Within the perspective of lifelong learning, adult and continuing education (ACE) is regarded to play a key role in the development of today’s knowledge society. Several policies on supra-, inter- and national governmental level are aiming at the qualitative and quantitative enhancement of participation in ACE. Participation rates in ACE are a crucial part of indicators and benchmarks monitoring the progress in this domain of education (e.g. OECD 2003, EC 2004). While the question about the efficiency of this type of indicator is an important issue, the purpose of this paper is to analyse the indicators’ quality with reference to its data source. Due to the structure and organisation of ACE evidence-based reporting is bound to regularly conducted representative surveys which give evidence about the participation rates in ACE on an individual level. In Germany there are several regularly conducted surveys suited for this purpose. Regarding an evidence-based educational policy it is problematically that the results of these surveys are everything but corresponding and participation rates differ in an extreme example up to 30 percent. To a certain degree these disparities can be explained by rather obvious terms, as the reference period on which they rely or differences in the operationalisation of participation in ACE (Bellmann 2004). Yet, a profound clarification including statistical analyses dealing with the differences of the data-sources is hard to find (Wohn 2006). In respect to an evidence-based policy and also in respect to a quantitatively orientated research in educational science, this clarification should be regarded as an essential desideratum of and substantial precondition for educational research in ACE. Referring to this demand, four German studies are contrasted with focus on the following criteria: Key figures, survey design, definition and results. Subsequently an empirical analysis deepens this comparison by the example of two studies (below-mentioned) that show the maximal overlap in question design. The effects of the differences in the initial participation rates are further observed by socio-demographic distinguishing features, descriptively and by their patterns of influence on participation and are commented on against the backdrop of basic survey research (e.g. De Vaus 2002). Ever since the implementation of the Adult Education Survey (AES) and its consequences for the respective national statistical systems on ACE, this topic in its main features – yet in each case a specific one – is of concern to several countries and their national statistics participating in the AES (Dollhausen 2008).

The selection of the different surveys is conducted by the aspect of maximal comparability, especially the conformity in reference period and question design: 1. Microcensus Germany, 2001 2. BIBB/IAB-Erhebung, 1998/99. Survey of the Federal Institute for Vocational Education. and Training/Institute for Employment Research 3. GSOEP – German Socio Economic Panel, 2000. 4. BSW - Berichtssystem Weiterbildung VIII, 2000. Reporting System on ACE. The four studies are characterised and compared by basic figures of survey research (e.g. sampling, response rate, data collection) and by their participation rates in ACE (non-formal - informal; job-related – non-job-related). The data of GSOEP and BSW are further compared within socio-demographic factors, descriptively and by logistic regressions to observe the influence of the deviations on the patterns of participation. In doing so the data of the two studies is assimilated as far as possible to diminish the deviations based on the different study designs.

The variations can be traced to deviations in the features of study design. Nevertheless, an explicit unsolved variation remains. Consequently refocusing elements of fundamentally survey research, particularly the analysis of nonresponse, can be instructive. The comparison to the distribution of the high-quality government survey ‘Mikrozensus’, shows an educational bias in both BSW and GSOEP towards the overrepresentation of higher educated respondents. Taking into account the leverage-salience theory whereupon the participant’s interest in the topic is determining his decision to take part in a study (Groves 2000), people with interest in the topic ACE might be overrepresented and might have distinct values (e.g. participation in ACE) compared to those with no interest. Hence, nonresponse is likely to cause nonresponse-bias. The influence of topic interest is more relevant to the BSW (one-topic-survey) than to the GSOEP (multi-topic-survey). As a result the participation rates in ACE of the BSW might be overestimated.