HELLENIC STATISTICAL AUTHORITY

LABOUR STATISTICS SECTION

PEIRAIOS 46 & EPOINTON str, GR 185 10, PIRAEUS, GREECE

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.

Greece, total country.

2007

2011

2016

2022

Not applicable.

Number, EUR.

12-month-period before the date of the interview.

2022 AES fieldwork period: 15 May 2023 - 31 July 2023

At European level:

Basic legal act: Regulation (EU) 2019/1700

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

At national level:

An "Approval, announcement, assignment, and distribution of the costs of conducting the statistical survey regarding the participation of the country's population in educational activities (Adult Education Survey), reference year 2022, approval of the use of statistical instruments and determination of their remuneration" was published in Government's Official Jοurnal (FEK 177/B/18.01.2023).

Not applicable.

The issues concerning the observance of statistical confidentiality by the Hellenic Statistical Authority (ELSTAT) are arranged by articles 6, 7 and 8 of the Law 3832/2010, as amended by article 90 paragraph 8 of Law 3842/2010 and by article 10 of Law 3899/2010, as well as by article 8 of Law 2392/1996, which was brought back into force, in accordance with article 90 paragraph 8 of Law 3842/2010.

Furthermore, ELSTAT disseminates the statistics in compliance with the statistical principles of the European Statistics Code of Practice and in particular with the principle of statistical confidentiality.

ELSTAT protects and does not disseminate data it has obtained or it has access to, which enable the direct or indirect identification of the statistical units that have provided them by the disclosure of individual information directly received for statistical purposes or indirectly supplied from administrative or other sources. ELSTAT takes all appropriate preventive measures to render impossible the identification of individual statistical units by technical or other means that might reasonably be used by a third party. Statistical data that could potentially enable the identification of the statistical unit are disseminated by ELSTAT if and only if:

a) these data have been treated, as it is specifically set out in the Regulation on Statistical Obligations of the agencies of the Hellenic Statistical System (ELSS), in such a way that their dissemination does not prejudice statistical confidentiality

or

b) the statistical unit has given its consent, without any reservations, for the disclosure of data.

The confidential data that are transmitted by ELSS agencies to ELSTAT are used exclusively for statistical purposes and the only persons who have the right to have access to these data are the personnel engaged in this task and appointed by an act of the President of ELSTAT.

Issues referring to the observance of statistical confidentiality are examined by the Statistical Confidentiality Committee (SCC) operating in ELSTAT. The responsibilities of this Committee are to recommend on:

a) the level of detail at which statistical data can be disseminated, so as the identification, either directly or indirectly, of the surveyed statistical unit is not possible;

b) the anonymization criteria for the microdata provided to users;

c) the granting to researchers access to confidential data for scientific purposes.

For more information, refer to the website of the Hellenic Statistical Authority:

1. Provision of anonymized microdata of statistical surveys
2. Access to confidential data for scientific purposes

At the end of December of each year ELSTAT publishes a release calendar announcing the precise release dates for the next year. The calendar is distributed to the press and is available free of charge to all interested parties.

The calendar is disseminated on ELSTAT's website (http://www.statistics.gr) at the link: "Press Releases Calendar”.

The data are released simultaneously to all interested parties through the Press Release entitled ''Adult Education Survey'' which is made available to the media at 12:00. At the same time, the press release is distributed electronically to the subscribers. The press release is subsequently posted on the website of ELSTAT (http://www.statistics.gr).

There is no internal government access to the data before their release to the public.

Every 6 years.

A press release with 2022 AES results for Greece was published in 26/10/2023.

A number of tables with data on participation in education is uploaded at https://www.statistics.gr/en/statistics/-/publication/SJO18/-, together with the survey questionnaire.

No online database.

Not applicable.

Micro-data are available to users for free upon request to the Hellenic Statistical Authority, Statistical Information and Publications Division. Path: Products and Services>Statistical data request http://www.statistics.gr/en/statistical-data-request.

Confidentiality of the data is assured by the implementation of anonymization criteria.

Users that need tailor-made tabulations of data can apply using an on-line form in ELSTAT's website.

Not applicable.

Quality report and survey questionnaire are uploaded in ELSTAT's website.

Not applicable.

Quality measures and other survey information are included in annex “EL - QR tables 2022 AES (excel)”.

Adult Education Survey complies with Eurostat standards concerning the selection and the size of the sample, the design of questionnaires, and the information collected.

Interviewers are trained before the beginning of the fieldwork and collected questionnaires are checked for completeness and quality of information. In case of important problems in the completed interviews respondents were recontacted by ELSTAT personnel.

The final data base is checked for data validity by ELSTAT and Eurostat.

Data were collected exclusively by electronic means (CAPI), thus helping to significantly accelerate the production of research results.

Basic quality indicators (sampling errors, response rates) are computed and included, see annex “EL - QR tables 2022 AES (excel)”.

Response rates are at an acceptable level (see item 13.3.3).

The indicator "Participation rate in non‐formal education and training (age 25‐69)" fulfils the criterion provided by AES regulation while indicator “Participation rate in formal education and training (age 18‐24)” misses the regulation criterion (see item 13.2.1).

The main users of Adult Education Survey data are international organizations (European Commission, OECD, Unesco, IMF) and a large number of national authorities and institutions (Ministries, Universities, Institutes) but also the press, researchers and the general public.

ELSTAT does not carry out a user satisfaction survey specifically for the users of the Adult Education Survey. However, there is continuous communication with the main users of survey results and their opinion and remarks are taken into account in the development, the production and dissemination of its results.

The Adult Education Survey dataset covers all variables that are included in the relevant Council and Commission Regulations.

Not applicable.

As in any sample survey, the main source of error is the fact that we infer population values from a random subsample of the population and the bias introduced by non-response or the inadequate design of the sample.

See item 13.2 for further information.

2022 Adult Education Survey's sample is a random sample of persons, selected among the members of the households that were surveyed for the Labour Force Survey during the period 2^nd Quarter 2021 – 4^th Quarter 2022.

The household’s members were grouped by NUTS 3 area, sex and age group (“young” and “old”) and a systematic sample of persons was selected from each group.

SPSS Complex Samples was used for the computation of standard errors, confidence intervals and coefficients of variation.

For the computation, strata were the NUTS 3 areas and clusters the four groups defined by sex and age-group in each NUTS 3 area.

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

The main issues concerning non-sampling errors in 2022 AES are:

1. The shortcomings of the sampling frame: The original frame for the AES sample was the households selected and surveyed for the LFS survey. As a result, there are two sources of error: a) the coverage issues with the "original" LFS sampling frame and (b) the problem with the selected for the LFS households that did not respond in LFS and, consequently, they were not included in the AES sampling frame.
2. Non-response in AES.
3. Problems due to recording of incorrect information (due to proxy interviews, misunderstanding of questions, etc.).

The sampling frame used for AES, for the first stage of sample selection, is an area frame which comprises all building blocks enumerated during census (and included at least 1 dwelling that was used as permanent residence of a private household).

The main shortcomings of such a frame are:

1. The number of dwellings where a private household is residing in an area at the time of the interview is not equal with the number of households residing during census. This may result in households with zero probability of selection. It affects also the probability of selection of primary sampling units which is proportional to the numbers of households residing in a particular group of blocks at the census (and not at the time of the survey).
2. Dwellings that are not used for household residence are included in the frame, resulting in over-coverage.
3. Does not include information on household characteristics (number of household members, age of household members) other than the household address.

In the second stage of sample selection for AES 2022, the sample was a subset of the persons surveyed (and who responded) for the LFS. In this second stage, there were no out of scope persons (since the selection took into account the age of the respondent at the time of the interview) and the over-coverage resulted from persons/households that did not live in the dwelling at the time of the interview.

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

Not applicable.

In general, main sources of measurement error are poor design of a questionnaire, response bias (e.g., social desirability bias), and proxy interviews. In order to avoid - to a possible extent - measurement errors we followed the next actions:

• The questionnaire of the survey followed closely Eurostat's model questionnaire, with few deviations.
• All interviewers that worked in data collection attended a seminar, where the survey questionnaire was presented and detailed instructions were provided.
• Proxy interviews were allowed but efforts were made to keep their percentage low.
• Editing programs identified illogical or improbable answers, and in many cases respondents were recontacted in order to resolve the issue.

Variables with highest non-response were variables associated with the characteristics of 3rd, 4th, and 5th non-formal activities and variables related with the cost of non-formal activities. It is probable that the high non-response of these variables is related to proxy interviews (it is higher for proxies) and also to the burden from the repeated question on the characteristics of the non-formal activities.

There was no treatment of item non-response, with the exception of variable HHINCOME.

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

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

Data entry programs included a full set of checks identifying errors in the flow or in the data values. As a result, the initial database was free of such errors.

A full set of data checks was implemented which allowed to identify major logical inconsistencies in the data set and correct them by re-contacting the surveyed persons.

No such models were used.

Data collection took part during May-July 2023.

The results of the survey were published on 26 October 2023, thus 3 months after the end of the data collection period.

Not applicable

Data collection took part during May-July 2023.

The results of the survey were published on 26 October 2023, thus 3 months after the end of the data collection period.

Data was sent in September 2023, 3 months after the end of the data collection period.

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

Not applicable.

There were no differences in the implementation of the survey in different areas of Greece, since the same definitions, type of sampling, questionnaire and data treatment was used throughout the whole country. The only difference between the implementation of the survey in different geographical areas is the response rate.

There are no differences or deviations from the AES concepts and definitions. The Greek questionnaire followed closely the Eurostat's model questionnaire.

No additional variables related to COVID-19 were collected.

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

Not applicable.

There have been 2 main changes in the implementation of the AES survey, compared to the 2016 AES:

a) Data collection mode was CAPI (while in 2016 AES was mainly PAPI) and

b) 2022 AES sample was a sample of individuals (selected among the members of the households that were surveyed for LFS) while the 2016 AES sample was a sample of households.

As far as it concerns definitions and questionnaire design there were no significant differences between the 2 implementations of the survey.

In general, there is no indication of important breaks between the 2 rounds of the survey.

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

Not applicable.

There are no available administrative sources that could be compared with AES results.

AES results can be compared with results from other surveys, like CVTS and LFS - taking of course into account the important differences between AES and these surveys.

The most recent CVTS results have as reference period the year 2020, and as a result any comparison with the 2022 AES can refer only to the order of magnitude. According to CVTS, about 182,000 persons employed (in enterprises with 10 and more persons employed) participated in CVTS during 2020, while, according to AES, 189,000 employees that were working in a local unit with 10 or more persons, and participated in a non-formal activity that was job-related, paid by the employer and the respondent was employed when the activity started. If we assume that the participation in CVTS programs during 2022 was not substantially different from 2020, we may conclude that the AES estimate seems plausible.

The comparison with LFS data reveals that LFS underestimated (compared to AES) the adult participation in learning activities. This is probably due to the difference in the purpose of the two surveys, to the fact that AES questions on participation in education are more detailed, as well as to the methodological differences between the surveys (e.g., sample size, sampling unit).

See also table 15.3 “Coherence - cross-domain” in annex “EL - 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.

The cost of data collection was 71.100 euros.

Not applicable.

Not applicable.

Not applicable.

The sampling frame that was used were the households surveyed (and which responded) to the Labour Force Survey in the period Q2_2021 - Q4_2022. To be precise, the sampling frame comprised of the 6th rotation group of the total sample of each quarter.

All the persons in these households were assigned in different groups (defined by NUTS 3 area, sex and age-group) and a random systematic sample of persons was selected in each of these groups. The sampling interval at this stage was set up so as to end up with the desired number of persons in the total sample.

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

Every 6 years.

Data were collected by personal interviews (either telephone or face to face interview) using CAPI in both cases. Since the persons participating in the survey, have already provided information for the LFS, data for some characteristics (for example, respondent's country of birth) were imputed from LFS data.

The data entry program included checks for the plausibility of the answers concerning HATYEAR and FEDSTARTYEAR.

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

After data entry the data file was processed by a program that incorporated the rules for flow and accepted values for each variable as well as plausibility checks.

Coding was implemented at 2 stages: The information for certain variables (field of highest education, field of formal and non-formal activities, sector of economy, and occupation) were collected through open questions and the (verbal) descriptions were entered into the data base. Then, coding programmes assigned automatically codes to the verbal descriptions. The cases where the programs failed to assign a code, were coded manually.

Imputation was used only for variable HHINCOME. This variable was collected either as exact value or as broad categories of income. Exact values of HHINCOME for the non-respondents and for those that reported their household income in bands were produced using a generalized linear model.

The final personal weights were computed in 4 steps:

1) A “design weight’ - equal to the LFS design weight - was assigned to each person.

2) The design weight in step 1 was multiplied by a coefficient equal to N_group/S_group where

• N_Group = The number of household members in each particular group (that is, combination of NUTS 3, sex and age-group)
• S_Group = The number of persons originally selected in each particular group

3) A non-response correction was computed for each Group (persons originally selected/persons responded).

4) Finally, post stratification correction factors were computed for each individual so that estimations from the Adult Education Survey are consistent with estimated (from the 4ht quarter estimations of the Labour Force Survey) population totals for post-stratification cells - that is, NUTS 2, sex and age-groups combinations.

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

Not applicable.

Not applicable.

None.