Consumers attitudes towards cross-border trade and consumer protection 2016

Transcription

1 Consumers attitudes towards cross-border trade and consumer protection 2016 Technical report Contract n under FWC EAHC 2013/CP/03 Lot 1 Written by: GfK Date: January 2017 Justice and Consumers

2 EUROPEAN COMMISSION Produced by Consumers, Health, Agriculture and Food Executive Agency (Chafea) on behalf of Directorate-General for Justice and Consumers Directorate E Consumers Unit E.1 (Consumer Markets) JUST-CONSULT-E1@ec.europa.eu European Commission B-1049 Brussels

3 EUROPEAN COMMISSION Consumers attitudes towards cross-border trade and consumer protection 2016 Technical Report Directorate-General for Justice and Consumers EU Consumer Programme 2017 EUR [number] EN

4 Europe Direct is a service to help you find answers to your questions about the European Union. Freephone number (*): (*) The information given is free, as are most calls (though some operators, phone boxes or hotels may charge you). This report was produced under the EU Consumer Programme ( ) in the frame of a service contract with the Consumers, Health, Agriculture and Food Executive Agency (Chafea) acting under the mandate from the European Commission. The content of this report represents the views of the contractor and is its sole responsibility; it can in no way be taken to reflect the views of the European Commission and/or Chafea or other body of the European Union. The European Commission and/or Chafea do not guarantee the accuracy of the data included in this report, nor do they accept responsibility for any use made by third parties thereof. More information on the European Union is available on the Internet ( European Union, 2017

5 TABLE OF CONTENTS 1. SURVEY METHODOLOGY Coordination Target population Interviewing method Languages of interviewing Sampling frames, sample sizes and sampling design Questionnaire translation and scripting Pilot Fieldwork Response rate improvement measures ANALYTICAL METHODOLOGY Data cleaning, processing and validation Weighting Trend data Computation of derived indicators Estimation of standard errors and statistical significance... 18

6 EUROPEAN COMMISSION 1. SURVEY METHODOLOGY This section details the methodology implemented for the survey, from the survey design to the data collection. 1.1.Coordination Sara Gysen, the Contract Project Manager, assigned 3 key roles within the central coordination team to ensure ease and efficiency in communications and coordination throughout the execution of the survey on consumers attitudes towards cross-border trade and consumer-related issues As Deputy Project Manager, Karen Lefever and Snezha Kazakova had key roles in the overall management of the project, being the main point of contact with the Contracting Authority as well as ensuring a smooth project flow within GfK Belgium and its network. Nancy Heremans took up a leading role in the coordination of the national teams. The central coordination team consisted of the same members that also conducted the Market Monitoring Survey 2015 (MMS 2015), which is part of the same Framework Contract as the current survey. 1.2.Target population The target population includes all people aged 18 and above, resident in the country surveyed and having sufficient command of (one of) the respective national language(s) to answer the questionnaire. In addition, only people living in private households are interviewed, excluding prisoners, residents of retirement homes, etc. which are difficult to contact in a telephone survey. Sample sizes were set at 1000 consumers per percountry in most of the EU countries and Norway. In Luxembourg, Cyprus, Malta and Iceland, the target was 500 consumers per country. No quota was set for socio-demographic variables but the overall sample intake was monitored daily, to follow up on the overall composition of the sample on gender, age, region and the possession of a mobile and/or a fixed phone in accordance with the sampling approach adopted. 1.3.Interviewing method Computer Assisted Telephone Interviewing (CATI) was used as the preferred survey method because of the high overall telephone penetration in the EU28 countries, Iceland and Norway ensured the representativeness of the results. Interviewers conducted the survey at national level using a central programme recording directly all survey answers and storing them in one location. 1.4.Languages of interviewing Interviews were conducted in 28 languages: the 24 official European Union languages, Luxembourgish, Russian, Icelandic and Norwegian. Language Bulgarian Croatian Czech Danish Dutch Country Bulgaria Croatia Czech Republic Denmark Belgium Netherlands 6

7 Language Country English Ireland Malta United Kingdom Estonian Estonia Finnish Finland French Belgium France Luxembourg German Germany Austria Luxembourg Greek Greek Cyprus Hungarian Hungary Icelandic Iceland Irish Ireland Italian Italy Latvian Latvia Lithuanian Lithuania Luxembourgish Luxembourg Maltese Malta Norwegian Norway Polish Poland Portuguese Portugal Romanian Romania Russian Estonia Latvia Slovak Slovakia Slovene Slovenia

8 EUROPEAN COMMISSION Language Spanish Swedish Country Spain Sweden Finland 1.5.Sampling frames, sample sizes and sampling design The sampling approach was identical to the one used for the Market Monitoring Survey in The survey took place in the EU28 Member States as well as Iceland and Norway. The target population includes all people aged 18 and above, resident in the country surveyed and having sufficient command of (one of) the respective national language(s) to answer the questionnaire. In addition, we only select persons living in private households, excluding prisoners, as well as residents of retirement homes, etc. who are difficult to contact in a telephone survey. In every country, a random sample representative of the national population aged 18 or over was drawn, i.e. each person belonging to the target universe had a chance to participate in the survey. For some countries, suitable telephone number register(s) are available for both fixed and mobile lines, whilst for other countries only register(s) for either fixed or mobile lines can be used or even no register exists at all. In case no register was available, RDD 1 -numbers were generated. The following variables for stratification were used: gender, age, region and level of urbanisation, as far as the information was available in the sample frame(s). A dual sampling frame was introduced: Mobile sample: potential respondents within a given country that can be reached via a mobile line (regardless of whether they can also be reached via a fixed line). As such, this sample includes respondents from both the mobile only and mixed population. % Mobile sample = Proportion of mobile lines Total population of phone numbers = M + MF (M + MF) + (F + MF) Fixed sample: potential respondents within a given country that can be reached via a fixed line (regardless of whether they can also be reached via mobile line). As such, this sample includes respondents from both the fixed line only and mixed population. %Fixed line sample = Proportion of fixed lines Total population of phone numbers = F + MF (M + MF) + (F + MF) F = fixed only; M = mobile only; and MF = mobile and fixed For example, Germany was set to have the following proportions in the study: 83% mixed, 9% fixed only, 8% mobile only. Therefore the local teams composed a gross 1 Random Digit Dialling. With RDD, software is used to generate new telephone numbers, starting from a list of starting numbers. New telephone numbers are created and used by adding and subtracting digits in the existing telephone number. The composition of the starting number is important here for obtaining sufficient geographical spread. 8

9 sample of 50% fixed numbers, defined as: ((83%+9%)/(83%+9%)+(83%+8%)) and 50% mobile numbers ((83%+8%)/(83%+9%)+(83%+8%)). In order to further guarantee the representativeness of the sample, the time of calling was predominantly weekday evenings, with interviewing before only authorised upon specific request with a motivated rationale. In case of interviews conducted during the weekend or appointments set up upon respondent request, calls could take place all day long. Also, the birthday rule question was included for landlines to ensure a random selection procedure and minimise potential bias related to the person who would answer the call. No quota was set for socio-demographic variables such as gender or age. However, during fieldwork the overall sample intake was monitored daily, to follow up on the overall composition of the sample on gender, age, region and the possession of a mobile and/or a fixed phone in accordance with the sampling approach adopted. 1.6.Questionnaire translation and scripting GfK conducted a translation and review of the new and amended questions. Seven working days were foreseen for national experts to review the translations of the new questions. An additional 2 days were available to the Contracting Authority to review the comments on the translations before passing them on to GfK. The suggested linguistic changes were checked with the professional translators and, when necessary, researchers in the national agencies. Whenever the requested changes were deemed appropriate, the changes were implemented in the questionnaire. When the feedback was related to the content of the question, instead of the translation itself, no changes were made in order to maintain consistency in the question wording. Some small final linguistic changes were carried out based on the scripting checks conducted by the national agencies and the feedback received after the pilot interviews. Translations for trend questions were provided by the Contracting Authority. In order to guarantee comparability with the previous editions of the survey, GfK inserted those existing translations. Translations of those trend questions were sent for revision to the national experts as these questions should remain unchanged. 1.7.Pilot Following approval of the translated scripts, the scripted questionnaire was piloted in all participating fieldwork countries by the members of the GfK Network. The pilot took place between 23 and 30 March 2016 The Spanish pilot interviews were conducted a few days earlier than foreseen as the field provider moved offices in the period originally foreseen for the pilot. Overall, the large majority of the interviews were completed on the 29 th of March. The aim of this pilot was to test that the survey questionnaire, translation and script are all appropriate and correct, before the survey is run on a full scale. This pilot survey consisted of 25 interviews per country, using the CATI script that will be used in the fully launched survey. This way, the fieldwork tools could be fully tested in a live environment. In order to ensure that the survey tools were tested with a range of respondent types per country, the pilot will use a quota methodology to ensure a spread of respondents according to age, gender and education level. The table below outlines the quotas that were set per country in the piloting phase. By interviewing a spread of respondent types in the pilot phase according to gender, age and education level GfK Belgium PS was able to assess the questionnaire according to different respondent types.

10 EUROPEAN COMMISSION Criterion Gender Age Level of Education Targets per country Min. 10 male Min. 10 female Min. 7 aged Min. 7 aged Min. 7 aged 55+ Min. 7 Low level Min. 7 Medium level Min. 7 High level The results of the pilot indicated that there were no problems with the script in most countries, with the exception of Romania and Spain, where some small changes to the script were necessary that would compromise the interviews comparability across different waves. In agreement with the Contracting Authority, the pilot interviews for these two countries were, therefore, discarded. Since no issues were observed in the majority of the surveyed countries, GfK Belgium PS fully launched the main stage of fieldwork in all participating countries as originally planned. It was agreed that the pilot interviews conducted in all countries except for Romania and Spain would be included in the main sample. The interviews were conducted without boosting certain groups. Furthermore, although GfK proposed some optimisations related to trend questions, new questions and the translations in a Pilot Report submitted on 4 April to the Contracting Authority. These changes would not significantly impact the content or meaning of the questions and the pilot interviews would remain comparable with the main fieldwork sample. The tables below present the completed interviews per country, age group and education level, including the ones conducted in Romania and Spain which were discarded from the final sample. The pilot targets were not reached for all age and education subgroups in all countries. In particular, respondents in the youngest age group and those with low education were problematic to reach in some countries. 10

11 Table 1 Number of pilot interviews per country Country Number of completed interviews Belgium 30 Denmark 30 Germany 34 Greece 35 Spain 33 Finland 30 France 36 Ireland 30 Italy 30 Luxembourg 38 Netherlands 32 Austria 34 Portugal 30 Sweden 30 United Kingdom 30 Bulgaria 46 Cyprus 31 Czech Republic 30 Estonia 29 Hungary 33 Latvia 36 Lithuania 33 Malta 29 Poland 35 Romania 34 Slovakia 33 Slovenia 38 Iceland 30 Croatia 32 Norway 30 Total 981 Table 2 Number of pilot interviews per country by gender and age groups

12 EUROPEAN COMMISSION Gender Age Country Male Female years years 55+ years Belgium Denmark Germany Greece Spain Finland France Ireland Italy Luxembourg Netherlands Austria Portugal Sweden United Kingdom Bulgaria Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Romania Slovakia Slovenia Iceland Croatia Norway Total

13 Table 3 Number of pilot interviews per country by education level groups Education level Country Low Medium High (ISCED 0-2) (ISCED 3-4) (ISCED 5-8) Belgium Denmark Germany Greece Spain Finland France Ireland Italy Luxembourg Netherlands Austria Portugal Sweden United Kingdom Bulgaria Cyprus Czech Republic Estonia Hungary Latvia Lithuania Malta Poland Romania Slovakia Slovenia Iceland Croatia Norway Total

14 EUROPEAN COMMISSION 1.8.Fieldwork The main fieldwork launched on 7 April 2016 in all countries. The fieldwork continued until 8 May without interruption, although the majority of the countries did complete fieldwork earlier. Particularly for countries where the Orthodox Easter was celebrated, fieldwork was scheduled to finish a week in advance. The countries that completed fieldwork in three instead of the initially foreseen four weeks due to Orthodox Easter holidays were Poland, Romania, Cyprus and Greece. There were also three countries that took a few extra days than foreseen to complete fieldwork: Sweden, the Netherlands and Cyprus 2. A necessary extension was given to Cyprus due to a fire incident in the call center which rendered it inoperationsal for a short while. Figure 1 presents an overview of the proportion of completed interviews (of the total target) per week per country, while figure 2 provides an overview of the number of completed interviews per day during the main fieldwork period. Even though the field providers in all countries were thoroughly briefed that no interviewing should take place on Sundays, providers in Sweden and Norway neglected to follow this rule on the first Sunday of fieldwork (10 April), resulting in 100 interviews conducted on that day. This issue was immediately picked up by our field coordinators and was not present for the rest of the field period. Figure 1 Proportion of completed interviews per week Figure 2 Proportion of completed interviews per day (overall sample) 2 Cyprus is listed twice in the country overview. This is a consequence of the Orthodox holidays making fieldwork in Cyprus shorter in combination with a fire accident in the calling centre which prolonged field with a couple of days. As a result, fieldwork in Cyprus was completed after the Orthodox holidays and required 2 additional days. 14

15 Number of interviews daily breakdown APRIL/MAY # Interviews In all 30 countries a total of interviews were realised of which were completed within the EU28. Detailed information about the number of interviews achieved and the fieldwork end date per country is shown in the table below.

16 EUROPEAN COMMISSION Table 4 Number of market interviews and fieldwork dates per country Country # interviews Population 18+ Fieldwork end Belgium MAY-2016 Denmark MAY-2016 Germany MAY-2016 Greece APR-2016 Spain MAY-2016 Finland MAY-2016 France MAY-2016 Ireland MAY-2016 Italy MAY-2016 Luxembourg MAY-2016 Netherlands MAY-2016 Austria MAY-2016 Portugal MAY-2016 Sweden MAY-2016 United Kingdom MAY-2016 Bulgaria MAY-2016 Cyprus MAY-2016 Czech Republic MAY-2016 Estonia MAY-2016 Hungary MAY-2016 Latvia APR-2016 Lithuania MAY-2016 Malta MAY-2016 Poland MAY-2016 Romania APR-2016 Slovakia MAY-2016 Slovenia MAY-2016 Croatia MAY-2016 Iceland MAY-2016 Norway MAY-2016 EU TOTAL Response rate improvement measures A range of measures were put in place in order to minimise non-response for the survey, as follows: The pilot and the implementation of the pilot feedback further ensured the quality of the questionnaire, both in terms of content and technical aspects. Interviewers all followed thorough briefings on the survey in addition to generic interviewer trainings, and were instructed on how to minimise the non-response rate. 16

17 2. ANALYTICAL METHODOLOGY This section presents the methodology used for the data processing, weighting and estimation of the standard errors. 2.1.Data cleaning, processing and validation All data processing and analyses were centralised at GfK Belgium in order to ensure the quality of the data and analyses. The need for data editing was minimised by the preliminary measures implemented during the fieldwork, such as automatic controls on the responses and warnings on the screen for the interviewers to prevent incorrect answers being stored. The main stage of the data cleaning process consists of thorough quality controls on the data, including consistency and missing answers checks. Following the data cleaning stage, the raw data were processed for the analysis and reporting stages. The specific indicators and all breaks were computed in advance in order to produce the various data files required for the analysis. 2.2.Weighting The final weighting process consisted of the following three steps: Post-stratification weight, taking into account: age and gender Factor representing the population distribution across countries A post-stratification and design weight approach was applied, incorporating the telephone ownership dimension identical to the weighting approach used for the Market Monitoring Survey in However, due to difficulties experienced with achieving the required weighting efficiency it was decided to apply the standard weighting approach, stratifying the sample by age and gender and applying a factor post-hoc to reflect the population distribution across countries. These are explained in detail in the Evaluation Report. 2.3.Trend data The Contracting Authority provided the trend results, more specifically the micro-data and data tables. GfK created a single SPSS file containing all trend data up to 2014 and two separate files: one containing 2014 data for comparison with 2016 and the other containing currently collected 2016 data. These files served as a single basis for all analyses conducted allowing full control over the data increased the efficiency with which GfK was able to respond to specific requests from the Contracting Authority regarding trend data. A possible issue with the target population in previous editions of the survey was discussed during the meeting on reporting. Specifically, the results from previous editions of the survey are based on a sample of 15+ compared to 18+ respondents. To address this issue regarding trend data it was agreed that for 2014 data a new weight was to be computed based on age and gender distributions to compare to 2016 data. In addition, data from all waves before 2014 was used for trend comparisons based on existing samples (population 15+) and using the existing weights provided by the Contracting Authority (also based on the 15+ population distribution). When comparing 2014 data to previous waves, the original weight was used based on the 15+ population distribution. The difference in sampling was clearly communicated in the final report when trend results were reported.

18 EUROPEAN COMMISSION Weight computations based on age and gender for the 18+ respondents on the 2014 data indicated that the overall efficiency of the weights was only around 64% in Ireland. This was due to filtering out relatively many year olds which made the weights relatively high for this category. In spite of the low efficiency this was accepted since the remaining sample in this age category is limited. To summarise, the following weighting procedures will be used to calculate the results presented in the final report, additional analyses, and country profiles: 2016 trend comparisons: Population, gender & age weighting (18+) 2014 trend comparisons: Population, gender & age weighting (18+) 2014 trend comparisons: Population, gender & age weighting (15+) 2012 trend comparisons: Population, gender & age weighting (15+) 2011 trend comparisons: Population, gender & age weighting (15+) 2010 trend comparisons: Population, gender & age weighting (15+) 2009 trend comparisons: Population, gender & age weighting (15+) 2008 trend comparisons: Population, gender & age weighting (15+) 2006 trend comparisons: Population, gender & age weighting (15+) 2.4.Computation of derived indicators A list of composite indicators to be focused on in the final report (computed in accordance with the results presented in the Consumer Conditions Scoreboard 2015) has been provided to GfK by the Contracting Authority. These indicators will serve as the core variables of interest in the Final Report structure, based on the three pillars used in the Scoreboard: knowledge and trust, compliance and enforcement, and complaints and dispute resolution. GfK has computed the composite indicators in accordance with the guidelines supplied and provided the syntax outlining the computations to the Contracting Authority, pending final approval. No issues were faced during the computations. 2.5.Estimation of standard errors and statistical significance The most common formula for the calculation of the standard error was applied: the standard deviation divided by the square root of the sample size (SD/ n). The same formula was used for the Market Monitoring Survey Significance testing will be done by using the same formulas as for the Market Monitoring Survey 2015: The formula for proportions/percentages is based on the percentages (p1 and p2) and sample sizes (n1 and n2), whilst the formula for means is based on the means (m1 and m2), sample sizes (n1 and n2) and standard deviations (std1 and std2). The correct formula will be selected as needed. t = p1*(1-p1) n1 p1 -p2 + p2*(1-p2) n2 t = std1 2 n1 m1 -m2 + std22 n2 18

19 3. RESPONSE RATES 3.1.Response rates per country The following distinctions were made regarding the response information: Eligible units (belonging to the target population) o o o Full responses Only partial responses Non-response Non-responding units with unknown eligibility Non-eligible units (not belonging to the target population) Response rates were then computed using the by AAPOR 3 defined calculations: RR1 = RR3 = RR4 = I (I+P)+(R+NC+O)+(UH+UO) I (I+P)+(R+NC+O)+e(UH+UO) (I+P) (I+P)+(R+NC+O)+e(UH+UO) RR = Response rate; I = Complete interview; P = Partial interview; R = Refusal and break-off; NC = Non-contact; UH = Unknown if household/occupied housing unit; UO = Unknown, other In RR3 and RR4, an estimate e is introduced, which is the estimated proportion of cases of unknown eligibility that are eligible 4. The default estimate was used in the calculated response rates. This estimate is based on the proportion of eligible respondents among all contacts in the sample for which a status was obtained. Whereas only completed contacts are considered as interviews in RR1 and RR3, RR4 also counts partial interviews in the numerator. 3 American Association for Public Opinion Research 4 AAPOR Response Rate Calculator in Excel, accesible via Resources/For-Researchers/Poll-Survey-FAQ/Response-Rates-An-Overview.aspx

20 EUROPEAN COMMISSION The three response rates per country are listed in the table below. Country Response rate RR1 RR2 RR3 Austria 4% 4% 4% Belgium 6% 7% 8% Bulgaria 17% 20% 22% Cyprus 22% 22% 23% Czech Republic 6% 7% 8% Germany 4% 4% 4% Denmark 12% 14% 15% Estonia 26% 27% 28% Greece 12% 13% 14% Spain 4% 4% 5% Finland 12% 13% 13% France 4% 4% 4% the UK 4% 4% 4% Croatia 5% 5% 7% Hungary 13% 14% 14% Ireland 4% 4% 4% Italy 27% 28% 28% Lithuania 18% 19% 19% Luxembourg 3% 3% 3% Latvia 21% 22% 22% Malta 26% 26% 29% the Netherlands 4% 7% 8% Poland 6% 6% 6% Portugal 4% 14% 16% Romania 7% 7% 13% Sweden 6% 7% 7% Slovenia 12% 12% 16% Slovakia 9% 9% 24% Norway 5% 5% 6% Iceland 34% 35% 35% The highest response rates are observed in Iceland, Italy, Estonia and Malta. In contrast, Luxembourg, Austria, Germany, Spain, the UK and Ireland recorded the lowest response rates. By estimating the amount of eligible contacts amongst those with unknown eligibility, as done in RR3 and RR4, response rates increase especially in Portugal, Slovakia, and Romania. Furthermore, RR4 shows that a relatively high proportion of interviews were only partially completed in Slovenia, Slovakia, and Romania. Those countries benefit from also including partial interviews in the numerator. 20

21 doi:[number]