HOW STATS CANADA DISTORTED THE PUBLIC PERCEPTION OF
VIOLENCE AGAINST
WOMEN
Reena Sommer, Ph.D.
University of Manitoba
Winnipeg, Manitoba
John Fekete, Ph.D.
Trent University
Peterborough, Ontario
A paper presented at the 4th International Family Violence Research
Conference, July 22, 1995
ABSTRACT
In November, 1993, Statistics Canada released the findings of the
Violence
Against Women Survey (VAWS). Since then, the survey has been hailed by
government ministers as well as the mass media, as the definitive
research
on the topic. Among the many findings, Stats Can reported that 51% of
women
had been abused by a man at some point since the age of 16 years.
This, and
the statement that "measures of violence for the VAWS were restricted
to
Criminal Code definitions of assault and sexual assault" left the
public
with the impression that the average woman living in Canada was at
risk of
ongoing assaults by men of a magnitude similar to that seen at women's
shelters.
This paper addresses flaws in the survey's design, sampling and method
of
reporting which have led to a gross distortion in public's perception
of the
problem of violence against women. By reviewing the information
contained in
the User's Guide for the VAWS, we demonstrate that both what is
contained
in, and also what was omitted from the report contribute to this
distortion.
Among the issues discussed are: (1) the nonrepresentative nature of the
sample, (2) the use of "double barrelled" questions, (3) the
interchangeable
use of % of women and % of marital relationships, (4) the use of
projected
figures to represent findings without indicating what proportion of the
population sample represent, (5) the neglect of the finding of a low
level
of victimization of women in the immediate 12 months prior to the
survey,
(6) the methodological backwardness of a one-sex victimization survey
of the
general population, and (7) the mis- impression created by use of
Criminal
Code definitions as measures of abuse. We conclude that the Stats Can
survey
trivializes the experiences of women who are victims of serious abuse
and
impedes our understanding of the nature of intimate and conflictual
relationships in contemporary society.
INTRODUCTION
As a preface to this discussion, I would like to point out that while
the
issues raised in this discussion originate from John Fekete and myself,
there is consensus among a number of academics whose backgrounds
include
anthropology, physiology, political studies, psychology, sociology and
family studies that Statistics Canada created a picture of violence
against
women that is inconsistent with the experiences of the general
population of
women living in Canadian. This consensus is not only based on empirical
evidence but also on what is considered logical.
It should also be pointed out that Statistics Canada has come under
considerable criticism regarding their survey from a number of
sources. To
date, they have written both John Fekete and myself lengthy letters
whose
purposes were to answer our criticisms. Unfortunately, neither letter
got to
the point of our concerns and instead dealt with other peripheral
matters.
Stats Canada has emphatically denied any suggestion that the design,
sampling and the reporting of the results from their survey was
politically
motivated. We on the other hand reject such denials, and will
demonstrate
through several examples drawn from their own report entitled Family
Violence in Canada and VAWS codebook that Statistics Canada distorted
the
public perception of violence against women. The issues I will deal
with
specifically are:
the methodological backwardness of a one-sex victimization survey of
the
general population
the nonrepresentative nature of the sample
the use of "double barrelled" questions
the interchangeable use of % of women and % of marital partnerships
the use of projected figures to represent findings without indicating
what
proportion of the population sample represent
the neglect of the finding of a low level of victimization of women in
the
immediate 12 months prior to the survey
the misimpression created by use of Criminal Code definitions as
measures of
abuse
Let me begin by briefly giving you some background information about
the
development of the survey. The VAWS followed soon after the Montreal
Massacre in which 14 women were killed by Marc Lepine a man who said he
hated women. The VAWS was conducted by Statistics Canada from February
to
June in 1993. The cost of the project was $1.9 million.
The Objectives of the Survey
Provide reliable estimates of the nature and extent of violence against
women by male partners, acquaintances and strangers
Examine women's fear of violence in order to support current and future
federal government activities
The assumptions made by the investigators are:
Macleod (1980) report based on the experiences of battered women found
that
1/4 are abused
Canadian Panel of Violence Against Women (1993) determined that 98% of
the
women they heard from suffered some form of abuse
Uniform Crime Reporting Survey
Homicide Survey
National Survey on Transition Homes
Study on Dating Violence conducted by DeKeseredy and Kelly (1993)
found that
81% of female students suffered some form of abuse in their dating
relationships
What the investigators ignored however, were:
the Canadian and U.S. general population research showing that men and
women
perpetrate abuse at equivalent rates
Although males make up the bulk of arrests for violent crimes, crimes
statistics showing that the rate of female arrests for violent crimes
from
1983-1993 rose by 130.9% whereas the rate of male arrests for violent
crimes
during the same time period rose 96.2%. These figures came from the
Canadian
Centre for Justice Statistics the same agency responsible for writing
the
Family Violence in Canada Report that includes the findings of the
VAWS
Homicide rates which show that depending on the year, females are
responsible for 10% to 33% of all murders.
Population The target population for the VAWS was all women 18 of
years of
age and over in Canada EXCLUDED:
residents of the Yukon and Northwest Territories
women who spoke languages other than English and French
women who held visas
women who did not have telephones
women with handicaps that interfered with their participation in a
telephone
survey
Of the 22,319 households contacted, 19,309 were eligible households
(86.5%).
Of those, 12,300 women completed the survey (63.7% response rate).
Their
overall response rate not taking into account the women not surveyed
in the
territories was 55.1%. Most important, excluded from this survey were
Aboriginal, Inuit and immigrant women (groups that have been shown to
be
particularly at risk.
Double Barrelled Questions
We are all too aware that questions that contain two distinct concepts
cannot be considered valid. This is because it is impossible to
determine
whether a respondent is responding to one or both concepts contained
within.
Examples of these types of questions are:
Now I'd like to read a list of statements that may apply to your
PREVIOUS
husband(s)/partner(s), and I'd like you to tell me whether each
statement
describes him/any of them.
Over Inclusive Questions Forced Sexual Activity:
"Since the age of 16, has a MALE STRANGER ever forced you or attempted
to
force you into any SEXUAL activity by threatening you, holding you
down or
hurting you in some way" 7.49% (n=921) (weighted factor=741,078)"
Sexual Harassment:
"Sometimes women receive other types of unwanted attention. In this
case I
mean anything that DOES NOT include touching such as catcalls,
whistling.
leering, or blowing kisses. Have you ever received unwanted attention
from a
MALE STRANGER?"
ever - 60% (n=7377) (weighted factor=6,278,447)
past 12 months - 27% (n=3311) (weighted factor=2,860,403)
Partner Abuse:
The percent noted in this column (48%) represents the proportion of
women
who had a previous partner and who experienced abuse in that
relationship.
This constitutes 2216 women or 18% of the sample. What we do see
however is
a decline in the reports of abuse between past and current
relationships
indicating that a large proportion of women are freeing themselves of
abusive relationships. This point however is missing from the report.
Definition of Abuse in Terms of the Criminal Code of Canada:
Defined according to Section 265 which according to a law professor I
consulted with stated these actions have to be without the person's
consent.
Given that general population research finds that 50% of the abuse
reported
is mutual and given the findings that there is considerable
inconsistencies
in husbands' and wives' reports of abuse, we believe that a criminal
designation to these behaviours is premature and inappropriate.
Estimates of Abuse:
Report highlights the prevalence of abuse but neglects the incidence of
abuse
Interchangeable Use of % of women and % of relationships
Language of the report:
Over and above the issues already discussed, we found examples of
written
text which also demonstrate bias in reporting. For example, in the
section
discussing perpetrators of child abuse and neglect, the following was
found:
"While children of either sex were equally likely to be abused by a
female
perpetrator (53% boys and 47% girls), female children were PREYED UPON
BY
MALES (my emphasis) in 70% of the cases." (p.78, Statistics Canada,
1993)".
CONCLUSIONS
The examples presented clearly suggest that there are a number of
problems
inherent in the VAWS and in the Family Violence in Canada report which
documents its findings. While the sampling technique indicates that
not all
Canadian women are represented by virtue of the systematic exclusions
already noted, we do not view this as the most serious problem since
one can
place limitations on generalizeability of findings. I might add at this
point that this is something that is not done the discussion of their
findings. On the other hand, we feel the more serious problem resides
in how
the results have been presented and more importantly, in the data that
have
not been presented.
In correspondence I received from Bruce Petrie, the Assistant
Statistician
from Stats Canada, he states:
"It is our practice to report figures based on the population "at
risk". We
believe it is more relevant to present figures describing the estimated
number of abused women in the population who contacted shelters and
not the
sample counts of either abused women or all women in the sample."
Given this practice, it appears that the survey did not meet its first
objective which was to achieve reliable estimates of partner and
acquaintance abuse. Focusing on the "at risk" population does not
bring us
any closer to understanding the experiences of Canadian women in
general
than previous research conducted on clinical samples of battered
women. In
light of the report's focus on chronically abused women,constituting a
small
proportion of Canadian women, the message delivered by this report and
reiterated by the media is the average Canadian woman is at risk of
ongoing
abuse of a very severe nature.
The data we have presented reflects just a sample of the many ways
Statistics Canada distorted the perception of violence against women.
Due to
time constraints we cannot present them all. An expanded discussion of
the
distorted perception of violence against women will follow in a paper.
Because of the manner in which the results of the VAWS are reported
and the
subsequent omissions discovered, we conclude that Stats Can survey
trivializes the experiences of women who are victims of serious abuse
and
impedes our understanding of the nature of intimate and conflictual
relationships in contemporary society. Portraying Canadian women as
victims
of domestic in the face of data which indicate that only a small
proportion
are affected does little to empower women. We feel that a much better
use of
the data would have presented a clear picture of family life for
Canadian
women which according to their unreported data is for the vast
majority is
violence free. In terms of findings solutions to this very serious
social
problem we suggest that examining the differences between abused and
nonabused women as well as those who are currently abused and others
who no
longer are abused might be particularly useful. We encourage interested
researchers to avail themselves of the data tapes that are now been
released
for public use.
Quelques commentaires sur les questions
posées pour l'enquête de 1993
Questionnaire
Enquête
sur la violence envers les femmes ou EVEF (Violence Against
Women Survey),
Statistics Canada, 1993
Dans cette
enquête,
tous les intervieweurs étaient des femmes avec une formation
«spéciale» pour les
rendre plus sensibles à la violence
subie par les femmes. en d’autres termes, elles avaient
été formées à présumer
que l’interlocutrice avait été une victime, ce qui peut
influencer la façon
dont l’entrevue est conduite (et par ricochet, les réponses des
répondantes,
étant donné les nombreux appel à la collusion que
contient le questionnaire de
l’EVEF).
Question
A1, p. 1
(Collusion entre l’intervieweur et la participante dès la
première question par
l’emploi du «nous», c’est-à-dire «most of
us», «our personal safety» ;
voir John Fekete). La collusion peut créer
un fort sentiment de «désirabilité sociale»
en vertu de laquelle la répondante
fournira des réponses qu’elle croira conforme aux attentes de
l’intervieweur.
Question
B1, p. 5 (La
répondante es confronté à une affirmation de fait
avéré sur le sujet de la
section de l’entrevue; autre possible biais de
désirabilité sociale, car la
répondante pourra être hésitante à fournir
des réponses allant à l’encontre de
l’expérience de la majorité)
Que
signifient
exactement les mots «unwanted», «unconfortable»
et «inappropriate» dans ces
deux questions. Chacune des répondantes peut interpréter
ces termes de manière
différente.
Que signifie
exactement le mot «unconfortable»?
Que signifie
exactement le mot «unnecessarily»?
Les questions B19-B22
définissent Sexual Harassment (Harcèlement sexuel)
Question
C1, p. 10 (le
sujet est la violence des hommes contre les femmes. La
répondante est incitée à
s’identifier à l’ensemble des femmes par l’usage du
«we» et la mention opposant
les hommes et les femmes quant à la violence;
possibilités de biais de
désirabilité sociale). Le biais possible s’applique
à toute la section qui suit
ensuite.
La
mesure des
agressions sexuelles
Question
C2, p. 10.
Question
C7, p. 11
Question
C13, p. 12.
Question
C19, p. 13.
Question
C25, p. 14.
Question
(commentaire)
D27 (à la fin de la section sur les agressions physiques.
L’utilisation du «we»
constitue un appel possible supplémentaire à la collusion
entre la répondante
et l’interervieweur), p. 24.
Question
(Commentaire)
J1, p. 35.
Question
J15, p. 37
Agression sexuelle de la part du conjoint actuel
Question
L1, p. 41
