D3.2 Sampling Report

Transcription

1 D3.2 Sampling Report 1st Reporting period WP3 Fieldwork Responsible Partner: Ipsos Contributing partners: N.A. Due date of the deliverable: March 31 st 2013 Actual submission date: July 8 th 2013 Dissemination level: PU PACT - Public perception of security and privacy: Assessing knowledge, Collecting evidence, Translating research into action Project co-funded by the European 1 Commission within the 7 th Framework Programme SECURITY theme

2 Document Management PROJECT FULL TITLE PROJECT ACRONYM Public Perception of Security and Privacy: Assessing Knowledge, Collecting Evidence, Translating Research into Action PACT GRANT AGREEMENT STARTING DATE 01/02/2012 Collaborative Project funded under Theme SEC The relationship between human privacy and security DURATION 36 months D3.2 Sampling report Task: 3.2 Sampling Leader: Ipsos Other contributors: N.A. Authors: Andrew Johnson, Elena Lucica All rights reserved. No part of this publication may be reproduced, distributed or utilized in any form or by any means, electronic, mechanical, or otherwise, without the prior permission in writing from the PACT consortium. Download and print of the electronic edition for non-commercial teaching or research use is permitted on fair use grounds. Each copy should include the notice of copyright. Source should be acknowledged PACT 2

3 Note on the delivery of D3.2 (Sampling report) According to the original timing, deliverable D3.2 (Sampling report) was originally due in Month 14 (March 2013). The deliverable was submitted at a later date (Month 18 July 2013) due, on the one hand, to the delays in the completion of other tasks which have an impact on the sampling design (more specifically, the submission of the final draft of the survey questionnaire), and on the other hand, to recent changes in the survey methodology, as an outcome of the pilot study. Delays in finalising the survey questionnaire, due in Month 13 (February 2013), first draft submitted in Month 15 (April 2013), and post-pilot version submitted in Month 18 (July 2013), as well as discussions on questionnaire content and its impact on the sampling approach (whether to interview the general public or to target only people who use the internet and who take public transport) made the two tasks (questionnaire design and sampling) interdependent. The pilot study provided important input on different aspects which impact the sampling design, such as interview length (time taken to answer the survey, implementation of the online/offline methodologies, respondents understanding of the questionnaire). Based on the outcomes of the pilot, the questionnaire administration technique was refined, and several adjustments were made to the planning as well as to the methodological approach (carrying out a second pilot in order to further test and perfect the survey questionnaire, and carrying out the fieldwork in Estonia online rather than face-to-face). Waiting for the outcomes of the pilot study (Month 17 June 2013) was therefore crucial in order to define the sampling approach. 3

4 Contents 1 Executive summary Methodology Sampling approach online survey Austria Belgium Denmark Finland France Germany Estonia Ireland Italy Luxembourg The Netherlands Spain Sweden UK Sampling approach face-to-face survey Bulgaria Cyprus Czech Republic Germany Greece Hungary Italy Lithuania Malta Poland Portugal Latvia Romania Slovakia Slovenia

5 1 Executive summary This document presents the sampling techniques which will be implemented for the PACT survey. As the methodology will vary between the different countries where the survey will take place (online, face-to-face), the sampling design will be adapted accordingly, all while taking into account countryspecific aspects. Our sampling approach will aim, on the one hand, to ensure that the survey results are representative of the general population in the different countries, and on the other hand, to maximise fieldwork efficiency and to reduce possible risks. The survey will be carried out in 27 countries of the European Union. Approximately 1000 people 1, aged 18 and older, will be interviewed in each country. In terms of methodology, the survey will be carried out online in 12 countries, and face-to-face in 13 countries. In two of the countries (Italy and Germany), the survey will be carried out through a mixed approach, as 500 interviews will be conducted online, and 500 will be carried out face-to-face, enabling us to control comparability between the two methods during the fieldwork process. In countries where the survey will be carried out online, the sampling design will be structured by national representative quotas, reflecting the structure of the offline population in each country by age, gender and region. The sample will be drawn from online panels in each country. Panellists are recruited from different sources (opt-in s, banner and text links, referral programs). The panel is constantly renewed with updated panellist contact information and demographics. Each potential respondent will be directed to a unique URL where they can access the questionnaire. Those who have not completed the survey within a certain pre-defined schedule will receive reminders. Fieldwork progress will be monitored centrally, on a daily basis. In countries where the survey will be carried out face-to-face, the sampling will designed following a three-step approach: stratification and selection of local area sampling points, selection of addresses within each of the sampling points, selection of individuals within each household. As a first step of the process, in each country, the population will be stratified by region and urbanisation degree. Sampling point selection (total number of sampling points, number of sampling points per cell) will be done according to the structure of the obtained matrix. Depending on the country, a certain number of interviews will then be allocated for each sampling point. In each sampling point, starting addresses will be selected. Households will then be selected through predefined random walk procedures. The number of addresses skipped between two interviews (random walk interval) can vary by country, as well as by type of settlement (whether rural or urban). Finally, at a household level, respondents will be selected according to national representative quotas, reflecting the structure of the general population in each country by age and gender (either crossed quotas or simple quotas, depending upon local procedures in each country). Only one respondent will be interviewed in each household. 1 The sample size will be of 1000 interviews in each country, apart from Luxembourg, Malta and Cyprus, where 750 interviews will be carried out 5

6 2 Methodology The target of the survey will be the general population, aged 18+, in each of the 27 countries. The survey will be conducted through online and face-to-face methodology, as following: Online: 12 countries Face-to-face: 13 countries Mixed methodology (500 interviews online, 500 interviews face-to-face): 2 countries Country Methodology Internet access 2 (%) Sample size Netherlands Online Luxembourg Online Sweden Online Denmark Online Finland Online UK Online France Online Austria Online Belgium Online Ireland Online Spain Online Estonia 3 Online Germany Online, face-to-face (500 online, 500 faceto-face) Italy Online, face-to-face (500 online, 500 faceto-face) Malta Face-to-face Slovenia Face-to-face Slovakia Face-to-face Poland Face-to-face Czech Republic Face-to-face Lithuania Face-to-face Hungary Face-to-face Latvia Face-to-face Cyprus Face-to-face Portugal Face-to-face Greece Face-to-face Romania Face-to-face Bulgaria Face-to-face The survey will be conducted online in countries with high internet penetration, and face-to-face in those where internet penetration is lower. In Germany and in Italy, both methods will be used, as in each of the two countries, 500 interviews will be carried out online, and 500 face-to-face. This 2 Internet access Eurostat In the initial planning, fieldwork in Estonia was to be carried out through face-to-face methodology. Due to decisions taken after the pilot study regarding methodology and planning (conducting a second pilot study in one country in order to test and perfect the survey questionnaire), and taking into account the high internet penetration rate in this country, it was agreed that fieldwork in Estonia would take place online. 6

7 approach will enable us to control the comparability of the two methods throughout the fieldwork process. 3 Sampling approach online survey In countries where the survey will be conducted online, a quota sampling approach will be implemented. The survey will target members of the general population, aged 18+. In order to ensure representativeness, quotas by age, gender and region will be applied in each country, reflecting the structure of the overall population in each country. The sample will be drawn from online panels in each of the participating countries. Ipsos is an industry leader in building and maintaining the highest quality panels. Panellists are recruited from a variety of sources, including numerous opt-in lists, banner and text links, and referral programs. The panel is continually refreshed with updated panellist contact information and panellist demographics. Non-responders are systematically purged, allowing us to maintain a very high response rate. Within each country, the sample to receive an invitation to participate in the survey will be representative of the offline population in that country, based on Eurostat data where available. For surveys conducted within a country, the matrix is based on the latest population statistics and balanced according to response rates for each demographic group, ensuring the sample is reflective of the target population according to gender, age, and region. Each potential survey respondent will be directed to a unique URL within our secure online research environment where they access the survey questionnaire. Respondents who are interrupted or choose to suspend the survey before they have had an opportunity to complete the questionnaire can at their convenience access the site again using their unique URL and finish the survey. To ensure the integrity of the data, and to avoid a situation in which an individual could fill out the survey multiple times, the unique URL provides password-protected entry to our site and expires after the completion of a single survey (one survey for one Unique URL). Once the survey invitation has been sent, reminders will be distributed to those who have not completed the survey at a schedule pre-determined based on the study plan. Ipsos will monitor the status of the online research twice daily (and more frequently during the first several days of fieldwork. Any issues that arise during the fieldwork will be flagged as soon as possible. Online fieldwork will be centralised within the Ipsos Interactive Service Bureau (IIS). All the work conducted by IIS is managed centrally, with one scripting, data collection and data delivery process. The following tables present the approximate description of the sample structure for each of the countries where the survey will take place online. 7

8 3.1 Austria Austria population 18+ ; N=1000 National representative quotas Gender Men 48,17% Women 51,83% Age ,61% ,84% ,58% ,25% ,21% ,52% Region Ostösterreich 43,22% Südösterreich 21,29% Westösterreich 35,48% Source: Eurostat Belgium Belgium population 18+ ; N=1000 National representative quotas Gender Men 48,51% Women 51,49% Age ,87% ,15% ,46% ,42% ,51% ,59% Region Center (Bruxelles) 9,98% North (Flandre) 58,19% South (Wallonie) 31,83% Source: Eurostat

9 3.3 Denmark Denmark population 18+ ; N=1000 National representative quotas Gender Men 49,11% Women 50,89% Age ,99% ,04% ,23% ,87% ,40% ,47% Region Jylland 32,93% Syddanmark 21,49% København 30,90% Sjælland 14,68% Source: Eurostat Finland Finland population 18+ ; N=1000 National representative quotas Gender Men 48,58% Women 51,42% Age ,80% ,97% ,46% ,52% ,31% ,93% Region Pohjois-Suomi 11,62% Itä-Suomi 12,35% Länsi-Suomi 25,33% Etelä-Suomi, Åland 50,70% Source: Eurostat

10 3.5 France France population 18+ ; N=1000 National representative quotas Gender Men 47,66% Women 52,34% Age ,32% ,78% ,81% ,43% ,13% ,53% Region Region Parisienne 18,47% Nord-Ouest 22,91% Nord-Est 22,66% Sud-Ouest 11,21% Sud-Est 24,75% Source: Eurostat Germany Germany population 18+ ; N=500 National representative quotas Gender Men 48,63% Women 51,37% Age ,85% ,32% ,95% ,48% ,78% ,62% Region Nielsen I 16,02% Nielsen II 21,62% Nielsen IIIa 13,52% Nielsen IIIb 12,98% Nielsen IV 15,20% Nielsen V(a&b) 4,33% Nielsen VI 8,23% Nielsen VII 8,11% Source: Eurostat

11 3.7 Estonia Estonia population 18+ ; N=1000 National representative quotas Gender Men 45% Women 55% Age ,2% ,2% ,6% ,6% ,3% ,1% Region Tallinn 30% Northern Estonia 14% Ida-Virumaa 13% Central Estonia 8% Western Estonia 12% Southern Estonia 23% Total 100% 3.8 Ireland Ireland population 18+ ; N=1000 National representative quotas Gender Men 48,97% Women 51,03% Age ,77% ,67% ,99% ,19% ,78% ,59% Region Border, Midland and Western 26,75% Southern and Eastern 73,25% Total 100% Source: Eurostat

12 3.9 Italy Italy population 18+ ; N=500 National representative quotas Gender Men 47,92% Women 52,08% Age ,64% ,95% ,30% ,72% ,99% ,41% Region Nord-Ovest 26,83% Nord-Est 19,28% Centro (I) 19,95% Sud 22,96% Isole 10,98% Source: Eurostat Luxembourg Luxembourg population 0+ ; N=750 National representative quotas Gender Men 49,32% Women 50,68% Age ,55% ,14% ,43% ,23% ,03% ,63% Region Canton Capellen 8,04% Canton Esch 29,76% Canton Luxembourg 29,59% Canton Mersch 5,32% Canton Clervaux 3,01% Canton Diekirch 5,56% Canton Redange 3,15% Canton Vianden 0,86% Canton Wiltz 2,70% Canton Echternach 3,25% Canton Grevenmacher 5,00% Canton Remich 3,76% Source: Statistics Luxembourg 2011; available only for population 0+ 12

13 3.11 The Netherlands The Netherlands population 0+ ; N=1000 National representative quotas Gender Men 49,05% Women 50,95% Age ,00% ,25% ,38% ,97% ,68% ,73% Region Noord & Oost Nederland 31,20% West-Nederland 47,04% Zuid-Nederland 21,76% Source: Eurostat Spain Spain population 18+ ; N=1000 National representative quotas Gender Men 48,76% Women 51,24% Age ,11% ,90% ,41% ,38% ,41% ,79% Region Noroeste and Noreste 19,62% Madrid and Centro 26,07% Este 28,93% Sur and Canarias 25,38% Total 100% Source: Eurostat

14 3.13 Sweden Sweden population 18+ ; N=1000 National representative quotas Gender Men 49,42% Women 50,58% Age ,96% ,52% ,16% ,37% ,82% ,17% Region Norrland 9,46% Mellansverige 25,61% Stockholm 21,46% Västsverige 19,94% Södra Sverige 23,53% Source: Eurostat UK United Kingdom population 18+ ; N=1000 National representative quotas Gender Men 48,64% Women 51,36% Age ,12% ,50% ,25% ,28% ,99% ,87% Region North&Yorkshire 12,77% North West 11,12% Midlands 15,92% South West & Wales 13,44% South East & Anglia 22,94% Greater London 12,49% Scotland 8,52% Northern Ireland 2,79% Source: Eurostat

15 4 Sampling approach face-to-face survey In countries where the survey will be carried out face-to-face, the following three-step approach will be implemented: Stratification and selection of local area sampling points Selection of addresses within each of the sampling points Selection of individuals within each selected household, according to quotas by age and gender, ensuring that the sample is balanced by key characteristics. In each country, the sampling design will be implemented according to the following procedure: 1. Stratification In each country, the population will be stratified by region (NUTS 4 2/3), and urbanisation degree 2. Selection of sampling points Based on the obtained matrix, a certain number of sampling points will be selected in each country According to the structure of the sampling points in each country, a certain number of interviews will be allocated for each sampling point 3. Address selection In each sampling point, a certain number of starting addresses will be selected Households will then be selected, either through random walk procedures, either based on an address list 4. Respondent selection At a household level, individuals will be selected according to national representative quotas by age and gender The procedure will vary by country, as some countries will follow crossed quotas (in which case, age categories will be regrouped), while others will follow simple quotas (in which case, detailed age categories will be used) Only one individual will be interviewed in each household. The following sections provide a detailed description of the sampling procedures, as they will be implemented in each of the countries where the survey will take place face-to-face, as well as the suggested sampling structure (approximate number of sampling points and of interviews per sampling point, approximate targets in terms of age and gender) in each country. 4 The Nomenclature of Territorial Units for Statistics 15

16 4.1 Bulgaria 1. Sample size A total of 1000 face-to-face interviews will be conducted 2. Stratification The sample will be stratified by region (NUTS 2) and type (district/ city, town and village) The Census data 2011 will be used as information source A matrix of 84 cells will be obtained as an outcome of the stratification 3. Selection of sampling points Sampling points will be selected randomly within each cell The total number of sampling points will be 100 Approximately 10 interviews will be carried out in each sampling point 4. Address selection Starting addresses will be selected randomly from a listing of streets Approximately 20 addresses will be selected in each sampling point Households will be selected using pre-defined random walk procedures The number of addresses skipped between two interviews (random walk interval) will be 3 for rural areas and 5 for urban areas 5. Respondent selection Respondents will be selected using crossed quotas (gender/age). 16

17 Selection of sampling points District city Town Vilages Total North West Vidin Montana Vratza Pleven Lovetch North Central Ruse Veliko Tarnovo Gabrovo Razgrad Silistra North East Targovishte Shumen Dobritch Varna South West Sofia capital Sofia district Pernik Kyustendil Blagoevgrad South Central Pazardjik Plovdiv Smolyan Kardjali Haskovo South East Stara Zagora Sliven Yambol Burgas Total

18 Target number of interviews per cell District city Town Vilages Total North West Vidin Montana Vratza Pleven Lovetch North Central Ruse Veliko Tarnovo Gabrovo Razgrad Silistra North East Targovishte Shumen Dobritch Varna South West Sofia capital Sofia district Pernik Kyustendil Blagoevgrad South Central Pazardjik Plovdiv Smolyan Kardjali Haskovo South East Stara Zagora Sliven Yambol Burgas Total Target - Respondent selection (crossed quotas) Sample structure Age Men Women Total N % N % N % ,00% ,10% ,10% ,00% ,50% ,50% ,00% ,40% ,40% Total ,00% ,00% ,00% 18

19 4.2 Cyprus 1. Sample size A total of 760 interviews will be conducted 2. Stratification The sample will be stratified by region (taking into account the five main districts: Nicosia, Larnaca, Famagusta and Paphos) and according to their urbanization degree (urban and rural) Crossing both variables generates a matrix of 124 cells All statistical information are based on the latest Population Census (2011), published by the Statistical Service Department, Ministry of Finance 3. Selection of sampling points A total of 76 sampling points will be selected randomly, with a probability of selection proportionate to their size The distribution of the sampling points selected among the 5 districts and among urban & rural areas within each district will be proportionate to their respective size In each sampling point approximately 10 interviews will be carried out 4. Address selection Starting addresses will be selected randomly from a street list; during the selection process, maps will be consulted in order to make sure that the starting point is effective and practical The number of addresses per sampling point will be set to 25. As a back-up, 10 extra addresses in each sampling point will be allocated A sampling point can only be replaced in extreme situations Households will be selected through random-walk procedures 5. Respondent selection Respondents will be selected using simple quotas (age, gender). 19

20 Selection of sampling points < Total Nicosia-urban Nicosia Municipality Agios Dometios Municipality 1 1 Egomi Municipality 2 2 Strovolos Municipality Aglantzia Municipality 2 2 Lacatamia Municipality Latsia Municipality Dali Municipality 1 1 Tseri Municipality 1 1 Nicosia-rural Nicosia District - Rural Famagustarural Famagusta District - Rural Larnacaurban Larnaca Municipality Aradippou Municipality 2 2 Levadia 1 1 Dromolaxia 0 0 Meneou 0 0 Larnaca - rural Larnaca District - Rural Limassol - urban Limassol Municipality Mesa Geitonia Municipality 1 1 Agios Athanasios Municipality 1 1 Germasogia Municipality 1 1 Pano Polemidia 0 0 Ypsonas 1 1 Kato Polemidia Municipality 2 2 Limassol - rural Limassol District - Rural Paphosurban Paphos Municipality Geroskipou Municipality 1 1 Konia 0 0 Chloraca 1 1 Emba 0 0 Paphos - rural Paphos District Rural Total

21 Target number of interviews per cell < Nicosia-urban Nicosia Municipality Total Agios Dometios Municipality Egomi Municipality Strovolos Municipality Aglantzia Municipality Lacatamia Municipality Latsia Municipality Dali Municipality Tseri Municipality Nicosia-rural Nicosia District - Rural Famagustarural Famagusta District - Rural Larnaca-urban Larnaca Municipality Aradippou Municipality Levadia Dromolaxia 0 0 Meneou 0 0 Larnaca - rural Larnaca District - Rural Limassol -urban Limassol Municipality Mesa Geitonia Municipality Agios Athanasios Municipality Germasogia Municipality Pano Polemidia 0 0 Ypsonas Kato Polemidia Municipality Limassol - rural Limassol District - Rural Paphos- urban Paphos Municipality Geroskipou Municipality Konia 0 0 Chloraca Emba 0 0 Paphos - rural Paphos District Rural Total

22 Target Respondent selection (simple quotas) Sample structure (N) (%) Men ,98 Women ,02 Total ,00 Age , , , , , ,60 Total , Czech Republic 1. Sample size A total of 1000 interviews will be conducted 2. Stratification The sample will be stratified by region (NUTS 2) and urbanisation degree The information source will be the Czech Statistical Bureau data ( ) The stratification will produce 40 cells 3. Selection of sampling points The total number of sampling points will be 125 Approximately 8 interviews will be carried out in each sampling point The list of UTJ by Czech Statistical Bureau will be used for sampling point selection (UTJ - Regional technical unit / Územně technická jednotka, Czech Republic includes such units) The required number of sampling points (UTJs) within each stratification cell will be selected through random sampling, from the list of all UTJ and according to the number of inhabitants 4. Address selection In each sampling point, one starting address will be selected No more than 20 addresses will be visited from the starting address The interviewers will first visit 10 addresses, using the random route procedure; if less than 7 interviews will be completed, they will visit another 5 addresses; if less than 7 interviews will be completed, they will visit another 5 (20 addresses in total) The above mentioned procedure will ensure that each address can be visited at least 3 times, and that the number of interviews in most of the sampling points would vary between 7-9 (8 on average) 22

23 Households will be selected through pre-defined random-walk procedures. The random walk interval will be different in the rural areas (one housing unit skipped between two interviews) and urban areas (4 skipped housing units between two interviews) Every apartment will be considered as a separate housing unit 5. Respondent selection Respondents will be selected using crossed quotas (age, gender) Selection of sampling points Total Praha Střední Čechy Jihozápad Severozápad Severovýchod Jihovýchod Střední Morava Moravskoslezsko Total Target number of interviews per cell Total Praha Střední Čechy Jihozápad Severozápad Severovýchod Jihovýchod Střední Morava Moravskoslezsko Total

24 Target - Respondent selection (crossed quotas) Sample structure Age Men Women Total N % N % N % ,7% ,0% ,7% ,6% ,9% ,5% ,3% ,5% ,8% Total ,6% ,4% % 4.4 Germany 1. Sample size A total of 500 face-to-face interviews will be carried out 2. Stratification Sampling units will be stratified according to counties (Kreise) and BIK community-types. Within these layers, the sampling units are ordered according to their size, i.e. specific weights 5 3. Selection of sampling points On the basis of the ADM-Sampling System, 52 sample points (electoral districts of the 2009 Federal Elections) will be randomly selected, proportional to the regional structure of the population Each sampling-unit is allocated a specific weight based on the number of all German private households 4. Address selection In each sampling point, the starting address will be drawn at random From this starting address, households will be selected through random route procedure (every third household) 5. Respondent selection Respondents will be selected using simple quotas (age, gender). 5 Note concerning the ADM Sampling system: Ipsos is member of the "Arbeitsgemeinschaft Stichproben des ADM" (i.e. the representation of private market and social research agencies in Germany). The members of the ADM and MMC, Frankfurt/Main, own the ADM-Samplingsystem. This system aims at the universe of private households and their respective household members, living at the location of their main residency in all 16 Länder in the Federal Republic. The spatial grid of the system is based on existing community borders and electronically-aided geographical delineations. As far as possible the geographical delineation is based on the smallest available administrative unit. Those areas are then further separated (using a geographical information retrieval system and the resident information) so that each unit contains on average 700 and a maximum of addresses with minimized variance. In order to maintain a minimum size of 350 inhabitants per unit, some of the sampling units are merged into synthetic units. This process produces selection units which form the basis for drawing each sample. 24

25 Selection of sampling points > Schleswig- Holstein Hamburg Niedersachsen Bremen Nordrhein- Westfalen Hessen Rheinland-Pfalz Baden- Württemberg Bayern Saarland Berlin Brandenburg Mecklenburg- Vorpommern Sachsen Sachsen-Anhalt Thüringen Total Total

26 Target number of interviews per cell > Schleswig- Holstein Hamburg Niedersachsen Bremen Nordrhein- Westfalen Hessen Rheinland-Pfalz Baden- Württemberg Bayern Saarland Berlin Brandenburg Mecklenburg- Vorpommern Sachsen Sachsen-Anhalt Thüringen Total Total Target - respondent selection (simple quotas) Sample structure Gender (N) (%) Men ,8% Women ,2% Total ,0% Age ,8% ,4% ,4% ,2% ,8% ,4% Total ,0% 26

27 4.5 Greece 1. Sample size A total of 1000 face-to-face interviews will be carried out 2. Stratification The sample will be stratified by Region (NUTS 2) and urbanisation degree Crossing both variables will generate a matrix of 52 cells All statistical information is based on information from the Census Data 2001 (ELSTAT) Each cell is attributed a number of interviews which is proportional to the general population NUTS2/urbanization degree 3. Selection of sampling points Sampling points will be selected randomly within each cell In each sampling point approximately 10 interviews will be carried out Sampling points will be selected from the official and complete register of ELSTAT containing all cities by NUTS2 and urbanization degree The total number of sampling points will be Address selection In each of the selected sampling points, one building block (as registered by ELSTAT) will be drawn at random from the listing of building blocks in the sampling point. The most North-eastern building of this building block will be the starting address Approximately 20 households will be taken into account in each sampling point The random walk interval applied in each sampling point will be calculated taking into account the number of 20 households (estimated to be needed in order to carry out 10 interviews) and the household population of the specific building block In case it is not possible to carry out the 10 interviews, the fieldwork supervising team will be notified and the interviewers will be instructed to proceed following the random walk rules to the next building block for 5 household contacts at a time 5. Respondent selection Respondents will be selected according to crossed quotas (age/gender) 27

28 Selection of sampling points < Total Anatoliki makedonia Kentriki makedonia Dytiki makedonia Ipiros Thessalia Ionia nisia Dytiki ellada Sterea ellada Peloponnisos Attiki Voreio aigaio Notio aigaio Kriti Total Target number of interviews per cell < Total Anatoliki makedonia Kentriki makedonia Dytiki makedonia Ipiros Thessalia Ionia nisia Dytiki ellada Sterea ellada Peloponnisos Attiki Voreio aigaio Notio aigaio Kriti Total

29 Target - Respondent selection (crossed quotas) Sample structure Age Men Women Total N % N % N % ,60% ,60% ,20% ,70% ,90% ,60% ,70% ,50% ,20% Total ,00% ,00% ,00% 4.6 Hungary 1. Sample size A total of 1000 interviews will be carried out 2. Stratification The sample will be stratified by Region (NUTS 2) and urbanisation degree. All statistical information is based on information from the Census Data 2011 (KSH Hungarian Central Statistical Office) Crossing both variables will generate a matrix of 21 cells 3. Selection of sampling points Sampling points will be selected randomly within each cell The total number of sampling points will be 100 In each sampling point, 10 interviews will be conducted on average 4. Address selection In each of the selected sampling points, 10 addresses will be selected randomly from the listing of streets in the geographical area of the sampling point Households will be selected using the pre-defined random-walk procedure (every second address) 5. Respondent selection Respondent selection will be done according to crossed quotas (age/gender) 29

30 Selection of sampling points 1 - Budapest 2 - County town 3 - Town 4 - Village Total 1 Budapest Pest Fejér Komárom Veszprém Győr Vas Zala Baranya Somogy Tolna Borsod Heves Nógrád Hajdú Szabolcs Szolnok Bács Békés Csongrád Total

31 Target number of interviews per cell 1 - Budapest 2 - County town 3 - Town 4 - Village Total 1 Budapest Pest Fejér Komárom Veszprém Győr Vas Zala Baranya Somogy Tolna Borsod Heves Nógrád Hajdú Szabolcs Szolnok Bács Békés Csongrád Total Respondent selection crossed quotas Sample structure Age Men Women Total N % N % N % ,00% ,30% ,30% ,80% ,60% ,40% ,80% ,50% ,30% Total ,60% ,40% ,00% 31

32 4.7 Italy 6. Sample size A total of 500 face-to-face interviews will be carried out 7. Stratification The sample will be stratified by Region (NUTS 2) and urbanisation degree Crossing both variables will generate a matrix with 80 cells All statistical information regarding town size is based on information from the Census Data 2006 (ISTAT) 8. Selection of sampling points Sampling points will be selected randomly within each cell In each sampling point approximately 10 interviews will be carried out Sampling points are selected from official and complete registers The total number of sampling points will be Address selection In each of the selected sampling points, the starting address will be drawn at random from the listing of streets in the geographical area of the sampling point The number of addresses per sampling point is set to 20 If it is not possible to achieve the target, interviewers will need to contact the fieldwork supervisor; 10 additional addresses will be allocated to the random route in batches of 5 depending on the number of interviews that have already been conducted in the sampling point A sampling point can only be replaced in extreme situations Households will be selected using pre-defined random walk procedures 10. Respondent selection Respondents will be selected according to crossed quotas (age, gender) 32

33 Selection of sampling points < Total Valle D'Aosta Piemonte Lombardia Liguria Emilia Romagna Friuli Venezia Giulia Trentino Alto Adige Veneto Lazio Marche Toscana Umbria Abruzzo Basilicata Calabria Campania Molise Puglia Sardegna Sicilia Total

34 Target number of interviews per cell < Total Valle D'Aosta Piemonte Lombardia Liguria Emilia Romagna Friuli Venezia Giulia Trentino Alto Adige Veneto Lazio Marche Toscana Umbria Abruzzo Basilicata Calabria Campania Molise Puglia Sardegna Sicilia Total Target - Respondent selection (crossed quotas) Sample structure Age Men Women Total N % N % N % ,0% 60 12,0% ,0% ,0% 90 18,0% ,0% ,0% ,0% ,0% Total ,0% ,0% ,0% 34

35 4.8 Lithuania 1. Sample size A total of 1000 face-to-face interviews will be carried out 2. Stratification The sample will be stratified by region (NUTS3) and by the urban/rural ratio provided by the Lithuanian department of statistics (source: Crossing the two variables generates a matrix of 20 stratified cells 3. Selection of sampling points The sampling points will be selected according to the proportion of the population by district The sampling design will include at least 80 sampling points (the current sampling design includes 100 sampling points) Approximately 10 interviews will be carried out in each sampling point (15 interviews per sampling point at the most) 4. Address selection The starting addresses will be selected randomly in each sampling point The approximate number of addresses in each sampling point will depend on the location of the sampling point Households will be selected using pre-defined random-walk procedure; The random walk interval (number of addresses skipped between interviews) will be 2 5. Respondent selection Respondents will be selected according to crossed quotas (age, gender) Selection of sampling points Urban Rural Total Alytaus Kauno Klaipėdos Marijampolės Panevėžio Šiaulių Tauragės Telšių Utenos Vilniaus Total

36 Target number of interviews per cell Urban Rural Total Alytaus Kauno Klaipėdos Marijampolės Panevėžio Šiaulių Tauragės Telšių Utenos Vilniaus Total Target - Respondent selection (crossed quotas) Sample structure Age Men Women Total N % N % N % ,05% ,30% ,30% ,03% ,60% ,40% ,79% ,50% ,30% Total ,87% 551 0, ,00% 4.9 Malta 1. Sample size A total of 750 face-to-face interviews will be carried out 2. Stratification The sample will be stratified by region and urbanisation level Malta falls under a single region in the NUTS 3 classification; however, regional division will be based on six geographical regions as defined by the National Statistics Office (NSO) Information is based on the Census carried out in 2005 by the NSO Urbanisation levels have been determined internally by MISCO International (due to lack of official publically available data issued by the NSO) Crossing the two variables will generate a matrix of 14 cells 3. Selection of sampling points Sampling points will be selected randomly from each cells 36

37 In each sampling point 5 interviews will be conducted The project will initiate with a total of 150 sampling points. Additional sampling points may be issued during fieldwork if necessary 4. Address selection The starting addresses will be indicated to the interviewers for each sampling point A maximum of 20 addresses will be selected in each sampling point; 5 additional addresses can be allocated if necessary If the required 5 interviews are not completed within the original sampling point an additional sampling point will be issued Households will be selected using pre-defined random walk procedure 5. Respondent selection Respondents will be selected according to simple quotas (age, gender) Selection of sampling points Total Northern Northern Harbour Southern Harbour South Eastern Western Gozo Total Target number of interviews per cell Total Northern Northern Harbour Southern Harbour South Eastern Western Gozo Total

38 Target - Respondent selection ( simple quotas) Sample structure (N) (%) Men ,2% Gender Women ,8% Total ,0% Age ,75% ,04% ,46% ,64% ,60% ,51% Total ,00% 4.10 Poland 6. Sample size A total of 1000 face-to-face interviews will be carried out 7. Stratification The sample will be stratified by region (NUTS2) and by urbanisation degree Crossing the two variables generates a matrix of 96 cells 8. Selection of sampling points In the first stage a probability sample of 197 sampling points will be selected from a list of 2479 communities (gminas) with probabilities proportional to the size of their populations. Approximately 5 interviews will be conducted (4-6) in each sampling point 9. Address selection Starting addresses in each location will be drawn according to the following procedure: finding the address of a school or church; if not available, finding a bus station; if not available, finding the address of the city hall One starting address will be appointed in each sampling point In terms of household selection, two different address search procedures will be used, depending on whether the interviews are conducted in urban or rural areas: o In urban areas in which addresses are defined by street name and number, this procedure will consist of counting off apartment/ house numbers and streets o In rural areas and settlements where addresses are not precisely defined, the procedure will consist of counting off residential properties (residential buildings) 38

39 The random walk interval will vary by type of area (5 in urban areas and 3 in rural areas) 10. Respondent selection Respondents will be selected according to crossed quotas (age, gender) Selection of sampling points Village < Total Dolnośląskie Kujawsko-Pomorskie Lubelskie Lubuskie Łódzkie Małopolskie Mazowieckie Opolskie Podkarpackie Podlaskie Pomorskie Śląskie Świętokrzyskie Warmińsko-Mazurskie Wielkopolskie Zachodniopomorskie Total

40 Target number of interviews per cell Village < Total Dolnośląskie Kujawsko-Pomorskie Lubelskie Lubuskie Łódzkie Małopolskie Mazowieckie Opolskie Podkarpackie Podlaskie Pomorskie Śląskie Świętokrzyskie Warmińsko-Mazurskie Wielkopolskie Zachodniopomorskie Total Target - Respondent selection (crossed quotas) Sample structure Age Men Women Total N % N % N % , , , , , , , , ,9 Total ,60% ,40% ,00% 4.11 Portugal 1. Sample size A total of 1000 face-to-face interviews will be carried out 2. Stratification The sample will be stratified by Region (NUTS 2) and urbanisation degree Crossing both variables generates a matrix of 22 cells. All statistical information regarding town size is based on information from the INE Census 2011 Each cell will be attributed a number of interviews which is proportional to the population aged 18 years and older within the cell 40

41 3. Selection of sampling points Sampling points will be selected randomly within each cell In each sampling point, between 6 and 10 interviews will be carried out Sampling points will be selected from official and complete registers The total number of sampling points will be 126 sampling points (plus 20 additional reserves ) 4. Address selection In each of the selected sampling points, a starting address will be drawn at random from the listing of streets in the geographical area of the sampling point The number of addresses per sampling point will be set to 20 If it will not be possible to carry out the required number of interviews within the 20 addresses, interviewers will need to contact the office; 10 additional addresses will then be assigned, in batches of 5, depending on the number of interviews that have already been carried out in the respective sampling point Only in extreme situation a sampling point can be replaced Households will then be selected using the pre-defined random-walk procedure. The random walk interval will depend on the area: every fifth housing unit (i.e. houses or apartments) for urban areas, and every third housing unit for rural areas 5. Respondent selection Respondents will be selected according to simple quotas (age, gender) Selection of sampling points <2000 hab Total Norte Centro Lisboa Alentejo Algarve Total Target number of interviews per cell <2000 hab Total Norte Centro Lisboa Alentejo Algarve Total

42 Target - Respondent selection (simple quotas) Sample structure (N) (%) Men ,00% Gender Women ,00% Total ,00% Age ,30% ,40% ,40% ,20% ,30% ,40% Total ,00% 4.12 Latvia 6. Sample size A total of 1000 face-to-face interviews will be carried out 7. Stratification The sample will be stratified by Region (NUTS 2) and urbanisation degree Crossing both variables generates a matrix with 36 cells All statistical information regarding settlement size is based on information from the Census Data 2012 and The Office of Citizenship and Migration Affairs Selection of sampling points Sampling points will be selected randomly within each cell In each sampling point, approximately 10 interviews will be conducted Sampling points will be selected randomly from official registers (postal codes) The number of sampling points will be 100 (minimum) 9. Address selection In each of the selected sampling points, the starting address will be drawn at random All the starting locations of the sample will be determined using maps The number of addresses per sampling point will be set to 20 If it will not be possible to achieve the target, interviewers will need to contact their fieldwork supervisor; the procedure in this case will be to add 10 additional addresses to the random route in batches of 5, depending on the number of interviews that have already been achieved in the sampling point 42

43 Only in extreme situation a sampling point can be replaced Households will be selected using the pre-defined random-walk procedure (at a step of 3) 10. Respondent selection Respondents will be selected according to simple quotas (age, gender) Selection of sampling points < Total Rīga Vidzeme Kurzeme Zemgale Latgale Pierīga Total Target number of interviews per cell < Total Rīga Vidzeme Kurzeme Zemgale Latgale Pierīga Total Target - Respondent selection (simple quotas) Sample structure Gender (N) (%) Men ,6% Women ,4% Total Age ,8% ,6% ,5% ,6% ,1% ,4% Total % 43