Statbel (Algemene Directie Statistiek Statistics Belgium) - Federale Overheidsdienst Economie, KMO, Middenstand en Energie.

Statbel (Direction générale Statistique Statistics Belgium) - Service Public Fédéral Économie, PME, Classes moyennes et Énergie.

Thematic Division Society

Thematische Directie Samenleving

Direction thématique Société

North Gate - Koning Albert II-laan 16 - 1000 Brussels

The Adult Education Survey (AES) covers adults’ participation in education and training (formal - FED, non-formal - NFE and informal learning - INF). The 2022 AES focuses on people aged 18-69. The reference period for the participation in education and training is the twelve months prior to the interview.

Information available from the AES is grouped around the following topics:

• Participation in formal education, non-formal education and training and informal learning
• Volume of instruction hours
• Characteristics of the learning activities
• Reasons for participating
• Obstacles to participation
• Access to information on learning possibilities and guidance
• Employer financing and costs of learning
• Self-reported language skills

For further information see the 2022 AES legislation (http://ec.europa.eu/eurostat/web/education-and-training/legislation) and the 2022 AES implementation manual (http://ec.europa.eu/eurostat/web/education-and-training/methodology).

- Classification of Learning Activities (CLA, 2016 edition)
- International Standard Classification of Education 2011 (ISCED 2011)
- Classification of Occupations 2008 (ISCO 08)
- Classification of economic activities Rev. 2 (NACE Rev. 2)

AES covers all economic sectors.

Definitions as well as the list of variables covered are available in the 2022 AES implementation manual (http://ec.europa.eu/eurostat/web/education-and-training/methodology).

Individuals, non-formal learning activities.

Individuals aged 18-69 living in private households.

Belgium

AES 2008 (02/2008 - 06/2008)

AES 2011 (10/2011-03/2012)

AES 2016 (11/2016 - 03/2017)

AES 2022 (09/2022 - 03/2023)

Not applicable.

Number, EUR.

The reference period for the 2022 AES is the 12 months prior to the interview.

The survey ran from 01/09/2022 to 30/03/2023.

5 months before the survey, background variables were extracted from the national register.

At European level:

Basic legal act: Regulation (EU) 2019/1700

Implementing act: Commission Implementing Regulation (EU) 2021/861

At national level:

There is no specific Act at the national level to organize the AES.

The AES is organized under the Belgian Statistics Act: https://statbel.fgov.be/en/about-statbel/who-we-are/legislation. 

Not applicable.

The confidentiality is regulated in the Belgian Statistics Act: https://statbel.fgov.be/en/about-statbel/who-we-are/legislation

Privacy policy explained at our site: https://statbel.fgov.be/en/about-statbel/privacy/privacy-gdpr 

The dissemination of pseudonymised study data is strictly regulated. The procedure is described on our website: https://statbel.fgov.be/en/microdata-research. In order to get the permission of Statbel's Data Protection Officer team and finally as data controller, Statbel's director-general, the third party should follow a procedure and sufficiently motivate the proportionality and relevance of its request. The more confidential the information requested, the better the need for it should be motivated.

Results were disseminated in November 2023: https://statbel.fgov.be/en/themes/work-training/training-and-education/adult-education-survey.

Micro-data are disseminated to national users from November 2023 onwards.

https://statbel.fgov.be/en/calendar

Press release at same time as publication of aggregated data on website.

Every 6 years.

See:

https://statbel.fgov.be/fr/themes/emploi-formation/formation-et-enseignement/education-des-adultes

https://statbel.fgov.be/nl/themas/werk-opleiding/opleidingen-en-onderwijs/volwasseneneducatie 

See:

https://statbel.fgov.be/fr/themes/emploi-formation/formation-et-enseignement/education-des-adultes

https://statbel.fgov.be/nl/themas/werk-opleiding/opleidingen-en-onderwijs/volwasseneneducatie#figures 

No Belgian online database.

Not applicable.

The dissemination of pseudonymised study data is strictly regulated. The procedure is described on our website: https://statbel.fgov.be/en/microdata-research. In order to get the permission of Statbel's Data Protection Officer team and finally as data controller, Statbel's director-general, the third party should follow a procedure and sufficiently motivate the proportionality and relevance of its request. The more confidential the information requested, the better the need for it should be motivated.

Not applicable.

Not applicable.

See this quality report, sections 13 and 18.

Not applicable.

See:

https://statbel.fgov.be/fr/themes/emploi-formation/formation-et-enseignement/education-des-adultes#documents

https://statbel.fgov.be/nl/themas/werk-opleiding/opleidingen-en-onderwijs/volwasseneneducatie

Quality indicators are calculated for all statistics on a yearly basis - for AES this is every 6 years. Not publicly available.

Quality assessment is annually performed at Statbel for every survey. 

Before publication all data are examined by the validation unit. This includes checks on internal coherence of results and comparison with last year results and publication of similar data by other institutes.

Quality assessment is annually performed as mentioned in section 10.7.

Before publication all data are examined by the validation unit. This includes checks on internal coherence of results and comparison with last year results and publication of similar data by other institutes.

Relevance: medium high, given that this survey is only organized every 5/6 years there is less relevancy after a few years as LFS also obtains some info about education and training. Only the information on languages is still relevant after a few years as this information is not surveyed elsewhere.

Accuracy: medium, standard error +/- 0.5% for total figures, figures for multiple breakdowns offer only rough estimates.

Timeliness: high, data of 2022-2023 are available to public by November 2023 of reference period.

Punctuality: high, data delivery within deadline set by the legislation.

Comparability: medium, different questions in LFS on training & education, time series deliver plausible trends although different survey and sampling techniques are used every year.

Coherence: no other (high quality) data source available.

AES statistics are considered to be of good quality thanks to a harmonised production process (i.e. legislation, AES manual). However, like any other survey, it is based on a sample of the population meaning that results are subject to the usual statistical errors of measurement.

National quality reports provide users with basic information on quality at national level and give further explanations about the possible weaknesses of the sampling methods used at national level and of the final national effective sample of the survey.

The regional statistical offices (Statistics Flanders: https://www.statistiekvlaanderen.be/, Statistics Wallonia: https://www.iweps.be/, Statistics Brussels: https://ibsa.brussels/) publish regional figures and special reports on lifelong learning.

Not administered.

All variables as required by the legislation are included.

Not applicable.

See below.

Random error due to sample size.

The sampling error reflects the fact that only a particular sample was surveyed rather than the entire population.

The sampling frame (see 18.1.) adequately covers the population under study, being the population living in the Belgian territory in private household aged 18-69 years.

It is estimated by the standard error and can be expressed by the square root of the estimate of the sampling variance.

The estimation of the sampling variance should ideally take into account the sampling design (e.g. the stratification).

See table 13.2.1 “Sampling errors - indicators for 2022 AES key statistics” in annex “BE - QR tables 2022 AES (excel)”.

More information on methodology for calculating precision estimates is detailed in the sections below.

See below.

See below.

See table 13.3.1.1 “Over-coverage - rate” in annex “BE - QR tables 2022 AES (excel)”.

Not applicable.

1) Measurement errors: unknown, of course people can deliberately report false values.

2) Questionnaire design and testing: questionnaire is controlled for routing and other errors by a test panel before launching the survey.

3) Interviewer training: interviewer training was organized for the CAPI interviews.

See below.

See table 13.3.3.1 “Unit non-response - rate” in annex “BE - QR tables 2022 AES (excel)”.

See table 13.3.3.2 “Item non-response rate” in annex “BE - QR tables 2022 AES (excel)”.

No information.

No information.

See below.

Results of AES 2022-2023. Fieldwork ended in March 2023, results sent to Eurostat in September 2023, results published November 2023 (both national and at Eurostat).

Not applicable.

See table 14.2 “Project phases - dates” in annex “BE - QR tables 2022 AES (excel)”.

Not applicable.

See table 15.1 “Deviations from 2022 AES concepts and definitions” in annex “BE - QR tables 2022 AES (excel)”.

No additional variables related to COVID-19 were collected.

Not applicable.

Results are comparable over the years 2007-2023. 

See table 15.2 “Comparability - over time” in annex “BE - QR tables 2022 AES (excel)”.

Not applicable.

The product of AES is to comprehensively measure participation in education and training. However, this research is only measured every six years. The annual Labour Force Survey (LFS) focuses on participation in the labour market. Since LFS is organized more frequently, LFS is better suited for e.g. to monitor lifelong learning.

The differences in recruitment for the surveys and the way lifelong learning (LLL) is measured results in the fact that we do not achieve identical results. The method of data collection differs, formal and non-formal education and training are surveyed differently, only participation in education and training in the last year is measured, and guided on-the-job training is not included in LFS. Only the last factor already leads to 6.7 pp difference in LLL in 2022 AES.

See also table 15.3 “Coherence - cross-domain” in annex “BE - QR tables 2022 AES (excel)”.

Not applicable.

Not applicable.

AES results for a given data collection round are based on the same microdata and results are calculated using the same estimation methods, therefore the data are internally coherent.

1) Costs and burden of the survey

Salary cost: ca. 130.000 €

Printing and sending questionnaires: 115.000 €

Renumeration interviewers: 2000€ (for online interviewer trainings)

ICT: 170 days programming work

A Eurostat Grant (2021-BE-AES - 101051499) to change data-collection method covered parts of these expenses.

2) Average time used for answering the survey questionnaire

Total: average 20 min

CAWI: average 22 min

CAPI: average 12 min

No formal and non-formal education followed: average 12 min

Formal and/or non-formal education followed: average 24 min

3) Measures taken to reduce the cost and burden of the survey

First, we offered a CAWI, and among the non-respondents a sample is drawn for the CAPI.

HHINCOME is calculated using administrative sources.

HATMOTHER/HATFATHER/BIRTHMOTHER/BIRTHFATHER is only surveyed if administrative data is lacking.

Not applicable.

Not applicable.

The source of the raw data is described with more detail in the technical report and paragraphs below.

1) The sampling frame is constructed from the National Population Register (July, 2022) which allows to identify individuals and the household they are living in. The starting frame is a list of all citizens with residence in Belgium (i.e. 11.286.064 persons), excluding people living in collective households.

Further eligibility criteria are based on age. We calculated each individual's age at the beginning of the field work period.

The National Population Register allows to include basic socio-demographic characteristics such as sex, age and place of residence into the sampling frame.

2) The individuals for the 2022 AES sample have been drawn from the sampling frame by stratified simple random sampling. 

The targeted net sample size for this survey was 9,130 individuals: 3,310 young adults (18-24 years old) and 5,820 adults (25-69 years old). To balance cost considerations and profile diversity, we opted for a survey employing two data collection methods. Therefore, 75% of the net sample was planned to be obtained through web surveys (CAWI), while the remaining 25% was planned to by collected through face-to-face surveys (CAPI).

Due to field constraints, the initial step involved selecting primary sampling units (PSUs), which are segments of Belgian municipalities containing more than 300 individuals each. Out of a total of 2,237 available PSUs, we randomly chose 319 PSUs.

To enhance regional precision (NUTS2), we opted to increase the number of PSUs drawn in smaller regions. To determine the allocation of PSUs to each region, we used the square root of the population size (number of individuals) in each region as a basis. The PSUs were then ordered based on their median fiscal income, and a systematic draw method was employed to make the final PSU selections.

Within each age group, optimal (i.e. Neyman) allocation is applied to calculate the desired respondent sample sizes by region and sex: the desired respondent sample sizes are proportional to the products of the sampling frame sizes and the standard deviations of a study variable. The study variable chosen for this purpose is F-NF (formal or non-formal education activity). The standard deviation for F-NF by sampling stratum are calculated from AES 2016 and LFS 2021, provided age groups are 18-24 and 25-69.

The choice of variable F-NF results of a comparison of various allocation rules: optimal allocation using indicator variable FED (Formal Education activity) or NFE (Non-Formal Education activity), N-proportional allocation and square root N-proportional allocation (where N designates the sampling frame stratum sizes). It was found that all three optimal allocation rules and the N-proportional allocation rule yielded very similar results.

The final step of calculating the initial sample sizes consists of dividing the calculated respondent sample sizes by appropriate (estimated or expected) response rates. 

For our CAWI survey, we assumed a response rate of 25% for the young adults and 30% for the adults, based on estimates derived from previous CAWI surveys conducted by Statbel. The response rate turned out to be significantly lower than anticipated, standing at around 15% instead of the expected 30%. Consequently, we found it necessary to draw a second sample of individuals while retaining the same primary sampling units (PSUs). As a result, our overall gross sample expanded to encompass 44,851 individuals.

3) The weighting procedure was performed in one step. Calibration was also used to correct for non-response. It was done at individual level using truncated linear method. The calibration variables are:

• ISCED classification (0-200, 300-400, 500-800)
• Full crossing of 3 regions, 4 age groups and 2 sexes

Calibration was made using CALMAR2 in SAS.

See also table 18.1 “Source data” in annex “BE - QR tables 2022 AES (excel)”.

Every 6 years.

1) Methods used to gather data:

Self-administered web survey (CAWI) or computer-assisted personal interviewing (CAPI) with an interviewer.

2) Short description of the survey method:

The data were cleaned and only if the respondent was too old or the respondent did not answer the FED/NFE questions, the survey was considered as non-response.

Data were pre-validated by the EDAMIS system, that mainly checks for inconsistencies after data cleaning. 

All data are subjected to our internal validation team before publication and dissemination.

Criteria for validation:

Not applicable.

See table 18.5.1 “Imputation - rate” in annex “BE - QR tables 2022 AES (excel)”.

Not applicable.

Not applicable.

None.