«NTEGRATED I H ED OUSESUOLYS H RV AMONG OPULMPLTIGONS MA PPROACH TO SAA IN RO ONE POSSIBLE AP NDP-WORLD ...»
MA PPROACH TO SAA IN
ONE POSSIBLE AP NDP-WORLD BANK-EC
USED IN THE U ROMA SURVEY 2011
Roma Inclusion Working Papers
ONE POSSIBLE APPROACH TO SAMPLING
USED IN THE UNDP-WORLD BANK-ECREGIONAL ROMA SURVEY 2011 Roma Inclusion Working Papers UNDP Europe and the CIS Bratislava Regional Centre, 2012 This publication was prepared with the support from the European Union © UNDP, 2012 ISBN (printed version): 978-92-95092-52-5 ISBN (electronic version): 978-92-95092-53-2 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted, in all forms by any means, mechanical, photocopying, recording or otherwise prior permission.
Cover design and layout: Yassen Panov Print: Valeur, s. r. o., Slovak Republic The content of this publication can in no way be taken to reflect the view of UNDP or the European Union.
To be cited as: Ivanov, A., Kling, J. & Kagin, J. (2012). Integrated household surveys among Roma populations: one possible approach to sampling used in the UNDPWorld Bank-EC Regional Roma Survey 2011. Roma Inclusion Working Papers. Bratislava: United Nations Development Programme.
Authors Andrey Ivanov, PhD, Senior Policy Advisor at the Bratislava Regional Centre, UNDP Jaroslav Kling, Policy Analyst at the Bratislava Regional Centre, UNDP Justin Kagin, PhD candidate at the University of California, Davis.
Acknowledgements The authors and the team involved in this research are grateful to Balázs Horváth for his support and the methodological guidance throughout the entire project.
The authors are grateful to Christian Brüggemann, Ioannis Dimitrakopoulos, Sevinç Rende and Ilona Tomova for the valuable comments received on various drafts of this paper. We are also grateful all the colleagues involved in the 2004 and 2001 surveys for their dedication, enthusiasm and ideas that made this endeavor possible. It should be noted that the all this wouldn’t have been possible without the nancial support from the DG “Regional Policy” of the European Commission, and the cooperative attitude from the colleagues in the Fundamental Rights Agency, the World Bank, IPSOS (who administered the 2011 UNDP survey), GALLUP Europe (who administered the 2011 FRA surveys) and TNS BBSS (who administered the 2004 UNDP survey).
Table of Contents Introduction
Part 1: Who’s Roma? De ning the universe of study
Between self-identi cation and external identi cation
The second best option: focusing on the communities where Roma are overrepresented
Part 2: The methodology and sampling procedure of the 2011 survey............... 11 Experience with Roma targeted surveys
The 2011 survey: the sampling
Part 3: The data-set used in the UNDP working papers
Introduction The Regional Roma Survey 2011 was completed in cooperation with the United Nations Development Programme (UNDP), World Bank, the European Commission (EC) and in coordination with the European Union’s Agency for Fundamental Rights (FRA).
Two parallel and complementary surveys were carried out in 2011 in an e ort to map the current situation of Roma in the EU: One was focusing on social and economic development aspects and carried out by the UNDP and the World Bank (funded by the European Commission,1 UNDP and the Nordic Trust Fund at the World Bank), and one focusing on the ful llment of key fundamental rights carried out by the EU Agency for Fundamental Rights (FRA).
The UNDP/WB/EC survey was conducted in May-July 2011 on a random sample of Roma and non-Roma households living in areas with higher density (or concentration) of Roma populations in the EU Member States of Bulgaria, Czech Republic, Hungary, Romania, Slovakia, and the non-EU Member States of Albania, Bosnia and Herzegovina, Croatia, FYR of Macedonia, Montenegro, Republic of Moldova and Serbia. In each of the countries, approximately 750 Roma households and approximately 350 non-Roma households living in proximity were interviewed.
The FRA survey was conducted in May-July 2011 on a random sample of Roma and non-Roma households living in areas with concentrated Roma populations in the EU Member States of Bulgaria, Czech Republic, Hungary, Romania, Slovakia, France, Greece, Italy, Poland, Portugal and Spain. In most of the countries the FRA sample consists of 1,100 Roma households and approximately 500 non-Roma. In France, about 700 gens du voyage and 300 Roma Migrant households in the greater Paris area were surveyed.
In Poland and Italy, the sample size was reduced to 600 and 700 Roma households respectively In total 16,648 persons (11,140 Roma and 5,508 non-Roma persons) were interviewed.
The survey questionnaire was designed jointly by a team from UNDP, the World Bank and the FRA. Each survey used di erent questions and a core common component composed of key questions on education, employment, housing, health, free movement and migration issues, and discrimination experiences. The questions in the common core were identical.
The UNDP/WB/EC survey was implemented by the IPSOS polling agency and the FRA survey through Gallup Europe. Both surveys applied the same sampling methodology in countries of overlap allowing for the development of a common dataset on core indicators and ensuring comparability and consistency of results. The combined UNDP/ WB/EC Regional Roma Survey 2011 and FRA Roma Pilot Survey 2011 has a total of 1/ Directorate General for Regional and Urban Policy of the European Commission funded the survey in the EU Member States.
20,018 Roma households (87,717 household members living in these households) and 9,782 non-Roma households living nearby (28,214 household members) covering 18 European countries.
This paper describes the methodology, dataset, and methods used behind the UNDP/ World Bank/EC Regional Roma Survey 2011 while an additional paper by the FRA will provide details of the FRA Pilot Roma Survey. The two surveys were performed in conjunction and have similar methodologies, however some questions were di erent and their geographic scope of study also di ered. The UNDP 2011 Survey covers 11 Eastern European countries including ve EU countries; Albania, Bosnia and Herzegovina, Montenegro, Croatia, Macedonia, Moldova, Serbia and the EU countries of Bulgaria, the Czech Republic, Slovakia, Romania, and Hungary. The FRA Roma Pilot Survey encompasses 11 EU countries; Greece, Spain, France, Italy, Poland, Portugal, including the ve covered by the UNDP 2011 Survey (FRA 2012). Forthcoming UNDP publications will note when they use the pooled data set from both surveys or data exclusively from the UNDP/World Bank/EC Regional Roma Survey 2011 when survey questions di ered.
Five European Union countries - Bulgaria, the Czech Republic, Slovakia, Romania and Hungary - represent the area of overlap.
The paper is divided into three parts. The rst part provides an idea of the general problems related to sampling of Roma populations stemming from the uid nature of Roma identity. The second part gives a background of the context of the survey and describes its methodology, the sampling procedure and the eldwork. The third part gives information on the data set produced by the survey and used for the UNDP working papers.
1 Who’s Roma?
De ning the universe of study Roma inclusion2 is increasingly visible on political agendas both of governments and international institutions. It entails many challenges which are usually reduced to its practical aspects - what to do and how to do it, so that real progress is achieved. Against the background of those practical aspects, one issue is gaining less attention than it should – a clear de nition of who are the “Roma” that are to be “included”? The task of de ning becomes even more challenging if we take into consideration the fact that both “Roma” and “inclusion” are vague – and interrelated – concepts. They are intellectual and political constructs to which di erent people (Roma or non-Roma, politicians or ordinary citizens) usually attribute di erent meanings. In addition, given the interrelated nature of the two concepts, the practical content of “inclusion” varies depending on the meaning attributed to “Roma” (Ivanov, 2012).
De ning the universe of study is even more critical in sampling research. “A sample of an unde ned universe” is an oxymoron. Strictly speaking, one cannot de ne a sample not knowing exactly what is being sampled. In the case of the Roma, however, the task of precisely de ning the universe presents important challenges. Census data (the source usually used by the government which needs formalized data sources) notoriously and signi cantly di er from “experts’ estimates”. On average, the di erence can be as large as four-fold.3 Depending on the speci c circumstances, it can change in scope and coverage. If the circumstances suggest that there is a certain risk associated with “being Roma”, the estimates get lower; if there are some potential bene ts (preferential access to services for example) – the estimates get higher.4 The need for more precise statistics on Roma has become ever more acute, as governThe terms – “integration” and “inclusion” – are often used as synonyms despite the important di erences between the two. The former entails the involvement of the representatives of the minority in dominating structures with limited elements of diverse identity retained (and usually the resulting involvement is limited as well). In that case the external system is more tolerant to diversity but remains static – it accepts certain elements of diversity but this acceptance doesn’t entail change in the system itself. “Inclusion” on the other hand entails a dual track process in which both the minority and the system adjust – the former preserves the core markers of its identity (but not all) and the latter becomes responsive and accommodative to those elements of unique identity. It’s interesting to note that in everyday policy jargon “inclusion” is also reserved for people with disability or indigenous groups, whereas “integration” is attributed to migrants. In the EU policy language the increasingly dominating term “integration” (the EC is calling for “National Roma Integration Strategies” for example) is used although the real meaning is “inclusion” (see UNDP 2012, pp. p-6). However the di erences between the two terms– as important as they might be – go beyond the direct scope of this paper.
ments have begun to develop special programs related to Roma (Petrova 2004, p. 5).
This constitutes a fundamental challenge because demographics is at the core of both sampling and of the denominator of any indicator (even if the number of Roma unemployed could be estimated, the “unemployment rate” would di er depending on the estimated size of the Roma labor force). What is more important, precise (to the extent possible) numbers are needed for practical responses: it matters whether you are planning a resource allocation for 100,000 or for 400,000 people to be targeted in whatever intervention.
Between self-identi cation and external identi cation
Two approaches are usually used to de ne one’s a liation (ethnic or other): self-identi cation or external identi cation. One is a “result of choice”, the other – of “ascription” (Rughiniş 2011). In the rst case the individual respondents are asked directly “To what ethnic group do you belong?” or indirectly “With which group/culture/community do you a liate?”5 In the second case, outsiders make a judgment on the identity of the person or the entire community – “Is he or she a Roma?” or “is the neighborhood over there a Roma one?”. Both approaches, which are used in various surveys, produce different results (the universe of “self-identi ed Roma” is often smaller than the “externally identi ed as Roma”, for various reasons. One – and most obvious – is the stigma associated with “belonging to Roma” and the experience of past and present misuse of ethnic data (Makkonen 2007, p 50). But the choice of di erent identity is often driven by more pragmatic reasons. It is easier to integrate with other minorities constituting a majority at the local level (like the Turks in Bulgaria – in which case the Roma living in Turkish-dominated settlements usually self-identify as Turks). Another could be better protection of group rights (for example, in countries such as Greece, where Roma would self-identify as ‘Muslims’ rather than Roma, since only Muslims have speci c minority rights).
Thus each of the two approaches is reasonable but re ects part of a complex reality.
This is why they are often used as complementary. The censuses are the largest-scale data collection e orts that rely on self-reported a liation. It is usually thought that 3/ One of the most frequently cited source of population estimates on “Roma” is the Council of Europe. It provides estimates of the “Roma” population for countries of CoE area, for EU member states and Europe in total. The ratio between the “minimum estimate”, “maximum estimate” and “average estimate” to and the o cial census data (for countries which register main ethnicity in their censuses) is respectively 2.7, 5 and 4. www.coe.int/t/dg3/romatravellers/Source/documents/stats.xls. See also Liegeois 1997.
4/ This phenomenon called “strategic ethnicity” is not unique for Roma. Examples as distant as Jews in 1939 Germany and being in a train with hooligans from the oppositional football team share the same logic.
5/ In some countries (like the US) individuals can choose multiple identities (as in the case of children from mixed marriages). This approach however is not used in countries with large Roma minorities.
INTEGRATED HOUSEHOLD SURVEYS AMONG ROMA POPULATIONS