EWCS 2010 - Methodology

In an effort to provide comparable and reliable data on working conditions across Europe, Eurofound has developed a unified methodological approach and quality assurance system. Over the years, the methodology has been improved, new concerns have been integrated and the geographical scope has expanded with the enlargements of the European Union.

For more information please see the Technical report (in English only).

The sample used in the European Working Conditions Survey is representative of those aged 15 years and over (16 and over in Spain, the UK and Norway) who are in employment and are resident in the country that is being surveyed. In each country, a multi-stage, stratified random sampling design was used.

In the first stage, primary sampling units (PSUs) were sampled, stratifying according to geographic regions (NUTS 2 level or below) and level of urbanisation. Subsequently, in each PSU households were sampled. In countries where an updated, high quality address or population register was available, this was used as the sampling frame. If such a register was not available a random route procedure was applied. In the 5th EWCS, for the first time, the enumeration of addresses through this random route procedure was separated from the interviewing stage. Finally, a screening procedure was applied to select the eligible respondent within each household.

The target number of interviews was 1,000 in all countries, except Slovenia (1,400), the UK, Italy and Poland (1,500), Germany and Turkey (2,000), France (3,000) and Belgium (4,000). The Slovenian, French and Belgian governments made use of the possibility offered by Eurofound to pay for an addition to the initial sample size. The number of interviews completed in each country is summarised in the following table.

Number of completed interviews:

For further information, please see the Sampling implementation report.

The 5th EWCS included a few open questions, to record the respondents’ occupation and the economic activity of the organisation or company they work for. The answers were coded according to international classification systems for occupation (ISCO88 and ISCO08) and activity of companies and organisations (NACE Rev 1.1 and 2.0).

For both ISCO and NACE the answers were coded using both the current and the previous version of the classification system. The coding using the previous versions is needed to make comparisons with previous waves of the EWCS.

For the 5th EWCS, ISCO coding was carried out on the 4-digit level, allowing for very specific distinctions between occupations. One of the reasons for doing this was to enable an appropriate transition from the old ISCO classification to the most recent one.

These open questions were not the only ones to require coding. The question about the respondent’s level of education was recoded into ISCED categories, in order to make the country-specific education categories internationally comparable. Also, the income questions referred to the national currency in each country, which was afterwards converted into Euro according to the exchange rates at the time of conversion (01 March 2010).

For more information, please see the Coding report.

As in previous waves of the EWCS three types of weights need to be applied to ensure that results based on the 5th EWCS data can be considered representative for workers in Europe.

1. Selection probability weights
   Because of the way the sampling process is designed, people in households with fewer workers have a greater chance of being selected into the sample than people in households with more workers. For example, in a household with one person in employment, the probability of this person to be selected is 100%, whereas it drops to 50% for persons in a household with two people in employment. Selection probability weights (or design weights) are constructed to correct for this.
2. Post-stratification weights
   Because of differences in the willingness and availability to participate in the survey certain groups are overrepresented and other groups are underrepresented in the EWCS sample. To ensure that the results accurately reflect the population of workers in each country post-stratification weighting is needed. The weights are calculated by comparing the EWCS to the Labour Force Survey (Eurostat) with regard to the gender, age, region, occupation and sector of economic activity of the respondents. 
3. Supra-national weights
   The differences between countries in the size of their workforce, are not (fully) reflected in the sample size in each country. To ensure that larger countries weigh heavier in the EU level results, supra-national weights need to be applied when performing analyses on the European level.

To ensure high quality of the 5th EWCS data each stage of the survey was carefully planned, closely monitored and documented, and specific controls were put in place. When compiling the draft questionnaire and planning the survey management, recommendations outlined in the Quality report of the 4th European working conditions survey for further improving the survey methodology as well as Eurofound’s experiences with previous surveys were taken into account.

Quality control mechanisms were included in the tender specifications to ensure the survey was implemented in accordance with ‘best practice’ and that each stage was documented in detail. The contractor for the 5th EWCS has compiled a Quality assurance report which provides a summary of the quality assurance activities performed during the survey preparation, fieldwork and data processing.

Finally, after the fieldwork an External quality assessment report took place to analyse the extent to which the quality criteria outlined in the European Statistical System (ESS) were met. It concluded that the 5th EWCS was based on a sound quality assurance framework and that notable improvements had been made compared to the previous wave, particularly with regard to questionnaire development and translation, sampling, data collection, and data processing. It commended the fact that process indicators were systematically monitored and recorded, both in real-time and after the fact, enabling users to make optimal use of the data as well as helping Eurofound to make further improvements in possible next waves of the survey.

The main problems identified in the assessment refer to the long data collection period, and the relatively low response rates in some countries; issues that need close attention should the survey be repeated.