EUROSTAT
Directorate D: External cooperation, communication and key indicators
Unit D-1: Statistical Cooperation with European and Mediterranean countries /
First meeting of the
Technical Coordination Group for the
Censuses in South East Europe
4 - 5 March 2010
Vienna, Grand Hotel, Kärntner Ring 9
Start: 14:00
Minutes
Revised version taking into account the comments received on a draft version circulated 23 March 2010
1
1.General issues
1.1.Welcome and adoption of the agenda
Mr Steurer (Eurostat- chairing) opened the meeting and welcomed the participants. The creation of the TCG has been decided at the last Policy Group Meeting in October 2009 in Budva. Objectives of the group are to issue technical advice and recommendations, to promote the use of harmonised definitions for the censuses in South East Europe and to propose a common approach when enumerating critical topics such as ethnic belonging and religion.
Mr Maragos (DG Enlargement)expressed the support of the European Commission to this meeting. He underlined the necessity that population censuses in candidate and potential candidate countries follow internationally recognised standards and produce high quality results. Census results are an important input in the accession negotiations, e.g. for the calculation of regional indicators that will influence financial allocations.
The agenda was adopted as proposed.
The group decided to make all documents from the meeting public and share them on the CIRCA website.
1.2.Discussion of the mandate of the TCG and key census issues
The mandate of the TCG (document TCG 02/2010) was adopted as proposed.
Mr Aalandslid (Eurostat) summarised the content of the matrix with thekey issues prepared by the countries (TCG doc 03/2010). Concerning planning and management, census laws are adopted in Serbia and Turkey, laws are planned to be adopted by June 2010 inall other countries except Bosnia and Herzegovina where the situation is still uncertain. The budget for the census is secured in Albania, Serbia and in the former Yugoslav Republic of Macedonia. For the other countries,a decision on the budget for the census is conditional to the adoption of the census law.Turkeyfirst has to decide on the census method before being in a position to prepare a budget.Pilot censuses took place in 2009 inAlbania, Croatia, the former Yugoslav Republic of Macedonia, Montenegro and Serbia and in Kosovo[1] in 2008. The 2 other countries, Bosnia and Herzegovina and Turkey, will have a pilot census in 2010.
Considering census methodology,except for Turkey where the census is planned in October 2011 and for Bosnia and Herzegovina where no decision has been taken yet, all countries will use 31 March 2011 as reference date. All countries have a traditional census based on face-to-face interviews, except Turkey where registers will be used. Census maps will be updated in 2010. All countries will include questions on ethnic belonging and religion (except Turkey).
In census logistics and IT, extensive plans have been prepared for the training of enumerators and other census personnel. Croatia has finalised the questionnaire whereas it is still pending in other countries (waiting for consultations with users and results of pilot census). The Optical Character Recognition (OCR) method will be used in Serbia, Albania, Croatia and Bosnia and Herzegovina. Other countries use manual data entry, Turkey has not yet decided on data entry.
The group decided that the census matrix will be updated with new information for the next meeting of the TCG.
All countries gave an overview of the main challenges ahead in the preparation of their population census.
Albania will organise a pilot census at the end of April 2010. Digital maps will be used for enumeration areas. The pilot census will be used to test OCR data entry. The questionnaire is almost ready, consultations with users and stakeholders are going on. A final version will be available by end of June 2010. It is so far not foreseen to translate the questionnaire into any minority languages, as almost all the population speaks and reads Albanian language. A website dedicated to the census is available. There were requests to enumerate Albanians emigrants. Contacts will be taken with the main hosting countries to obtain this information.
The census law is not yet adopted in Montenegro nor the census budget. The last census had already to be postponed from 2001 to 2003 due to the lack of financial resources. It was confirmed that the agriculture census will take place in June 2010 and not in April as planned. This postponement is not anticipated to have consequences for the population census. A roadmap for the census activities has been prepared. A new law on territorial organisation, that may affect the census mapping, is being discussed in the Parliament.Communication towards minorities is organised, and contacts are established with the Minority Council. Mr Maragos (DG Enlargement) insisted that in case of problems and eventual delays in the census taking, Montenegro should immediately inform the Commission so that the necessary measures can be taken.
Serbia does not expect problems in the organisation of the census. In December 2009 the Parliament adopted the census law, unfortunately one week too late to be taken into account in the 2010 budget. This means that census activities have to be financed in 2010 with the reserve budget. The geodetic institute, in charge of producing the maps for the census, is faced with the same problem and has no budget for this year. In addition, the statistical office has to reduce the staff to 485 memberswhich means a reduction by120 persons. This may affect the organisation of the population census.
Turkey plans a register-based census, supplemented with a simple questionnaire covering the whole population. A sample of the populationwill be interviewed with a full questionnaire. The method will be developed based on existing registers (address system, education database, employment database, etc.). The sampling method for the full questionnaire is not yet defined. Turkey will conduct a round of methodological tests in April and May 2010 and based on the results from these tests, the enumeration method will be chosen.
In Kosovosatellite images are usedfor census cartography. Consultations with users and stakeholders are ongoing in order to draft the final version of the census questionnaire.
The pilot census in the former Yugoslav Republic of Macedoniashowed that interviewers should be more trained on questions related to education and employment. The census law is scheduled to be discussed in the Parliament in May 2010. Bilingual questionnaires are used for the population census. This solution has been successfully tested in the previous 2002 census.
No particular problems are to be mentioned in Croatia. Laptops were envisaged for the enumeration but finally traditional questionnaires will be used and prepared in Croatian language and Latin letters. The questions will be translated into 20 languages and will be available in a separate booklet for the respondents during the enumeration. OCR is used for data entry.
The census law is not yet adopted in Bosnia and Herzegovina, due to political disagreement about questions on ethnic belonging and enumeration of Bosnians living abroad.Bosnia and Herzegovina may use the example of Kosovo for the enumeration of the diaspora, i.e. to attach a separate form to the census questionnaire where all members of the family living abroad are listed.A new version of the census law will be discussed with representatives of the Parliament on 9 March 2010. Cooperation between the 3 statistical institutes works well.
2.Place of usual residence
Mr Lanzieri (Eurostat) presented a reminder of this concept, with special focus on the requirements of the EU regulations on population and housing censuses 2011. The group discussed several detailed issues related to the definition of usual residence. Some potentially challenging cases were discussed such as: persons with multiple residences, children, students, temporary workers, diplomatic personnel etc.
The difference between enumeration of the population (meaning the data collection from all persons) and the inclusion in the official population count, based on the concept of usual residence, was clarified.
The group specifically discussed where to enumerate students and workers who study/work in another region of one country or in a different country and regularly return home in weekends. In order to avoid double-counts there need to be clear rules for where to enumerate these groups and these should be harmonised in the region.
It was stressed that the population census is not an ideal tool for the estimation of the number of emigrants. A possible option is to gather statistics on native-born living abroad from the census data in other countries.
3.Enumeration of ethnic/religious/language groups
3.1.Introduction by Donatella Zindato, Istat
Mrs Zindato described the international recommendations on how to measure ethnic/religious/language groups, followed by a review of national data collection practices in the last census round. The UN principles and recommendations state that the decision to collect these data in a census is dependent on the national needs for such data and the suitability and sensitivity of asking such questions in a country’s census. Not all countries collect these data (in the 2000 Census around half of the countries in Europe had questions about ethnicity).
If a country decides to include questions about ethnicity/religion/language the general recommendations are that the questionnaires should include an open question, and interviewers should refrain from suggesting answers to the respondents. Furthermore respondents should be free to indicate more than one ethnic affiliation or a combination of ethnic affiliations if they wish so. The practice of presenting only one pre-coded category (for the most common ethnic group) should be avoided. It is also important that respondents are allowed to indicate “none” or “not declared/prefer not to answer” when asked for their ethnicity and census instructions should specify how to determine the ethnicity of children from mixed couples.
When disseminating statistics on ethnicity/language/religion the categories used in the tables should be extensive and discussed with the minorities themselves. Classifications of ethnic groups at the finest level should be comprehensive and as for aggregated classifications, aggregations should be documented and fully transparent. Results should be presented also for multiple ethnic affiliations and combinations of ethnic affiliations.
3.2.Presentations given by the countries
The introduction was followed by prepared presentations by some of the countries.
The former Yugoslav Republic of Macedonia uses open, not pre-coded, questions on ethnicity/language/religion. The enumerators are instructed to write down the answer as the respondent has declared and respondents are allowed to refrain from declaring ethnic affiliation. For children up to 15 years of age the answer is given by the parents, adoptive parents, or the tutor. Following the presentation there was a discussion about how to assign ethnic affiliation for children of couples with mixed ethnicities. The question of how to disseminate data on mixed ethnicities and whether to list all or to group mixed categories together was also discussed.
The presentation from Montenegro showed the historic development of how questions about ethnicity/religion/language have been covered in earlier censuses.Montenegro also uses open, not pre-coded, questions on ethnicity/language/religion. Based on experiences from the pilot census,Montenegro has chosen to change the question about 'ethnic belonging' to ask about 'national and ethnic belonging' as this was thought to be better understood by the respondents.
Serbia now uses open, not pre-coded, questions in the questionnaires about ethnicity/language/religion. Based on experiences from the pilot, and after consultations with organisations representing the minorities, Serbia has changed the framing of these questions from partly pre-coded questions to open questions Based on consultations with the minority organisations Serbia also plans to translate the census questionnaires into seven languages.Serbia also elaborated on their plans to include the Roma population in the census and hire and train Roma as enumerators. In earlier censuses the Roma community has claimed to be more numerous than what the official statistics imply as they have not declared themselves as Roma in the census.
In the first version of questions on ethnicity,Albania had eight categories to choose from. In the second version Albania has now chosen to have only two: 'Albanian' and 'other'. The questionnaire was modified after discussions with minority groups who claimed that the population was more diverse than what the alternatives listed in the first version indicated and that it was difficult to agree on a fixed number of ethnicities to be listed in the questionnaire.
For collection of data on ethnicity, Croatia will have pre-coded groups: 'Croatian', 'other' and 'not declared' as well as an open field. The questions on language will have only two categories 'Croatian' and 'other' with open question, and for religion: “Catholic”, 'other' with open question, “Atheist” and “Not declared”.Croatia will use OCR and automatic text recognition for the coding of the open categories.The questions will be translated into 20 languages and will be available in a separate booklet for the respondents during the enumeration.
Kosovo has eight pre-coded alternatives representing what is thought to be the largest minority groups. It was unclear to what degree minority organisations were consulted in the creation of ethnic categories.
The issue of how to measure second languages spoken in the region was also discussed in this session.
Conclusions:
As these questions can be potentially sensitive, the free and open declaration of the respondents is of essential importance. Interviewers should refrain from suggesting answers to the respondents and the census instructions should emphasise that it is voluntary to answer these questions.
When collecting data on ethnicity, the group recommended using a number of pre-coded categories covering the large majority of the population. These categories should be agreed in consultation with minority groups in the country. From a data entry point of view it is a clear advantage to have pre-coding. If it is not possible to agree upon a fixed number of categories, the group recommended an entirely open question, although the approach with open questions creates challenges with data entry and special attention and resources are needed in the scanning/data entry phase.
Countries should allow for mixed categories or multiple answers in the response on ethnic affiliation. In the dissemination process, multi-ethnic categories should also be published, as well as the categories ‘no ethnicity’ and ‘not declared/prefer not to answer’. The practice of grouping mixed categories into one single category should be avoided, even if some groups will be small. In case of open questions, it should be made clear to the respondent that he/she has also the possibility to answer with more than one category.
The ethnicity of children below 15 years of age should be decided by the parents or tutor, this is especially important for children of mixed ethnicity couples. In many countries in the region information about children's ethnicity is essential for planning purposes (schools etc.).
In order to ensure transparency and the correct understanding of the questions it is recommended that representatives of ethnic, language and religious groups are consulted during the drafting of the census questions, the definition of classification rules and the conduct of the census. All the census material (including instructions manuals) should be made available in all the principal minority languages.
4.Experiences with the Pilot Surveys
4.1.Introduction by Gabor Rozsa, Icon
As an introduction Mr Rozsa pointed out the ideal situation for census tests. Most of the census phases and elements should be tested: mapping, addresses, delineation of enumeration areas (EA); questionnaire form and content, clarity of questions; duration of enumeration; instruction manuals; training of (non-professional) interviewers; logistics; data entry, coding and processing (incl. tabulation). Mr Rozsa continued with a summary of the progress in the census preparations in the beneficiary countries.
4.2.Experiences/lessons learnt from the pilots
Following the introduction the beneficiary countries presented experiences/lessons learnt from their pilots.
Serbia made a presentation on their two pilot surveys, the first a general one on a reduced sample (50 EAs instead of the planned 250), testing questionnaires, instruction manual, data capture, census cartography; the second (with funding by SIDA) to test the possibility to include Roma in census activities; one of the main points is to consolidate cooperation with the Geodetic Authority, aiming at digitising maps until November 2010, in line with the relevant Law.
The pilot census in Montenegro was run from 01-15 June 2009. It was divided into nine different phases (Census Questionnaires, Guidelines for enumerators, sample for pilot census, field work, data entry software, data entry, analysis of data, final data update and mapping). It is characterised by nine sectors (Demographic, Geographic, Migration, Educational,Economic, Ethno-cultural, Disability, Family and household and Dwellings. At this stage Montenegro is analysing the data of the pilot census and updating all maps.
In Turkey, various quality checks were conducted of the National Address Database (NAD) and the Address Based Population Registration System (ABPRS). The first quality check study was carried out in August 2008. The second quality check study was carried out November-December 2009. Field applications of both studies were realised by TurkStat Regional Offices.
A first pilot census will be conducted in April-May 2010. In the pilot census all above aspects will be tested. Related to the field application methodology, data collection method (canvasser or householder/self completion method (mail out, mail back, via internet, via governmental organisations) and data capture method (keyboard data entry, optical mark reading, optical character reading and imaging processing) will be tested. Related to the questionnaire sampling (selecting a systematic subset of households to receive long questionnaire or selecting a sample of enumeration areas to receive long questionnaire) will be tested and estimation level of the economic, demographic and social indicators will be decided.