Victims of Crime Research Digest No. 10

Missing and Murdered Indigenous Women and Girls: The importance of collaborative research in addressing a complex national crisis

By Marsha Axford

Marsha Axford, MCA, is a Researcher with the Research and Statistics Division, Department of Justice Canada. She has research experience on a range of issues related to both offenders and victims of crime.


The overrepresentation of Indigenous Canadians in the criminal justice system (CJS), both as victims and as accused, is a serious and ongoing challenge. Federal departments, provincial justice systems, social service organizations, and academic researchers have examined the issue for many years (see Solicitor General 1996; Rudin 2005; and The Office of the Correctional Investigator 2013 for an overview). Although some steps have been taken to address Indigenous overrepresentation as accused in the CJS (for example, through national and provincial inquiries and the consideration of Gladue factors in court), little measurable progress has been made.

The Homicide Survey

The Homicide Survey was introduced in 1961 and has been revised and expanded multiple times in its history. From the outset, the Survey collected information from investigative officers about the relationship between the victim and the accused involved in their homicide. Prior to 1991 however, the Survey collected this information via a write-in text field with possible relationship types defined in the Survey User Guide. The options for relationship type were expanded over time, and the write-in text field was changed to a list of options from which the responding investigative officer can now choose. Changes to relationship types occurred in 1991, 1997 and 2015. The Homicide Survey began to include “casual acquaintance” as a relationship type in 1997.

According to the Survey User Guide, a “casual acquaintance” is someone known to the victim, but who has no romantic, sexual or close friendship with the victim. This relationship type is to be used when none of the other “acquaintance” relationship types (i.e. close friend, neighbour, authority figure, business relationship or criminal relationship) are appropriate. The Homicide Survey’s relationship types are hierarchical; the closer the relationship to the victim, the higher the relationship is on the hierarchy. Further, in cases where multiple relationship types describe the link between accused and victim, only the closest relationship type is selected. For example, if the accused and victim were spouses and also business partners, only the spousal relationship would be reported. If the accused were the uncle of the victim and also her neighbour, only the familial relationship would be reported. In the Homicide Survey, the only relationship type considered more distant than “casual acquaintance” is “stranger.”

The Homicide Survey features separate questionnaires focused on the homicide incident, the accused and the victim. In some cases, there are multiple accused and victim questionnaires for a single incident questionnaire. The victim questionnaire collects the relationship between victim and the closest accused person.Footnote 13 The incident questionnaire includes a free-text “narrative” section: a summary provided by the investigative officer of the circumstances leading up to and surrounding the homicide incident. The officer determines how much detail to include in the narrative; the intent is to summarize relevant information about the incident, victim and accused.

Recently, attention has focused on the overrepresentation of Indigenous peoples among victims of violent crime. Many national and international organizations have made recommendations to address the issue of missing and murdered Indigenous women and girls in Canada (Native Women’s Association of Canada 2010; Pearce 2013; United Nations 2014; Royal Canadian Mounted Police 2014; Royal Canadian Mounted Police 2015; United Nations 2015; Truth and Reconciliation Commission of Canada 2015;). After the 2015 federal election, the new government issued mandate letters to all Ministers outlining expectations and goals for the next four years. The Ministers of Justice, and Indigenous and Northern Affairs, with support from the Minister of the Status of Women, were mandated to develop an approach to an inquiry into Missing and Murdered Indigenous Women and Girls (MMIWG). While initial investigations had begun in other public agencies (for example, The Royal Canadian Mounted Police [RCMP] reports released in 2014 and 2015) the mandate letter reaffirmed the government’s commitment to move the issue forward.

In 2014, Statistics Canada worked with the RCMP and the wider Canadian police community to gather information on the AboriginalFootnote 14 identity of female victims of homicide. These data were used to produce the RCMP’s report Missing and Murdered Aboriginal Women: A National Operational Overview and to update Statistics Canada’s Homicide Survey database. The Homicide Survey includes a section for responding police services to report the Aboriginal identity of victims and accused persons, although historically this information has not been routinely reported. The RCMP-Statistics Canada initiative provided Statistics Canada with the Aboriginal identity of female victims of homicide between 1980 and 2014; 2014 was the first reporting year that Aboriginal identity was reported on the Homicide Survey for all victims and accused persons, regardless of sex.

Analysis of the newly updated dataset revealed that a relatively large proportion of female victims between 1980 and 2015 were killed by a “casual acquaintance” (12%), and that Aboriginal female victims (18%) were more likely than non-Aboriginal female victims (11%) to be killed by a “casual acquaintance.” This difference between Aboriginal and non-Aboriginal females, contextualized in the wider focus on Missing and Murdered Indigenous Women and Girls, was the impetus for initiating deeper analysis of the Homicide Survey dataset.

Interdepartmental collaboration

In the spring of 2016, Statistics Canada’s Canadian Centre for Justice Statistics (CCJS) and the Department of Justice Canada’s Research and Statistics Division (RSD) began to collaborate on a special analysis of Homicide Survey data. This partnership provided Statistics Canada with additional resources to publish new analyses within a short period of time. The goal was to further examine specific aspects of “casual acquaintance” murders, particularly the relationships between accused persons and female victims, and the circumstances surrounding the homicides. The project involved reading and analyzing a free-text portion of the Homicide Survey called a “narrative,” which is provided by responding officers on the Homicide Survey, for cases in which the victim was murdered by a “casual acquaintance.”

Officials from Statistics Canada and the Department of Justice discussed the scope of the special analysis and the resources required to complete the project. The officials decided to assign two Department of Justice employees with previous file-coding experience to the project. The two became temporary deemedFootnote 15 employees of Statistics Canada, in accordance with Statistics Canada legislation. The officials felt that two researchers working together could:

  1. discuss and analyze the content of police narratives throughout the re-coding process;
  2. allow for additional insight into possible new relationship types that a single researcher working alone may not have considered; and
  3. provide peer support throughout the process, as analyzing a large number of homicide narratives within a short period of time can be difficult and result in vicarious trauma (Campbell 2002).

The goal was to include the special analysis in Statistics Canada’s annual publication of detailed homicide data: the Juristat article “Homicide in Canada, 2015.”Footnote 16 Work on recoding of the data and building a supplementary dataset began in May 2016 and was completed in August 2016. The two Department of Justice employees worked at Statistics Canada two days per week on these tasks.

Once the majority of cases had been reviewed and re-coded, one of the employees continued to work at Statistics Canada part time until the Juristat article’s release on November 23rd, 2016. The employee incorporated feedback from internal and external review processes, re-analyzed data where necessary, contributed to the editing process, and assisted in release-day activities.

Project methodology

A Memorandum of Understanding outlined the activities, services and products expected of the Justice Canada employees assigned to Statistics Canada for the project. Activities included developing a coding structure and method for analysis, creating a supplementary dataset based on the information found in the Homicide Survey narratives, and incorporating statistical findings into the Juristat article “Homicide in Canada, 2015.”

As part of the process to be deemed Statistics Canada employees, the researchers took an oath as per the Statistics Act. The oath binds employees to the requirements of the Act and all policies related to the use of data collected and held by Statistics Canada, particularly those related to confidentiality. All work on the project was completed on protected networks in Statistics Canada offices, where the researchers accessed only the datasets and files required to complete the project.

A total of 755 victims were originally identified for inclusion in the study, based on a number of criteria, including:

  1. The victim was female;
  2. The homicide was recorded in a Homicide Survey submitted to Statistics Canada between 1980 and 2015; and
  3. The relationship between the accused and the victim was reported as “casual acquaintance.”Footnote 17

The CCJS team pulled a subsample of 100 of the narratives that met the above criteria for cursory analysis by the Justice employees. This initial examination served to orient the researchers to the types of information contained in the narratives and to develop a preliminary list of relationship types that might provide further context to the homicide incidents. After this initial overview, the two researchers met with the CCJS team to discuss methodology, the relevance of context and specific relationship types identified in the narratives, technical details of the existing Homicide Survey dataset and other research-specific issues.

All 755 narratives (including the 100 previously examined) were read by both researchers simultaneously to identify recurring themes and contexts. During the review process, the CCJS team and the two researchers consulted methodologists from Statistics Canada’s Household Survey Methods Division (HSMD). The methodologists helped select appropriate thresholds that justified the creation of new relationship types (1% of the entire sample) and provided direction on inter-coder verification to ensure that the new categories were statistically accurate and robust.Footnote 18

Themes and contexts for which 1% or more of the total sample were identified were assigned descriptive names and analyzed according to Aboriginal identity. Themes identified among fewer than 1% of sample were not changed from “casual acquaintance” as the evidence was insufficient to support the creation of a new relationship type.Footnote 19

During the subsample analysis, the researchers identified a number of relationship types and contexts that did not meet the criteria for inclusion (1%), or were determined to be related more to motive than to relationship type, or were considered too context-specific. In some cases, the information available was too vague to be considered reliable. For example, the length of time that a victim and accused person were acquainted was identified as a theme in some of the narratives (e.g., the victim and accused were acquainted for less than 24 hours before the homicide occurred). Theoretically, the length of a relationship could help define the difference between “stranger” and “casual acquaintance.” This is done in other jurisdictions,Footnote 20 but since the Homicide Survey does not collect this information, and the availability of these data in the narratives was variable, it was excluded from the study. The potential category of a predatory/targeted “relationship” was also identified during the subsample analysis, but was determined to relate more to motive than relationship, and was not included in the final coding structure, in part because the Homicide Survey already collects information about motive.


Due to incomplete information, seven homicide victims were excluded from the final dataset, for a total of 748 female victims killed by a casual acquaintance between 1980 and 2015. For about half of the victims (390 or 52%), the police narratives included enough information to support further analysis of the nature and context of the “casual acquaintance” relationship between accused and victim. The researchers determined that in 17% of the 748 cases, the relationship could be better described by a relationship type currently collected on the Homicide Survey (e.g., neighbour, or other intimate partner). In many of these cases, the “better” relationship type did not exist at the time the Homicide Survey was completed. Often the Just over one-third (35%) of all accused-victim “casual acquaintance” relationships could be described by a new relationship type.

In 48% of the 748 cases, the researchers did not reclassify the “casual acquaintance” relationship because the narratives featured either enough information to justify the original classification, or not enough details to justify reclassification.

Overall, the researchers identified eight new relationship types among the “casual acquaintance” cases:

The relationship type of co-substance user was identified most frequently: 18% of the 748 victims were killed by someone with whom they were consuming intoxicating substances in a bar, private residence, or public outdoor location immediately prior to the homicide. Aboriginal female victims were much more likely to be killed by a co-substance user (38%) compared to non-Aboriginal female victims (12%). Of the 136 victims killed by a co-substance user, 39% were murdered after they left the original location of substance use with the accused. Again, there was a considerable difference between the two groups, with non- Aboriginal women being much more likely to leave with the accused than Aboriginal women (51% compared to 27%).

Much smaller proportions of victims were killed by someone they knew outside of the co-substance user context:

Other differences were noted between the homicides of Aboriginal and non-Aboriginal females within these less frequently reported categories. For example, 5% of non-Aboriginal victims were killed by an other institutional member, compared to 1% of Aboriginal victims. Non-Aboriginal females were also more likely to be killed by an other non-family household member (6% compared to 3% for Aboriginal victims).

What does this mean?

This analysis would not have been completed without the collaborative efforts of Statistics Canada and Department of Justice Canada. The project provides a deeper level of understanding of the contexts and relationships related to homicides of females, and adds to a growing body of knowledge of the risk factors associated with violence and how they may differ between Aboriginal and non-Aboriginal females. In the context of MMIWG, no single department or organization will be able to address the myriad of relevant factors and none have access to all of the expertise necessary for such an undertaking. Collaboration is necessary and important to address such a complex national problem, and the research completed by Justice Canada and Statistics Canada demonstrates what this type of cooperation can produce.