See the opening page for a list of tables, and an explanation of the terms, definitions and methods used.
The Mikrozensus is an official statistical survey in Germany (see footnote 1). Its organisation and technical preparation is done by the Federal Statistical Office (Statistisches Bundesamt), while it is carried out in practice by the statistical offices in the individual Länder (Statistische Landesämter). It has been conducted since 1957 on a yearly basis, except in 1975, 1983 and 1984; i.e. its major advantage is the regular time series information on the stocks of the adopted variables, although some of the variables are addressed only every two years (see footnote 2). The questions contained in the survey concern mainly the economic and social position, employment and education of the selected participants. The principal reason for choosing the Mikrozensus data is that the Mikrozensus survey asks the participants questions on both the highest general schooling exam and the last vocational qualification, so that a cross-tabulation of these two answers allows to produce a single and consistent ranking of qualifications. Other data sources, like the statistics of the education departments of the Länder or the publications of the chambers of industry and commerce, do not combine these two 'tiers' (see footnote 3).
The Mikrozensus is a representative survey in which 1% of the German households take part; i.e. it includes more than 350000 households with approximately 800000 individuals (see footnote 4). The households which come from specified geographical areas are selected by a random sample. They take part in the survey for four years. One quarter of the households are replaced every year (see footnote 5). The data refers to one week, the so-called 'reference week' (Berichtswoche) during which the survey is carried out.
Basically, the survey is carried out by interviewers who fill in the survey form in co-operation with the participants. This procedure guarantees not only a quick response, but also a high quality of the answers, as the questions can be explained more easily to the participants. It has been necessary, however, to allow households to prepare the answers by themselves and to reply by mail, as nobody may be available in the household during general office opening hours. In 1991, 11.2% of all participants chose this option, while 86.2% of all participants were interviewed in the 'conventional' way (see footnote 6). The possibility of response by mail is, of course, a source of errors. On the other hand, it helps to ensure a very high unit-response of the sample population and, therefore, contributes to the relative reliability of the Mikrozensus (see below).
In the Mikrozensus survey the data are collected for all members of the selected households. In effect, the questions are answered by one adult person in the household on behalf of the others, insofar the other members of the household do not insist on answering the question themselves.
For reasons of data protection, the minimum threshold for the inclusion of a figure in the print-outs from the Mikrozensus is 50 occurrences. This means that if an occurrence in a category or a combination of categories is less than 50, this category is left blank in the specially designed tables (Sonderauswertungen) that we worked with (see footnote 7). This leads to slight inconsistencies when different categories are added up or subtracted. Therefore, the sums of the data in the columns of the tables hardly match 100%; they are between 99 and 101%. In the tables with a more detailed breakdown, i.e. in the tables that specify age groups and industrial branches, the deviations can be even somewhat larger, as the denominators of the percentage calculations may be very small.
The central feature of the Mikrozensus is the legal obligation of the selected households to take part in the survey and to answer the questions truthfully. This obligation has been reinforced by the 'Mikrozensus acts' (Mikrozensusgesetze und -verordnungen) since 1957 and the judgements of the Federal Constitutional Court (Bundesverfassungsgericht) (see footnote 8). The obligation concerns the majority of the questions, while the response to a minority of the questions is not mandatory. With a unit-nonresponse rate of less than 3% on average, the statistical quality of the Mikrozensus as a whole is very high, judged by international standards (see footnote 9). The quality of the survey is further raised by plausibility checks of the data which test the logical consistency of the answers. However, due to the rigidity of German data protection laws there is no check of the actual content of the data, for example by comparison with former surveys or other data sources.
Unfortunately, since 1991 the answers to some of the questions which formerly were compulsory have become optional. Therefore, the Mikrozensus is currently being carried out on two form sheets, one of which contains the compulsory questions and the other one the optional questions. The questions that became optional include those on general and vocational qualifications. This change has resulted in distortions of the data. First of all, the item-non-response rate to the questions concerned have increased, as the following graph shows:
It is also apparent from the tables in the data set that the item non-response rates have soared in 1991: In the tables, the non-respondents are included among 'others', together with those who have basically no schooling or vocational qualification (for details see the glossary). This group has increased from 1.21% of the active population in 1989 to 11.11% in 1991.
It has to be assumed that the non-responses are not randomly distributed, but that they disturb the information on education and qualification categories. Therefore, T. Riede and D. Emmerling have compared the 'new' figures (following the change in 1991) with two constructed proxies for the 'true' values: The first of these proxies is the position in the workplace which can be assumed to have some relation to the acquired qualifications. Answering the relevant questions on the position in the workplace is compulsory. The second proxy is based on a cohort analysis and compares the answers to the questions about education and qualification of certain age groups in 1991 with the respective cohorts in 1989, when answering the questions was still mandatory. The results of the two analyses show that the answers to the very low and the very high qualification categories (like Hauptschulabschluß and University degree respectively) are probably somewhat understated, while the data for the categories in the middle of the scale (like Fachhochschulreife or Fachhochschulabschluß) may actually overstate the facts. It is difficult, however, to assess by how much the individual categories may be under- or overstated.
What is clear from the tables is that from 1989 to 1991 the category 'others' has increased by 10 or more percentage points, and that these people are very likely to have a general and / or a vocational qualification, but chose not to answer. It is not absolutely clear out of which categories these non-respondents come, but the data suggest that up to half of those (4-5 percentage points) come from the category with a Hauptschulabschluß only. Another 30% or so of them probably come from the category 'Apprenticeship or Berufsfachschulabschluß with Hauptschulabschluß'. It looks as if the rest comes from all the other categories.
Despite these difficulties it should be pointed out that the Mikrozensus
is a relatively reliable tool which gives a fair impression of
the distribution of qualifications.
