Page 1

Fifth Regional Workshop on MED-HIMS Methodology and Tools

Meeting jointly organised byMEDSTAT III, the World Bank, UNFPA and UNHCR and in collaboration with DoS-Jordan

Landmark Hotel, Amman

4-8March 2012

Final Report

(7 May 2012)

A.Introduction

  1. The MED-HIMS Regional Workshop held on 4-8March 2012 inAmman was jointly organised by MEDSTAT III, the World Bank, UNFPA and UNHCR for discussing with the Mediterranean Partner Countries (MPCs) and other international institutions the latest versions of MED-HIMS Manual 1 - Model Questionnaires (MQs) and other data collection manuals as well as the sampling issues linked to the prospects of coordinated national surveys. The meeting followed a series of prior workshops on developing the MQs and several technical meetings and distance consultationsbetween the experts and collaborating institutionsand countries held since February 2009.
  1. The meeting was the first event organised in the framework of MED-HIMS Phase One, the project recently launched for coordinated implementation of HIMS operations in Egypt, Jordan, Lebanonand another country to be defined in 2012-2014. It was attended by the statistical experts of the National Statistical Offices (NSOs) and partner institutions of Algeria, Egypt, Jordan, Lebanon, Morocco, the occupied Palestinian territory and Tunisia and experts representing EUROSTAT, MEDSTAT III, the World Bank, UNFPA, UNHCR, ESCWA, ILO, IOM and the League of Arab States.Samir Farid, the MED-HIMS Chief Technical Adviser and main author of MQs and data collection manuals generally chaired the meeting.
  1. This report makes reference to the latest versions of Manual 1 – Model Questionnaires, Manual 2 – Survey Design and Organisation and Manual 3 – Instructions for Field Supervisors which were distributed to the participants before and during the meeting.

B.Opening session

  1. The meeting was opened by Mr. Fathi Nsour, the Director General of the Jordanian Department of Statistics (DoS),who emphasized the needsfor measuring international migration in the region and particularly in Jordan, the usefulness of detailed data from the MED-HIMS surveys, the wide national support for the Jordan-HIMS, and DoS commitment and readiness to start the survey within the next few weeks. On the other hand, the representatives and expertsof international institutions and projects supporting the MED-HIMS briefly reported on their plans for promoting the national surveys, coordination, and fund raising.
  1. Representatives of other international organizations indicate their enthusiasm for the projects to improve the measurement and understanding of international migration in the region and willingness to be part of the MED-HIMS initiative, although they would not be able to provide financial support.
  1. Further interventions highlighted the high relevance of the proposed survey programme and the need for exchanging experiences between participating countries.
  1. Finally, following the request of some national experts, it was decided to devote extra time to discussing the sampling design for the MED-HIMS surveys.

C.Review of Manual 1 - Model Questionnaires (MQs)

  1. The meeting considered the following set of MQs:

MQ-1: The Household Questionnaire

MQ-2: Individual Questionnaire for Out-migrant

MQ-3: Individual Questionnaire for Return Migrant

MQ-4: Individual Questionnaire for Non-migrant

MQ-5: Individual Questionnaire for Forced Migrant

MQ-6: The Household Socio-economic and Environmental Conditions Questionnaire

MQ-7: The Community Characteristics Questionnaire

MQ-7 was provided in draft form for discussion.

  1. The main points of discussionamongexperts concerned the following:
  2. A series of clarifications on methodological aspects and decisions. These includesreference to the situation of out-migrants at the time of first migration since the start of the reference period as well asat the time of the last migration, the number of individual questionnaires to administer in each sample household, and the linkage of questionnaires within households.
  3. Eligibility criteria for assigning the individual questionnaires.
  4. The adoption of international classifications, coding or thresholds for questions concerning occupation and education.
  5. The need of considering the varying frequency of migratory flows such as the increasing returning migration (e.g., from Libya to both Egypt and Tunisia).

c.Whether there should be any distinction between ‘core’ and ‘optional’ questions.

  1. The opportunity to add more questions to the MQ-5 on forced migration.
  2. The need to shorten and simplify MQ-7, and still consider whether it would be used in some contexts/countries.
  3. The composition and size of field teams for optimal efficiency of field work, taking into account the expected duration of interviews, and whether each field team should have both a field supervisor or and a field editor, or whether most field editing would be done later in the central office. .
  4. The modalities of the pre-tests and the expectation of using for estimating the data collection workload, planning the size and composition of field teams, and planning the total duration and cost of the total survey in each country, taking into account country conditions and normal practices of statistical offices.
  5. The progress made on translating the latest versions of MQs from English to Arabic by CAPMAS and DoS.
  6. The sampling plan and its suitability to identify the various target groups in the survey
  1. Among the main conclusions are the following:

a. MQ-1: The Household Questionnaire:

-To restructure MQ-1: The Household Questionnaire: would be altered to take into account the household roster would have already been initiated by a household lister to classify the household as a non-migrant household, an out-migrant household, and a return migrant household to draw the sample. Thus the Household Questionnaire wouldestablish eligibility for the various individual questionnaires and complete the household roster information on marital status, education and economic activity.

-To add to the first part of MQ-1 the questions of the brotherhood and sisterhood method for the indirect estimation of the ‘stock’ of international migration.

-To add at the end of the first part of MQ-1 detailed instructions to the interviewer on the eligibility for the various individual questionnaires (see below).

-To include deterioration in the natural/environment as a possible cause for forced migration

b. MQ-2, MQ-3 and MQ-4:

- A number of changes in the wording of questions were agreed upon during the meeting.

c. MQ-5: Forced Migrant Questionnaire:

-It was agreed to keep the MQ-5 inits current version as a ‘core’ while additional questions dealing with specific situations could be added at the country level according to data needs.

d. MQ-6: Household Characteristics:

-To discuss and decide, from distance,on some adaptations to the MQ-6.

e. MQ-7: Community Characteristics Questionnaire:

-The meeting reviewed the second draft of this questionnaire. Some participants questioned the relevance of gathering community data in a migration survey, arguing that the decision to migrate is made mainly on an individual and household basis.

-DoS (Jordan) said they would probably not use MQ-7 in any case, as the rural sector represents only 18% of the total population.

-CAPMAS (Egypt) and other countries said they would not use the MQ-7 in its current form, that must be shortened and simplified.

-CAS (Lebanon) said they would like to gather community data in their MED-HIMS survey but that major changes needed to be made in both content and size.

-Discussion ensued on the nature of the migration decision, whether it would be useful to collect community-level data at all. The international expert explained that it is now widely recognized in the research world that contextual factors such as community characteristics have effects on individual decisions such as migration, that statistical models that fail to include and thereby control for such factors are likely to lead to biased estimates of the effects of the other individual and household factors.

-There was also discussion of the size of communities for which MQ-7 should be applied, and agreement that countries could have some flexibility to decide the threshold for community/population size, to apply it in only rural communities, though some country participant recommended it also be applied in urban areas in urban neighborhoods.

-In conclusion, there was a general consensus that major changes needed to be made to the MQ-7 draft, particularly to shorten and simplify sections on population/migration,agriculture and economic activities, as well as precode more responses After this, countries would re-assess their interest in applying the questionnaire but it would not be obligatory for countries to include it.

f. Other conclusions

-To noto pre-code questions on occupation; i.e. to leave the occupation questions as ‘open questions’ (then to code in the office using3 digits) in all model questionnaires.

-To possibly pre-code all the questions on economic activity by economic sector.

-To revise and finalise as soon as possible the MQs according to the outcomes of the workshop, including circulation of an adapted version of MQ-7.

-To review and update the Arabic translation of the MQs prepared by CAPMAS and DoS.

D.Manual 2 – Survey Design and Organisation

  1. The contents of this manual are as follows:

1.Objectives

1.1Scope and Purpose of the MED-HIMS

1.2LongRange Objectives

1.3Immediate Objectives

1.4Use of Manual

2.Background and Justification for the Project

2.1Changing MENA Demographics

2.2The Need for Scientific Data on International Migration

2.4Potential Users of MED-HIMS Data

3.The MED-HIMS Instruments

3.1Strategy

3.2The MED-HIMS Basic Documentation

3.3The MED-HIMS Model Questionnaires

3.4Concepts and Definitions

3.5Outline of the Model Questionnaires

3.6Sampling Strategy

3.7MED-HIMS Outputs and Examples of Data Utilization

3.8Outputs and Dissemination

4.Organization of MED-HIMS at the International Level

4.1Strategy

4.2Structure

4.3Functions of MED-HIMS Project Implementation Unit

4.4Technical Assistance

5. Implementation of MED-HIMS at the National Level

5.1General Strategy

5.2The National Executing Agency

5.3The National Project Implementation Unit (NPIU)

5.4Basic Survey Decisions

5.5Adaptation of the Survey Documents

5.6The Pre-test

5.7Sample Design and Implementation

5.8Training of Field Staff

5.9Publicity

5.10Listing and Data Collection

5.11Data Management

5.12Data Analysis and Reporting of Results

5.13Dissemination of Research Findings5.14Advanced Analysis

5.15Regional Comparative Analysis

6.Project Monitoring and Evaluation

7.Ethical Statement and Data Protection

8.National Follow-Up Action

Annex 1.General Form of National Survey Proposal

Annex II:Logical Framework Matrix for MED-HIMS at the National Level

Annex III: Timetable for MED-HIMS Activities at the National Level

  1. The manual includes also a series of exhibits focusing on the main responsibilities of the fieldworkers and other topics.
  1. The meeting reviewed the draft Manual 2 and agreed on the eligibility criteria for the various individual questionnaires. It was also agreed to expand the section on training and to add examples of timetable and guidelines forrevising the questionnaire, conducting a the pre-test and training field staff.
  1. Chapters 4 and 5, describing the organization of the survey both at international and national level would be revised in the future to better reflect the reality of implementing the survey, following the actual experiences of the first two countries,Jordan and Egypt.

E.Manual 3 – Instructions to Field Supervisors

  1. The contents of this manual are as follows:

1.Introduction

2.Preparing for Fieldwork

3.Organising and Supervising Fieldwork

4.Monitoring Interviewer Performance and Ensuring Data Quality

5.Dealing with Non-Response

6.Importance of Confidentiality and Data Security

7.Administrative procedures

Annex 1. Statement of Confidentiality

Annex 2. Letter of Authorization

Annex 3. Supervisor’s Control Sheet

Annex 4. Interviewer’s Cluster Control Sheet

Annex 5. Summary of Field Work Results Sheet

Annex 6. Checklist for Observing Interviews

  1. There was some discussion between experts with regard to the optimum number of interviewers in a team, the extent of editing the questionnaires in the field, and the transportation.It was decided to conduct full editing of completed questionnaires in the field, to take full advantage of the possibility of returning to the household whenever there is missing or dubious data. Accordingly, it was agreed that each field team could includeaField Supervisor,Field Editor and 6 interviewers (or alternatively one supervisor-editor and four interviewers).It was also agreed to add to the draft manual a detailed section on editing instructions and to add a clear reference to the ‘Editors’ in the manual.

F.Manual 4 – Instructions to Interviewers

  1. This manual is currently being prepared in Arabic. The meeting was briefed about the content of the manual, which includes three main parts:

Part I: Conducting the Interview

-Introduction to the MED-HIMS programme

-Interviewer’s Duties

-General Interviewing Procedures

-Specific Field Procedures

-Outline of the Model Questionnaires

Part II: Completing the Questionnaires

-MQ-1

-MQ-2

-MQ-3

-MQ-4

-MQ-5

-MQ-6

-MQ-7

Part III: Control Sheets

-Critical Items Checklist

-Filling in Control Sheets

G.Sampling

  1. The following summarizes considerablediscussion among the experts had in the workshop and an experts’ meetings held back-to-back as well as from a distance in the following months. Two notes prepared after the workshop are annexed to this document to illustrate a selection method based on sex-ratios proposed by Youssef Courbage (Annex 3) and the position of Richard Bilsborrow on that approach (Annex 4).
  1. The session on sampling started with a power-point presentation by Richard Bilsborrow, first showing that data of the UN Population Division (2009) show the relative rarity of international migration in populations all over the world, with few countries having stocks of migrants (measured as residents who had been born abroad) as high as 15% of their population—and these are stocks of “lifetime migrants” accumulated over many decades. When flow statistics are considered, especially pertaining to a fixed time interval, such as the most recent 5 or 10 years, the rarity of migration is striking. Thus in developing countries (almost all of net emigration), the mean percentage of the population emigrating in 2005-2010 was only 0.05 per year, or one quarter of one percent total over the five year period. In the eight countries in the study region (excluding Jordan and Syria, countries of net immigration in 2005-2010, mainly due to refugees), annual net emigration varied from --0.04 to -0.27 percent per year in 2005-2010. In addition, households with emigrants tend to be concentrated in particular locations due to the importance of migrant networks in migration.
  1. It was pointed out that these two facts have direct implications for the sample design of surveys on international migration (focusing on emigration) in the MED-HIMS countries, and illustrate the need to use a sampling approach appropriate for “rare elements”. This approach, in the context of international migration, involves two components. First, seek data (when possible) from a recent census on households with a former member emigrating in recent previous years (or having a former member living abroad) to stratify administrative areas of the country according to the proportion of the population constituted by emigrants (or by households with emigrants). Select Primary Sampling Units (PSUs), such as provinces or districts, by oversampling areas with higher proportions of emigrants. Repeat for as many stages as necessary in order to select areas with greater likelihood of encountering households with emigrants at each stage, down to the final small Ultimate Area Units (UAUs). Second, in each UAU, use two-phase sampling, to first conduct a quick listing operation of around 100 (50 to 150 maximum) households to identify which ones have emigrants (or return migrants). Prepare separate lists of households with emigrants, households with return migrants, and households with neither (the majority). Then select the sample of households for actual interview by oversampling from the first two lists compared to non-migrant households. This will concentrate - and make more efficient - the fieldwork.
  1. With the assistance of Nadia Touihri, the process of stratification was illustrated for Tunisia, showing how governorates, delegations within governorates, and finally Enumeration Areas (EAs) within governorates could be selected using oversampling at three stages. It was also pointed out that the existing Master Sample of EAs in Tunisia could be stratified so that the entire sampling process could be carried out in a single stage by stratifying and oversampling EAs. However, this will lead to a very dispersed selection of UAUs and more expensive field operation than using multiple stages of stratification.
  1. The ensuing discussion focused on the problems posed by the fact that some MED-HIMS countries do not have any recent data on emigrants in their census (or only unreliable data), so what other data could be used to establish an adequate sampling frame to use for forming strata. Several countries indicated they had survey data, or will soon have it, that could provide some indicator of the location of households with emigrants. Both Egypt and Jordan have a question that identifies whether a sampled household has an emigrant in their first quarter labor force surveys in 2012. It is expected that this will provide data by April or May. Morocco has data from its longitudinal survey of 110,000 households involving three visits to households every 6 months in 2009-2010.
  1. Youssef Courbage suggested to use a demographic method which might be applied in MED-HIMS countries where adequate census or survey data are lacking. The rationale is to use the sex ratio of the adult population, say 15 to 49 years of age, as an indicator of the extent of emigration from areas, since the vast majority of emigrants in the region’s countries are male adults. This is explained in the note at Annex 3, together with some evaluations for the case of Egypt based on the 2006 Population Census. A first reaction to this proposal by Richard Bilsborrow is provided in Annex 4. At this stage, the experts will first wait for the results/data from the surveys being implemented in the priority countries,Egypt and Jordan, to make a decision, which may differ for the two countries, also considering some possible combination of methods.A provisional draft of guidelines on sampling will be available by June 2012, which will be finalized for other countries following the actual experience of sampling in the first two countries.
  1. During the last day Bilsborrow presented an example to show how weights would be computed for households in a stratified sample, since the sample would not be self-weighting, as is usual in household surveys (e.g., labour force, DHS). The example used was of Nigeria from the Africa (International) Migration Project of the World Bank, illustrating how the weights would be created from each stage of a three-stage stratified sample, including for EAs as the UAUs. There two-phase sampling was used, involving first listing 50-80 households in each EA and then oversampling those with migrants. This presentation was followed by further discussion of the problems of conducting listing operations in MED-HIMS countries and the need to address this issue and for the expert to design listing sheets.
  1. The way to select respondents for administrating the MQ-5 on forced migration raised particular interest and requests. Concerning the sampling of forced migration, Tarek Abou Chabake described the UNHCR request to not only catch a nationally representative sample of forced migrants in the main household survey but drawing a supplementary sample using data available at UNHCR allowing identification of target areas with high concentrations of forced migrant respondents, at leastin the capital city.

H.Future activities