ESRA logo

Tuesday 16th July       Wednesday 17th July       Thursday 18th July       Friday 19th July      

Download the conference book

Download the program





Tuesday 16th July 2013, 11:00 - 12:30, Room: No. 7

Educational attainment in cross-national surveys: instrument design, data collection, harmonisation and analysis

Convenor Dr Silke Schneider (GESIS - Leibniz Institute for the Social Sciences)

Session Details

Educational attainment is of key interest for both academic research and education and social policy at the national and international levels. It is usually measured by the highest formal education certificate achieved and is one of the most used, but also most difficult to harmonise, socio-economic variables in survey research. Compared to labour market and occupational measures, the comparative measurement of education has received much less attention in sociological research and official statistics. Consequently, there is no consensus yet on how to best conceptualise and measure educational attainment: Different surveys, even within countries, implement different concepts, measurement and coding procedures.

Many surveys however try to improve their methodologies: For example, in round 5 of the European Social Survey, a new measurement procedure was implemented. The Survey of Health, Ageing and Retirement in Europe has also improved its measurement instruments in 2012. Finally, official bodies such as Eurostat and OECD are currently working hard on the implementation of the new International Standard Classification of Education (ISCED) adopted by the Unesco General Conference in late 2011 in official surveys such as the European Labour Force Survey.

There are signs of both convergence and continuing disagreement between academic and official surveys concerning the issue of education measurement. The ESRA conference 2013 is thus a good opportunity to
- reflect on the ongoing changes across surveys and stakeholders,
- review and evaluate recent changes in measurement procedures in specific surveys,
- support coordination across stakeholders, and
- work towards a cross-national standard across surveys.

This session intends to bring together researchers and practitioners working for and with different cross-national surveys or official data concerning instrument design, data collection, harmonisation and analysis of education attainment and related concepts.


Paper Details

1. Validity of Measuring Educational Attainment - Education and Value Orientation in the European Values Study 2008

Ms Verena Ortmanns ( GESIS - Leibniz Institute for the Social Sciences)

These days various schemes measuring educational attainment exist like age completed education, years of schooling, CASMIN, and ISCED. We know that each scheme has its advantages and disadvantages but how large are the differences in the end when it comes to analyse data? Do the schemes convey different findings?
I assume that ISCED-97 in general delivers better results than the other schemes. But the performances of the schemes will be different in each country due to specifics in national education systems. This is tested for eight selected European countries by using the dataset of the European Values Study 2008. Items on the topic of hostility/ prejudices towards immigrants were chosen as dependent variables for assessing the predictive performance of each educational scheme.
Building on preliminary findings with regard to the explained variance the outcome is that in most of these countries ISCED-97 delivers the highest percentages. For the Netherlands, Sweden, and Great Britain also years of schooling lead to quite good results.


2. International vs. national classifications of education: Advantages and limitations in explaining values and attitudes

Professor Zbigniew Sawinski (Educational Research Institute, Poland)

Two types of classification of education are disseminated with the data from European Social Survey: (i) a common classification based on the ISCED (7 categories); (ii) the national classifications reflecting the specific structure of education systems in each country (from 11 to 26 categories). A question arises whether the international classification can be used instead of the national ones, which, as it can be assumed, better explain the role of education in societies. To verify this, in each ESS country the same questions concerning values and attitudes have been cross-classified both with the national and international classification. The results confirm that the national classification better explains values and attitudes than the international one. There are two types of educational outcomes which are not detected by the international classification. The first is the heterogeneity of the lowest category of education ("less than lower secondary" in ESS). When this category is divided into different types of primary schools and educational levels, the explained variance significantly increases, especially in countries with low educational attainments. The second boils down to differences between general and vocational schools in countries with highly selective tracks at the secondary level. The conclusion is that national classifications provide a better insight into the role of education at a country level. However, the international classification seems to be a good solution for cross-national analysis. Differences between countries are so distinct that they can be detected by a common classification, even if it ignores the specificity of education systems.



3. Measuring educational attainment in the case of disadvantaged groups: the example of the Roma

Dr Sabine Springer (European Union Agency for Fundamental Rights)
Dr Vida Beresneviciute (European Union Agency for Fundamental Rights)

In 2011, the European Union Agency for Fundamental Rights conducted a survey in 11 European countries interviewing Roma regarding their employment, education, health and housing situation and their discrimination experiences. The Roma are the biggest European minority group with very heterogeneous socio-economic living conditions. The sample was drawn from areas where Roma live in higher than national concentration. It therefore reflects a part of the Roma population, which often combines disadvantages for the educational outcome such as overcrowded households, poverty and high unemployment. Other educationally marginalised national and migrant groups are often facing similar problems of social exclusion and discrimination.

Based on examples from the Roma survey, it will be shown that the EUROSTAT indicators on educational attainment (tertiary education, upper secondary education and early school leaving) are only of limited use in the context of very low educational attainment. Indicators used in the context of the Roma survey, and in a comparative perspective, trying to capture educational attainment or achievements at a low level will be proposed for discussion.


4. Measuring education in official labour market statistics

Mr Ralf Dorau (University of Bonn)

The German sample of Integrated Labour Market Biographies (IABS) of the German Institute for Employment Research contains data that were collected from 1975 to 2008 with approx. 34,900,000 entries and more than 1,500,00 individuals (an average of more than 20 entries per individual). The dataset is based on three different sources: (1) notifications of employment to social security bodies (more than 25,800,000 entries), (2) data on the receipt of unemployment benefits (more than 5,400,000 entries) and (3) registered job-seekers at the Federal Employment Agency (almost 3,600,000 entries), the latter contains data from 2000 to 2004 and from 2007 to 2008 only.

Only (1) and (3) permit an identification of the education level. The education variable of (1) comprises information on secondary and tertiary schooling and vocational education, but has two drawbacks: missing values and inconsistencies. The education information of (3) is relative reliable and contains in addition details such as in-firm training, full-time vocational school and it includes a separate variable only for secondary schooling.

For improving the education variable one has to impute the most likely education from past or future information. We must therefore take into account how many and which spells are missing or inconsistent and which source they originate from. For running the imputation procedure we also can use other variables in the dataset such as occupational status, age and justifications from the companies the individuals have been employed.






5. Computer-Assisted Measurement and Coding of Educational Qualifications in Surveys (CAMCES)

Dr Silke Schneider (GESIS - Leibniz-Institute for the Social Sciences)

The individual's educational attainment is a core social background variable in standardised surveys. It is usually measured by the highest educational qualification the individual has obtained, using closed questions and showcards showing a limited number of response categories. However, increasing differentiation of educational systems and education and work-related migration complicate the measurement of educational attainment in surveys with limited response categories considerably, especially in cross-national and panel studies. Post-hoc harmonisation is highly problematic given the lack of standardisation and detail in most measurements.
This presentation will provide an overview of a new project to develop a tool for measuring educational qualifications in computer-assisted surveys, moving from a closed to an open question format using targeted probes. The tool will be based on 1) an international database of educational qualifications, 2) optimised questionnaire instruments not relying on a closed question with limited response categories anymore and using probes, and 3) an interface to directly access the database for use in computer-assisted surveys. For this session, the most interesting question is how such an instrument can be designed in order to produce valid and reliable results in the absence of a fixed set of response categories, e.g. using probes, nested questions and database feedback.