Reference metadata describe statistical concepts and methodologies used for the collection and generation of data. They provide information on data quality and, since they are strongly content-oriented, assist users in interpreting the data. Reference metadata, unlike structural metadata, can be decoupled from the data.
Eurostat, the statistical office of the European Union
1.2. Contact organisation unit
F4: Quality of life
1.3. Contact name
Confidential because of GDPR
1.4. Contact person function
Confidential because of GDPR
1.5. Contact mail address
2920 Luxembourg
1.6. Contact email address
Confidential because of GDPR
1.7. Contact phone number
Confidential because of GDPR
1.8. Contact fax number
Confidential because of GDPR
2.1. Metadata last certified
20 March 2025
2.2. Metadata last posted
20 March 2025
2.3. Metadata last update
20 March 2025
3.1. Data description
The main purpose of the EU survey on gender-based violence (EU-GBV) is to assess the prevalence of violence in order to address the requirements of the Istanbul Convention. The survey covers psychological, physical and sexual violence by intimate partner, physical and sexual violence by non-partner, sexual harassment at work, violence experienced in childhood and stalking by any perpetrator.
The data collection for the first wave (year 2021) was conducted in voluntary bases and took place between September 2020 and March 2024 in the EU countries, based on their national timetables. Eurostat coordinated data collection in 18 Member States (BE, BG, DK, EE, EL, ES, FR, HR, LV, LT, MT, NL, AT, PL, PT, SI, SK, FI). Italy shared main indicators from their national survey. The European Union Agency for Fundamental Rights (FRA) and the European Institute for Gender Equality (EIGE) launched a joint data collection in the eight Member States not covered by Eurostat (CZ, DE, IE, CY, LU, HU, RO, SE) following the EU-GBV manual.
The target population of the EU-GBV survey is persons aged 18-74 who live in private households, with a focus on women. However, men can be included in the target population by countries willing to do so. Out of all EU, 11 Member States (BE, EE, FR, HR, LV, LT, MT, NL, PT, SI, FI) covered in the survey men, and accordingly, data for violence experienced by total population are available for them. The disseminated indicators focus on violence by perpetrator, disaggregated by type of violence, frequency, severity, seriousness and reporting of the experienced violence and by personal characteristics of the respondent as age, level of education, activity limitation and degree of urbanisation.
It is necessary to point out that survey data might only be a close proxy to real prevalence as survey data depends on the willingness of the respondent to disclose any violence experienced. Therefore, to understand the prevalence of violence and disclosure rates by survey respondents, it is important to take into account the extent to which violence is tolerated in the wider community. For example, in cultures where people are ready to talk about their painful experiences, their answers may reflect more accurately their own experiences rather than community norms.
Based on the literature review and also results of the survey, both, women and men are experiencing violence. However, the experiences of women and men are very different considering type of perpetrator and type of violence as well as seriousness and consequences of the experienced violence. Even more, as the survey questions were focused on the victim’s experience of certain violent acts, both women and men could experience such violence due to victimising or putting up a defence against the perpetrator. Therefore, it is important to consider the seriousness of the violence experienced to better understand the victimisation.
3.2. Classification system
The country codes conform to the ISO 3166 (International Organisation of Standardisation), and regional codes to the NUTS 2 classification. Educational level of respondent and her partner are based on ISCED 2011. Variables on respondent occupation and economic activity are optional, however, if collected, the classification of occupation uses ISCO-08 (two-digit) and the classification of economic activity uses NACE Rev. 2 (two-digit).
Type of perpetrator of violence are defined according to relationship.
Violence by intimate partner covers psychological, threats, physical and sexual violence. Intimate partners are persons with whom a respondent has or had an intimate relationship. Regarding the disseminated indicators by type of perpetrator, the prevalence of intimate partner violence is calculated out of persons who have ever been in intimate partnership.
Violence by non-partner, covers threats, physical and sexual violence. Non-partners are all perpetrators with whom a respondent does not have or has never had an intimate relationship:
‘Family member or relative’ includes blood relatives like parents and children, and other blood relatives that can be cohabitating or non-cohabitating, as well as other household members or relatives by marriage or adoption (e.g. siblings, grandparents, aunts, uncles, cousins, nephews, nieces, in-laws, etc.);
‘Non-family or relative but other known’ as friends, family friends, schoolmates, colleagues, co-workers, supervisor, boss, professor, teacher; any person with some authority as army or police officer, priest, religious leader, judge, doctor; any other person known;
‘Any known’ includes family member or relative or other known explained above;
‘Stranger’ is someone completely unknown to the respondent.
Multiple victimisation is considering type of perpetrators as current partner, former partner(s), relative(s), friend(s), supervisor(s), other person(s) with authority, other(s) known, stranger(s) or other person(s). Respondent experienced violence by one type if one of them committed violent act, but not others. For example, if respondent experienced violence by more than one former partner, but not by current partner or any non-partner, it is considered as experienced violence by one type of perpetrator.
Regarding disseminated indictor by type of perpetrator, for intimate partners it is calculated out of persons who have ever been in intimate partnership and for non-partners and any it is calculated out of all persons.
Type of violence
Psychological violence by intimate partner includes a range of behaviours, encompassing acts of emotional abuse and controlling behaviour towards the respondent: belittling and humiliating; forbidding the respondent to see friends or family, or to engage in hobbies or other activities; tracking the respondent via GPS, phone or social network; forbidding the respondent to leave the house without permission or locking the respondent up; constantly accusing the respondent of being unfaithful or getting angry if the respondent speaks to another person; forbidding the respondent to work; controlling the finances of the whole family and the respondent’s personal expenses; keeping or taking away the respondent’s ID card/passport to control the respondent; yelling and smashing things or behaving in a certain way with the aim of scaring or intimidating the respondent; threatening to hurt the respondent’s children or other people close to the respondent; threatening to take away the respondent’s children or to deny custody; and threatening to harm them self if the respondent leaves them.
Threatening means behaviour involving fear, such as threatening to harm the respondent in a way that really frightened respondent. Regarding the disseminated indictors by type of violence, threatening means only threatening and not physical or sexual violence.
Physical violence refers to a range of violent types of behaviour or acts involving harm and fear, such as pushing or shoving the respondent, pulling their hair, slapping or throwing something at them; punching the respondent or beating them with an object; kicking; burning (with fire or acid or by some other means); trying to choke or strangle the respondent; threatening to use or actually using a knife, gun, acid or something similar; or using force against the respondent in some other way with the aim of hurting them. Regarding the disseminated indictors by type of violence, physical violence means only physical and not sexual violence.
Sexual violence includes unwanted sexual intercourse through force or physical violence or by exploiting a situation in which the respondent is not able to refuse sexual intercourse because they are under the influence of alcohol or drugs. It also includes unwanted sexual intercourse which the respondent is too afraid to refuse and cases in which the respondent is forced into unwanted sexual intercourse with another person or persons. Attempts to carry out any of the above acts or any other unwanted sexual behaviour that the respondent finds degrading or humiliating are included as well. Finally, this type of violence also covers unwanted sexual touching by non-partners.
Sexual harassment at work covers the following unwanted behaviours with sexual connotations that happen in a work context: inappropriate staring or leering; being exposed to sexually explicit images or videos; indecent sexual jokes or offensive remarks about a person’s body or private life; inappropriate invitations to go out on a date or suggestions for sexual activity of any kind; unsolicited physical contact; inappropriate advances on social networking websites or sexually explicit emails or text messages; threatening with unpleasant consequences if sexual proposals or advances are refused; and any other similar behaviour with sexual connotations that took place at work or work-related settings and that offended, humiliated, or intimidated the respondent. Prevalence of sexual harassment at work is calculated out of persons ever worked.
Repeated violence
Repeated violence (series of episodes) refers to similar violent episodes repeated by the same person(s), during which similar thing(s) are done under the same circumstances more than once. For instance, a woman might be beaten by her partner in several episodes over a period of three years.
Reporting or telling other people about violent experiences
Estimating unreported violence is an important aspect in defining targeted policies to eradicate violence. Indicators on reporting of violence show whether the respondent talked with anyone about the experienced violence. Reported violence experienced by intimate partner or by non-partner (to whom it was reported) includes:
‘Close person’ that may be family or friend and should be understood in its widest meaning.
‘Health or social service’ includes reporting or talking to any doctor or nurse, or any other health, medical or social worker.
‘Support service’ means any organisation or official body providing help to victims of violence, e.g. support services, helplines, shelters.
‘Police’ covers the situation when respondent herself reported to the police any episode of experienced violence.
Share of persons reported or talked about experiences of violence by to whom is calculated out of persons experienced this type of violence.
Consequences of experienced violence
The consequences of violence are an important consideration both for the victim and for society. Regarding violence experienced by intimate partner or non-partner, indicators for following consequences are disseminated:
‘Psychological consequences’ means for instance, depression, panic attacks, problems with concentration, problems in sleeping or eating or other consequences.
‘Physical injury’ is damage to the body caused by external force as cuts, scratches, bruises, burns, broken bones or fractures, head, internal or genital injuries, miscarriage, pregnancy due to rape or any other type of injury due to physical or sexual violence.
‘Felt that their life was in danger’ means that the respondent felt that her life was in danger when any of violent episodes happened.
Share of persons experienced this type of violence and consequences due to this is calculated out of all relevant persons: for intimate partner violence out of persons ever been in intimate partnership; for non-partner and any perpetrator out of all persons.
Personal characteristics
Indicators are disseminated by following personal characteristics:
Disability status is measured through a concept of general activity limitation (limitation in activities people usually do because of health problems for at least the past 6 months).
Educational attainment level refers to the highest level in the international standard classification of education (ISCED) which an individual has successfully completed.
Degree of urbanization of the area where a person has their usual residence is classified in local administrative units at level 2 as cities, towns and suburbs, or rural areas, based on the share of local population living in urban clusters and in urban centres.
Indicators are calculated as share of persons with a certain personal characteristic (e.g. severely limited) who experienced violence (e.g. by intimate partner) out of all persons with this personal characteristic (e.g. severely limited).
National difficulties of implementation of the common EU-GBV questionnaire
Countries were requested to translate and adopt the common EU-GBV survey questionnaire in their national context: relevant languages to be used and software used to collect the data. The translation and adaption of the questions, terms, definitions and concepts was done via pretesting. In general, countries did not report particular difficulties in translating the definitions and concepts. However, few countries had to find proper translation for the word ‘episode’.
Issues appeared with implementation of the questionnaire:
FR had an issue with implementation of the routing and therefore not all relevant questions were asked when applicable. Therefore, indicators are not disseminated for reporting and consequences of non-partner violence, and accordingly of violence by any perpetrator.
3.5. Statistical unit
Statistical unit of EU-GBV survey is individual. Only one person per household can be interviewed and proxy is not allowed. The reason is the safety of the respondent and the interviewer, as it is not uncommon for the abuser to be part of the victim’s household.
3.6. Statistical population
The EU-GBV results cover individual persons, residing in the territory of the Member States and participating countries at the time of data collection. Persons living in collective households and in institutions are excluded from the target population as individuals living in institutions (e.g., hospitals, prisons or religious institutions) are often not listed in the household/person survey frames and this makes it difficult to contact them.
Table 1. Gross sample size by sex of respondent.
Country
Total
Women
Men
BE
23177 CAWI phase + 9343 CAPI phase
19246 CAWI phase + 7782 CAPI phase
3931 CAWI phase + 1561 CAPI phase
EE
16700 persons
6800
9900
FR
Phase II (in-depth): 15000
Phase II (in-depth): 11474
Phase II (in-depth): 3526
HR
22695
11430
11265
LV
10407 (including sample error)
6245
4133
LT
10151 (final gross sample)
5020
5131
MT
12000 (final gross sample)
7200
4800
NL
28250 (including oversampling)
15635
12615
PT
21030 dwellings
not applicable
not applicable
SI
5000
2452
2548
FI
20000 (including aged 18-74)
15000
5000
3.7. Reference area
EU-GBV survey includes all EU Member States, but out of them 11 Member States (BE, EE, FR, HR, LV, LT, MT, NL, PT, SI, FI) covered in the survey men, and accordingly, data for violence experienced by total population are available for them.
3.8. Coverage - Time
Data collection for the first wave covers the 2020-2024 period. The duration of the fieldwork was recommended as four-months period based on national timetables (wave 2021). However, the sensitive nature of the survey means that staff may encounter many refusals and may need to make return visits, ring people back, or even do re-sampling and therefore, fieldwork period could be prolonged in order to ensure high quality data.
Table 2. Fieldwork period.
Country
Fieldwork period
BE
July 2021 - August 2022
EE
October 2021 - August 2022
FR
March 2021 - July 2021
HR
October 2022 - February 2023
LV
August 2021 - January 2022
LT
April 2021 - September 2021
MT
February 2022 - May 2022
NL
October 2020 - December 2020
PT
July 2022 - September 2022
SI
September 2020 - December 2020
FI
October 2021 - February 2022
3.9. Base period
Restricted from publication
Results of the EU-GBV survey are disseminated in prevalence rates (e.g. % of victims who experienced violence). The denominator (out of whom) has to be checked for each indicator: it could be relevant population (e.g. ever partnered persons, ever working persons) or victims of certain violence.
The reference time of experienced violence is divided into the last 12 months, 1-5 years ago or earlier than 5 years ago, according to the occurrence of the last episode. However, the indicator on occurrence of the last episode is disseminated only for women.
Data covering experiences over the last year and the last five years can give an indication of the extent and the nature of current levels of violence and an estimate of the number of people who may require help. Lifetime experiences, by contrast, provide an indication of the total number of people ever affected by such forms of violence.
6.1. Institutional Mandate - legal acts and other agreements
To develop efficient and effective policy and legal responses to limit/stop gender-based violence, accurate data are necessary. Thus, the Gender Equality Strategy of the European Commission for 2020-2025 calls for comprehensive, updated and comparable data for policies on combating gender-based violence. The EU-GBV survey is not covered by an official EU legislation and the first wave of the data collection in 2020-2024 takes place on a voluntary basis.
Data collection on gender-based violence is implemented based on the Commission implementing decision on the financing of the Programme for single market, competitiveness of enterprises, including small and medium-sized enterprises, and European statistics and the adoption of the work programme for 2021-2024. Even more, Regulation (EU) 2021/690 of the European Parliament and of the Council of 28 April 2021 requests to provide high quality, timely and reliable statistics to support the European Pillar of Social Rights and the Union Skills Policy, including statistics on the labour market, employment, education and training, income, living conditions, poverty, inequality, social protection, gender-based violence, undeclared work, and satellite accounts on skills.
Data on GBV are collected in order to address the requirement of Istanbul Convention (Council of Europe, Convention on preventing and combating violence against women and domestic violence, Istanbul 2011). The Gender Equality Strategy announced measures to obtain the same objectives as the Istanbul Convention. Therefore, on 8 March 2022, the Commission adopted a legislative proposal on combating Violence against Women and Domestic Violence. Accordingly, the Directive (EU) 2024/1385 of the European Parliament and of the Council of 14 May 2024 on combating violence against women and domestic violence includes a provision to conduct population-based surveys at regular intervals to assess the prevalence and trends.
6.2. Institutional Mandate - data sharing
Restricted from publication
7.1. Confidentiality - policy
Regulation (EC) No 223/2009 on European statistics (recital 24 and Article 20(4)) of 11 March 2009 (OJ L 87, p. 164) stipulates the need to establish common principles and guidelines. It ensures the confidentiality of data used for the production of European statistics and the access to those confidential data, with due account for technical developments and the requirements of users in a democratic society.
7.2. Confidentiality - data treatment
Publication rules:
estimates should not be published if based on fewer than 20 sample observations or if the non-response for the item concerned exceeds 50%;
estimates should be published with a flag if based on 20 to 49 sample observations or if non-response for the item concerned exceeds 20% and is lower or equal to 50%;
estimates shall be published in the normal way when based on 50 or more sample observations and non-response for the item concerned is lower than 20%.
Flags:
u unreliable (i.e. due to small number of observations)
8.1. Release calendar
Indicators on gender-based violence against women based on EU-GBV survey were published on the 25th of November 2024, covering the EU-27 MS and the estimated EU-average. Data on violence experienced by total population were disseminated on the 20th of March 2025, covering 11 Member States (BE, EE, FR, HR, LV, LT, MT, NL, PT, SI, FI), that covered in the survey men.
In line with the Community legal framework and the European Statistics Code of Practice, Eurostat disseminates European statistics on Eurostat's website (see section 10 - 'Accessibility and clarity'), respecting professional independence and in an objective, professional and transparent manner in which all users are treated equitably. The detailed arrangements are governed by the Eurostat protocol on impartial access to Eurostat data for users.
Information is provided in the Methodological manual for the EU survey on gender-based violence against women and other forms of inter-personal violence (EU-GBV) (available here) as well as in Statistics on crime and criminal justice dedicated group in CIRCABC.
10.6.1. Metadata completeness - rate
Restricted from publication
10.7. Quality management - documentation
Not applicable
11.1. Quality assurance
Common questionnaire and the methodology have been developed for the EU-GBV survey that are followed by the countries in order to implement the survey. Please see more details in the Methodological manual for the EU survey on gender-based violence against women and other forms of inter-personal violence (EU-GBV).
EU-GBV survey questionnaire and methodology are developed based on the results of the pre-testing and pilot survey. Testing was conducted in 14 countries over the period 2018–2019.
11.2. Quality management - assessment
Standardisation is achieved with the common questionnaire and common list and content of target variables. This is complemented by Eurostat consistency and integrity checks on the micro data. In addition, countries should report to Eurostat any deviation from the standard.
EU-GBV statistics have overall high quality. National surveys are considered as reliable sources applying high standards with regard to the methodology. However, the EU-GBV, like any survey, is based upon a sample of the population. The results are therefore subject to the usual types of errors associated with random sampling. Based on the sample size and design in the various Member States and participating countries, Eurostat implements basic guidelines intended to avoid publication of figures that are unreliable and to give warning for low reliability.
12.1. Relevance - User Needs
The relevance of an instrument has to be assessed in the light of the needs of its users. As for the EU-GBV survey results, the main users are the following:
institutional users like DG JUST of the Commission and other Commission services;
statistical users in Eurostat or in Member States and participating countries;
end users - including the media - interested in data about experienced violence
12.2. Relevance - User Satisfaction
Eurostat carries out a general Eurostat User Satisfaction Survey every year to collect feedback on the quality of its statistics. The survey is usually addressed to the registered Eurostat users who are mainly students, academic, private users, business, government and international organizations.
12.3. Completeness
EU-GBV results cover national territories of the participating countries.
12.3.1. Data completeness - rate
Restricted from publication
13.1. Accuracy - overall
Data were to be based on a nationally representative sample of the population residing in private households within the country, irrespective of language, nationality or legal residence status. The sampling frame and methods of sample selection were left to the countries’ experiences with other population surveys in which they could ensure that every individual in the target population was assigned a known and non-zero probability of selection. The main condition that needed to be fulfilled is that only one person can be interviewed per household.
The indicative effective sample size defined for the EU-GBV is 5000 respondents per country. However, countries are allowed to increase or decrease the sample size in order to collect high quality data.
Table 3. Gross and net sample size.
Country
Gross sample (selected for survey)
total/men/women
Net sample (accepted interviews)
total/men/women
Share of responded (net/gross*100)
total/men/women
Sampling and selected mode of data collection
BE
23177 / 3931 / 19246
5494 / 965 / 4529
23.7% / 24.5% / 23.5%
First sample: for CAWI phase from the national register, higher share was sampled in Brussels than in Flanders and Wallonia. After CAWI, subsample was drawn from not responded part for CAPI phase, CATI/CAWI also allowed.
EE
16700 / 9900 / 6800
10414 / 5841 / 4573
62.4% / 59.0% / 67.3%
Probability stratified sampling by sex and age group was used, sampling frame based on the statistical register of residents. CAPI and CAWI used.
FR
15000 / 3526 / 11447
9064 / 2175 / 6889
60.4% / 61.7% / 60.2%
Sample drawn from tax database “the Housing and individual demographic files” (Fideli), around 40 out of 101 NUTS 3 areas were overrepresented in order to be representative at NUTS3 level in phase 1 (screening); sample for phase 2 (follow-up survey with EU-GBV questionnaire) drawn out of responded for phase 1. Phase 2 conducted with CATI/CAWI.
HR
22695 / 11265 / 11430
6171 / 2755 / 3416
27.2% / 24.5% / 29.9%
Sampling frame: Census 2021 data linked with the telephone book. CATI/CAWI used.
LV
10378 / 4133 / 6245
6269 / 2328 / 3941
60.4% / 56.3% / 63.1%
Automate tools for build sample frame used, which combines information from various registers and additional sources. CATI/CAWI used.
LT
10151 / 5131 / 5020
5504 / 2318 / 3186
54.2% / 45.2% / 63.5%
Sampling frame: Population Register. CAWI/CAPI used.
MT
11990 / 4796 / 7194
5005 / 1991 / 3014
41.7% / 41.5% / 41.9%
Stratified random sampling where the strata were constructed using District, Gender and Age Group, frame is based on population statistics. CATI used.
NL
28250 / 12615 / 15635
7536 / 3352 / 4184
26.7% / 26.6% / 26.8%
Sampling frame: Personal Records Database (BRP). CAWI used.
PT
21030
11346
54.0%
Sample is selected from a sampling frame (BA) extracted from the National Dwellings Register (FNA), following a stratified and multistage sampling scheme where the primary units. Eligible respondent selected by the last birthday method. CAWI/CAPI/CATI used.
SI
5000 / 2548 / 2452
2430 / 1148 / 1282
48.6% / 45.1% / 52.3%
Sampling frame: Demographic database, Central Register of Population (CRP). CAWI/CAPI/CATI used.
FI
20000 / 5000 / 15000
6014 / 1417 / 4597
30.1% / 28.3% / 30.6%
Source of sample: Population according to urban-rural classification by age and sex. CAWI used.
Notes:
BE: Based on phase I
FR: Based on phase II (already contacted respondents)
PT: Sampling is based on dwellings, therefore total population is applicable
13.2. Sampling error
Standard errors of key indicators are commonly used as a measure of the reliability of data collected through sample survey. EU-GBV survey was designed to provide measure of prevalence of violence during lifetime aged 18-74 persons. The indicative effective sample size was defined taking into account this accuracy requirement.
13.2.1. Sampling error - indicators
Restricted from publication
13.3. Non-sampling error
A non-sampling error is an error in survey estimates which cannot be attributed to sampling fluctuations. Such errors can either be coverage errors, measurement errors, non-response errors, processing errors or model assumption errors.
13.3.1. Coverage error
Coverage errors arise due to divergences that exist between the target population and the sampling frame.
13.3.1.1. Over-coverage - rate
Restricted from publication
13.3.1.2. Common units - proportion
Restricted from publication
13.3.2. Measurement error
Measurement errors occur at the time of data collection. There are a number of sources for these errors, such as the survey instrument, the information system, the interviewer and the mode of collection.
Data collection is based on the common questionnaire, definitions and concepts. Due to sensitivity of the topic, safety and well-being of survey participants and interviewers was highlighted in the methodological manual and countries were requested to report methods used in the metadata.
It was highlighted in the methodological manual that only one member of the household should be interviewed alone in safe and secure conditions in order for both the respondent and interviewer to be safe and have no distractions while conducting the interview. Not feeling safe or being in the same place with the potential abuser makes the interview impossible to conduct and therefore leads to refusals to participate to the survey.
Different data collection modes were allowed for this data collection, however, face-to-face and computer-assisted methods were most recommended and possibility to provide self-completion was highlighted. Advantages and disadvantages for each data collection mode were provided in the methodological manual. Countries were recommended to use most appropriate method in national context, combined methods were allowed.
Table 4. Data collection mode.
Country
Computer Assisted
Personal Interviewing
(CAPI)
Computer Assisted
Web Interviewing
(CAWI)
Computer Assisted
Telephone Interviewing
(CATI)
Comment for multiple mode
BE
√
√
√
First CAWI (phase 1), then sample was drawn out of not responded for CAPI and CATI, CAWI was also allowed (phase 2).
EE
√
√
√
First CAWI, if not completed despite reminders, then CAPI; interviewers were allowed to conduct interviews via phone (CATI).
FR
√
√
For the phase 2, sample was divided between CAWI and CATI.
HR
√
√
First CAWI, if not completed despite reminders, then CATI.
LV
√
√
First CAWI, if not completed despite reminders, then CATI.
LT
√
√
√
First CAWI, if not completed despite reminders, then CAPI; interviewers were allowed to conduct interviews via phone (CATI).
MT
√
Not applicable
NL
√
Not applicable
PT
√
√
√
First CAWI, if not completed despite reminders (if email address was available) then, CATI or CAPI. CAWI was available through full fieldwork period.
SI
√
√
√
First CAWI, if not completed despite reminders, then CATI or CAPI.
FI
√
Not applicable
13.3.3. Non response error
Non-response errors arise due to an unsuccessful attempt to obtain the desired information from an eligible unit. Two main types of non-response errors are considered:
unit non-response, which refers to the absence of information of whole units (households and/or persons) selected into the sample;
item non-response, which refers to the situation where a sample unit has been successfully enumerated, but not all the required information has been obtained.
13.3.3.1. Unit non-response - rate
All surveys have to deal with non-response, i.e., information missing for some of the sample units. Based on the national metadata, total response rate (not taking into account impact of data collection mode) is calculated as share of completed and accepted interviews out of gross sample size.
Table 5. Total response rate for interviewer assisted and not assisted modes.
Interviewer not assisted mode (CAWI): out of total sample
Country
Accepted interviews
Completed but not accepted
Interruption after beginning
BE
17.5%
0.8%
4.5%
EE
17.9%
0.07%
0%
FR
59.9%
7.3%
3.9%
HR
18.1%
0.005%
0%
LV
5.8%
0.14%
0%
LT
13.7%
0.1%
0%
NL
26.7%
0.8%
70.9%
PT
14.6%
0.8%
1.8%
SI
44.4%
3.7%
0.4%
FI
30.1%
0.7%
3.6%
Notes:
NL: explained that CAWI interruption includes also cases when respondent just opened the questionnaire;
MT: not applicable as CAWI was not used
Interviewer assisted modes (CAPI/CATI): response rates out of contacted; not ccontacted out of total planned for this mode
Country
Accepted interviews
Completed but not accepted
Interruption after beginning
Refusal
Not contacted
BE
33.0%
0.2%
0.8%
10.1%
57.0%
EE
62.2%
0.07%
0%
26.4%
16.9%
FR
63.0%
9.6%
6.0%
16.3%
n.a.
HR
12.5%
0%
0%
54.5%
11.3%
LV
76.0%
0.2%
0%
21.6%
23.1%
LT
51.2%
0.1%
0%
27.9%
9.2%
MT
68.1%
0.2%
0.6%
8.8%
38.7%
PT
90.7%
1.7%
0%
1.5%
49.2%
SI
25.9%
0.6%
0%
62.3%
68.6%
Notes:
FR: response rates are provided based on phase II (with respondents participated in Phase I);
NL, FI: not applicable as only CAWI was used
13.3.3.2. Item non-response - rate
Item non-response typically happens to questions the interviewee does not answer because they consider them personal or not easily understandable. Regarding EU-GBV survey, several countries had issues with implementing the routing properly in the interviewing software due to complexity of the questionnaire. In case the question was applicable but not asked due to routing issues, the value 9 – not known is used in the variable.
In the EU-GBV survey, non-response was not allowed for several questions:
most important background questions of respondent and her partner
questions that were used to consider if section of the questionnaire was applicable, e.g. current activity status (for section on ‘Sexual harassment at work’) or actual marital status (for section on ‘Intimate partner violence’)
However, due to nature of the national survey, Portugal allowed non-response for all questions.
Secondly, as the main purpose of the survey is to measure the prevalence of violence, the set of screening questions for each type of violence covered in the survey could not be completely not responded (at least one screening question had to be responded as ‘yes’ or ‘no’ in each set of screening questions). In case the rule was not met, the interview was not accepted in the database. Therefore, item non-response regarding screening questions is not relevant when disseminating the results. However, item non-response is provided in order to illustrate the sensitivity of the screening questions.
Item non-response is provided for the variables used to calculate disseminated indicators. Composite variable used for calculation of the indicators is considered as not responded if all relevant and applicable variables are not responded. For example, ‘occurrence of intimate partner violence’ is considered as not responded if respondent experienced violence only by current partner and occurrence of last episode by current partner is not responded or respondent experienced violence only by former partner and occurrence of last episode by former partner is not responded or respondent experienced violence by both partner and occurrence of last episode by both partners are not responded. Item non-response is calculated out of all applicable respondents. For example, ‘occurrence of intimate partner violence’ is calculated out of all respondents who experienced violence by intimate partner.
Processing errors occur in post-data-collection processes such as data entry, keying, editing and weighting.
13.3.5. Model assumption error
Restricted from publication
14.1. Timeliness
Not applicable
14.1.1. Time lag - first result
Restricted from publication
14.1.2. Time lag - final result
Restricted from publication
14.2. Punctuality
Not applicable
14.2.1. Punctuality - delivery and publication
Not applicable
15.1. Comparability - geographical
A high level of comparability across the participating countries is namely ensured by:
(a) the use of the same definitions for all countries;
(b) the transmission to Eurostat of the same list of variables with the same coding;
(c) the use of common classifications;
(d) the central validation of data done by Eurostat.
15.1.1. Asymmetry for mirror flow statistics - coefficient
Restricted from publication
15.2. Comparability - over time
Restricted from publication
15.2.1. Length of comparable time series
Restricted from publication
15.3. Coherence - cross domain
EU-GBV survey data collection follows international standards: ISCO, NACE, ISCED, degree of urbanisation, standardised variables for social surveys.
15.3.1. Coherence - sub annual and annual statistics
Restricted from publication
15.3.2. Coherence - National Accounts
Restricted from publication
15.4. Coherence - internal
Restricted from publication
Not available
17.1. Data revision - policy
Historical data; no longer updated or revised.
17.2. Data revision - practice
Historical data; no longer updated or revised.
17.2.1. Data revision - average size
Restricted from publication
18.1. Source data
The EU-GBV is a random sample survey of persons living in private households. The sampling units are dwellings, households or individuals depending on the sampling frame. Different schemes are used to sample the units, ranging from the simple random sampling method to complex stratified multi-stage sampling methods of clusters.
Participation in the survey is voluntary for all participating countries.
18.2. Frequency of data collection
Not yet agreed.
18.3. Data collection
There are no constraints on the most appropriate mode of data collection, which can vary from country to country. Nevertheless, due to the sensitive nature of this survey, and research and data-based findings, most recommended mode is face-to-face and computer-assisted methods over other methods. The possibility of self-completion is also supported.
18.4. Data validation
Prior to the dissemination of transmitted national data, Eurostat checks the data quality and consistency. Validation rules are provided in the Eurostat methodological manual.
18.5. Data compilation
Not applicable
18.5.1. Imputation - rate
Restricted from publication
18.6. Adjustment
Not applicable
18.6.1. Seasonal adjustment
Restricted from publication
No further comments.
The main purpose of the EU survey on gender-based violence (EU-GBV) is to assess the prevalence of violence in order to address the requirements of the Istanbul Convention. The survey covers psychological, physical and sexual violence by intimate partner, physical and sexual violence by non-partner, sexual harassment at work, violence experienced in childhood and stalking by any perpetrator.
The data collection for the first wave (year 2021) was conducted in voluntary bases and took place between September 2020 and March 2024 in the EU countries, based on their national timetables. Eurostat coordinated data collection in 18 Member States (BE, BG, DK, EE, EL, ES, FR, HR, LV, LT, MT, NL, AT, PL, PT, SI, SK, FI). Italy shared main indicators from their national survey. The European Union Agency for Fundamental Rights (FRA) and the European Institute for Gender Equality (EIGE) launched a joint data collection in the eight Member States not covered by Eurostat (CZ, DE, IE, CY, LU, HU, RO, SE) following the EU-GBV manual.
The target population of the EU-GBV survey is persons aged 18-74 who live in private households, with a focus on women. However, men can be included in the target population by countries willing to do so. Out of all EU, 11 Member States (BE, EE, FR, HR, LV, LT, MT, NL, PT, SI, FI) covered in the survey men, and accordingly, data for violence experienced by total population are available for them. The disseminated indicators focus on violence by perpetrator, disaggregated by type of violence, frequency, severity, seriousness and reporting of the experienced violence and by personal characteristics of the respondent as age, level of education, activity limitation and degree of urbanisation.
It is necessary to point out that survey data might only be a close proxy to real prevalence as survey data depends on the willingness of the respondent to disclose any violence experienced. Therefore, to understand the prevalence of violence and disclosure rates by survey respondents, it is important to take into account the extent to which violence is tolerated in the wider community. For example, in cultures where people are ready to talk about their painful experiences, their answers may reflect more accurately their own experiences rather than community norms.
Based on the literature review and also results of the survey, both, women and men are experiencing violence. However, the experiences of women and men are very different considering type of perpetrator and type of violence as well as seriousness and consequences of the experienced violence. Even more, as the survey questions were focused on the victim’s experience of certain violent acts, both women and men could experience such violence due to victimising or putting up a defence against the perpetrator. Therefore, it is important to consider the seriousness of the violence experienced to better understand the victimisation.
Type of perpetrator of violence are defined according to relationship.
Violence by intimate partner covers psychological, threats, physical and sexual violence. Intimate partners are persons with whom a respondent has or had an intimate relationship. Regarding the disseminated indicators by type of perpetrator, the prevalence of intimate partner violence is calculated out of persons who have ever been in intimate partnership.
Violence by non-partner, covers threats, physical and sexual violence. Non-partners are all perpetrators with whom a respondent does not have or has never had an intimate relationship:
‘Family member or relative’ includes blood relatives like parents and children, and other blood relatives that can be cohabitating or non-cohabitating, as well as other household members or relatives by marriage or adoption (e.g. siblings, grandparents, aunts, uncles, cousins, nephews, nieces, in-laws, etc.);
‘Non-family or relative but other known’ as friends, family friends, schoolmates, colleagues, co-workers, supervisor, boss, professor, teacher; any person with some authority as army or police officer, priest, religious leader, judge, doctor; any other person known;
‘Any known’ includes family member or relative or other known explained above;
‘Stranger’ is someone completely unknown to the respondent.
Multiple victimisation is considering type of perpetrators as current partner, former partner(s), relative(s), friend(s), supervisor(s), other person(s) with authority, other(s) known, stranger(s) or other person(s). Respondent experienced violence by one type if one of them committed violent act, but not others. For example, if respondent experienced violence by more than one former partner, but not by current partner or any non-partner, it is considered as experienced violence by one type of perpetrator.
Regarding disseminated indictor by type of perpetrator, for intimate partners it is calculated out of persons who have ever been in intimate partnership and for non-partners and any it is calculated out of all persons.
Type of violence
Psychological violence by intimate partner includes a range of behaviours, encompassing acts of emotional abuse and controlling behaviour towards the respondent: belittling and humiliating; forbidding the respondent to see friends or family, or to engage in hobbies or other activities; tracking the respondent via GPS, phone or social network; forbidding the respondent to leave the house without permission or locking the respondent up; constantly accusing the respondent of being unfaithful or getting angry if the respondent speaks to another person; forbidding the respondent to work; controlling the finances of the whole family and the respondent’s personal expenses; keeping or taking away the respondent’s ID card/passport to control the respondent; yelling and smashing things or behaving in a certain way with the aim of scaring or intimidating the respondent; threatening to hurt the respondent’s children or other people close to the respondent; threatening to take away the respondent’s children or to deny custody; and threatening to harm them self if the respondent leaves them.
Threatening means behaviour involving fear, such as threatening to harm the respondent in a way that really frightened respondent. Regarding the disseminated indictors by type of violence, threatening means only threatening and not physical or sexual violence.
Physical violence refers to a range of violent types of behaviour or acts involving harm and fear, such as pushing or shoving the respondent, pulling their hair, slapping or throwing something at them; punching the respondent or beating them with an object; kicking; burning (with fire or acid or by some other means); trying to choke or strangle the respondent; threatening to use or actually using a knife, gun, acid or something similar; or using force against the respondent in some other way with the aim of hurting them. Regarding the disseminated indictors by type of violence, physical violence means only physical and not sexual violence.
Sexual violence includes unwanted sexual intercourse through force or physical violence or by exploiting a situation in which the respondent is not able to refuse sexual intercourse because they are under the influence of alcohol or drugs. It also includes unwanted sexual intercourse which the respondent is too afraid to refuse and cases in which the respondent is forced into unwanted sexual intercourse with another person or persons. Attempts to carry out any of the above acts or any other unwanted sexual behaviour that the respondent finds degrading or humiliating are included as well. Finally, this type of violence also covers unwanted sexual touching by non-partners.
Sexual harassment at work covers the following unwanted behaviours with sexual connotations that happen in a work context: inappropriate staring or leering; being exposed to sexually explicit images or videos; indecent sexual jokes or offensive remarks about a person’s body or private life; inappropriate invitations to go out on a date or suggestions for sexual activity of any kind; unsolicited physical contact; inappropriate advances on social networking websites or sexually explicit emails or text messages; threatening with unpleasant consequences if sexual proposals or advances are refused; and any other similar behaviour with sexual connotations that took place at work or work-related settings and that offended, humiliated, or intimidated the respondent. Prevalence of sexual harassment at work is calculated out of persons ever worked.
Repeated violence
Repeated violence (series of episodes) refers to similar violent episodes repeated by the same person(s), during which similar thing(s) are done under the same circumstances more than once. For instance, a woman might be beaten by her partner in several episodes over a period of three years.
Reporting or telling other people about violent experiences
Estimating unreported violence is an important aspect in defining targeted policies to eradicate violence. Indicators on reporting of violence show whether the respondent talked with anyone about the experienced violence. Reported violence experienced by intimate partner or by non-partner (to whom it was reported) includes:
‘Close person’ that may be family or friend and should be understood in its widest meaning.
‘Health or social service’ includes reporting or talking to any doctor or nurse, or any other health, medical or social worker.
‘Support service’ means any organisation or official body providing help to victims of violence, e.g. support services, helplines, shelters.
‘Police’ covers the situation when respondent herself reported to the police any episode of experienced violence.
Share of persons reported or talked about experiences of violence by to whom is calculated out of persons experienced this type of violence.
Consequences of experienced violence
The consequences of violence are an important consideration both for the victim and for society. Regarding violence experienced by intimate partner or non-partner, indicators for following consequences are disseminated:
‘Psychological consequences’ means for instance, depression, panic attacks, problems with concentration, problems in sleeping or eating or other consequences.
‘Physical injury’ is damage to the body caused by external force as cuts, scratches, bruises, burns, broken bones or fractures, head, internal or genital injuries, miscarriage, pregnancy due to rape or any other type of injury due to physical or sexual violence.
‘Felt that their life was in danger’ means that the respondent felt that her life was in danger when any of violent episodes happened.
Share of persons experienced this type of violence and consequences due to this is calculated out of all relevant persons: for intimate partner violence out of persons ever been in intimate partnership; for non-partner and any perpetrator out of all persons.
Personal characteristics
Indicators are disseminated by following personal characteristics:
Disability status is measured through a concept of general activity limitation (limitation in activities people usually do because of health problems for at least the past 6 months).
Educational attainment level refers to the highest level in the international standard classification of education (ISCED) which an individual has successfully completed.
Degree of urbanization of the area where a person has their usual residence is classified in local administrative units at level 2 as cities, towns and suburbs, or rural areas, based on the share of local population living in urban clusters and in urban centres.
Indicators are calculated as share of persons with a certain personal characteristic (e.g. severely limited) who experienced violence (e.g. by intimate partner) out of all persons with this personal characteristic (e.g. severely limited).
National difficulties of implementation of the common EU-GBV questionnaire
Countries were requested to translate and adopt the common EU-GBV survey questionnaire in their national context: relevant languages to be used and software used to collect the data. The translation and adaption of the questions, terms, definitions and concepts was done via pretesting. In general, countries did not report particular difficulties in translating the definitions and concepts. However, few countries had to find proper translation for the word ‘episode’.
Issues appeared with implementation of the questionnaire:
FR had an issue with implementation of the routing and therefore not all relevant questions were asked when applicable. Therefore, indicators are not disseminated for reporting and consequences of non-partner violence, and accordingly of violence by any perpetrator.
Statistical unit of EU-GBV survey is individual. Only one person per household can be interviewed and proxy is not allowed. The reason is the safety of the respondent and the interviewer, as it is not uncommon for the abuser to be part of the victim’s household.
The EU-GBV results cover individual persons, residing in the territory of the Member States and participating countries at the time of data collection. Persons living in collective households and in institutions are excluded from the target population as individuals living in institutions (e.g., hospitals, prisons or religious institutions) are often not listed in the household/person survey frames and this makes it difficult to contact them.
Table 1. Gross sample size by sex of respondent.
Country
Total
Women
Men
BE
23177 CAWI phase + 9343 CAPI phase
19246 CAWI phase + 7782 CAPI phase
3931 CAWI phase + 1561 CAPI phase
EE
16700 persons
6800
9900
FR
Phase II (in-depth): 15000
Phase II (in-depth): 11474
Phase II (in-depth): 3526
HR
22695
11430
11265
LV
10407 (including sample error)
6245
4133
LT
10151 (final gross sample)
5020
5131
MT
12000 (final gross sample)
7200
4800
NL
28250 (including oversampling)
15635
12615
PT
21030 dwellings
not applicable
not applicable
SI
5000
2452
2548
FI
20000 (including aged 18-74)
15000
5000
EU-GBV survey includes all EU Member States, but out of them 11 Member States (BE, EE, FR, HR, LV, LT, MT, NL, PT, SI, FI) covered in the survey men, and accordingly, data for violence experienced by total population are available for them.
The reference time of experienced violence is divided into the last 12 months, 1-5 years ago or earlier than 5 years ago, according to the occurrence of the last episode. However, the indicator on occurrence of the last episode is disseminated only for women.
Data covering experiences over the last year and the last five years can give an indication of the extent and the nature of current levels of violence and an estimate of the number of people who may require help. Lifetime experiences, by contrast, provide an indication of the total number of people ever affected by such forms of violence.
Data were to be based on a nationally representative sample of the population residing in private households within the country, irrespective of language, nationality or legal residence status. The sampling frame and methods of sample selection were left to the countries’ experiences with other population surveys in which they could ensure that every individual in the target population was assigned a known and non-zero probability of selection. The main condition that needed to be fulfilled is that only one person can be interviewed per household.
The indicative effective sample size defined for the EU-GBV is 5000 respondents per country. However, countries are allowed to increase or decrease the sample size in order to collect high quality data.
Table 3. Gross and net sample size.
Country
Gross sample (selected for survey)
total/men/women
Net sample (accepted interviews)
total/men/women
Share of responded (net/gross*100)
total/men/women
Sampling and selected mode of data collection
BE
23177 / 3931 / 19246
5494 / 965 / 4529
23.7% / 24.5% / 23.5%
First sample: for CAWI phase from the national register, higher share was sampled in Brussels than in Flanders and Wallonia. After CAWI, subsample was drawn from not responded part for CAPI phase, CATI/CAWI also allowed.
EE
16700 / 9900 / 6800
10414 / 5841 / 4573
62.4% / 59.0% / 67.3%
Probability stratified sampling by sex and age group was used, sampling frame based on the statistical register of residents. CAPI and CAWI used.
FR
15000 / 3526 / 11447
9064 / 2175 / 6889
60.4% / 61.7% / 60.2%
Sample drawn from tax database “the Housing and individual demographic files” (Fideli), around 40 out of 101 NUTS 3 areas were overrepresented in order to be representative at NUTS3 level in phase 1 (screening); sample for phase 2 (follow-up survey with EU-GBV questionnaire) drawn out of responded for phase 1. Phase 2 conducted with CATI/CAWI.
HR
22695 / 11265 / 11430
6171 / 2755 / 3416
27.2% / 24.5% / 29.9%
Sampling frame: Census 2021 data linked with the telephone book. CATI/CAWI used.
LV
10378 / 4133 / 6245
6269 / 2328 / 3941
60.4% / 56.3% / 63.1%
Automate tools for build sample frame used, which combines information from various registers and additional sources. CATI/CAWI used.
LT
10151 / 5131 / 5020
5504 / 2318 / 3186
54.2% / 45.2% / 63.5%
Sampling frame: Population Register. CAWI/CAPI used.
MT
11990 / 4796 / 7194
5005 / 1991 / 3014
41.7% / 41.5% / 41.9%
Stratified random sampling where the strata were constructed using District, Gender and Age Group, frame is based on population statistics. CATI used.
NL
28250 / 12615 / 15635
7536 / 3352 / 4184
26.7% / 26.6% / 26.8%
Sampling frame: Personal Records Database (BRP). CAWI used.
PT
21030
11346
54.0%
Sample is selected from a sampling frame (BA) extracted from the National Dwellings Register (FNA), following a stratified and multistage sampling scheme where the primary units. Eligible respondent selected by the last birthday method. CAWI/CAPI/CATI used.
SI
5000 / 2548 / 2452
2430 / 1148 / 1282
48.6% / 45.1% / 52.3%
Sampling frame: Demographic database, Central Register of Population (CRP). CAWI/CAPI/CATI used.
FI
20000 / 5000 / 15000
6014 / 1417 / 4597
30.1% / 28.3% / 30.6%
Source of sample: Population according to urban-rural classification by age and sex. CAWI used.
Notes:
BE: Based on phase I
FR: Based on phase II (already contacted respondents)
PT: Sampling is based on dwellings, therefore total population is applicable
Results of the EU-GBV survey are disseminated in prevalence rates (e.g. % of victims who experienced violence). The denominator (out of whom) has to be checked for each indicator: it could be relevant population (e.g. ever partnered persons, ever working persons) or victims of certain violence.
Not applicable
The EU-GBV is a random sample survey of persons living in private households. The sampling units are dwellings, households or individuals depending on the sampling frame. Different schemes are used to sample the units, ranging from the simple random sampling method to complex stratified multi-stage sampling methods of clusters.
Participation in the survey is voluntary for all participating countries.
Not applicable
Not applicable
A high level of comparability across the participating countries is namely ensured by:
(a) the use of the same definitions for all countries;
(b) the transmission to Eurostat of the same list of variables with the same coding;
(c) the use of common classifications;
(d) the central validation of data done by Eurostat.
