ANNEX A.1
FRA2-2007-3200-T04
Discrimination and Victimisation in EU Member States: Experiences and Attitudes of Immigrants and Other Minorities
TECHNICAL SPECIFICATIONS
1. Background Information concerning the caLL
1.1 FRA mandate and data collection
On 1st March 2007 Council Regulation (EC) No 168/2007 came into effect establishing the European Union Agency for Fundamental Rights (FRA). With this, the FRA became the legal successor to the European Monitoring Centre on Racism and Xenophobia (EUMC), which had been operating since 1998.
The FRA took over the work of the EUMC to encompass a wider mandate covering fundamental rights. Under Council Regulation (EC) No 168/2007, paragraph 9 of the preamble states that: ‘The Agency should refer in its work to fundamental rights within the meaning of Article 6(2) of the Treaty on European Union, including the European Convention on Human Rights and Fundamental Freedoms, and as reflected in particular in the Charter of Fundamental Rights’, and in paragraph 10 that ‘the work of the Agency should continue to cover the phenomena of racism, xenophobia and anti-Semitism, the protection of rights of persons belonging to minorities, as well as gender equality, as essential elements for the protection of fundamental rights’.
Continuing the work of the EUMC, the primary mandate of which was to provide the European Community and its Member States with objective, reliable and comparable data and information on racism and xenophobia, paragraph 12 in the above preamble states that: ‘The [FRA] Agency should collect objective, reliable and comparable information on the development of the situation of fundamental rights, analyse this information in terms of the causes of disrespect, consequences and effects and examine examples of good practice in dealing with these matters’.
The FRA’s primary tool for data collection on racism and xenophobia is the RAXEN (Racism and Xenophobia) network, which was established under the EUMC. RAXEN consists of 27 national focal points (NFPs), one in every Member State. Each NFP is contracted by the Agency to collect data and information on relevant legislation covered by the Agency’s mandate, as well as information about the extent and nature of and responses to ‘racist’ violence and related crime from both government and non-government sources. NFPs also collect information about the situation in their Member State with respect to the treatment of and services provided to different groups, including examples both of discrimination and ‘good practices’, in key areas of social life; namely: employment, education, housing, and health and social care. This data collection serves to inform the Agency’s key stakeholders with respect to the development of strategies to combat racism and xenophobia. Yet, in certain fields, such as racist violence and crime, NFPs, and in turn the Agency, are severely hampered by a lack of comprehensive official criminal justice data that is able to inform evidence-based policies about vulnerable minorities’ exposure to crime and the impact that it has on their lives.
1.2 Pilot Victim Survey
1.2.i
Background Information
As part of its mandate to collect and publish data on racist crime and violence, the FRA’s predecessor, the EUMC, published a report in 2005 on ‘Racist Violence in 15 EU Member States: a comparative overview of findings from the RAXEN NFP Reports 2001-2004’[1]. The report indicated that both official and unofficial data collection on racist crime and violence is woefully inadequate in the majority of Member States. The report also showed that data, where it is available, cannot be compared directly between Member States as ‘incidents’ or ‘crimes’ are defined and recorded differently. Although the report focused on the EU15 in the period 2001-2004, evidence from the ‘new’ Member States also paints a picture of inadequate data collection in some.
The chapter on ‘racist violence and crime’ in the Agency’s Annual Reports also shows that official criminal justice data collection remains inadequate or absent in many Member States[2]. This assertion is supported by information made available on the Agency’s on-line InfoBase[3].
Given the absence of official data on racist crime and violence in much of the EU, one of the key recommendations in the racist violence report was to ‘develop crime/victim surveys’ that can capture ‘vulnerable’ communities’ experiences of criminal victimisation, including any incidents that are ‘racially’, ethnically or religiously motivated. In turn, the Agency’s report on ‘Policing Racist Crime and Violence’[4], which was written by an independent expert and published in 2005 as a follow-up to the racist violence report, also recommended the development of an ‘EU-wide victim survey’.
Questionnaire-based victim surveys offer an alternative to official police/prosecution data because they directly ask members of the public about their experiences of victimisation. To this end, victim surveys are able to capture a greater range of incidents than those reported to the police because they are not reliant on people’s willingness to report to the police or on the police’s ability or willingness to record incidents.
Victim surveys are now a well established research instrument in a number of Member States; with, in 2005, the first ‘European Crime Survey’ being conducted in the old EU15 and three new Member States. The European Crime Survey, using the same questionnaire in each country, offers an overview of people’s responses in the different Member States surveyed, and provides for the identification of patterns in responses according to respondents’ characteristics, such as gender and age. Yet, as with the mainstay of victim surveys in the EU, important and particularly ‘vulnerable’ groups – namely, ethnic minorities, foreigners and/or immigrants – are not captured in sufficient numbers by the survey’s sampling frame.
1.2.ii
Launch of Pilot Survey
In response to the lack of European-wide data on ethnic minorities and foreigners/immigrants’ experiences of criminal victimisation, policing and discrimination, both from official and unofficial data sources (including existing victim surveys), the EUMC launched a pilot victim survey initiative in June 2006 on ‘ethnic minorities and foreigners/immigrants’ experiences of criminal victimisation and policing’. The pilot examined the experiences and attitudes of two to three minority respondent groups in each of the six Member States taking part in the research, with the net sample size ranging from 500 to 900 per Member State.
MEMBER STATE / No. of respondents / Groups surveyedAustria / 700 / Turkish, Ex-Yugoslav, Black African
Belgium / 499 / Turkish, North African, Italian, Black African
Bulgaria / 900 / Turkish, Roma
Italy / 603 / Albanian, North African, Romanian, Black African
Romania / 600 / Hungarian, Roma
Slovakia / 605 / Hungarian, Roma
The fieldwork for the pilot was staggered over a period of a few months between the end of 2006 and the beginning of 2007, with the results submitted at the end of May 2007.
Given that a standardised criminal victimisation survey specifically on immigrants and ethnic minorities has never been undertaken in Europe, the primary objective of the pilot exercise was to test sampling frames and questionnaire application in the Member States taking part in the research.
1.2.iii
Population identification and sampling
Unless a survey of the general population is very large, the numbers of ‘minorities’ picked up through standard random sampling procedures will be too small for any meaningful break-down of results according to respondents’ immigrant or ethnic minority background. For example, the British Crime Survey (BCS), which is a rolling survey established in the 1980s, has a household sample from the majority population of more than 55,000 in each 12 month period. Yet, even with this scale of sampling, the BCS has still found it necessary to conduct booster samples of ethnic minorities to extend its coverage of minority respondents who are not readily identified through standard random sampling of the majority population. In comparison, many victim surveys only have a sample in the range of 1,000-1,500 for the majority population only.
Given the inability of standard surveys on the majority population to pick-up enough minority respondents through random sampling, and given the agency’s mandate to collect comparable data on manifestations of racism in the EU, the challenge of ‘how best’ to survey immigrants and ethnic minorities presented itself; hence the need for a dedicated victim survey, which can be initially tested through a pilot.
The core of this challenge lies in how to effectively sample non-majority populations with a view to producing a random sample that can be said to:
(a) represent these groups, and
(b) not be skewed to reflect the experiences and attitudes of a particular sub-population within the non-majority group or groups being investigated.
In this regard, and in an effort to enhance the credibility and quality of results, the use of quota or snowball sampling, which is often used in surveys where criteria about the representativeness of survey respondents is not a top priority, was discouraged in the pilot survey. With this in mind, the FRA pilot survey tested the validity of different probability random sampling approaches on immigrant and minority populations in six EU Member States, in an effort to develop knowledge and expertise for future survey research on difficult to survey groups in Member States which differ from each other in terms of immigrant population, composition and spread.
However, probability random sampling of minority groups, such as immigrants and ethnic minorities, requires information about the numbers of different groups within the general population - at the local level - in order to construct a proper sampling frame.
In this regard the pilot survey faced a number of challenges for sampling, that are primarily a reflection of inadequate population data on ethnic minority and immigrant populations in Member States, which can be broken down at the local level; for example:
§ Member States in the EU collect information on immigrants, national minorities and ethnic minorities (which includes second and third generation citizens) differently. For example, while the UK does collect information on immigrant and self-identified ethnic background, France collects information on immigrant status but not on ethnicity.
§ Member States have different mechanisms in place for collecting data about their resident populations; for example, censuses, population registers, residence registers etc.
§ Some population data collection mechanisms are collected more regularly than others; for example, census data can be 10 years old, while population registers are collected on a regular basis.
§ The level of detail provided at the local geographical level differs greatly from one Member State to the next, with some restrictions in place concerning the level of localised information that can be used for survey research purposes.
1.2.iv
Sampling Approach and Preliminary Findings
The pilot survey tested two main sampling frames in the six Member States – (1) random digit dialling and focused enumeration and (2) random route cluster sampling.
Face-to-face interviewing was used in all countries once eligible respondents were identified.
NOTE: Face-to-face interviewing, rather than telephone interviewing, was adopted for the pilot survey for a number of reasons; for example, the sensitive nature of the survey subject, and to enhance the overall quality of interview responses (including the use of different languages and non-verbal communication).
- Random digit dialling and focused enumeration
§ This method was tested in Austria, Belgium and Italy.
§ The method was tried out in the capital and the second (and sometimes third) largest city with a sizeable immigrant/minority population.
Random digit dialling does not provide an adequate coverage of the target population without being combined with other sampling approaches. Therefore, a two-stage probability random framework was tested using CATI (computer assisted telephone interviewing) for the screening, and focused enumeration at the fieldwork stage.
Potential respondents were screened using CATI and asked to identify whether they considered themselves as belonging to one of the selected minority groups for interviewing in a Member State, and whether they would be willing to take part in an interview at a convenient time. Interviewers were dispatched to interview addresses, and were required to make up to three call backs for conducting an interview.
At the end of each completed interview (both CATI screening interviews and face-to-face full interviews) interviewees were asked to indicate whether any of their neighbours – on the basis of rigid selection criteria – were from one of the groups selected for interviewing in each Member State. Interviewers then called on identified neighbours and asked whether they would take part in the survey. This process – focused enumeration – was applied in turn to each interviewee.[5]
NOTE: Existing random digit dialling based on landlines is increasingly problematic as a questionnaire screening and application method as landline usage declines and mobile phone usage increases, and, as it appears, more people are refusing to take part in surveys over the phone. It is also problematic in a comparative perspective due to different (declining) landline penetration rates in the Member States, and a different demographic and socio-economic profile of mobile-only households. As a method for identifying respondents it is even less reliable when it comes to many minority groups who are less likely to have fixed landlines – such as the Roma. As a reflection of this, surveys on ethnic minority and immigrant populations have to consider alternative random sampling methods – such as random route cluster sampling.
Results of tested sampling approach:
It appears that a combination of CATI screening and focused enumeration (FE) worked well in Austria, as it resulted in the identification of eligible respondents. In comparison, this approach does not appear to have been very successful in Belgium and Italy (for some of the reasons outlined in the previous paragraph). In this regard, an alternative sampling approach or a combination of sampling approaches might be more applicable in some countries other than CATI and FE.
- Random Route Cluster Sampling
§ This method was tested in Bulgaria, Slovakia and Romania.
§ The method was applied across large parts of the country, which was made possible due to the cheaper cost of survey work (face-to-face interviewing) in these Member States.
The Roma were the main group for interviewing in Bulgaria, Romania and Slovakia. Their low landline telephone coverage means that they could not be identified by means of a CATI screener, and therefore random route cluster sampling was adopted as the most appropriate sampling method.