An Estimation of the Economic Impact of Spousal Violence in Canada, 2009

2. Methodology

This section introduces the definitions of important terms, presents the framework, describes the data sources, explains the limitations and provides a discussion of the important methodological issues.

2.1 Definitions

Spousal and Spousal Violence: In this study, the term “spousal” refers to relationships of married, common-law, separated, or divorced partners of at least 15 years of age. Both current and former marriage and common-law relationships are captured in these definitions, and both heterosexual and same-sex relationships are considered. Following this, “spousal violence” refers only to violence perpetrated by a spouse against another spouse in any one of those types of relationships.

Much of the recent literature in this area uses the term intimate partner violence (IPV); IPV includes a broad range of relationships, and includes dating couples. The terms domestic violence (DV) and family violence (FV) also potentially include violence against household members other than spouses, such as children.Footnote 4 Therefore, the scopes of IPV, DV and FV are too broad for the purposes of this study, and the corresponding violence is not examined in this report.

The consequences of spousal violence are not limited to the direct victim (partner) who experiences abuse, and research on victimization shows that there can be primary, secondary, and tertiary victims (Hill 2009). However, in this study, the term “victim” refers only to the primary victim, that is, the spouse who has experienced violence at the hands of his or her partner. Children exposed to spousal violence, as well as family members who might be secondary and tertiary victims, are all classified as third parties in this report.
Technically, the term “offender” would only be used upon conviction in the criminal justice system as, until that time, the allegations of violence have not been proven in court. Terms such as “alleged perpetrator” or “accused” (once charges are laid) may be used. For ease of reading, this report uses the word “offender” in most instances, but the distinctions noted above must be acknowledged.
Some situations of spousal violence may involve reciprocal violence, where the victim uses violence in a defensive manner. In such situations, spouses may be classified as both victims and offenders.
Stalking/Criminal Harassment:
Criminal harassment is the term used in Canada to describe a specific Criminal Code offence that includes stalking, which is the term more widely used. Both terms are used in this report interchangeably to refer to behaviour that “consists of repeated conduct that is carried out over a period of time and that causes victims to reasonably fear for their safety but does not necessarily result in physical injury”. Importantly, criminal harassment may be a precursor to subsequent violent acts (Department of Justice 2004).
Economic Impact and Costs:
The term “economic impact” is synonymous with the term “cost” in this study. Anything resulting from an incident of spousal violence that otherwise would not have resulted is considered an economic impact, or cost, of spousal violence. This definition includes tangible expenditures, such as police costs, and intangible costs, such as the value of pain and suffering. While many of the impacts do not actually involve a financial transaction, an estimate of financial magnitude is calculated regardless. Quantifying all impacts in monetary terms allows for a direct comparison between the different impacts of spousal violence and between spousal violence and other social issues.
Terminology of References:
Note that when referencing academic research, government reports, and surveys, the language used in the referenced work is used in this report for accuracy. For example, when Allen (2011) is referenced in the Introduction, the term “intimate partner violence” (IPV) is used in place of “spousal violence” because the referenced work uses IPV; when Weinbaum et al. (2001) is referenced in Section 4, the term “battered” is used to describe victimization because this is the term used in the referenced work.

2.2 Framework

A comprehensive framework is designed to capture as much of the economic impact of spousal violence as possible. This framework binds together the many cost-estimation models unique to each cost item. The framework essentially defines the scope of the project: the sources of impact; what regions, time periods, data types, and crimes are included; the breakdown of costs by gender; and the categorization of costs.

2.2.1 Sources of economic impact

Spousal violence incidents (victims) represent the key source of economic impact in this report. The costs associated with the impact of spousal violence on victims and on others, such as the justice system, family members, and employers, are included. However, the costs to the offender, such as lost freedom, are not included in this model for the reasons discussed below.

Some researchers, such as Cook (1983, 374), make the case that offenders should be perceived as normal members of society, with the implication that their utilities should be included in the social welfare function. Others, such as Trumbull (1990), argue that offenders are not entitled to be considered as full members of society as a consequence of contravening the law. Though useful, this debate in economic theory is not sufficient to convince the authors to include or exclude offender costs. An analysis of the cost items that might be considered if offenders were included in the social welfare function might provide further persuasion.

First, society may lose the productive capacity of individuals who are incarcerated. Unfortunately, data of the prior earning histories of incarcerated offenders are insufficient; the Correctional Service of Canada's Intake Assessment Process collects information on the employment status immediately prior to the incarceration of offenders who have been admitted to federal facilities, but no data are collected on income levels or industry of employment.

Second, there is debate over whether the value of lost freedom to offenders sentenced to custody should be included in such cost analyses. Becker (1968, 179-80) supports inclusion, but others question why lost freedom should be considered, since incarceration is the punishment decided by the courts, and its very purpose is to curtail the freedom of the offender. This same debate occurs for other civil rights, such as voting and employment. Data for these intangible costs is also very limited.

Other potential costs to offenders include the disruption of the lives of the offender's family and possible injury or death of the offender during incarceration. Data are very limited in these areas as well, making an estimate with confidence virtually impossible.

Therefore, due to a combination of data limitations and the inconclusive debate on the subject by economic theorists, this report does not include the costs to offenders.

2.2.2 Geography

This study attempts to encompass all ten of Canada's provinces and all three territories; however, the lack of available and useable data for the territories (from the GSS) has necessitated the exclusion of the territories from calculating certain impacts on the civil justice system, victims, and third parties.As prevalence rates and service delivery costs are significantly higher in the territories than in the provinces (see Perreault and Mahony 2012; Perreault and Brennan 2010; de Léséleuc and Brzozowski 2006),Footnote 5 any cost estimates which are based on the GSS data may be underestimated.

Note that the data for calculating the impact on the criminal justice system come predominantly from data sources other than the GSS, including police reported data, court reported data, and correctional service data. Given the national coverage of these data sources, the Criminal Justice System section includes complete calculations for both the provinces and the territories.

2.2.3 Time period

The purpose of this report is to measure the impact (or costs) of all spousal violence that occurredin a given time period, regardless of when those costs were incurred or will be incurred. Therefore, all incidents of spousal violence that were reported (either self-reported to a survey or reported to police) in 2009 are taken into account, and all costs associated with these incidents are included, whether the costs were realized in 2009 or at some later date.Footnote 6 For example, if an offender committed a spousal violence crime in 2009 and received a custodial sentence of 5 years, the entire cost of keeping that offender in custody for 5 years is counted.

While incidents in 2009 may result in costs that extend for multiple years, many of the impacts of spousal violence are the result not of just one violent episode, but of a culmination of several violent episodes spanning over several years (e.g., 2007 to 2009). In these cases, only part of the impact and associated costs should be attributed to incidents that occurred in 2009, as violence in previous years may also have contributed. Examples of this are separation and divorce and the negative impacts on children exposed to spousal violence. The decision to end a spousal relationship in 2009 may be the result of multiple years of violence, such that only part of the costs associated with these divorces should be attributed to incidents in 2009. Similarly, conditions such as physical aggression or mental health issues in children might also result from the accumulated impact of long-term exposure to spousal violence. As such, claiming all costs for 2009 will overestimate the true impact of the 2009 incidents. However, depending on estimation methods, these two situations are treated differently in the report, as explained below.

Divorce costs in this report are estimated based on the number of divorce cases initiated in 2009 that were primarily caused by spousal violence. As stated earlier, this report is concerned with the impact of spousal violence incidents in 2009, not with the costs incurred in 2009 that were the result of spousal violence occurring in previous years. However, if certain assumptions are maintained, it is argued that these two measures will generate equivalent measures, which would justify the calculation method used. For example, suppose a victim stays in an abusive relationship for 2 years (2008 and 2009) before making the decision to divorce (in 2009). If spousal violence in each year contributes equally to the decision, then half of the consequences and associated costs in 2009 are the result of the 2009 spousal violence incidents and half are due to the 2008 incidents. If there is another victim in a similar situation, but with violence occurring in 2009 and 2010, then half of the divorce costs incurred in 2010 should be attributed to 2009 incidents. Adding all of the costs attributable to spousal violence incidents in 2009 (one half plus one half) gives a total of one divorce, which is equivalent to the divorce costs incurred in 2009 if divorce costs are assumed to be the same in both 2009 and 2010.Footnote 7 The same conclusion holds for longer time periods (greater than 2 years) and also applies to child protection systems, but to a lesser extent.

The economic impact on children exposed to spousal violence is estimated based on the number of children exposed to spousal violence in 2009 who have likely developed or will likely develop negative conditions due to the exposure. For example, if the GSS reported that 100 children were exposed to spousal violence in 2009, and it is known that 10% of children exposed to spousal violence develop a negative condition, then 10 (=100*10%) children have developed or will develop a negative condition at some point in time. However, attributing all costs associated with the 10 children observed in 2009 will overestimate the true impact of the 2009 incidents.

To illustrate this, suppose that 50 of the 100 children were newly exposed to spousal violence in 2009 (i.e., without prior exposure) and 50 have had one previous year of exposure. Also, suppose that all 100 children are exposed to spousal violence for 2 years in total and that 10% will develop a negative condition at the end of the second year of exposure. Therefore, 5 (=50*10%) children who were first exposed to violence in 2009 should develop some negative condition in 2010, after the second year of exposure, but only half of that impact should be attributed to 2009 incidents (assuming that exposure in both years contributes equally to the development of the condition). Another 5 (=50*10%) children who had been exposed to violence in 2008 and 2009 will develop some negative condition in 2009, after the second year of exposure. Again, only half the impact on these 5 children should be attributed to 2009 incidents. Adding the impact attributable to exposure in 2009 only (five halves plus five halves) results in an impact equivalent to the impact on five children in total, not the 10 children initially estimated in 2009. This is because a portion of these children develop a negative condition partially as result of exposure to violence in 2008 and another portion develop a negative condition partially as a result of exposure to violence in 2010.

As this report only accounts for the impacts resulting from spousal violence incidents in 2009, an adjustment must be made if the number of children exposed is used as the estimation base. Ideally, this adjustment would be the division of the total costs based on children who are exposed to spousal violence in 2009 by the number of years of exposure that contributed to the development of any negative conditions. However, as no such data exists, the average number of years that children are exposed to spousal violence is used as a proxy measure.

2.2.4 Offences

Offence categories examined in the Justice System section are captured by the UCR2, including 1st degree murder, 2nd degree murder, manslaughter, sexual assault, assault, using a firearm in the commission of an offence, robbery, forcible confinement, criminal harassment, uttering threats, and other violent violations. Note that traffic violations are excluded from the analysis.

The GSS captures both physical and sexual violence as defined by the Criminal Code and includes violent acts such as being threatened, pushed, grabbed, beaten, choked, threatened with a gun or knife, or forced into sexual activity. All of these violent acts, as well as stalking, are examined in this report. Respondents to the GSS are also asked about emotional and financial abuse that they might have experienced at the hands of a current or ex-spouse. However, this study does not estimate the costs of emotional and financial abuse as separate incidents for several reasons. First, the main question on the survey asks respondents about lifetime emotional or financial abuse, making it impossible to estimate incidents of abuse in any particular year.Footnote 8Second, all spousal violence situations can involve some degree of emotional abuse (e.g., the emotional and psychological impact of any abuse or violence by someone you love, trust, have committed to, etc.); hence, costing emotional abuse as an individual incident will lead to double counting, as emotional costs (pain and suffering) are already estimated in various sections. And third, the decision to exclude self-reported acts of financial and emotional victimization at the hands of a spouse is consistent with the practices of other government agencies that report on spousal violence (Brennan et al. 2011).

2.2.5 Gender

To reflect the well-documented differences between male and female victims of spousal violence, the estimates presented here have been disaggregated according to the victim's gender. Since there is disagreement over the frequency and severity of spousal violence victimization in the gender groups, we have not relied on one prevalence rate per gender for the entire report, but instead use the appropriate data to calculate different rates for each cost item. For example, the breakdown of criminal justice system costs by gender is derived from police-reported information, and the breakdown of hospitalization costs by gender is calculated directly from GSS respondents who stated specifically that they were hospitalized for an injury.

2.2.6 Cost categories

The cost items are grouped into three main categories: the impact borne by the justice system (criminal and civil), the impact borne by the victim, and the impact borne by third parties. Within each of these categories, several sub-categories have been delineated, some of which are individual cost items themselves (e.g. police costs) and some of which are further divided into individual cost items (e.g. corrections is divided into incarceration costs and conditional sentence costs, among others).

Cost items fall under the category of the party that bears the actual impact, not the category of the party that bears the actual financial burden of the cost item. For example, medical costs resulting from injury to the victim fall under the “victim costs” category because the victim bears the impact of the injury, even though a third party (health system) bears much of the financial cost of the medical treatment. Section 6 provides an additional breakdown of the economic impact by the party who actually bears the financial cost. In that section, the costs are attributed to the state, private sector and individuals. The three main impact categories are described in more detail as follows. Impact borne by the justice system – criminal and civil justice system costs

The Canadian criminal justice system is comprised of three main institutional bodies (the police, the courts, and corrections), each with a specific role in protecting victims and the public, applying the rule of law and ensuring offenders pay their debt to society and are assisted in rehabilitation. In the context of spousal violence, police are tasked mainly with protecting victims and detecting and investigating criminal acts. Criminal courts provide a forum where the accused can be tried and, if convicted, sentenced. The corrections system places serious offenders in confinement and limits the actions of other offenders. As a whole, the justice system has the responsibility of preventing crime by acting as a deterrent, of ensuring respect for the rule of law and the maintenance of a peaceful society, and of denouncing criminal activity by placing restrictions on offenders. The three interconnected institutions all involve widely disparate processes, equipment, infrastructure, and personnel. Previous studies have attempted to estimate the costs of violence across the entire criminal justice system at an aggregate level only (Walby 2004; Brand and Price 2000). Following Cohen et al. (1994), this study estimates the costs of spousal violence at each stage of the criminal justice system.

Table 2.1: Categories of Justice System Costs
Criminal Justice System Civil Justice System
Police Civil Protection Orders
Court Divorce and Separation
Prosecution Child Protection Systems
Legal Aid  
Conditional sentences  

Table 2.1 lists the sub-categories that are included in the criminal justice system analysis. The three institutional bodies are all represented: police; courts, including prosecution, and legal aid sub-categories; and corrections, further separated into incarceration, conditional sentence, probation, and fines. As fines are considered a cost to offenders, and as costs to offenders are not considered in this report, fines will not be counted in the main analysis. Fines will, however, be considered in the analysis by who pays in the conclusion of the report, counted as a revenue to the state and a cost to individuals.Footnote 9 Other criminal justice activities such as non-legal aid defence costs and expenditures related to Part XX.1 of the Criminal Code (mental disorder and review boards) have not been considered due to data limitations.

The criminal justice system is the most relevant legal system for most violent crimes, but in the case of spousal violence, the civil justice system is also an important part of the legal response. The civil justice system is concerned primarily with matters that arise between private parties, such as an action in damages or a family law matter. Although civil redress may take many forms, including torts and civil damages, the two main types of civil legal actions that are covered in this report are civil protection orders and divorce and separation, as listed in Table 2.1.

Civil protection orders are available either through the common law, as general injunctive relief, or under specific legislation. The superior courts have inherent jurisdiction to grant injunctive relief, including restraining orders (Christopher 2009). Restraining or no-contact orders are also available under provincial or territorial family law if the victim is undergoing a separation or divorce. Moreover, emergency protection orders or emergency intervention orders are available in most provinces and territories under specific civil family violence or domestic violence legislation.Footnote 10 They can grant the victim temporary exclusive occupation of the home, remove the abuser from the home, set limits on contact and communication with the victim, as well as other remedies.

A divorce or legal separation can be sought if the victim decides to permanently leave the intimate relationship. Ongoing violence against a spouse may create a dangerous or harmful atmosphere for a child, and authorities may be forced to remove the child from the home to ensure personal safety. Exposing a child to domestic violence is grounds for child protection involvement in most provinces and territories.Footnote 11 Children are often themselves victims of violence perpetrated by a parent figure, but the scope of this report permits only cases involving spousal violence, whether violence against children is present or not. Impact borne by primary victims – victim costs

While unreported crimes incur no immediate or measurable costs to the criminal justice system, the impact of unreported incidents on the victim may be just as significant as the impact of reported incidents. For this reason, victim costs are calculated separately based on self-reported data such as the GSS in order to capture the prevalence and costs of spousal violence more accurately.

Victims bear the direct impact of spousal violence and they may sustain both physical and emotional injuries. Physical injuries may be immediate and short-term in nature, but chronic violence can result in long-term physical health effects such as headaches, ulcers, and reproductive problems (Johnson and Dawson 2011). Psychological and emotional effects caused by spousal violence include fear and anxiety, depression, and addiction issues (Johnson and Dawson 2011; Bradley et al. 2002), and spousal violence can diminish one's sense of safety, self-esteem, and confidence as well. Table 2.2 below lists the specific items in this section.

Table 2.2: Categories of Victim Costs

  • Health Care,
  • Lost education,
  • Physician visits,
  • Lost childcare services,
  • Emergency department visits,
  • Other Personal Costs,
  • Acute hospitalization,
  • Damaged or destroyed property,
  • Mental Health Issues,
  • Divorce and separation (legal costs),
  • Medical services,
  • Special phone features,
  • Work loss,
  • Moving expenses,
  • Suicide attempts (medical costs),
  • Intangible Victim Costs,
  • Productivity Losses,
  • Pain and suffering,
  • Lost wages,
  • Loss of life,
  • Lost household services

Victims sustaining physical injuries may seek medical attention. The costs of visits to physicians and emergency departments are included, as are the costs of overnight hospitalizations. Mental health issues caused by spousal violence may also necessitate the use of medical services, including counselling, or have a detrimental effect on productivity. The victim may not be able to perform normal daily activities, which may result in lost wages, lost household productivity, lost education, or lost child care. If the parties decide to end the relationship through separation or divorce, the victim might need to pay the necessary legal fees. The offender may attempt to intimidate or get revenge on the victim by destroying or damaging the victim's personal property. The victim may be forced to activate special phone features in order to avoid the offender, and the victim may even relocate if the safety of the victim or his or her children is threatened.

While all of the effects mentioned thus far incur a direct cost, other effects are abstract. For example, pain and suffering are hidden, but are no less real than tangible impacts. Emotional trauma caused by the situation may lead to a deterioration of mental health and subsequent use of mental health care services. The most severe cases of spousal violence result in the loss of a life, either at the hands of the offender or by the victim's own volition. All of these impacts present financial costs, physical costs, and emotional costs to the victim. Impact borne by third parties and others – third-party costs

The impact of spousal violence ultimately extends to every member of society. Any party with a relationship to the victim, from the children to employers, may consciously be aware of how the violence affects them. Persons and entities with no direct relationship to the victim are also affected, at the very least through the allocation of public funds, but may be unaware in what way or to what extent. Table 2.3 below lists the costs to all third parties.

Table 2.3: Categories of Third-Party Costs

  • Funeral Expenses,
  • Losses to Employers,
  • Loss of Affection and Enjoyment to Family Members,
  • Lost output,
  • Costs to Other Persons Harmed/threatened,
  • Tardiness and distraction,
  • Health care,
  • Administration costs,
  • Productivity losses,
  • Negative Impact on Children Exposed to SV,
  • Social Service Operating Costs,
  • Medical costs,
  • Shelters and transition homes,
  • Missed school days,
  • Crisis lines,
  • Lost future income,
  • Support centres,
  • Delinquent acts against property,
  • Victim services,
  • Other Government Expenditures,
  • Other federal expenditures,
  • Other provincial and territorial expenditures

In the event of the victim's death, the most immediately affected is the victim's family, who is not only burdened by administrative matters and costs such as funeral expenses, but is also deprived of affection and enjoyment. Children who are exposed to the violence can be considered secondary victims. It has been documented that children exposed to violence by one parent against another often suffer from emotional, social, cognitive, and behavioural maladjustment problems, and that they are at greater risk of developing these problems than children who are not exposed to spousal violence (Fortin 2009; Wolfe et al. 2003; Jaffe et al. 1990; Margolin 1998; Peled and Davis 1995; Holden et al. 1998). These negative impacts on children exposed to spousal violence can continue into adulthood (Dube et al. 2002; Zametkin et al. 1990; Harrington et al. 1990; Loeber and Hay 1997) and the costs associated with a child developing these problems may accrue over the remainder of the child's life. Furthermore, Fantuzzo and Mohr (1999) suggest that individuals exposed to domestic violence in childhood are also at risk for repeating the pattern of violence – as victims or perpetrators – in their own adult intimate relationships, but the costs of second-generation spousal violence are beyond the scope of this study.

Some violent incidents involve a third party sustaining physical injuries that require health care services or that force suspension of daily activities. Government and non-profit organizations are active in providing support for spousal violence victims in many ways: establishing social services such as shelters, crisis lines, and support centers, and providing a variety of victim services in collaboration with police, courts, and the community. Federal, provincial, and territorial governments also provide funding for a variety of prevention and awareness programs. As stated in the description of victim costs, the victim may miss work because of either physical or emotional distress, but actual work performance may also suffer. The victim may be consistently late or may not be able to concentrate on work tasks, and this kind of lost productivity will reflect in the employer's output.

2.2.7 Cost item-specific models

The framework sets out the scope of the analysis and provides guidance for developing the many item-specific cost estimation models. One major difficulty is the disparity in the nature of each type of cost, which precludes a standard, uniform estimation method for all cost items. For example, the elements that comprise the police costs (e.g. salaries for police investigating crime, spending on tools used in investigations) are fundamentally different than elements that comprise the medical costs (e.g. ambulance use, medical staff salaries, overnight hospitalization costs). It follows that the sources and methods for calculating police costs are different from that for calculating medical costs. The effect that data limitations have on developing rigorous estimation methods is another challenge. In many cases, the best estimation method is not feasible because the appropriate data are not available.

Together, these two difficulties force each specific cost item to be examined individually. Each cost estimation model is explained in detail in the appropriate section of the report: 3. Justice System Costs, 4. Victim Costs, and 5. Third-Party Costs. In addition, complete and detailed calculations for each cost item are presented in the appropriate appendix: Appendix A: Justice System Costs, Appendix B: Victim Costs, and Appendix C: Third-party Costs.

2.3 Data Sources

The UCR2 is the main data source for the costs to the criminal justice system, while the 2009 GSS, Cycle 23 is the main data source for the victim and third-party costs. Police-reported data and self-reported data complement each other, and are both used to attain as comprehensive an account of spousal violence as possible. Other reliable sources have also been utilized including Statistics Canada publications, government- commissioned surveys and academic research.

2.3.1 Uniform Crime Reporting Survey 2 (UCR2)

The UCR2 survey is an administrative survey that captures detailed information on all Criminal Code violations reported to and substantiated by police services across Canada. The individual-level micro data indicates the relationship of the accused to the victim, which allows identification of spousal and common-law relationships, as well as whether the parties were divorced or separated at the time of the incident. The UCR2 survey records incidence, accusation, and charge statistics for each Criminal Code violation, and specifies the genders and ages of the offender and victim. The survey is comprehensive in both details and in geographic scope, as national coverage of the UCR2 survey was 99% in 2009. This study uses UCR2 data to find the number of incidents requiring police resources, which is then used to estimate the number of criminal court cases.

2.3.2 General Social Survey (GSS), Victimization, Cycle 23

The General Social Survey (GSS) is a regular annual survey conducted by Statistics Canada; each year or cycle focuses on a particular social subject. These surveys provide insight into social phenomena of current or emerging interest. The victimization cycle was first conducted in 1988 and now occurs every five years. It is the only national survey of self-reported victimization.

The target demographic is Canadians aged 15 and over who can conduct an interview in one of the official languages and who can be reached by landline telephone. Data collection for the territories in 2009 was carried out differently from the territories in that a portion of interviews were conducted in person in an attempt to improve response rate and representativeness. Microdata for the territories were not accessible for 2004 or for 2009 and as such, references to the GSS are for provincial data only.

GSS respondents are asked about their personal experiences of criminal victimization, regardless of whether or not the incident was reported to police. Other questions inquire of the respondents' perceptions of crime, safety, and the justice system.

As not all crimes are brought to the attention of the police, the GSS provides an important complement to police-reported crime statistics found in the UCR2 survey. Since victimizations not reported to police have no less of an effect on victims, family, and employers than those incidents that are reported to police, self-reported data as recorded in the GSS gives a more comprehensive and accurate reflection of the prevalence and impact of spousal violence in Canada. The GSS victimization cycle contains two questionnaires and corresponding data files; the main file includes all questions relating to victimization of violence perpetrated by a current or former spouse, while the incident file includes questions relating to non-spousal victimization. Demographic information relevant to the cost items, such as gender, age, work status, annual income, education and family size, is recorded. The main file, which is used in this report, has a count of 19,422 respondents, and each respondent represents an average of approximately 1,400 non-institutionalized Canadians.

2.3.3 Other major data sources

2.4 Limitations

Cost analysis of a widespread, complex social issue such as spousal violence involves limitations related to data availability, data reliability, and scope and method. Although efforts are made to include all reasonable impacts of spousal violence, the paucity of data has rendered this impossible. Some cost items are underestimated as a precaution while others are omitted completely. Underestimation occurs where it is necessary to make an assumption because of insufficient data and where more than one assumption is possible; in these cases, the assumption resulting in a more conservative estimate is used in order to ensure a lower bound.

2.4.1 Data availability

The main limitation of this study is the lack of available data. Information gaps lead to omission or underestimation of some cost items, and assumptions adopted prohibit a fully accurate overall estimate of costs. For example, calls to crisis lines are anonymous and so it is impossible to determine how many calls a single caller makes; as such, an average number must be assumed based on qualitative data obtained from frontline workers. The lack of data may alternatively force the use of next best available data. For example, the number of children who have developedor will develop hyperactivity solely as a result of exposure to spousal violence in 2009 is unknown, but prevalence rates for child witnesses and non-witnesses can be calculated from a Statistics Canada publication (Dauvergne and Johnson 2001) that uses regression analysis to compare hyperactivity in the two groups. The hyperactivity prevalence rates of the two groups can then be used in the calculation of the number of children exposed to violence in 2009 who develop hyperactivity. Making necessary assumptions and using the next best available data are both common problems. A high-quality analysis of the economic impact of spousal violence would require very specific data that would be difficult to collect even through targeted studies or surveys.

2.4.2 Data reliability

The reliability of available data sources is also a major issue. The UCR2 survey collects data from police jurisdictions across Canada. Having individual police services report their own data may result in an aggregation of dissimilar data in the UCR2 survey. In addition, each jurisdiction has its own procedures and policies. For example, while all jurisdictions have some form of pro-charging or pro-prosecution policy for domestic violence, the individual responsible for the decision to lay a charge differs by jurisdiction. In British Columbia and Québec, the Crown makes the decision to lay a charge, while in New Brunswick it is the police that lay the charge based upon advice from the Crown (Department of Justice 2003, 11).

As noted previously, the UCR2 survey only records incidents that are reported to police. Based on self-reported statistics, the UCR2 therefore does not include the majority of spousal violence incidents. Victims and suspecting third parties may not report the incidents to police because of the individual's “circumstances, feelings, beliefs and level of knowledge about family violence” (Department of Justice Canada, “Family Violence: Department of Justice Canada Overview Paper” 2009, 6). However, the police-based nature of the survey does make it suitable to examine the criminal justice system costs as the actual criminal justice system spending is based on the incidents that are brought to the attention of police.

The main limitation of the GSS is its lack of complete and comprehensive coverage. The GSS coverage may exclude certain demographic and socio-economic groups, which may lead to a misrepresentation of these groups after weighting techniques are applied.

One particular source of underrepresentation is that the survey is conducted solely by landline telephone. Recent social and technological developments make it costlier and more difficult to conduct representative telephone surveys (Tourangeau 2004). These developments include the increased use of landline telephone features such as caller identification, call blocking, and answering machines. In addition, many individuals and households cease using landline telephones altogether in favour of cellular phones. For instance, the percentage of households relying solely on cellular phones increased from 6.4% to 13.0% from 2007 to 2010.Footnote 12 In the 2009 GSS, persons with only cellular telephone service were excluded from sampling. While major disparities between the sample and the population in terms of gender, age, income, education, and marital status are partially addressed by the use of weighting techniques and quota sampling, the bias resulting from non-representative coverage cannot be completely mitigated by these techniques. Specifically, young people are more likely to have cellular phones only (Arcturus Solutions 2008) and therefore younger demographic groups are more likely to be underrepresented by the 2009 GSS.

The failure of the GSS to capture persons who are institutionalized at the time of the survey is a second reason for the possible misrepresentation of demographic groups. Individuals staying in shelters, retirement homes, hospitals, correctional facilities and other non-traditional accommodations will not be accessible through the methods used in the GSS. People in these institutions may have different victimization rates than people in the non-institutionalized population, and as a result, the GSS may not accurately estimate the prevalence of spousal violence.

Another related problem of misrepresentation may result from the criterion that respondents must complete the survey in English or French. This excludes a significant number of people from consideration to be a GSS respondent, as 1.6% of Canadians could not speak either official language in 2006 (Canadian Heritage 2010). Language minorities, such as immigrants or refugee claimants who are not proficient in either official language or Aboriginal persons in remote communities who speak only their native language, are the most obvious groups affected by this requirement.

The issues discussed above are contributors to imperfect coverage by the GSS. In addition, note that the non-response rate for the 2009 GSS was 38%. Little or nothing is known about these non-responding cases, and so the results may be biased to the extent that the non-responding cases differ from those that provided responses.

Another limitation of the GSS is its reliance on respondents to accurately recall events that may have been emotionally traumatizing. Over time, events may become altered in memory and victims may not remember details of victimization experiences (see Perreault and Brennan 2010). Some respondents may not be candid intentionally. The interpretation of questions, a fundamental element of surveys, can vary between individuals despite extensive focus group and pre-testing.

The GSS uses a structured, closed-ended questionnaire. Respondents have no opportunity to describe the incidents, and instead must respond to questions posed by the telephone interviewer. Spousal violence is highly contextual. For example, a man could describe an incident where he was assaulted by his spouse, when in reality, his spouse may have been acting out of self-defence. It is important to bear this in mind when interpreting GSS results.

Issues also arise from the structure of GSS questions. Specifically, there is insufficient information to determine whether outcomes of violence were a result of one single incident or a series of repeated incidents. For example, one victim may be hospitalized once as a result of a single violent incident while another victim may be hospitalized multiple times as a result of multiple violent incidents. However, both of these cases are recorded as a one-time hospitalization, as the response to that survey question is only yes or no. The lack of clarity makes it impossible to identify the link between hospitalizations and incidents that caused those hospitalizations. As this problem extends to all of the questions regarding outcomes of violence, many cost estimates in this report are based on the number of victims, not the number of incidents. Given that almost one half of victims stated that they had been victimized on more than one occasion, and that injuries and psychological trauma can be worsened with repeat victimizations (as is often the case in spousal violence), this approach might result in cost underestimation.

One reliability issue shared by both the UCR2 survey and the GSS is the possibility that they present a distorted picture of the spousal violence landscape. There may be circumstances common to relationships in which spousal violence is present that consistently prevents one group of victims from being able to report their victimizations. For example, in situations involving complete control and dominance of one spouse over another, the victim may not have the ability (being physically prevented by injuries or lack of resources, or emotionally prevented by fear of reprisal or harm to children) to contact police to report the crimes committed by the offending spouse, or the offending spouse may not allow the victim to answer the phone, thus preventing the victim from taking part in the GSS. Should either of these situations typically involve one specific demographic of victims (women, for example), then both the UCR2 and the GSS would report lower numbers of victimizations for this group as compared to other groups (men).

2.4.3 Scope and method

Estimating the costs of crime is complicated by two initial necessary choices:

  1. the choice of what actually is a cost (the scope), and
  2. the choice of what method will result in the most accurate estimate.

The first choice is challenged by the nature of society. Every component of society is interconnectedwith every other component, so that any action may have an effect on the rest of society regardless of how unconnected that action may seem. In addition, the action may produce future actions, and this cycle may continue indefinitely. In this way, providing a cost estimate of one specific action (e.g. spousal violence) necessitates a decision to include only a certain range of the theoretically limitless effects, both present and future. The difficulty is in determining what actually should be considered a direct consequence of the action. Some costs are obvious, such as any medical treatment to address injuries sustained by a victim, or police and court costs associated with the investigation and trial of an assault. Other costs are less obvious or may not be limited to one cause, such as the increased mental health risk associated with victims of sexual assault or the lost work productivity caused by distraction of the victim of repeated assaults.

Once the cost items have been decided, a method of calculating each one must be devised. The difficulty here arises due to the heterogeneity of the different cost items. For example, the method for calculating the direct medical costs of an assault victim cannot be used as the method for calculating lost work productivity caused by distraction. Even within each cost item, decisions must be made: is it feasible to interview every victim about his or her injuries and the associated medical costs, or do average medical costs across the population have to be applied instead? Methods that are more accurate may not be used in several cases based on the constraints of time and resources.

2.5 Methodology Discussions

Economic impacts, or costs (which is the term more widely used), can be categorized in many ways.

A brief overview of commonly used cost classifications (definitions) is presented in this section.

2.5.1 Classifications of costs

Four common classifications of costs are: real vs. transfer; tangible vs. intangible; out of pocket vs. opportunity; and short-run vs. long-run. Access Economics (2004) and Laing and Bobic (2002) provide useful discussions of some of these classifications.

Real vs. transfer: Real costs diminish or prevent the growth of the overall production or consumption capacities of the economy. For example, a real cost is incurred if a victim of spousal violence is no longer able to work and exits the labour force, thus removing production capacity from the economy. Transfer costs are simply financial transfers from one economic agent to another that do not actually constrain economic growth. An example of a transfer cost would be the legal fees for a divorce caused by spousal violence that are paid from the victim to the lawyer, but neither has reduced their production capabilities because of the transfer.

Tangible vs. intangible: Tangible costs are actual expenditures that would not have occurred in the absence of spousal violence. For example, court costs for cases of spousal violence are tangible costs. Intangible costs are abstract and do not have a market price. Though they cannot be detected or precisely measured, they are negative realities that are caused by the spousal violence. The primary intangible costs are the pain and suffering of victims and the loss of life.

Out of pocket vs. opportunity: There are many parallels between tangible vs. intangible costs and out of pocket vs. opportunity costs. Out-of-pocket costs are direct, immediate expenditures that are easily measured because they have a price. In the case of spousal violence, out of pocket costs include everything from police costs (an out-of-pocket cost to the state) to medical costs for the victim (a combination of state and personal out-of-pocket costs) to funeral services (an out-of-pocket cost to the family of the victim). Opportunity costs are the costs of any good, service, or activity measured in terms of the value of the next best alternative forgone. For example, if a victim spends several hours with police on an investigation, that victim suffers from a lost opportunity to earn money or to enjoy leisure activities during these hours. In this particular case, even this off-work time is covered by the employer (such as leave with pay) – no actual wages were lost, the lost opportunity, valued at forgone hourly wages, would still be one measure of the opportunity costs of spending with justice system.

Short-run vs. long-run: Making the distinction between short-run and long-run costs is difficult because every individual situation is unique. Short-run costs are those immediate impacts that will likely eventually disappear, while long-run costs may only be incurred after the violence and are continually incurred over a long period of time. The difficulty is that costs considered as short-run, such as medical costs for a physical injury, may never stop incurring, while those impacts considered as long-run, such as mental health issues, may end relatively soon.

Every type of cost mentioned above is represented in this report, and there is great overlap among these classifications. However, this report does not explicitly categorize costs in any of these four ways, but instead separates costs in an original and unique way: justice system costs, victim costs, and third-party costs. This categorization allows for an examination of the effects that spousal violence has on different parties. As noted earlier, there is also a brief presentation of the costs categorized by who actually pays (the state, private sector, or individuals) at the end of this report.