drug use. Victimization surveys are also likely to underestimate the incidence of certain crimes involving deception and multiple victims (e.g., market fraud, price fixing, polluting) since victims can remain unaware of the offense even when the offender is apprehended and convicted. As well, certain sensitive acts, such as domestic abuse, may not be reported due to embarrassment, fear, or a failure of the victim to define the act as criminal. Victimization surveys can also result in inflated crime rates when participants report events outside the identified time frame. Nevertheless, the information gathered from victimization surveys provides an important context for interpreting UCR data. 

Crime Trends As Measured By Victimization Surveys

The type of research that best addresses the extent of changes in violence are victim surveys administered over time. Almost all of the repeated surveys come from the United States and Europe. In general, these studies find that victim reports of violent crimes have not significantly increased.

The United States' National Crime Survey began in 1973. Each year, residents aged 12 or more from randomly selected households, are interviewed (personally and by telephone). A review of trends in crime victimization between 1973 and 1988 shows that crimes in general and violent crimes in particular peaked in 1981, and decreased in subsequent years (U.S. Department of Justice, 1991). The overall decrease in crime victimization between 1981 and 1988 was 14% and for violent crimes the decrease was approximately 10%.

In Canada, there have been only three victimization surveys approaching national sampling (the Canadian Urban Victimization Survey and the 1988 and 1993 General Social Survey), and a small national sample taken as part of an international study of crime victimization (van Dijk, Mayhew & Kilias, 1990; van Dijk & Mayhew, 1992). The Canadian Urban Victimization Survey, conducted in 1982, comprised a random sample of the adult population (age 16 and over) from seven cities, including Canada's three largest cities. Over 61,000 telephone interviews were conducted and residents were asked to report a wide range of victimization experiences for the year 1981. The General Social Survey conducted telephone surveys for samples of 10,000 in 1988 and 1993 and focused on a more limited set of crimes (CCJS, 1990, 1994).

The findings from these studies showed that in general, and not unexpectedly, many crimes go unreported to the police (about 40%). Another noteworthy finding is that some property crimes are more likely to be reported than violent crimes. For example, 68% of break and enter offenses were reported to police compared to 10% of sexual assaults (CCJS, 1994). Not only does the UCR under-report crime in general, but it also appears to particularly under-estimate violent crimes.

To establish trends we require repeated surveys. In Canada there have been three victimization surveys that can speak to the issue of trends and whether violence is getting worse, better, or remaining about the same. The first such repeated survey was conducted in Edmonton. Approximately 10,000 residents were interviewed in 1981 and again in 1985 (Solicitor General Canada, 1987). Although the victimization and UCR methodologies are not strictly comparable (e.g., the victimization survey includes subjects age 16 and over while the police reports include all age categories) the findings are revealing. UCR data from Edmonton for this time period showed a 16% increase in violent crime but the victimization data showed no change.

In the second study, two thousand Canadians were interviewed in 1989 and 1992 as part of the International Crime Survey (van Dijk & Mayhew, 1992). This survey found that the percentage of respondents (28%) reporting one or more crimes remained stable during this time period. The survey found a small increase in assaults and a small decrease in sexual offenses. These changes, however, were statistically insignificant and attributable to sampling error.

Finally, the General Social Survey conducted telephone interviews in 1988 and 1993. The survey found that victimization rates changed little over the five year period (CCJS, 1994). There was no difference, for example, in the likelihood of being a victim of an assault or a theft, and the likelihood of being a robbery victim actually decreased by 31%.

In addition to the general crime victimization surveys, there have been focused surveys that address specific issues, such as spouse assault and sexual victimization. Such specialized surveys are important since the manner in which the questions are asked can influence the results obtained. In the Urban Victimization Survey, for example, four general questions about being attacked were embedded in a series of questions concerning other types of crime and crime control strategies (e.g., locking doors, neighbourhood watch). If the respondent indicated having been attacked, he or she was then asked to identify the attacker. This approach found an annual incidence rate of assaults by intimates of 0.4% for married women (Solicitor General Canada, 1985b). In contrast, when surveys ask women about specific violent acts inflicted upon them by their partners, the incidence rates for assault are found to be much higher: 10-14% (Smith 1988). The reasons for the differences are not entirely known, but one contributing factor may be the respondents' hesitance in identifying their partners' specific violent acts (e.g., being slapped, pushed) as being the type of "crime" requested in the Urban Victimization Survey.

For whatever reason, the rates of spouse abuse have been consistently higher in surveys that specifically addressed this topic. A study by Brinkerhoff and Lupri (1988) found a 24% incidence of husband-to-wife violence in Calgary in 1981. In 1987, Kennedy and Dutton (1989) found the annual incidence of husband-to-wife violence in Alberta to be 11%. In Toronto, the incidence of husband-to-wife violence was found to be 14% (Smith, 1988). A 1993 survey by Statistics Canada estimated that 3-4% of all women were assaulted by their male partners during the previous 12 months; for young women (ages 18-24), the annual incident rate increased to 18% (Statistics Canada, 1993). Despite the variation in methods and definitions of abuse, all the studies found wife assault to be much more pervasive than would be indicated by police reports or non-specialized crime victimization surveys.

Canada has no national studies that address changes in the rate of husband-to-wife violence. Canada's closest approximation to such a study would be a comparison between the Alberta data collected by Kennedy and Dutton (1989) in 1987 and the Alberta data collected in 1993 by Statistics Canada. Although the studies are not strictly comparable, rates in the more recent survey were lower than in the earlier survey (4.0% vs 11.2%). Repeated surveys in the United States have similarly found the rates of husband-wife violence to have decreased slightly from 12.1% in 1975 to 11.3% in 1985 (Straus & Gelles, 1986; Straus, Gelles, & Steinmetz, 1980). High rates of sexual victimization have also been found in specialized surveys (Bourque, 1989). The Statistics Canada (1993) survey found that 6% of all women reported a sexual assault (as defined by the criminal code) during the past 12 months, with the rates increasing to 18% in the highest risk age category (18-24). Dekeseredy and Kelly (1993) found 4-10% of Canadian female university students indicating that in the previous year they had engaged in sexual acts due to force or threats of physical force. In a national Canadian study conducted in 1983, half of the women and a third of the men reported being the target of some form of unwanted sexual act at some point in their lives (Badgley, 1984). Most of the sexual victimization happened during childhood, which suggests that even specialized surveys of adults are likely to seriously underestimate the incidence of sexual victimization.

The rates of childhood sexual abuse have only been assessed through retrospective reports of adults. Although such reports are subject to numerous sources of error, adult surveys can provide some information about changes in the rates of childhood sexual abuse. Badgley (1984) found that the rates of reported childhood sexual victimization did not change as a function of the respondent's current age. For example, the rate of childhood sexual victimizations based on reports of fifty year olds was similar to the rate based on reports by those in their twenties. Consequently, the report concluded that the rate of childhood sexual abuse had been reasonably stable for a number of decades.

A recent study by Bagley (1990) replicates the high rates of sexual abuse found in previous studies and suggests that the rate of childhood sexual abuse may actually be decreasing. Figure 2 plots the rates of reported childhood sexual abuse for 750 Calgary women according to their year of birth. Those born most recently reported less abuse than those born earlier, although the overall rates were still extremely high (20%-40%). 

As with officially recorded crime, victim surveys find that the serious forms of violent criminal acts have remained stable over time. This is true for general surveys and also the more specific surveys on sexual and spousal assault. 

Self-Reported Measures of Crime 

The third method of measuring crime involves asking members of the general population or known offenders themselves to report what criminal acts they have committed over a certain time period (usually in the past year). Beginning with the earliest use of self-report measures (Wallerstein & Wyle, 1947) researchers have consistently found that although many crimes go undetected most of these crimes are of a minor nature. Asking known offenders to report on criminal activities for which they were not apprehended also reveals an extraordinary level of crime. Nowhere are these findings more dramatic then with sex offenders. For example, Abel, Mittelman and Becker (1985) interviewed 411 sex offenders seeking treatment. These 411 offenders reported 218,900 sex crimes or an average of 533 sex crimes per offender. Even if we consider the possibility that these offenders exaggerated and misrepresented their sexual activities, it is clear that there is not only a high incidence of sexual crimes but they are committed by a relatively few offenders.

To our knowledge, there have been no systematic studies of crime trends using self-report methodology. Trend studies using this methodology have been conducted in the addictions field where the interest is monitoring alcohol and drug use. Nevertheless, what we do know from self-report studies echoes the findings from victimization surveys: many crimes go undetected and the incidence of serious crimes are relatively infrequent. 

Summary of the Measurement of Crime 

The results from victimization surveys and self-report studies show that official police statistics tap only a fraction of the violent offenses that occur each year. In particular, violent crimes against women and children are grossly under-reported. These studies also suggests that violence is not a new problem in Canadian society. The few repeated victimization surveys suggest that there has been neither a significant increase or decrease in violent crime. Contrast these findings with those from the UCR demonstrating a rising crime rate and it becomes apparent that other factors are operating to produce a different picture of violence.

The increase in officially recorded violent crimes, and this increase is only seen with official measures, can be traced to many possible sources: improved policing (which increases the likelihood of detection), increased reporting of violent offenses by victims, more accurate and uniform data collection, changes in public and criminal justice tolerance of certain crimes and, of course, changes in the law that have redefined the meaning of violent crime. All of these factors may contribute to a perceived increase in the rates of not just violent crime, but crime in general. There remains the possibility however, that part of the increase in violent crime as measured by the UCR may be due to a real rise in criminal victimizations in certain segments of the population (e.g., inner city, far north). 

The Perception of Violence

Opinion surveys repeatedly report that the fear of violence is a major public concern. It also appears that this fear has become an increasing preoccupation of the public over the past decade. This growing worry is often paired with the perception that society is being engulfed by a rising wave of crime that is threatening its very foundations. Given that there is little evidence that the crime rate is actually increasing, it remains doubtful that the type of information gathered in this report would be sufficient to allay public fears about violence. Consequently, it is useful to examine other possible factors that may be contributing to an increased concern about criminal violence.

The Reporting of Violence Statistics

The observed increase reported by the UCR is one obvious contributor to the perception of an increasing rate of violent crime. Despite its limitations (e.g., subject to policing policies, changes in legislation defining crimes, etc.), the UCR is still the only available index of crime that reports on a national level and over time. For many agencies and policy makers, the UCR is the source for reporting crime data in Canada. Consequently, changes in policy, legislation and policing practices or in the public's willingness to report crime can have a tremendous impact on the perception of crime rates in Canada.

Accepting the significant rise in UCR violent crimes rates, the manner is which these statistics are reported can augment the perception of an increase in crime. In reviewing various Juristats, reports produced by criminal justice agencies and professionals, and media accounts of crime, the following observations can be made:

1) Sometimes changes in the absolute numbers form the basis for discussion and sometimes rates are used. Depending upon which set of numbers is used, different impressions can be created. For example, the number of criminal offenses increased from 797,675 in 1961 to 2,848,091 in 1992, a 357% increase. The officially reported crime rate increased by 242% during the same time period (4,287 to 10,394 per 100,000 persons).

2) Trends can only be established through the collection of data over a long period of time. One of the problems this presents is that definitions of crimes and policing practises change over time. Comparing crime data from 1961 with data from 1991, for example, runs the risk of mistakenly assuming that what was recorded as a violent crime in 1961 is the same as it is today. Legislative changes over the years have altered the way we define and therefore count crime. Consider, for example, the 1983 changes in the definition of sexual offenses and assaults, or the way in which the Young Offenders Act in 1985 redefined adult statistics. These legislative changes have influenced what and how crimes are counted.

If we limit our conclusions on trends in the UCR data to after 1983 (the sexual assault legislation) or even 1985 (the Young Offenders Act) we would still find a 24.9% increase in the violent crime rate between 1988 and 1992. Certainly, 24.9% is not as "dramatic" as the changes in rates over longer periods of time, but at least it avoids some of the difficulties in making comparisons separated by 30 years.

3) There is a "statistical" problem that arises when we try to interpret changes in events that are relatively infrequent. The case of homicide serves to illustrate the problem. Homicide is a very rare event. If the number of homicides increases in one year from 605 to 680, the homicide rate rises from approximately 2/100,000 to 3/100,000. This represents a 50% increase as a result of 75 more homicides. Because the crime is relatively infrequent, very little is required in absolute numbers to produce a large percent change. The violence rate in 1992 was 1,122 per 100,000, a far cry from the homicide rate of 2-3 per 100,000. It would take many, many crimes of violence to achieve a 50% increase in rate. The violence rate, however, is dwarfed by the property crime rate of 6,110 and relatively fewer violent crimes are required to achieve a 50% increase as compared to property crimes.

4) The final shortcoming in the reporting of official violence statistics is the failure to place violent crime in the context of crime in general. When compared to the complete range of crimes, violent crime has comprised about 7 to 11% of all offenses for the past ten years. It is also important to note that the violent crime category consists of different crimes varying in seriousness. For example, the most frequent form of violent crime, assaults, is composed of different types. Level 1 assaults (no physical injury), the most common form of assault, has shown the largest rate increase (60%) since 1983. In contrast, level 3 assaults, the least common but most serious form of assault involving significant physical injury, have shown a 12% decrease in rates (CCJS, 1990a). As we noted earlier, a similar pattern can be seen with sexual assaults. The least serious forms of sexual assaults have risen since 1983 while the most serious sexual assaults (aggravated sexual assault) has shown no change. 

The Role of the Media

Without a doubt, the media plays a significant role in the shaping of public attitudes (Gebotys, Roberts, & DasGupta, 1988). With this role comes certain responsibilities for accurate reporting. When it comes to reporting on criminal justice issues, unfortunately, accounts of violence are disproportionately represented. A study of over 800 newspaper reports dealing with criminal sentencing found over half of the stories dealt with violent crimes and half of these violent crimes dealt with murder (Canadian Sentencing Commission, 1988). In reality, approximately 10% of all officially reported crimes in Canada are of a violent nature and less than one-half of 1% involve murder. Thus, it comes as little surprise that the public overestimates, for example, homicide rates (Doob, 1982).

The media's need "to sell newspapers" limits what criminal justice agencies can or wish to do in influencing media reports. What the agencies can do, however, is supply as accurate and reliable information as possible. As the previous discussion suggests, there is room for improvement in this domain. The image of violence in the public eye can be improved through both sound data and responsible reporting. Almost 30 years later, Biderman's (1966) admonishment to the media still holds today: that the media should support "constructive measures against crime" rather than lead to "despair and distrust of the mechanisms coping with it". In fairness to the media, there have been some balanced and factual reports on Canadian crime statistics (e.g., Hess, 1993) and hopefully we will see more of these. 

Demographic and Changing Social Values

There are several demographic and political factors that are likely contributing to an increased concern about violent crime. Victimization surveys have consistently found that the concern about violent crime is higher among women than among men, and that the fear of violent crime increases with age (Solicitor General Canada, 1985b, 1985c, 1987). Not only is Canadian society becoming older, but also the health and economic wealth of Canadian seniors have been improving. Consequently, it is likely that the concerns of older people will attract increasing political attention. Parenthetically, we are also led to wonder that if age and fear are related and growing older is an inescapable biological fact, was there ever a time in history when people did not think that crime was on the rise?

The increased voice of women has, and will continue to augment public attention to violent crime. The victimization surveys show that women have been subject to high rates of sexual assault and domestic abuse for many years. These types of offenses have not been adequately captured in official statistics, or even in the general crime victimization surveys. As values change, it is likely that the definition of violent crime will evolve to further unambiguously include the types of offenses that occur within the domestic environment (e.g., spousal assault, child and elder abuse). Given the combined influence of an aging population and the expanded voice of women, it is likely that the concern about such violence (and the official crime statistics) will grow even if there is a real and substantial decrease in actual crime rates. (Such a real decrease is possible considering that most violent crimes are committed by young males and demographic projections show a decrease in the number of males in the 15 to 25 age category). 

Fear of Violence: A Conceptual Framework 

Researchers in the area of environmental health risks have noted that providing people with statistical information on the impact of an environmental or health threat resulting from a toxic waste site, for example, has little impact on the fear people report from that threat (Wandersman & Hallman, 1993). Since providing the "facts" is not enough to alleviate fears in the area of environmental and health threats, there is little reason to suppose that "facts" will ease the public's fears about crime. Fear of violence will remain disproportionate to the actual probabilities of violence. Striking a more equitable and reasonable balance between perceptions of risk and the actual probabilities of risk requires a careful analysis of all the variables that contribute to the public's fear of violence.

Drawing upon communications theory, Kasperson and his colleagues (1988) have presented a model of the social amplification of risk. At its simplest level, the amplification (or attenuation) of the perception of risk depends not only on what the message is, but also upon who transmits the message and how the receiver decodes and evaluates the message. In the case of violence, these three general factors and their interactions can alter perceptions of risk and fear of violence. A brief illustration underscores how these factors may operate:

What - Violence is a highly dramatic act with serious consequences. These characteristics are capitalized upon by the media, which, in turn, transmit this information to the public. The emotionally charged nature of violent acts also facilitate encoding the information and ensuring saliency among the receivers of the message.

Official measures of violent crimes, although they underestimate the incidence of violent crime, overestimate increases in violence relative to other types of crime. Dependence upon this type of measure to show trends serves to amplify public fears.

Who - Many people transmit information about risk. As already noted, the media are a major transmitter. Other transmitters of information include experts (e.g., researchers, criminal justice professionals) and the government agencies responsible for the management of risk. Disagreements by the experts and the public's growing distrust of social agencies serve to heighten public apprehension. For example, recent revelations of sexual offenses committed within trusted institutions, such as church and social service agencies shake public confidence in these institutions. The widespread abuse of native children in residential schools is another tragic example. After repeated denial that such violations were possible, we are now faced with the challenge of not only regaining the public's confidence but also with implementing reforms in our institutions to protect future generations from similar abuse.

How - The information transmitted is judged and evaluated according to many criteria. Fear for one's personal safety has been the most widely considered criterion but it is only one. Other possible criteria for evaluating the threat and impact of violence may include judgements on the safety of their children, interference with daily activities, and loss of income (e.g., not being able to work). Grasmick and McGill (1994) have even suggested that religious beliefs may play a greater role in the fear of violence than previously thought. The list of identifying factors influencing the coding of information is just being explored. 

A Summary and Agenda for Change 

The general public, with urging from the media, will continue pressing governments and institutions to do "something" about violence in Canada. Exactly what will be done, and how effectively, will depend upon a host of factors. What will be done will be influenced by the demands of victims, the fear of violence by many citizens, the advice of professionals familiar with violent behaviour, and the actions of politicians listening to all of this.

The criminal justice system has invested significant resources to control crime through increased policing and correctional sanctions. Overall, the ratio of police personnel to the general population has steadily increased over the years (although there was a slight decrease in the major urban centres); in 1991, there was one police officer for every 475 people in Canada (The Daily, March, 1992). In 1991-92, policing cost $5.3 billion dollars, a 5% increase over the previous year. In 1991-92, there was an 8% increase in admissions to federal and provincial prisons, a 10% increase in probation caseloads and a 9% increase in parolees. Correctional services cost $1.9 billion, a 5% increase over the previous year. Yet, there is little evidence that crime rates have diminished in tandem with the increases in criminal justice resources. The recent report of the Standing Committee on Justice and the Solicitor General (1993) has recognized that increased policing and tougher sanctions is not the solution. Alternative strategies must be found.

In this paper, we have argued that:

1) Violence is a serious social problem that has probably existed at similar levels for the past 20 to 30 years.

2) The reported increases in violence can be traced to official measurements of crime (UCR). The official measures reflect increased detection, victim reporting and legislative changes in the definition of some violent crimes.

3) Statistically, the greatest risk for violence comes from family members and friends. Violence perpetrated by strangers is relatively infrequent.

4) The perception of risk is amplified by a number of sources of which the media is but one.

In order to attenuate the public's fear of violence, a number of strategies are possible. First, and the most obvious course, is to increase our efforts to diminish the incidence of violent acts. This involves continuing efforts to develop more effective rehabilitation programs for offenders and also better prediction methods to facilitate the identification of high risk, violent offenders (Andrews & Bonta, 1994). Focusing on the known offender, however, is unlikely to have a significant impact on violent crime rates. Society needs to promote early interventions that prevent the development of violent behaviour.

Early crime prevention programs have focused on controlling aggressive behaviour in young children, training parents in child management techniques and using schools to instill prosocial values and to teach non-aggressive conflict resolution skills to children. There have been a number of exemplary programs in these areas ranging from programs targeting bullying behaviour in the school yard (Olweus, 1991) to programs enriching academic achievement and increasing parental involvement with their children's school behaviour (Berrueta-Clement, Schweinhart, Barnett, & Weikart, 1987; Yoshikawa, 1994). Presently, many of these program have operated with little input from corrections. Corrections has made considerable progress in the development of effective offender rehabilitation programs and the knowledge gained can be usefully applied to crime prevention programs (Solicitor General Canada, 1993).

Finally, information about crime rates shows that crime is a serious problem. Certainly great care needs to be taken when relying upon UCR statistics without consideration of the political, social and administrative context in which the data is collected. However, additional types of information are required in order to understand the causes of crime. Research studies focusing on the specific correlates of violent behaviour are critical for building a knowledge base sufficiently comprehensive to construct new and effective crime control strategies. The statistical profiling of crime tells us we need to act. Research into the causes and control of violent crime can tell us how. 

REFERENCES 

Abel, G. G., M. S. Mittelman, and J. V. Becker
(1985). Sexual offenders: Results of assessment and recommendations for treatment. In M. H. Ben-Aron, S. J. Hucker, & C. D. Webster (Eds.), Clinical criminology . Toronto: M. & M. Graphics.

Adrian, M.
(1989). Alcohol-related casualty statistics in Canada. In N. Geisbrecht, R. González, M. Grant, E. Österberg, R. Room, I. Rootman and L. Towle, (Eds.), Drinking and casualties: Accidents, poisonings and violence in an international perspective. London: Tavistock/Routledge.

Andrews, D. A., and J. Bonta
(1994). The psychology of criminal conduct . Cincinnati, Ohio: Anderson.

Bagley, C.
(1990). Is the prevalence of child sexual abuse decreasing? Evidence from a random sample of 750 young adult women. Psychological Reports, 66 , 1037-1038.

Berrueta-Clement, J. R., L. J. Schweinhart, W. S. Barnett, and D. P. Weikart
(1987). The effects of early educational intervention on crime and delinquency in adolescence and early adulthood. In J. D. Burchard and S. N. Burchard (Eds.), Prevention of delinquent behavior . Newbury Park: Sage.

Biderman, A. D.
(1966). Social indicators and goals. In R. Bauer (Ed.), Social indicators. Cambridge, Mass.: M.I.T. Press.

Bonta, J., R. Parkinson, B. Pang, and L. Barkwell
(1994). Toward a Revised Manitoba Offender Classification System. Unpublished report available from Ministry Secretariat, Solicitor General Canada, Ottawa.

Bourque, L. B.
(1989). Defining rape. Durham: Duke University Press.

Brinkerhoff, M. B. and E. Lupri
(1988). Interspousal violence. Canadian Journal of Sociology, 13 , 407-434.

Burris, C. A. and P. Jaffe
(1983). Wife abuse as a crime. Canadian Journal of Criminology, 25 , 309-318. 

Canadian Sentencing Commission
(1988). Sentencing reform: A Canadian approach. Ottawa: Supply and Services.

CCJS
(1988). Uniform crime reporting manual. Ottawa: Statistics Canada, Canadian Centre for Justice Statistics.

CCJS
(1990a). Violent crime in Canada. Juristat, 10, #15. Ottawa, Statistics Canada.

CCJS
(1990b). Criminal victimization in Canada: The Findings of a Survey. Juristat, 10, #16. Ottawa, Statistics Canada.

CCJS
(1991). Weapons and violent crime. Juristat, 11, #12. Ottawa, Statistics Canada.

CCJS
(1992). Canadian crime statistics: 1992. Juristat, 14, #3. Ottawa: Statistics Canada.

CCJS
(1993). The violence against women survey. The Daily, Catalogue 11-001. Ottawa: Statistics Canada.

CCJS
(1994). Trends in criminal victimization. Juristat, June, Ottawa: Statistics Canada.

DeKeseredy, W., and K. Kelly
(1993). The incidence and prevalence of woman abuse in Canadian university and college dating relationships. Canadian Journal of Sociology, 18 , 137-159.

Evans, J. and A. Himelfarb
(1987). Counting crime. In R. Linden (Ed.), Criminology: A Canadian perspective. Toronto: Holt, Rinehart and Winston.

Gebotys, R. J., J. V. Roberts, and B. DasGupta
(1988). New media use and public perceptions of crime seriousness. Canadian Journal of Criminology, 30 , 3-16. 

Grasmick, H. G., and A. L. McGill
(1994). Religion, attribution style, and punitiveness toward juvenile offenders. Criminology, 32 , 23-46.

Hess, H.
(1993). Shedding light on Canadian Crime. Globe and Mail, September 27.

Kasperson, R. E., O. Renn, P. Slovic, H. S. Brown, J. Emel, R., Goble, J. X. Kasperson, and S. Ratick
(1988). The social amplification of risk: A conceptual framework. Risk Analysis, 8, 177-187.

Kennedy, L. W. and Dutton
(1989). The incidence of wife assault in Alberta. Canadian Journal of Behavioural Science, 21 , 40-54.

Maltz, M. D.
(1977). Crime statistics: A historical perspective. Crime and Delinquency, 23 , 32-40. 

McCaghy, C. H. and S. A. Cernkovich
(1987). Crime in American society . New York: MacMillan.

Olweus, D.
(1991). Bully/victim problems among schoolchildren: Basic facts and effects of a school based intervention program. In D. J. Pepler and K. H. Rubin (Eds.), The development and treatment of childhood aggression. Hillsdale, NJ: Erlbaum.

Reiss, A. J., Jr.
(1971). The police and the public. New Haven, Connecticut: Yale University Press.

Roberts, J. V.
(1990). Sexual Assault Legislation in Canada: An Evaluation. An Analysis of National Statistics: Report No. 4. Ottawa: Department of Justice.

Silverman, R. A., and J. J. Teevan, Jr.
(1986). Crime in Canadian society (3rd ed). Toronto: Butterworths.

Smith, M. D.
(1988). Women Abuse in Toronto: Incidence, Prevalence and Demographic Risk Markers. Toronto: Institute for Social Research, York University.

Solicitor General Canada
(1985a). Reported and unreported crimes. Canadian Urban victimization bulletin, 2 .

Solicitor General Canada
(1985b). Female victims of crime. Canadian Urban victimization bulletin, 4 .

Solicitor General Canada
(1985c). Criminal victimization of elderly Canadians. Canadian Urban victimization bulletin, 6 .

Solicitor General Canada
(1987). Patterns in violent crime. Canadian Urban victimization bulletin, 8 .

Solicitor General Canada
(1993). The Correctional Contribution to Community Safety and Crime Prevention. Report to the National Symposium on Community Safety and Crime Prevention, March 10-12, Toronto, Canada.

Standing Committee on Justice and the Solicitor General
(1993). Crime prevention in Canada: Toward a national strategy . Ottawa: Communications Group, Supply and Services.

Statistics Canada
(1993). Violence against women survey. Ottawa: Supply and Services.

Straus, M. A. and R. J. Gelles
(1986). Societal change and change in family violence rates from 1975 to 1985 as revealed by two national surveys. Journal of Marriage and the Family, 48 , 465-479.

Straus, M. A., R. J. Gelles, and S. K. Steinmetz
(1980). Behind closed doors: Violence in the American family . New York: Anchor.

The Daily
(1992). March 10 release.

U.S. Department of Justice
(1991). Criminal victimization in the United States: 1973-88 trends. Washington: Bureau of Justice Statistics.

van Dijk, J. J. M., P. Mayhew, and M. Killias
(1990). Experiences of crimes across the world: Key findings of the 1989 international crime survey . Boston: Kluwer. 

van Dijk, J. J. M. and P. Mayhew
(1992). Criminal Victimization in the Industrialized World: Key Findings of the 1989 and 1992 International Crime Surveys. The Hague: Ministry of Justice, The Netherlands.

Wallerstein, J. S. and J. Wyle
(1947). Our law-abiding lawbreakers. Probation, 25 , 107-112.

Wandersman, A. H. and W. K. Hallman
(1993). Are people acting irrationally? Understanding public concerns about environmental threats. American Psychologist, 48 , 681-686.

Yoshikawa, H.
(1994). Prevention as cumulative protection: Effects of early family support and education on chronic delinquency and its risk. Psychological Bulletin, 115 , 28-54.

Appendix A

Violent Crimes in the UCR

Offense 

Homicide

First degree murder

Second degree

Manslaughter

Infanticide 

Attempted Murder 

Assault

Aggravated sexual assault

Sexual assault with weapon

Sexual assault

Assault level 1

Assault with weapon or causing bodily harm level 2

Aggravated assault level 3

Unlawfully causing bodily harm

Discharge firearm with intent

Assault police officer

Assault other peace officer

Other assaults 

Other Sexual offenses 

Abduction

Of person under 14

Of person under 16

Abduction contravening custody order

Abduction no custody order 

Robbery

Firearms

Other offensive weapons

Other robbery 

Note: Assault level 1 = No visible physical injury

2 = Minor injury

3 = Major injury requiring medical attention

Also noteworthy is that some crimes that could be viewed violent are not in the violent category. Examples are possession of offensive weapons (e.g., silencer) and arson. Similarly, offenses such as robbery without the actual use of force is counted as a violent offence.