EU-MIDIS II The Second European Union Minorities and Discrimination Survey Surveying Roma Jaroslav Kling Research Officer April, 2018
EU-MIDIS II
The Second European Union
Minorities and Discrimination
Survey
Surveying Roma
Jaroslav Kling
Research Officer
April, 2018
2
Council Regulation (EC) 168/2007
▪ to provide assistance and expertise on fundamental rights issues to the European Union institutions and the Member States, when they implement European Union law
▪ to collect, record, analyse and disseminate relevant, objective, reliable and comparable information and data on fundamental rights issues in the EU
▪ to promote dialogue with civil society, in order to raise public awareness of fundamental rights and actively disseminate information about its work
FRA’s mandate
3
FRA’s surveys
▪ EU-MIDIS: European Minorities and Discrimination survey (2008) – EU-27 ➢ 23,500 ethnic minorities & immigrants
➢ 5,000 majority population in 10 EU MS – random sample
▪ Roma pilot survey (2011) – 11 EU MS➢ 22,000 respondents – random sample
▪ Survey on Discrimination and Hate Crime against Jewish people (2012) – 8 EU MS➢ 6,000 respondents – online (opt-in)
▪ LGBT survey (2012) – EU-28➢ 93,500 respondents – online (opt-in)
▪ Violence against Women survey (2012) – EU-28➢ 42,000 women – random sample (general population)
▪ EU-MIDIS II (2015-2016) – EU-28➢ 25,500 ethnic minorities & immigrants (including Roma) – random sample
▪ Fundamental Rights survey (2017-2019) – EU-28 ➢ general population
➢ Pre-test and feasibility study (2016); Pilot (2017-2018)
▪ Second Survey on Discrimination and Hate Crime against Jewish people (2018) online (opt-in) – 13 EU MS
▪ Roma and Travellers survey 2018 – 6 EU MS
4
EU-MIDIS II survey:
Roma – at a glance
➢ Roma: “umbrella” term that covers a wide diversity of Roma,
Sinti, Kale and related groups, including self-identifying Gypsies
➢ In 9 MS: BG, CZ, EL, ES, HR, HU, PT, RO, SK
➢ 7,947 selected Roma respondents
➢ in households with 33,785 individuals
➢ Based on self-identification, in areas 10% density and more
➢ Random route, and adaptive cluster sampling in low density
areas
See EU-MIDIS II Technical report – available online – detailed
description of the survey design and methodology
5
EU-MIDIS II survey:
countries and target groups
6
Definition of target population:
Roma
The survey sampled individuals aged 16 years and older who:
• self-define as Roma– When contacting the household: “Does anyone in the household meet - ALL
of the following criteria: a) Aged 16 or over; b) Is Roma?”
– Listing of all HH members who fulfil the above mentioned criteria
– Before the interview with respondent starts, instruction for the interviewer: PLEASE CONFIRM THAT THE RESPONDENT SELF-IDENTIFIED AS BELONGING TO THE TARGET GROUP ROMA ;
• whose usual place of residence is in the EU Member State surveyed;
• have been living in private households in the EU Member State surveyed for at least the last 12 months
For the purpose of the survey, the term ‘Roma’ refers to persons who self-identify as ‘Roma’ or as one of the other groups which are subsumed under the term ‘Roma’ (BG-show card of various groups).
7
Sample size:
Roma
Country Estimated Roma
population (all ages) as
per the sampling frame
Target sample size
(N)
Achieved sample
size
(respondents/
households)
Household
members
BG 325,343 1,050 1,078 4,278
CZ 237,865 850 817 3,245
EL 62,920 500 508 2,719
ES 550,000 700 776 3,059
HR 16,975 500 538 2,800
HU 315,583 1,050 1,171 4,941
PT 39,233 500 553 1,992
RO 621,573 1,450 1,408 5,764
SK 402,590 1,150 1,098 4,987
Total 2,572,082 7,750 7,947 33,785
8
Sampling strategies:
Roma
➢ Multi-stage area sampling, with primary sampling units (PSUs)
selected at the first stage.
• For the second stage sampling within sample areas the addresses identified via
random route plus Adaptive cluster sampling to increase the efficiency of the sample
(in 4 countries)
• Random selection of the respondent
➢ To improve the efficiency of the screening and fieldwork related efforts
and costs, country sampling plans were:
➢ optimised through stratification of the frame according to degree of concentration of
the target population (i.e. percentage of target population in the total population of
the PSU)
• excluding empty or low concentration strata from the sample by setting a minimum level of concentration ‘cut-off’
• oversampling more concentrated strata; this resulted in lower sample efficiencies
9
Sampling strategies by country:
RomaCountry Sampling method Cut-off level and selection criteria Population coverage
after cut off and
exclusion rules
BGRandom route with ACS (2b: indirect, multi-stage stratified)
Areas with > 10 % concentration and areas with > 20 Roma households
At least 70 %
CZ
Random route (2b: indirect, multi-stage stratified)
Areas with > 139 Roma persons (~ 30 Roma households) and, among areas with 139–500 Roma persons, areas with > 10 % concentration
79 %
ELRandom route (2b: indirect, multi-stage stratified)
Selection of regions of Attica, Western Greece, Central Macedonia, Thessaly
64 %
ESRandom route (2b: indirect, multi-stage stratified)
Areas with > 200 Roma households and < 10 % concentration
65 %
HRRandom route with ACS (2b: indirect, multi-stage stratified)
Areas with > 200 Roma persons (~ 40 Roma households)
68 %
HURandom route with ACS (2b: indirect, multi-stage stratified)
Areas with > 10 % concentration 61 %
PTRandom route (2b: indirect, multi-stage stratified)
Areas with > 60 Roma households and > 5 % concentration
70 %
RORandom route with ACS (2b: indirect, multi-stage stratified)
Areas with > 30 Roma households or > 10 % concentration
64 %
SKRandom route (2b: indirect, multi-stage stratified)
Areas with > 30 Roma households or > 10 % concentration
75 %
10
PSUs location:
Roma
11
Sampling frames (PSUs):
Roma
Country Sample frame/ source of data
used for sampling
Sample frame and selection stages Stratification variables
BG Census 2011 Census areas (PSUs) NUTS3, rural/urban
CZ Census 2011 & Total number of
Roma estimates
Municipality (PSUs, larger settlements
divided)
NUTS3, urban/rural
EL Roma Pilot Survey 2011 Municipality (PSUs, larger settlements
divided)
NUTS2, rural/urban
ES Gitanos.org 2007 data / 2015
update
Municipality ‘municipio’ (PSUs, larger
settlements divided)
NUTS2, settlement size
HR Census 2011 Settlements ‘naselje’ (PSUs, larger
settlements divided)
NUTS3, urban/rural
HU Census 2011 Enumeration district (PSUs) NUTS2, rural/urban
PT Roma Pilot Survey 2011 Municipality ‘MUNICÍPIO’ (PSUs, larger
settlements divided)
NUTS2, settlement size
RO Census 2011 Census districts (PSUs) NUTS3, urban/rural
SK Atlas 2013 Municipality (PSUs, larger settlements
divided)
NUTS3, urban/rural
12
Weighting
• Design weights – PSU selection stage
– Address selection stage
• Adaptive cluster sampling adjustment
– Respondent selection stage
• Non-response– Neighbourhood characteristics at household level
– Respondent characteristics at individual level
• Post-stratification– Not possible due to unavailability of reliable data
– Only for cross-country weights
13
EU-MIDIS II survey:
Questionnaire
14
Indicators to meet the 2013 Council Recommendation’s
goal for effective Roma integration
Roma: EU-MIDIS II
- approximation of standard indicators used by
Eurostat
- indicators specifically designed
General population: Eurostat – standard indicators
EU-MIDIS II survey:
Analysis - Roma
15
• Further development of methods in 2016
• Creation of additional weights for comparison
EU-MIDIS II survey:
Roma - comparison with 2011
16
• Importance of involving local contractors early
• Use of qualified interviewers + pay scheme taking into account the screening efforts
• Importance of involving local community
• Adaptive Cluster Sampling worked well in some countries, but challenging to implement
• Organisation of fieldwork in stages and centralised monitoring of fieldwork (under/overestimation of response rates
EU-MIDIS II survey:
Some lessons learned
17
EU-MIDIS II survey:
Publications
Publications:
• Roma Selected Findings - December 2016
• Muslims Selected Findings - September 2017
• Main results - December 2017
• Technical Report – December 2017
• Online visualisation – December 2017
FRA data explorer
Upcoming
• Thematic reports
• Microdata available
18
EU-MIDIS II
Selected results
_____________________
Living conditions (Roma)
19http://fra.europa.eu/sites/default/files/fra_uploads/fra-2018-anti-gypsyism-barrier-roma-inclusion_en.pdf
At risk of poverty
20
Paid work
http://fra.europa.eu/en/publication/2016/eumidis-ii-roma-selected-findings
21http://fra.europa.eu/sites/default/files/fra_uploads/fra-2018-anti-gypsyism-barrier-roma-inclusion_en.pdf
Youth neither in employment
nor in education
22http://fra.europa.eu/sites/default/files/fra_uploads/fra-2018-anti-gypsyism-barrier-roma-inclusion_en.pdf
Youth with at most lower
secondary education and not in education
23http://fra.europa.eu/sites/default/files/fra_uploads/fra-2018-anti-gypsyism-barrier-roma-inclusion_en.pdf
Basic sanitation
24http://fra.europa.eu/en/publication/2016/eumidis-ii-roma-selected-findings
Quality of dwelling
25http://fra.europa.eu/en/publication/2016/eumidis-ii-roma-selected-findings
… but also
Discrimintation
26http://fra.europa.eu/sites/default/files/fra_uploads/fra-2018-anti-gypsyism-barrier-roma-inclusion_en.pdf
Harassment
fra.europa.eu
Thank you!