Home > Statistics > Participation in adult education > 2006 > European comparison > Quality description: Adult Education Survey

Quality description: Adult Education Survey

• 1. Relevance of statistical information
• 2. Methodological description of the survey
• 3. Correctness and accuracy of data
• 4. Timeliness and promptness of published data
• 5. Accessibility and transparency/clarity of data
• 6. Comparability of statistics
• 7. Coherence and consistency/uniformity

1. Relevance of statistical information

The EU Adult Education Survey (EU-AES) results are part of the EU’s statistics on lifelong learning. The EU-AES is a sample survey undertaken in 29 EU, EFTA and EU candidate countries and coordinated by Eurostat. The data collection for this, the first such survey, was on a pilot basis and covered the period 2005–2008.

The countries participating in the survey were Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Estonia, Finland, France, Germany, Greece, Hungary, Italy, Latvia, Lithuania, Malta, the Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey and the United Kingdom.

The survey covered the following: participation in education and training by adults aged 25–64 (both formal education and training, i.e. leading to a qualification, and non-formal education and training, i.e. not leading to a qualification); special features of adult education and training (e.g. job-related v. non-job-related); informal learning; modules on social and cultural participation; foreign language skills; and IT skills.

The survey’s most important indicators are:

• Participation in formal, non-formal and informal education

• Non-participation and obstacles

• Field of learning

• Share of the job related non-formal education

• Volume of instruction hours in formal and non-formal education

• Employer financing and cost of learning in formal and non-formal education

• Language and ICT skills

• Social and cultural participation

The background variables common to all countries were sex, age, highest level of educational attainment, degree of urbanization (in the municipality where the individual resides) and labour force status.

The key concepts in the survey, namely formal education and training , non-formal education and training and informal learning , are based on Eurostat’s Classification for Learning Activities (2005).

The survey results will be used by many different bodies. For example, the survey provides the European Commission, international organisations ( OECD, ILO, IMF) and education and training researchers with a set of internationally comparable indicators. In addition, national institutions (e.g. government ministries) will have access to valuable information for the purposes of planning education and training, assessing education and training policy programmes and developing lifelong learning. Employers and trade unions, too, will be able to obtain information for the purposes of assessing the state of adult education and training and personnel training, and for considering the future needs of employees.

2. Methodological description of the survey

The EU-AES is a sampling survey. The data collection methods used for the period 2005–2008 different among the survey countries. Computer-assisted personal interviewing (CAPI) was used in the following countries: Cyprus, Estonia, Finland, France, Germany and the United Kingdom. Paper and pencil interviewing (PAPI) was used as the data collection method in Bulgaria, Greece, Hungary, Italy, Latvia, Lithuania, Poland, Slovakia and Spain. In Norway and Sweden the data collection method was a mix of CAPI and computer-assisted telephone interviewing (CATI).

The sampling method also differed among countries. Stratified simple random sampling was used in Austria, Cyprus, Estonia, Finland and Sweden. Simple random sampling was used in Latvia and Slovakia. In France, Greece, Hungary and the United Kingdom, the sampling method employed was multi-stage stratified sampling. Germany, Italy, Latvia, Norway and Poland and all used stratified multi-stage multi-stratified sampling. In Bulgaria and Spain, the sampling method adopted was two-stage stratified sampling. The central population register provided the sampling frame in the following countries: Austria, Estonia, Finland, Germany, Italy, Latvia, Lithuania, Norway, Poland, Slovakia and Sweden. In Bulgaria, Cyprus, France, Greece and Hungary, the sampling frame was the latest population census. In the United Kingdom, the sampling frame consisted of the Post-code Address File, which comprises all addresses that receive 50 items of mail per day or less.

The EU-AES target population consisted of persons aged 25–64 who are permanently resident in each country. The sample sizes are given in Table 1 of section 3.

The sample overcoverage comprised persons who had died since the most recent register update or population census, all persons abroad during the survey reference period, and persons living in institutions. The sample undercoverage consisted of immigrants aged 25–64 who had moved to the country since the most recent register update or population census.

Weighting coefficients are used in calculating the survey results in order to ensure that the results correspond to the target population aged 25–64 in each country. The weighting coefficients both reduce the bias caused by non-response and improve the efficiency of estimation. In calculating the weighting coefficients, the survey countries used age and sex as the principal variables. Where a calibration method was used in calculating the weighting coefficients, the variables used included municipality group, region (NUTS II), level of education and labour force status.

3. Correctness and accuracy of data

The accuracy of the data obtained in the sample surveys is affected by measurement errors, non-response and random variation caused by sampling. Measurement errors can arise from questions being understood or interpreted in a different way, or from respondents choosing not to declare certain information. Efforts are made to reduce measurement error by means of interviewer training and by testing the data collection forms in advance.

Non-response is divided into unit and item (or partial) non-response. Unit non-response occurs when an interview with a sampled individual cannot be carried out at all because of e.g. refusal by the individual or failure to contact the individual. Unit non-response is corrected by means of weighting coefficients. Item non-response refers to question-specific non-response. In this case the interview is carried out, but there is a lack of data in regard to some of the answers, for instance due to interruption of the interview or refusal to reply.

The EU-AES response rate for those countries for which data is currently available varies from 54.5 per cent for Germany to 94.6 per cent for Hungary (Table 1). Finland’s response rate was 65.2 per cent.

Table 1. EU-AES response rates, non-response and sample size, by country.

	Response rate, %	Non-response rate, %	Sample size, n
Austria	57,0	43,0	8 204
Bulgaria	81,5	18,5	6 458
Cyprus	75,8	24,2	6 347
Estonia	68,9	31,1	5 200
Finland	65,2	34,8	6 388
France 1)	..	..	..
Germany	54,5	45,5	16 602
Greece	84,3	15,7	4 210
Hungary	94,6	5,4	7 924
Italy	71,7	28,3	38 816
Latvia	60,4	39,6	3 830
Lithuania	82,6	17,4	4 473
Norway	68,1	31,9	4 429
Poland	68,7	31,3	19 247
Slovakia 2)	..	..	5 001
Spain	93,2	6,8	19 696
Sweden	73,8	26,2	4 922
United Kingdom	40,9	59,1	8 636

1) Data not available
2) Data not available

Despite the non-response rates, the interviewees are considered to be very representative of the target population in all countries. Minor distortions are balanced out with the aid of the weighting coefficients.

Partial non-response is an indication of how well the data collection form succeeded and of the quality of the questions. A high proportion of ‘do not know’ responses for a particular question indicates that either the question was difficult to understand or interviewees were reluctant to answer. Partial non-response was very low in the EU-AES in the case of the most important indicators and thus did not present a problem for the reliability or comparability of results.

The random variation caused by sampling refers to the fact that figures calculated from different samples vary somewhat from one sample to the next. This random variation is estimated using standard error. Standard error expresses how closely the observations cluster around the sample mean. The sample size and the variation in the values of variables affect the size of the standard error. The standard error can be used for calculating the confidence interval for different estimates of variables. The confidence interval expresses the probability that a variable value will lie within a certain range.

A confidence interval of (1-a)x100% refers to a defined interval within which the true value of a parameter is located with a probability of (1-a)x100%, or

Pr{pЄ(p-tαd(p), p+tαd(p))}=1–α

The tα value corresponding to a 95 per cent confidence interval is 1.96.

The confidence intervals calculated for the main EU-AES estimates are given in Table 2.

Table 2. Estimates of the participation rates in formal and non-formal education and training, together with the 95% confidence intervals, by country

Country	Education
	Formal education	Non-formal education
Austria	4,2 ± 0,3	39,8 ± 1,4
Bulgary	2,7 ±0 ,6	35,2 ± 1,9
Cyprus	2,9 ± 0,5	39,5 ± 1,4
Estonia	5,0 ± 0,7	40,2 ± 1,6
Finland	10,2 ± 0,9	51,2 ± 1,5
France 1)	1,7	34,1
Germany	5,2 ± 0,5	43,1 ± 1,2
Greece	2,3 ± 0,4	12,7 ± 0,8
Hungary	2,5 ± 0,4	6,8 ± 0,6
Italy	4,4 ± 0,2	20,2 ± 0,5
Latvia	5,4 ± 0,9	30,7 ± 1,9
Lithuania	6,3 ± 0,8	30,9 ± 1,5
Norway	9,9 ± 1,1	50,6 ± 1,8
Poland	5,5 ± 0,3	18,6 ± 0,5
Slovakia	6,1 ± 0,7	41,2 ± 1,4
Spain 2)	6,0	27,2
Sweden	12,7 ± 1,1	69,4 ± 1,5
United Kingdom	15,1 ± 1,2	40,3 ± 1,8

1) Data on confidence interval not available
2) Data on confidence interval not available

4. Timeliness and promptness of published data

The EU-AES was the first survey of its kind and, as such, constituted a pilot survey. The aim is to carry out the next EU-AES (specified by regulation) in 2011 or 2012.

The EU-AES fieldwork was undertaken in the period 2005–2008. The reference period for the survey was the 12 months prior to the interviews.

Table 3. EU-AES fieldwork and reference period, by country.

	Fieldwork	Reference period
Austria	04/2007 — 11/2007	05/2006 — 11/2007
Bulgaria	11/2007 — 12/2007	12/2006 — 12/2007
Cyprus	09/2006 — 12/2006	09/2005 — 12/2006
Estonia	09/2007 — 12/2007	10/2006 — 12/2007
Finland	03/2006 — 08/2006	04/2005 — 08/2006
France	01/2006 — 01/2007	02/2005 — 01/2007
Germany	03/2007 — 07/2007	04/2006 — 07/2007
Greece	10/2007 — 12/2007	11/2006 — 12/2007
Hungary	07/2006 — 09/2006	07/2005 — 08/2006
Italy	05/2006 — 08/2006	06/2005 — 08/2006
Latvia	05/2007 — 06/2007	06/2006 — 06/2007
Lithuania	03/2006 — 04/2006	04/2005 — 04/2006
Norway	05/2007 — 08/2007	06/2006 — 08/2007
Poland	10/2006 — 12/2006	11/2005 — 12/2006
Slovakia	08/2007 — 09/2007	09/2006 — 09/2007
Spain	02/2007 — 04/2007	02/2006 — 04/2007
Sweden	10/2005 — 03/2006	11/2004 — 03/2006
United Kingdom	10/2005 — 02/2006	11/2004 — 02/2006

Eurostat published the EU-AES results for 18 of the survey countries in November 2008. The results for the remaining 11 countries will be published before the end of 2009. The results are available only in tabular form and in a tabular database.

5. Accessibility and transparency/clarity of data

In November 2008, Eurostat published the EU-AES results in tabular form for 18 of the survey countries (Austria, Bulgaria, Cyprus, Estonia, Finland, France, Germany, Greece, Hungary, Italy, Latvia, Lithuania, Norway, Poland, Slovakia, Spain, Sweden and the United Kingdom). The results can be viewed in tabular form on the Eurostat website.

The results for the remaining 11 countries (Belgium, Croatia, the Czech Republic, Denmark, Malta, the Netherlands, Portugal, Romania, Slovenia, Switzerland and Turkey) will be published before the end of 2009.

The following can obtain the right to use EU-AES microdata for research purposes from Eurostat: universities, research institutions, national statistical agencies, central banks of European Union member states, individual researchers and the European Central Bank.

6. Comparability of statistics

The classification of education and training content that is used in the EU-AES is based on the Eurostat classification developed from the ISCED97 (Fields of Education and Training, 1999). The survey’s other classifications common to all countries are sex, age, level of education (ISCED97), degree of urbanization and labour force status.

The key education and training concepts in the survey ( formal education and training , non-formal education and training and informal learning ) are derived from the Eurostat’s Classification for Learning Activities (2005).

No comparison data from earlier years are available for the EU-AES, as the survey was the first of its kind. Comparisons between the countries participating in the survey must take into account the fact that the reference periods differ slightly among the countries and different data collection methods have been used.

The recommended data collection method for the EU-AES was computer-assisted personal interviewing (CAPI). The CAPI method was used for data collection in Bulgaria, Cyprus, Estonia, Finland, France, Germany, Spain and the United Kingdom. Paper and pencil interviewing (PAPI) was used for data collection in Austria, Hungary, Italy, Greece, Latvia, Lithuania, Poland and Slovakia, while in Norway and Sweden the data collection method was a mix of CAPI and computer-assisted telephone interviewing (CATI). The influence of the different data collection methods on the survey results has not yet been analysed. It is nevertheless considered that participation estimates obtained through personal interviews would be a little higher than those obtained through telephone interviews. This was the case in Sweden, for example. Since the interview method was not randomised, but, rather, the interviewee was able to select the method, it is not possible to conclude whether the difference observed is derived from the data collection method or, for instance, from the fact that persons not participating in adult education and training more readily selected the telephone interview option. As the telephone interview method was used only in Sweden and Norway, and only for a proportion of the interviews, this does not present any problem for the comparability of data.

The reference periods for the results published so far cover the years 2005–2007. Thus, the effect of small differences between reference periods on the survey results cannot be regarded as significant.

The EU-AES data collection was fully input-harmonised for some variables, while for other it was output-harmonised. The need to translate the data collection form, originally in English, into the different languages introduces the potential for comparability problems. Nevertheless, the main indicators describing participation in adult education and training and the volume of adult education and training are unambiguously defined, and so the translation of the questions into the different national languages does not present any problem for the comparability of data. However, the measurement of informal learning has apparently produced problems in some countries at the translation stage. As a consequence, in some countries so-called incidental learning has sometimes found its way into the informal learning results, despite the fact that the definition of informal learning in the Classification of learning activities manual does not include occasional learning. Eurostat is not therefore publishing the results for informal learning.

7. Coherence and consistency/uniformity

The EU-AES results for Finland differ somewhat from the results of Finland’s national Adult Education Survey 2006. Although the EU-AES data on Finland are based on the national Adult Education Survey, the datasets differ in certain respects. The greatest difference was in the survey population, as the target population for the EU-AES dataset consisted of persons aged 25–64, whereas the target population in the national Adult Education Survey also included persons aged 18–24.

A further difference concerns the follow-up questions on education and training received and the indicators based on these. In the national dataset, the follow-up questions (e.g. content of education and training, instruction hours, funding of education and training) were asked for all courses attended by the respondent, whereas in the internationally comparable dataset produced for Eurostat the data on the follow-up questions cover only three randomly chosen courses.

These dataset differences exist because it has been necessary to secure simultaneously Finland’s national data needs and the continuity of national time series data and international comparability.

The results of Finland’s national Adult Education Survey are published on the Statistics Finland website and in the following research publication: Pohjanpää - Niemi - Ruuskanen: Participation in adult education and training. Adult Education Survey 2006. Education 2008. Statistics Finland, Helsinki .

Source: EU Adult Education Survey, Eurostat

Inquiries: Timo Ruuskanen (09) 1734 3620, Irja Blomqvist (09) 1734 3221

Director in charge: Riitta Harala

---

Updated 24.4.2009