Methodology

The second Cedefop European skills and jobs survey (ESJS2) project was implemented during 2018-22. The concept and main English questionnaire was originally developed by Cedefop experts Konstantinos Pouliakas and Marco Serafini, with the support of the ESJS2 expert working group. The project’s contracting authority (KANTAR PUBLIC) provided support and expertise in developing, testing, and translating the second ESJS2, as well as carrying out the fieldwork in all countries and providing a first analysis and draft of the report.

Prior to launching the main survey, robust cognitive testing and a pilot survey were carried out. The cognitive testing took place in six countries (Germany, Ireland, Greece, Spain, France and Poland) between May and June 2020. Both survey modes, (computer assisted telephone interviewing (CATI) and computer assisted web interviewing (CAWI)), were tested on 180 participants (30 in each country), with the overall aim of determining the reliability and validity of the survey questions.

The master English questionnaire was then translated into national versions using a robust TRAPD methodology and translatability assessment.

A pilot survey (around 40 interviews per country) was conducted in all participating countries between 10-24 February 2021 as a dress rehearsal for the main survey.

The data of the ESJS2 were collected among adults aged 25-64 who are in wage and salary employment (i.e. paid employees, excluding those in self-employment and family workers), living in private households and whose usual place of residence is in a territory of each of the EU-27 Member States, Iceland and Norway. The main fieldwork took place between May and August 2021. A total of 46 213 interviews (both telephone and online) were conducted.

In all countries apart from Cyprus and Malta, a dual mode design (telephone and online) was implemented. In each country 500 respondents were sampled using a probabilistic telephone sample, except for Finland, Iceland and Norway where population registers were used. The telephone interviews were complemented by online interviews (CAWI) run by reliable panel providers using a quota methodology. In Cyprus and Malta, where the panels were not sufficiently representative of the underlying population, a single mode probabilistic telephone design was implemented.

The telephone survey was executed using mobile phones and landlines, both selecting a random sample from the population aged 25+ using the random digit dialling (RDD) technique. The probabilistic telephone sample surveyed for the ESJS2 ensures almost full coverage of the target population in all countries, with all phone owners/users having a non-zero probability of selection.

For the non-probabilistic online sample, the majority of panellists were sourced from the Kantar LifePoints/ Profiles access panel. The top up sample of online panellists helps maximise the total sample size per country, ensuring representativity of the adult employee population in demographic and skills dimensions. Quotas based on the adult working population aged 25 and over were set based on the Eurostat labour force survey 2019; strong quotas based on gender, age and region in each country and flexible/monitoring quotas based on education (ISCED), occupation (ISCO) and sector (NACE).

Overall, 30 701 responses were collected via CAWI and 15 512 via the CATI mode. ESJS2 sample sizes range from a minimum of about 1 000 observations in some countries (e.g. Cyprus, Denmark, Iceland, Malta) to a high of over 3 000 respondents in Germany, Spain, France, Italy and Poland - an average of about 1 600 adult workers per country.

Weights were calculated independently for each survey mode before being combined for the final responding sample. The benchmark population data used to calibrate the final net samples was taken from the most recent European labour force survey data. The sample was weighted to match population data on gender by age, gender by educational attainment, gender by industry, gender by occupation and region. Targeted adjustments were applied to account for differences between the telephone and online samples mainly attributed to selection effects. The final weights are adjusted so that countries with larger eligible populations have more influence (weight) on the pan European survey estimates than those with small eligible populations, reflecting differences in population size.

More information about the ESJS2 sampling methodology is available at the ESJS2 sampling methodology report.