Violence Perpetrated by Ex-spouses in Canada

2. Methodology

As the purpose of this report is to provide an update to Hotton's (2001) Spousal Violence after Marital Separation report, this report follows, where possible, the format of that report. As with Hotton's (2001) report, the data presented in this report are based on three different surveys: The General Social Survey - Victimization, the Incident-based Uniform Crime Reporting Survey and the Homicide Survey.

General Social Survey - Victimization

Approximately half of the General Social Survey - Victimization data presented in this report are based on the 2009 Public Use Microdata File. The Public Use Microdata Files from the 1999 and 2004 GSS were also used. Where data were not available in these data files, special requests for data were made to the Canadian Centre for Justice Statistics (CCJS).

The GSS data presented in this report are based on responses collected from a representative sample of Canadians. The 2009 GSS collected data from over 19,000 Canadians in the provinces, who represent approximately 27 million Canadians. Data were collected in five waves between February and November 2009.

The GSS on Victimization is conducted every five years by Statistics Canada. In the provinces, data are collected through a computer-assisted telephone survey and participating households are selected through random digital dialing.[5] The GSS includes Canadians 15 years of age and older who are not full-time residents of institutions (e.g. hospitals and prisons). In this report, only data collected from the provinces are examined and thus does not reflect a Canada-wide perspective.

The analyses for which tests of significance were conducted are noted throughout the report. In these cases, a Chi-Square Test for Independence[6] was conducted.[7] At each point in the text when the term "significant" is used, the Pearson's Chi-Square Test revealed a significant difference at less than p < .001.

Coefficient of Variation and Symbols

In regard to the coefficient of variation, please note the following:

The GSS data are based on information collected from a sample of the population and are therefore subject to sampling error[8]. Although the exact sampling error of the estimate cannot be measured from sample results alone, it is possible to estimate a statistical measure of sample error, the standard error. Because of the large variety of estimates that can be produced from a survey, the standard error is usually expressed relative to the estimate to which it pertains. The resulting measure, known as the coefficient of variation (CV) of an estimate, is obtained by dividing the standard error of the estimate by the estimate itself and is usually expressed as a percentage.

This report uses the coefficient of variation (CV) as a measure of the sampling error. For the purposes of this survey, an estimate with a coefficient of variation (CV) of higher than 33.3% is considered too unreliable to be published and the symbol "F" is printed in the corresponding cell of the data table or figure. When the CV of the estimate is between 16.6% and 33.3%, the corresponding estimate is accompanied by the symbol "E"[9] in the table or figure. These estimates should be used with caution to support a conclusion (Mihorean 2005, 14).

Estimates under 16.6% are considered to have moderate sampling variability and can be considered for general unrestricted release (Burns and Williams 2011).

For data that were derived from the Public Use Microdata Files of the 1999, 2004 and 2009 GSS, the coefficient of variation was derived from the Approximate Variance Tables provided in the Public Use Microdata File Userguides for each GSS cycle. The coefficient of variation for data that was provided through Statistics Canada was derived through the SUDAAN program[10]. Because the techniques to determine the coefficient of variation differ depending on the source of the data, there are likely situations in which the CV would differ if another data source was applied. As such, data that might be released using one technique might not be released using the other. For example, the CVs derived from the Approximate Variance Tables are more stringent than those that would be produced using the SUDAAN program and as such, there are likely situations in which data were not released if the CVs were derived from the Approximate Variance Tables but might have been released if the SUDAAN program was used.

Terminology

Findings based on GSS data are reported in Sections 3.1 to 3.6. In these sections, violence is defined as physical and/or sexual violence and is based on a set of 10 questions. Respondents who indicated that they had contact with a previous spouse in the past five years were asked:[11]

During the past five years, did your previous spouse/partner:

  1. Threaten to hit you with his/her fist or anything else that could have hurt you?
  2. Throw anything at you that could have hurt you?
  3. Push, grab or shove you in a way that could have hurt you?
  4. Slap you?
  5. Kick you, bite you, or hit you with his/her fist?
  6. Hit you with something that could have hurt you?
  7. Beat you?
  8. Choke you?
  9. Use or threaten to use a gun or knife on you?
  10. Force you into any unwanted sexual activity by threatening you, holding you down, or hurting you in some way?

Respondents who answered "yes" to any of these questions were deemed to have experienced violence by an ex-spouse.

Those who answered yes to any of the 10 questions above could have experienced violence by an ex-spouse either while living together or after separation (the violence could have occurred while they were living together or after they separated, or it could have also occurred both while they were living together or after separation).

Among these individuals, the violence could have occurred at a number of different points:

  • While Living Together (Note: the respondent could have also experienced violence after separation but only the violence while living together is captured in this variable)
  • After Separation (Note: the respondent could have also experienced violence while living with their ex-spouse but only violence that occurred after separation is captured in this variable)
  • Both while living together and after separation (the violence continued)
  • The violence ceased when they separated from their ex-spouse (only occurred while living together).

Incident-based Uniform Crime Reporting Survey

Section 3.7 highlights the findings regarding criminal harassment based on the 2011 Incident-based Uniform Crime Reporting Survey (UCR2). This information was gathered through a special request made to the Canadian Centre for Justice Statistics at Statistics Canada. "The Incident-based Uniform Crime Reporting (UCR2) Survey collects detailed information on criminal incidents that have come to the attention of, and have been substantiated by Canadian police services&133;In 2011, data from police services covered 99% of the population of Canada" (Sinha 2013, 92).

Homicide Survey

Section 3.8 explores homicides and is based on the Homicide Survey. These data were also gathered through a special request made to the Canadian Centre for Justice Statistics. "The Homicide Survey collects detailed information on all homicides that have come to the attention of, and have been substantiated by, Canadian police services&133;Coverage for the Homicide Survey has represented 100% of the population since recording began in 1961. The count for a particular year represents all homicides reported in that year, regardless of when the death actually occurred[12]" (Sinha 2013, 92).


  • [5] In the territories, data are collected through a supplementary survey by computer-assisted telephone interview and computer-assisted personal interviews.
  • [6] "The Chi-Square Test for Independence is used to determine whether two categorical variables are related. It compares the frequency of cases found in the various categories of one variable across the different categories of another variable" (Pallant 2007, 212).
  • [7] Note that the "Don't Know/Not Stated/Not asked" categories were excluded when running the Chi-Square tests.
  • [8] "The estimates derived from this survey are based on a sample of persons. Somewhat different figures might have been obtained if a complete census had been taken using the same questionnaire, interviewers, supervisors, processing methods, etc. as those actually used. The difference between the estimates obtained from the sample and the results from a complete count taken under similar conditions is called the sampling error of the estimate" (Burns and Williams 2011, 27).
  • [9] Co-efficients of variation accompanied by an "E" reflect high sampling variability and should be used with caution.
  • [10] The SUDAAN program "is a commercial software package [&133;] specifically for analysis of data from complex sample surveys&133;" (Burns and Williams 2011, 482).
  • [11] Hotton 2001: 3.
  • [12] Note that "because some homicides become known to police long after they occur, there are typically some homicides included in a given year's total that actually occurred in previous years" (http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SDDS=3315&Item_Id=1723&lang=en).
Date modified: