Data collection and
statistics on domestic violence challenges and lessons learned
Dr Henrica A. F. M. Jansen
(henriette.jansen@gmail.com)
International Conference
Joint Approach to Family Violence:
Legislation, Indicators, Enforcement
May 20-21, 2009, Bishkek
Outline
What are indicators?
 Why collect data and have statistics and
indicators on domestic violence/VAW?
 Data collection: Sources of data
 Understanding what the data tell us: issues
and biases
 Recommendations and lessons learned

What are indicators?
Need to provide a simple summary
of a complex picture (number, proportion,
percentage, trend...)
 Need to present features to support informed
decision making, policy and programs
 Need to be sensitive enough to measure
change (periodicity)
 Need to enable comparisons between groups

Why use indicators?
To know the magnitude of the problem and
to compare between different populations
 To understand underlying causes and to
understand effects
 To monitor State response and changes
over time: laws developed, cases reported,
persons arrested, persons referred for
services, etc.
 To measure impact of services and
interventions (are programs successful?)

Who are interested in indicators?
Programme/Service managers
 People who use services
 Advocates for health service users –
community
 Policy-makers
 International community

Sources for data on
domestic violence

Population based surveys:





National crime victimization surveys
Demographic and reproductive health surveys
Focussed specialized surveys
Short module added to other surveys
Records from police, courts, hospital, etc
Data collected on DV/VAW
in surveys
Almost all surveys give
indicators of prevalence.
 Many also on frequency
 Almost all on perpetrators
 Surveys carried out by many national
statistical offices and other institutions,
often as ad hoc activity
 Attitudes sometimes collected -- Issues
around usefulness

Many women told me that they
never talked about this with anyone,
not even with the neighbors, friends
or relatives, ‘because if I tell her, she
might tell her husband or her
mothers, and word will get around
and might reach my husband, which
would be terrible. If he found out he
would kill me.’
(female interviewer, Nicaragua)
Population-based surveys to
collect data on domestic
violence – Challenges:

Prevalence rates on violence are highly
sensitive to methodological issues

Research on violence raises major issues of
safety and ethics

Unrealistic to expect reduction in prevalence
in short or medium term
Suggestions for measuring
domestic violence in surveys






Define the study population broadly
Use behaviorally specific questions: specific acts
Specify discrete time frames (last year, ever)
Give multiple opportunities to
disclose
Cue respondent to different
contexts and perpetrators
WHO ethical guidelines for
violence research
Population based studies
are very useful for advocacy, policy
development and program design:
◦ Understanding the magnitude and
characteristics of violence
◦ Health burden of violence
◦ Risk and protective factors
But less useful for monitoring and
evaluation programs and services
Using criminal justice statistics for
program evaluation
ECLAC indicator on violence:

Number of women and girls
reporting sexual violence/ 100,000

Number of women and girls
reporting non-fatal injuries due to
domestic violence / 100,000

Source: Bodies that produce police,
judicial and forensic medical
statistics

“A falling value for the indicator
notes improvement”
Problems with this indicator

It is not representative

It is not “interpretable”

It sets the bar too high
Service based/Criminal justice
system indicators
According to
police records in
Nicaragua, 3,000
women reported
domestic
violence in 1995
According to
population based
surveys 150,000
women suffered
domestic violence in
1995
Service based records are not
easy to interpret…

In 1997 more
than 8,000 cases
were reported

Did rates of
violence
increase?

During this period
special police
stations for women
were opened
throughout the
country, and media
campaigns carried
out
More services and better quality of care
More women reporting violence
KNOW WHAT YOUR
DATA TELL YOU
Issues in measurement and
interpretation of the data collected
Representativeness
 Gender bias
 Underreporting

Gender bias
Community based prevalence rates from
surveys bias towards a symmetry in the
rates women and men are perpetrators or
victims of certain forms of domestic
violence
 Service based statistics tend to show that
men commit almost all violence and
overestimate women as victims of
domestic violence

Domestic Violence:
incidents and gender (British Crime Survey)
Women
Men
%
against
women
Ratio:
Women:
men
Victims
657,000
356,000
35%
1.8
Average number
incidents per
victim
20
7
Total incidents
12.9
million
2.5
million
3.9
84%
5.2
Prevalence and incidents
Prevalence use of ‘course of conduct’
might mean that a series of 20 incidents
may count only as one crime, thereby
underestimating the proportion of violent
crime that is dv/gender-based violence
 Prevalence: single events count, thereby
skewing the gender composition towards
image of symmetry

Injuries and severity

Injuries show better the gender
differences than actions
◦ Minor force (e.g. slap): 49% women 36%
men sustain physical injury
◦ Severe force (e.g. choke, weapon): 77%
women 56% men sustain physical injury
(UK data)
Police records
Example of gender bias:

According to
police records,
95% of child
abuse victims in
Nicaragua are
girls

According to
anonymous
population based
surveys, 70% of child
abuse victims are
girls and 30% are
boys
Police records
Example of gender bias:
In cases of couple violence, police
often finds it easier to act against
male perpetrator
 His arrest does not involve children
 Other offences also usually by
males
 Police themselves are often males

Police records:
issues around underreporting
Reporting is very
variable
 Also other violent
crime is underreported
 Women feel
embarrassed, ashamed,
fear for more violence,
economic dependence,
children

Correlation between
degree of underreporting and
 degree of intimacy
 degree of
seriousness
 nature of offence
(more stigma on
sexual violence)
Lessons to take home
No indicator is perfect
 It is important to be aware of
interpretation and potential weaknesses
 The range of indicators to be used changes
according to the purpose and context
 If you use of a group of indicators that reflect
different aspects of VAW than you get a
better idea of the bigger picture

Recommendations
Base indicators on existing information
where possible
 Should be action-oriented (relevant and
useful for programme or case
management)
 Disaggregation of indicators (sex, age,
rural/urban, etc.)
 Include severity and incidents
 Ownership by stakeholders in countries

henriette.jansen@gmail.com