Content uploaded by Alison R Hernandez
All content in this area was uploaded by Alison R Hernandez on Feb 12, 2016
Content may be subject to copyright.
Expert Opinions on Improving Femicide Data
Collection across Europe: A Concept Mapping
*, Isabel Goicolea
, Alison Hernández
, Belen Sanz-Barbero
Aisha K. Gill
, Anna Costanza Baldry
, Monika Schröttle
, Heidi Stoeckl
1Department of Community Nursing, Public Health and Preventive Medicine and History of Science,
Alicante University, Alicante, Spain, 2CIBER de Epidemiología y Salud Pública (CIBERESP), Barcelona,
Spain, 3Public Health Research Group, Alicante University, Alicante, Spain, 4Epidemiology and Global
Health Unit, Department of Public Health and Clinical Medicine, UmeåUniversity, Umeå, Sweden, 5Escuela
Nacional de Sanidad, Instituto de Salud Carlos III, Madrid, Spain, 6Department of Social Sciences,
University of Roehampton, London, United Kingdom, 7Department of Psychology, Second University of
Naples, Naples, Italy, 8Department of Rehabilitation Science, Technical University of Dortmund, Dormund,
Germany, 9Department of Global Health and Development, London School of Hygiene and Tropical
Medicine, London, United Kingdom
Femicide, defined as the killings of females by males because they are females, is becoming
recognized worldwide as an important ongoing manifestation of gender inequality. Despite
its high prevalence or widespread prevalence, only a few countries have specific registries
about this issue. This study aims to assemble expert opinion regarding the strategies which
might feasibly be employed to promote, develop and implement an integrated and differenti-
ated femicide data collection system in Europe at both the national and international levels.
Concept mapping methodology was followed, involving 28 experts from 16 countries in gen-
erating strategies, sorting and rating them with respect to relevance and feasibility. The
experts involved were all members of the EU-Cost-Action on femicide, which is a scientific
network of experts on femicide and violence against women across Europe. As a result, a
conceptual map emerged, consisting of 69 strategies organized in 10 clusters, which fit into
two domains: “Political action”and “Technical steps”. There was consensus among partici-
pants regarding the high relevance of strategies to institutionalize national databases and
raise public awareness through different stakeholders, while strategies to promote media
involvement were identified as the most feasible. Differences in perceived priorities accord-
ing to the level of human development index of the experts’countries were also observed.
“Femicide”is becoming recognized worldwide as the ultimate manifestation of violence against
women and girls . Diana Russell proposed the term for the first time at the International
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 1/14
Citation: Vives-Cases C, Goicolea I, Hernández A,
Sanz-Barbero B, Gill AK, Baldry AC, et al. (2016)
Expert Opinions on Improving Femicide Data
Collection across Europe: A Concept Mapping Study.
PLoS ONE 11(2): e0148364. doi:10.1371/journal.
Editor: Elizabeth W Triche, St Francis Hospital,
Received: May 5, 2015
Accepted: January 19, 2016
Published: February 9, 2016
Copyright: © 2016 Vives-Cases et al. This is an
open access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
Data Availability Statement: All relevant data are
within the paper and its Supporting Information files.
Funding: This study received funding from Umeå
Center for Global Health Research, funded by
FORTE, the Swedish Council for Working Life and
Social Research (grant no. 2006-1512). With this
funding it was possible to do the data collection and
analyses with the appropriate software.The authors
didn’t receive any other funding for the study design
and preparation of the manuscript.
Tribunal on Crimes Against Women in 1976 in order to name the intentional “killings of
females by males because they are females”, and it is now broadly used also at the UN level
. In relation with this definition of Femicide, different forms of women’killing are recog-
nized such as intimate partner-related killings , so-called “Honour” and Dowry-related
murders, forced suicide , female infanticide  and gender-based sex-selective foeticide ,
genital mutilation-related death cases , targeted killing of women at war  and in the con-
text of organized crime, among others . A broader definition of this term also refers to the
responsibility of States in women’s death-cases perpetuated by misogynous attitudes and/or
social discriminatory practices against women. In this broader definition are included, for
example, deaths associated to lack of accessibility to healthcare for women and girls, gender-
based selective malnutrition and trafficking of women for prostitution or drugs . As it is
difficult to decide in all cases of killings of women and girls if they had been killed because of
their gender, researchers often include all killings of women in the first step and then differenti-
ate between forms that are more or less relevant with regard to gendered backgrounds and
motives. This is also a reason why consensual research on prevalence and nature and character-
istics of victims and offenders and their relationship is needed.
Approximately 66,000 women every year from 2004 to 2009 were victims of killings, repre-
senting almost one-fifth of total homicide victims (396,000 deaths) in an average year world-
wide . It is also known that 38–70% of female homicides are perpetrated by their current or
former intimate partners, whereas 4–8% of male homicide are by their intimate partner [4,14].
Despite its high prevalence or widespread prevalence, they are based on incomplete data collec-
tion systems in roughly half of the countries in the world [4,12]. There is an evident need to
identify a systematic method to gather data on the incidence of femicide that will allow com-
parisons across countries. Without reliable information, policy makers and programmers at all
levels (national, regional or local) are unable to allocate resources so as to achieve the greatest
impact in preventing these killings and reducing the harm they do to the victims and their rela-
tives. Policy makers and programmers need information specific to their areas of concern .
In Europe, in 2011, the female death rate due to assault had great geographic variability, rang-
ing from 0.03 deaths/100.000 inhabitants in Latvia to 2.85 deaths/100.000 inhabitants in Lithua-
nia . The heterogeneity in the surveillance systems makes it difficult to estimate the
implications of these differences. Despite the fact that many countries have sex-disaggregated
data on homicides, and a few (e.g. Spain and France)collect intimate partner-related homicide
data in particular, the availability of specific data on all forms of gender-based homicide against
women is far less developed . Some of the challenges to accurate femicide surveillance across
Europe include the misunderstanding of the gender basis of crimes and the limited available
information in the existing homicide data registers about the relationship between victims and
perpetrators, factors surrounding crimes, the motives of the perpetrators and the modus ope-
randi . At the time this study was conducted, only few national monitoring systems offered
an example of how to overcome some of these challenges. The Finnish monitoring system, for
example, is a registry of preliminary police investigations on intentional murders, manslaughters,
killings, infanticides and negligent homicides that includes compulsory information related to
nearly 90 variables regarding victim’sandoffender’s characteristics, surrounding circumstances
of the crimes, perpetrators’behavior after the killing and spatial and temporal distribution .
Femicide is both a sociopolitical as well as a public health problem which has damaging
effects on the lives of all women and their social environment, and holds negative implications
for the whole society . In order for Governments to begin to take action, there must be a
common ground for understanding what femicide is, and the social costs, not only the individ-
ual ones, associated with it. If Governments were to understand this, not only will it help save
lives, but it will also reduce the annual costs related to justice, social, welfare and social system
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 2/14
Competing Interests: The authors have declared
that no competing interests exist.
. Implementation of national surveillance systems on femicide and harmonization of these
within Europe can facilitate a deeper understanding of this social and public health problem,
and provide evidence for policy development, monitoring and prevention.
In 2013, the first professional and research network with focus on the issue of femicide was
created within the platform of COST (European Cooperation in Science and Technology).
The COST Action entitled “Femicide Action IS1206”brings together top-level experts on femi-
cide and other forms of violence against women from 27 countries . This is the first pan-Euro-
pean coalition on femicide involving researchers already studying the phenomenon nationally,
with the purpose of advancing research clarity, agreeing on definitions, improving the efficacy of
policies for femicide prevention, and publishing guidelines for the use of national policy-makers.
The COST Action is organized in four working groups (WGs), and one of them (WG2) focuses
on ‘Reporting’. The WG2 members organize annual meetings where data collection systems
across Europe are analyzed, and data on femicide as well as on other related topics are compared
and discussed. Discussions in the initial meeting of WG2 in early 2014 confirmed that most Euro-
pean countries collect data on homicides disaggregated by sex, which provides an overall picture
of the magnitude of homicide against women across Europe. Furthermore, several countries col-
lect data on the relationship between perpetrators and victims of homicide, and some of them col-
lect data specifically on intimate partner homicide. However, it became clear that the collection of
differentiated data on femicide was underdeveloped in most European countries.
The aim of this study on the COST Action on Femicide was to assemble expert opinions
regarding strategies that might feasibly be employed to promote, develop and implement an
integrated and differentiated femicide data collection system in Europe at both the national
and international levels. The intended outcome of this process was to identify actions which
are needed to improve the availability, collection and better comparability of data on femicide,
taking into account different perceived needs across countries.
Material and Methods
Concept mapping, an integrated mixed methods approach, was used to examine the diverse
views of European experts on strategic actions needed to improve and systematize femicide
data collection systems in Europe. The integration of qualitative and quantitative data occurs
through sequential steps, beginning with generation of ideas (brainstorming), structuring of
ideas through sorting and rating, representation of structuring in maps based in multivariate
statistical methods, and finally collective interpretation of the maps . This methodology
was selected to meet the study objective based on its demonstrated usefulness in integrating the
input of broad expert panels to guide development and planning, and its capacity to enable
groups of actors to visualize their ideas around an issue of mutual interest and develop com-
mon frameworks [25,26]. The participatory, structured nature of the concept mapping process
was well-suited to the complexity of the task of integrating the views of femicide, Violence
against Women (VAW), and data registration system researchers from different disciplines
and European countries to develop policy recommendations for strategic action in the region.
The fact that concept mapping combines qualitative input with multivariate analysis to pro-
duce a visual display of how a group views a particular topic was also considered in the selec-
tion of this method. Unlike purely qualitative techniques, concept mapping provides a
structured approach for allowing participants to co-produce the content in focus in the study
and interpret visual representations of their group perceptions .
The study was done with the approval of the members of the coordination board of the COST
Action and written informed consent was asked of the participants. In addition, ethical approval
for this study was granted by the Ethical Committee of the University of Alicante (Spain).
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 3/14
Concept mapping activities were carried out from December 2014 until February 2015 in
three phases: 1) brainstorming, 2) sorting and rating, and 3) representation in maps and
Based on discussions with the coordinators of the working group on ‘Reporting’, the research-
ers developed the following focus question to orient the brainstorming activity: In what aspects
shall we improve our own country’s data collection systems to collect accurate data on femicide?
This question was sent via email, together with clear instructions of the entire concept map-
ping process to all of the 70 members of the whole COST Action. Participants were asked to
write down as many strategies as possible in response to the question, with each strategy con-
taining only one idea. The brainstorming phase was carried out from the 1
of December, 2014
until the 13th of January, 2015. Twenty five members from 16 countries provided strategies,
which the researchers checked to eliminate duplicates, and to divide complex strategies into
simpler ones(Table 1). We were not able to recruit participants from Bosnia and Herzegovina,
Denmark, Estonia, Finland, France, Iceland, Latvia, or Sweden.
The researchers sent the refined list of 69 strategies to all participants to give them the
opportunity to review and determine if their ideas were accurately reflected. In the next step,
the strategies were entered in the Concept System software , which was used to facilitate
the following steps of the process.
Sorting and rating
The sorting and rating phase was accomplished in January 2015at the annual working group
meeting in Rome. During the meeting, the first two authors presented the final list of strategies
Table 1. Participants in the brainstorming, sorting and rating by country of their institutions.
Brainstorming Sorting and rating
Spain 4 4
Israel 2 2
Italy 1 3
Belgium 0 1
Germany 1 2
Croatia 0 1
Lithuania 2 1
Macedonia 2 1
Malta 1 2
Poland 1 1
Portugal 1 1
Romania 1 6
Slovenia 2 1
UK 1 2
Austria 2 0
Greece 1 0
Netherlands 1 0
Cyprus 2 0
Total 25 28
Countries represented among participants in the different phases of the concept mapping study.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 4/14
to the whole working group and explained the sorting and rating activities. Sorting activities
consisted of experts organizing the strategies into meaningful groups, or thematic clusters, and
giving them a title. Rating activities consisted of giving each strategy two ratings, for 1) its rele-
vance to the goal of strengthening data collection systems and 2) its feasibility of being imple-
mented . Each strategy should be given a value from 1 to 6 for relevance and feasibility,
where 1 meant very low relevance or feasibility and 6 meant very high relevance or feasibility.
Experts who did not finish the sorting and rating in Rome or who were not present were able
to complete the sorting and rating online. Of the 45 experts invited to participate in the sorting
and rating, 28 from 14 countries, provided answers (Table 1).
Representation in maps and interpretation
In the representation and interpretation step, the gathered data was analyzed using concept
mapping techniques that facilitate visualization of thematic clusters, and identification of areas
of consensus for action. The sorting data was analyzed using multi-dimensional scaling to gen-
erate point maps, where strategies are plotted in a two-dimensional graph based on a similarity
matrix, which captures the number of times experts grouped them together. Strategies that
were more frequently sorted together are positioned closer to each other. The degree of fit
between the point map and the data in the similarity matrix is reflected by the stress index,
where a lower value indicates a better fit. The stress index of the final point map, which was
used as the base for identifying clusters, was 0.207. This score was in line with the results of
other concept mapping studies, as reflected by the results of a meta-analysis of concept map-
ping projects, which estimated an average stress value of 0.285 with a standard deviation of
Hierarchical cluster analysis was used to show options for aggregating strategies into clus-
ters based on their proximity to each other in the point maps. Cluster maps ranging from seven
to twelve clusters were evaluated through discussion among the researchers by reviewing the
strategies grouped together at successive levels of clustering based on their conceptual coher-
ence and the value of the precision offered at each level. For example, in moving from the level
of nine to ten clusters, the clusters “Public Awareness”and “Media coverage”were separated,
and the researchers decided this division was valuable in distinguishing the strategies. Ten clus-
ters were identified as the final solution, and names were assigned by the researchers through
consideration of the content of the clusters and the group names suggested by experts. This
level of division was similar to the average number of thematic clusters identified in the sorting
activity, where participants created an average of nine clusters (average = 8.8 clusters, standard
deviation = 4.2, minimum = 4, maximum = 27).
Rating data was analyzed to identify the priorities for action based on the views of the group
as a whole, as well as dynamics in the perception of priorities across regional sub-groups using
role-stratified averages. We have a variety of countries in the sample, and we wanted to explore
whether the level of general development of the country could influence the perceived rele-
vance of the strategies proposed. The rationale behind this assumption was that in countries
with better public systems, registers in general and also those related to femicide would be bet-
ter and so the priorities might differ from those of countries with less developed registers and
public institutions. We choose the Human Development Index(HDI) as a proxy for this. The
HDI measures the average achievements in a country in three dimensions of human develop-
ment: health dimension, measured by life expectancy at birth; education component, measured
by adult literacy rate and the combined primary, secondary and tertiary gross enrollment ratio;
the standard of living, measured by Gross Domestic Product per capita in purchasing parity
. The scores for the three HDI dimensions are then aggregated into a composite index
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 5/14
using geometric mean. This index score was used to classify countries into two groups due to
its association with different mortality causes [31–33], diseases distribution [34,35] and health
behavior [36,37]. Seven countries with higher HDI (Germany, Israel, Belgium, Slovenia, Spain,
Italy and the UK) were included in one group and seven countries with lower HDI (Malta,
Poland, Lithuania, Portugal, Croatia, Romania and Macedonia) were included in the other
After creating the maps, initial results of these analyses were discussed during the second
part of the working group meeting in Rome to facilitate their interpretation. During the meet-
ing, the participants worked in small groups to evaluate the appropriateness of the clusters gen-
erated by the analysis and were asked to determine if the strategies within each cluster were
coherent, if clusters could be joined or divided and if the titles were appropriate. Experts also
reflected on their own experiences to improve data collection systems and discussed connec-
tions among the clusters of actions. Notes from these discussions, as well as on-going dialogue
among the core research team, provided the base for finalizing the cluster map and identifying
domains of thematically related clusters.
Actions to promote improved data collection systems
Of the 69 strategies that were generated by the participants, the final clustering solution identi-
fied 10thematic areas of action: “Putting femicide on the public agenda”,“Media coverage”,
“Awareness raising on the importance of data collection”,“Definition”,“Quality of data collec-
tors”,“Institutionalizing national databases”,“Data collection structure”,“Variables to be col-
lected”,“Triangulation”, and “Qualitative follow-up”(Fig 1).
Based on relationships among these clusters, the cluster map was divided into two domains:
“Political awareness”and “Technical steps”. The six clusters in the domain “Technical steps”
include straightforward actions that would directly improve the content, structure, quality and
continuity of national data collection systems, as well as measures to improve detection and fol-
low-up investigation. The four clusters in the domain “Political awareness”include actions
directed towards enhancing public awareness of the existence and scale of femicide, as well sen-
sitizing key groups to their role in data collection on femicide, and clarifying a common under-
standing of what constitutes femicide. Table 2 presents an overview of the content of the
clusters and selected strategies (for a complete list of all strategies see S1 File).
The varying size of the clusters depicted in the map reflects the tightness of the conceptual
coherence of the strategies the clusters contain, while the proximity of clusters reflects per-
ceived relationship between the strategies they contain. The relatively larger size of the clusters
“Putting femicide on the public agenda”and “Raising awareness on the importance of data col-
lection”reflects the diversity in the nature of actions needed to generate public consciousness
and shift political will. While the concentration of many strategies in the smaller and closely
proximal clusters, “Variables to be collected”,“Quality of data collection structure”, and “Insti-
tutionalizing national databases”, reflects that the technical steps required to strengthen data
collection systems are numerous and closely interrelated.
Identifying priority actions
Analysis of the overall average ratings of the items that make up the clusters indicated that the
cluster “Institutionalizing national databases”was the most relevant (5.28), followed by “Vari-
ables to be collected”(5.20) and “Quality of data collectors”(5.10). The clusters “Qualitative
follow up”(4.56) and “Data collection structure”(4.66) were rated with the lowest relevance. It
is also noteworthy that items in the cluster “Definition”received a relatively low overall rating
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 6/14
(4.69). The ratings on feasibility were overall lower than those of relevance. The clusters with
the highest rating on feasibility included “Media coverage”(4.44), “Awareness raising on data
collection”(4.40) and “Putting Femicide on the Public Agenda”(4.29), while the lowest rated
were “Quality of data collectors”(3.74), “Data collection structure”(3.90), and “Definition”
(4.04) (Table 2 and S1 File).
Strategies with the highest ratings for both relevance and feasibility (items 51, 55, 32, 33, 28)
referred to ensuring that specific types of information are collected in a standardized and insti-
tutionalized way that provides a base for identifying cases and monitoring femicide cases at a
regional level. The strategies with the lowest ratings for relevance and feasibility referred to
strategies to conduct in-depth analysis of suspected cases of femicide (items 49, 64, 69, 63, 48)
Analyzing differences in priorities across countries
The comparison of average rating of the relevance of action clusters showed differences
between the priorities of countries with lower and higher HDI ranking. Countries with lower
HDI rated clusters in the domain of “Political action”most highly, whilst the countries with
higher HDI gave highest rating to clusters in the domain of “Technical steps”. The greatest dif-
ference in the perceived relevance across groups was found in the clusters “Putting femicide on
the public agenda”,“Awareness raising on data collection”, and “Media coverage”, and t-tests
showed that these difference were significant with t-values, degrees of freedom and levels of sig-
nificance of(-3.01, 14, p<0.01),(-3.59, 14, p<0.05), (-5.18, 8, p<0.001), respectively. Both
groups agreed on the lower relevance of clusters on “Qualitative follow up”,“Triangulation”,
Fig 1. Clusters of actions to promote improvement of data collection systems on femicide across Europe. Cluster map based on experts’thematic
grouping of action strategies.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 7/14
“Data collection structure”and “Definition”. Both groups also agreed on the very high rele-
vance of “Institutionalizing national databases”. Actions belonging to the “Media coverage”
cluster were rated with the lowest relevance by experts from countries with higher HDI, and
among the most relevant by those from countries with lower HDI. Despite these variations in
perceived relevance of actions, the sub-groups’assessments of feasibility were very similar (Fig
Table 2. Overview of the content of the clusters, selected strategies and average ratings of clusters items.
CLUSTERDESCRIPTION SELECTED STRATEGIES RELEVANCE FEASIBILITY
PHASE 1. POLITICAL AWARENESS
Putting femicide on the public agenda—Strategies to enhance
political will in order to make femicide a public priority.
- Ensuring political will and commitment (1). Putting the
concept of femicide into the academic, social, political
and legal agenda (4).
Media coverage- Strategies to improve the quantity and quality
of media coverage of the problem of femicide.
- Publicizing the information on femicide through
accessible communication venues (9). Training
journalists on how to report these cases properly (10).
Awareness raising on data collection—Strategies to raise
awareness of the importance of collecting accurate data on
femicide among data collectors and stakeholders.
- Alerting the public institutions, ministries and other
state authorities to the need to identify, register and
analyze the characteristics of femicide as a speciﬁc
crime (14). Increasing awareness among data
collection personnel (15).
Deﬁnition—Strategies to reach consensus on a deﬁnition of
femicide that captures the complexity of the phenomenon.
Establishing a clear deﬁnition of femicide across
PHASE 2. TECHNICAL STEPS
Quality of data collectors—Strategies to improve the quality of
the data collection systems on femicide, with special focus on
adequate sensitization and training of professionals involved in
the collection and reporting of data.
Training those in charge of collecting those data on the
importance of gathering correct information on all
relevant aspects (28).
Institutionalization of national data base—Strategies to ensure
that countries have a publicly funded, centralized and
sustainable data collection system on femicide.
- Establishing a database, publicly funded and
sustained, to collect information on all forms of violence
against women including femicide (32). Developing a
centralized system that gathers data from all relevant
Data collection structure—Strategies to ensure the quality of
data collection systems of femicide, with focus on structural
and organizational aspects.
- Standardizing data collection systems across police
and court data collection system (42). Deciding on what
information to collect based on the state of the art of
the issue (50).
Variables to be collected—Suggestions for speciﬁc,
standardized information to be gathered for every case of a
- Collecting basic socioeconomic data on victims and
offenders, including their age, education level,
employment status and/or occupational class, place of
birth, and area of residence (11). Ensuring that all types
of data collection systems (crime, court, etc.) collect at
least the following information: sex of both victim and
perpetrator, type of relationship between them, prior
history of domestic violence, previous institutional
Triangulation—Strategies to enhance triangulation across data
collection systems, both at the national level, to enhance case
detection, and at the regional level, to enhance comparability.
- Identifying a minimum set of variables covered that
allow us to know the situation in Europe and make
comparisons between countries (55). Triangulating
monitoring systems data with newspaper articles,
police and court statistics (59).
Qualitative follow up—Strategies to collect in depth information
on every suspicious case in order to diminish underreporting
and better understand the phenomenon.
- Developing qualitative research on motives, context
and background of the cases in order to ﬁnd out, if and
how these crimes could be prevented (65).
Description of the clusters depicted in Fig 1. Examples of strategies and the average rating of strategies within the cluster. Corresponding numbers of
example strategies are indicated in parenthesis.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 8/14
The strategies generated through the brainstorming step provided a broad range of recommen-
dations that may be applied within and across different countries in order to improve the avail-
ability, collection and monitoring of femicide data. These include political will, technical
specific requirements and the involvement of different actors—governments, mass media,
police bodies, courts and professionals, who are in charge of identifying, registering and moni-
toring. Priority clusters of actions were also identified within this range of strategies, and
according to experts’assessment, “Institutionalizing national databases”was found to be most
relevant, while “Media coverage”was rated most feasible. Variation in ratings across countries
indicated differing perceptions of priorities based on their current situation.
As has been observed in relation to other health information system issues [38,39], the
experts’responses reflected that promoting and implementing concrete changes requires not
only technical steps, but also socio-political processes. Strategies related to the latter emerged
despite the explicit emphases on data collection on femicide in the initial focus question. The
connection between technical change and socio-political processes is also reflected in the inter-
relationship among actions related to institutionalizing national databases, defining the vari-
ables to be collected and the quality of data collectors. In order for registers to function
continuously, they have to be institutionalized , which depends on political will backed by
adequate state funding. When registers are institutionalized, the definition of indicators is
enabled and priority is given to certain indicators  that will enable differentiation of several
forms of femicides from women´s homicides . The definition of variables and indicators,
as it has been shown with the implementation of health information systems  or in the
harmonization of European registers , is strongly connected to the ability to collect valid
Table 3. Summary of the most and the least relevant and feasible strategies to build femicide data col-
lection systems according to participants’opinions.
The 5 most relevant and feasible strategies The 5 least relevant and feasible strategies
51. Ensuring that all type of data collection systems
(crime, court, etc.) gather at least the following
information: sex of victim and perpetrator, type of
relationship between them, prior history of
domestic violence and previous institutional
49. Ensuring that cases where the court does not
have enough evidence to convict the offender
for a crime likely to be femicide are included in
the monitoring systems as suspicious cases of
55. Identifying a minimum set of variables covered at
least in the European context that allow us to
know the situation in Europe and make
comparisons between countries.
64. Reviewing past cases of women murdered to
identify if they are femicides or not.
32. Establishing a database, publicly funded and
sustained, to collect information on all forms of
violence against women including femicide.
69. Interviewing perpetrators, relatives, friends,
neighbors and acquaintances.
33. Ensuring that national data on femicide are
collected following international recommendations
and comparable with data collected in other
63. Tracking cases in which the perpetrator
commits suicide after committing the intimate
28. Training those in charge of collecting those data
on the importance of gathering correct information on
all relevant aspects.
48. Upgrading national records about the deaths
and causes of death with the information about
murder as a cause of death (Ministry of Health) and
using this source as a possible detector of those
murders that are committed before the perpetrator
commits a suicide.
Results of strategies rating based on experts’assessment of relevance and feasibility.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 9/14
and reliable data describing both the extent of its occurrence, its context and background of
risk factors to establish action lines for its prevention.
Feasibility ratings were lower overall than those of relevance. Experts in this study may have
clear ideas about actions needed to ensure improved data collection on femicide (relevance),
but they also have experience in the difficulties of achieving these changes (feasibility). Chal-
lenges include existing structures of policies and data collection systems, where those responsi-
ble may not see the importance of collecting this kind of data, or lack the necessary training,
budget or statistical data collection systems to do so. The difficulty of getting different agencies
to cooperate with each other to produce this data is another important challenge that is
reflected in the low feasibility scores given to “Definition”. Despite the relatively high perceived
relevance of actions to reach consensus on a definition of femicide, experts are aware that it
will be very difficult to harmonize definitions and also to have enough information for each
case of female homicide to allow the data collectors to define it as femicide.
In countries with lower HDI ranking, “Political action”was considered the most relevant
and necessary first step. In countries with higher HDI ranking, the “Technical steps”were
more relevant. This result could perhaps be explained by the observation that in countries
with lower HDI, topics of violence against women or femicide are not prioritized in policies or
public discourse. In Portugal, for example, as well as in most East-European countries,
women civil society groups addressing VAW are more recently formed than those located in,
for example, UK where women’s groups and public awareness about this problem started in
the early 1970s . Based on this pattern, it is expected that experts from the former group
of countries would perceive the relevance of actions to heighten public awareness and
strengthen the political will for data collection activities more strongly than experts from the
Fig 2. Comparison of average cluster relevance ratings by countries with higher and lower HDI. The clusters with significant difference in the domain
of “Political awareness”are highlighted in bold.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 10 / 14
The results of this study should be interpreted taking into account several study limitations.
The final samples of experts in both, brainstorming and sorting/rating steps do not fully repre-
sent all countries of the European region. Despite the fact that we asked 70 professionals from
all countries involved in the COST Action “Femicide across Europe”, only a part of them
accepted this invitation. We probably ended up engaging those participants that were more
interested in the topic. Unfortunately, it was not possible to recruit a representative from Fin-
land or France, where examples of femicide-related intimate partner violence and domestic
violence data registries have been further developed [17,19]. It would be important to gain
their perspectives in future research. We conducted this concept mapping study with the mem-
bers of the COST Action due to their professional experience in the topic of femicide, but this
expertise not always was focused in data collection systems. However, this profile may be also
considered as strength due to their understanding of issues surrounding the development of
femicide data collection systems (such as those related to political action) as well as those
related to technical aspects. The ecological perspective of this study limits the transferability of
our results to the specific situation of each country. Future research should be applied within
the specific contexts of each country.
Among the strengths, the use of the concept mapping method must be highlighted, as it
allowed us to structure and rate relevant aspects of a complex topic in a very short time and to
integrate various expert opinions from different countries. It thus contributed directly to clear
and manageable scientific results that may provide an important basis for improved data col-
lection systems. The timing of the study is also strength. Until the establishment of the COST
Action IS1206 “Femicide across Europe”in mid- 2013, European agencies had never recog-
nized the lethal act of femicide as a separate topic, although they had funded initiatives on gen-
der and violence where femicide was a rather side-topic. Nowadays with COST Action IS1206
operating for the past two years, the phenomenon of femicide in Europe is entering the public
agenda, and intermeshing with that of global institutions such as ACUNS(Academic Council
on the United Nations System) and EIGE(European Institute for Gender Equality).
In conclusion, the results of this study provide a concrete plan of the next (political and
technical) steps to be taken in order to improve data collection and monitoring on femicide in
and across European countries: Institutionalizing a national database on femicide, agreeing on
a minimum set of variables that have to be collected in each case, and investing in the training
of those professionals who are in charge of collecting the data. Furthermore, expert assessments
revealed that implementing and sustaining femicide data collection systems entails not only
technical data collection, but also a firm political commitment. Once in place, the evidence pro-
duced can contribute to increased public awareness and demand for a public health sector
response, as done with IPV, as well as providing concrete information on risk factors and risk
groups to guide police, legal, educational, and political forces in development of prevention
strategies and services.
S1 File. Complete list of strategies with individual ratings, organized by cluster.
To the members of the COST Action “Femicide across Europe”that participate in this study.
This study wouldn’t have been possible without their generous contributions.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 11 / 14
Conceived and designed the experiments: CVC IG. Performed the experiments: CVC IG AH
BS. Analyzed the data: CVC IG AH BS. Contributed reagents/materials/analysis tools: CVC IG
AH BS. Wrote the paper: CVC IG AH BS AG AB MS HS. Made substantial contributions to
the conception or design of the work; or the acquisition, analysis, or interpretation of data for
the work: AND. Drafted the work or revised it critically for important intellectual content:
AND. Final approval of the version to be published: AND. Agreement to be accountable for all
aspects of the work in ensuring that questions related to the accuracy or integrity of any part of
the work are appropriately investigated and resolved: CVC IG AH BS AG AB MS HS.
1. Laurent C, Platzer M, Idomir M. Femicide: A Global Issue that Demands Action: Academic Council on
the United Nations System (ACUNS) Vienna Liaison Office; 2013. Available: http://www.
2. Russell DE. Defining femicide and related concepts. Femicide in global perspective. 2001:12–28.
3. Femicide volumes by UN/ACUN. New 2nd Edition of Femicide: A Global Issue that Demands Action
Vienna Liaison Office: Academic Council on the United Nations System; 2014. Available: http://acuns.
4. Stockl H, Devries K, Rotstein A, Abrahams N, Campbell J, Watts C, et al. The global prevalence of inti-
mate partner homicide: a systematic review. Lancet. 2013; 382(9895):859–65. doi: 10.1016/S0140-
6736(13)61030-2 PMID: 23791474.
5. Nasrullah M, Haqqi S, Cummings KJ. The epidemiological patterns of honour killing of women in Paki-
stan. Eur J Public Health. 2009; 19(2):193–7. doi: 10.1093/eurpub/ckp021 PMID: 19286837.
6. Rastogi M, Therly P. Dowry and its link to violence against women in India: feminist psychological per-
spectives. Trauma Violence Abuse. 2006; 7(1):66–77. doi: 10.1177/1524838005283927 PMID:
7. Porter T, Gavin H. Infanticide and neonaticide: a review of 40 years of research literature on incidence
and causes. Trauma Violence Abuse. 2010; 11(3):99–112. doi: 10.1177/1524838010371950 PMID:
8. George SM. Millions of missing girls: from fetal sexing to high technology sex selection in India. Prenat
Diagn. 2006; 26(7):604–9. doi: 10.1002/pd.1475 PMID: 16856224.
9. Banks E, Meirik O, Farley T, Akande O, Bathija H, Ali M, et al. Female genital mutilation andobstetric
outcome: WHO collaborative prospective study in six African countries. Lancet. 2006; 367
(9525):1835–41. doi: 10.1016/S0140-6736(06)68805-3 PMID: 16753486.
10. Kubai A, Ahlberg BM. Making and unmaking ethnicities in the Rwandan context: implication for gender-
based violence, health, and wellbeing of women. Ethn Health. 2013; 18(5):469–82. doi: 10.1080/
13557858.2013.832012 PMID: 23998330.
11. World Health Organization and Pan American Health Organization. Understanding and addressing vio-
lence against women. 2012.
12. Radford J, Russell DE. Femicide: The politics of woman killing: Twayne Pub; 1992.
13. Secretariat GD. Global burden of armed violence 2011. Cambridge Books. 2011.
14. Krug EG, Mercy JA, Dahlberg LL, Zwi AB. The world report on violence and health. Lancet. 2002; 360
(9339):1083–8. doi: 10.1016/S0140-6736(02)11133-0 PMID: 12384003.
15. Holder Y, World Health Organization. Injury surveillance guidelines: World Health Organization
Geneva; 2001. 80 p.
16. Eurostat. Death due to homicide, assault, by sex—2011: Eurostat. Available: http://epp.eurostat.ec.
europa.eu/tgm/graph.do?tab = graph&plugin=1&language = en&pcode = tps00146&toolbox = type.
17. Corradi C, Stöckl H. Intimate partner homicide in 10 European countries: Statistical data and policy
development in a cross-national perspective. European Journal of Criminology. 2014; 11(5):601–18.
18. Heidi Stöckl CV-C, Belén Sanz-Barbero, Ecaterina Balica, Anna Costanza Baldry and Monika Schröt-
tle. Issues in measuring and comparing the incidence of intimate partner homicide within and across
European countries (Forthcoming).
19. Manjoo MR. Femicide and feminicide in Europe. Gender-motivated killings of women as a result of inti-
mate partner violence. New York: Alliance of NGOs on Crime Prevention & Criminal Justice, 2011.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 12 / 14
20. Dobash RP, Dobash RE. Who died? The murder of collaterals related to intimate partner conflict. Vio-
lence Against Women. 2012; 18(6):662–71. doi: 10.1177/1077801212453984 PMID: 22831847.
21. Walby S. The Cost of Domestic Violence: Up-date 2009: Lancaster University, UK; 2009. Project of the
UNESCO Chair in Gender Research, Lancaster University]. Available: http://www.lancaster.ac.uk/fass/
22. Cost European Cooperation in Science and Technology. About Cost: Cost. Available: http://www.cost.
23. Cost European Cooperation in Science and Technology. Femicide Across Europe. Cost Action IS-
1206: Cost. Available: http://www.femicide.net/.
24. Kane M, Trochim WM. Concept mapping for planning and evaluation: Sage Publications Thousand
Oaks, CA; 2007.
25. Waltz TJ, Powell BJ, Matthieu MM, Chinman MJ, Smith JL, Proctor EK, et al. Innovative methods for
using expert panels in identifying implementation strategies and obtaining recommendations for their
use. Implementation Science. 2015; 10(Suppl 1):A44. doi: 10.1186/1748-5908-10-S1-A44 PubMed
Central PMCID: PMCPMC4551718.
26. Trochim WM, Milstein B, Wood BJ, Jackson S, Pressler V. Setting objectives for community and sys-
tems change: an application of concept mapping for planning a statewide health improvement initiative.
Health promotion practice. 2004; 5(1):8–19; discussion 0. doi: 10.1177/1524839903258020 PMID:
27. Burke JG, O'Campo P, Peak GL, Gielen AC, McDonnell KA, Trochim WM. An introduction to concept
mapping as a participatory public health research method. Qualitative health research. 2005; 15
(10):1392–410. doi: 10.1177/1049732305278876 PMID: 16263919.
28. Concept Systems Incorporated. Concept Systems Software Ithaca, NY: Concept Systems, Incorpo-
rated; 2013. Available: http://www.conceptsystems.com/.
29. Witkin BR, Trochim WW. Toward a synthesis of listening constructs: A concept map analysis. Interna-
tional Journal of Listening. 1997; 11(1):69–87. doi: 10.1207/s1932586xijl1101_5
30. United Nations Development Programme. Human Development Reports. Human Development Index
(HDI): United Nations. Available: http://hdr.undp.org/en/content/human-development-index-hdi.
31. Hu Q-D, Zhang Q, Chen W, Bai X-L, Liang T-B. Human development index is associated with mortality-
to-incidence ratios of gastrointestinal cancers. World journal of gastroenterology: WJG. 2013; 19
(32):5261. doi: 10.3748/wjg.v19.i32.5261 PMID: 23983428
32. Rodriguez-Morales AJ, Castaneda-Hernandez DM. Relationships between morbidity and mortality
from tuberculosis and the human development index (HDI) in Venezuela, 1998–2008. Int J Infect Dis.
2012; 16(9):e704–5. doi: 10.1016/j.ijid.2012.04.011 PMID: 22721701.
33. Shah A. A replication of the relationship between elderly suicide rates and the human development
index in a cross-national study. Int Psychogeriatr. 2010; 22(5):727–32. doi: 10.1017/
S1041610210001110 PMID: 20497623.
34. Rodrigues-Junior AL, Ruffino-Netto A, de Castilho EA. Spatial distribution of the human development
index, HIV infection and AIDS-Tuberculosis comorbidity: Brazil, 1982–2007. Rev Bras Epidemiol.
2014; 17 Suppl 2:204–15. PMID: 25409649.
35. Roy A, Roe MT, Neely ML, Cyr DD, Zamoryakhin D, Fox KA, et al. Impact of Human Development
Index on the profile and outcomes of patients with acute coronary syndrome. Heart. 2015; 101(4):279–
86. doi: 10.1136/heartjnl-2014-306389 PMID: 25538134; PubMed Central PMCID: PMC4345920.
36. Dumith SC, Hallal PC, Reis RS, Kohl HW 3rd. Worldwide prevalence of physical inactivity and its asso-
ciation with human development index in 76 countries. Prev Med. 2011; 53(1–2):24–8. PMID:
37. Oliveira MG, Lira PI, Batista Filho M, Lima Mde C. [Factors associated with breastfeeding in two munici-
palities with low human development index in Northeast Brazil]. Rev Bras Epidemiol. 2013; 16(1):178–
89. PMID: 23681334.
38. World Health Organization. Country health information systems: a review of the current situation and
trends: World Health Organization; 2011.
39. Westra BL, Latimer GE, Matney SA, Park JI, Sensmeier J, Simpson RL, et al. A national action plan for
sharable and comparable nursing data to support practice and translational research for transforming
health care. J Am Med Inform Assoc. 2015. doi: 10.1093/jamia/ocu011 PMID: 25670754.
40. Castillo-Salgado C. Módulos de principios de epidemiología para el control de enfermedades. 2 a edi-
ción. OPS Washington. 2002:52–62.
41. World Health Organization. Monitoring, evaluation and review of national health strategies: a country-
led platform for information and accountability. World Health Organization, 2011.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 13 / 14
42. Dawson M, Gartner R. Differences in the characteristics of intimate femicides the role of relationship
state and relationship status. Homicide studies. 1998; 2(4):378–99.
43. Doiron D, Burton P, Marcon Y, Gaye A, Wolffenbuttel BH, Perola M, et al. Data harmonization and fed-
erated analysis of population-based studies: the BioSHaRE project. Emerg Themes Epidemiol. 2013;
10(1):12. doi: 10.1186/1742-7622-10-12 PMID: 24257327; PubMed Central PMCID: PMC4175511.
Femicide Data across Europe
PLOS ONE | DOI:10.1371/journal.pone.0148364 February 9, 2016 14 / 14