Metadata for Official Statistics

FAQ: Mikrozensus

The following pages contain a thematically classified collection of answers to the most common questions about the Microcensus.

General

What is the Microcensus?

The Microcensus is a representative population sample containing structural population and labour market data in which 1 percent of all households in Germany are involved (ongoing household sample).

Since when is there Microcensus?

The Microcensus has been introduced in the "former" states since 1957 and in the new states and East-Berlin since 1991.

Which functions does the Microcensus fulfill?

The Microcensus provides statistical information on the economic and social situtation of the population as well as on employment, the labour market and the education system (multipurpose sample). The Microcensus updates the results of the population census. Moreover, it serves the streamlining of other official statistics, e.g. the income and consumer panel. The sampling on labour of the European Union is also integrated in the Microcensus.

What can the results be used for?

The results of the Microcensus are an essential information source for Parliament, government, administration, science and the public throghout the country. They are used in government reports and in the annual expert opinions by the German Council for the appraisal of macroeconomic development. They also build the basis for ongoing labour market and occupation research, the annual annuity insurance reports of the government, and more. The results from the labour force sample of the EU are also used for the allocation of EU regional and social funds.

Are there any legal regulations which prescribe the implementation?

Yes. Legal basis is the „Mikrozensusgesetz“ of January 17, 1996 (BGB1. I S.34) in combination with the law on statistics for federal purposes from January 22, 1987 (BGB1. I S. 462, 565). It was changed last by adding article 2 of the law from January 17, 1996 (BGB1. I S. 34). The EU evaluation of workforce was regulated in the order (EEC) no. 3711/1991 of the council from December 16, 1991 for conducting an annual evaluation on workforce in the European Community. It was substituted by the order of the council from March 9, 1998 for conducting an evaluation on workforce in the European Community (EC) no. 577/98.

In which interval is the survey repeated?

Until 2004 the Microcensus was conducted once a year. Since this year (2005) the Microcensus is conducted continously throughout the whole year.

Who conducts the survey?

The organizational and technical preparations for the Microcensus take place in the Federal Statistical Office. The interviews and the processing are the responsibility of the statistical offices of the different states (decentralized statistics).

What is the selection procedure for participating households?

The Microcensus is a random sample in which all households have the same probability to be selected. Out of the whole country, certain regions are chosen in which then all households and persons get interviewed. The way in which the regions are selected are based upon the procedures of the census in 1987. For the new states a comparable selection was established based upon the “Bevölkerungsregister Statistik”. With the help from the statistics on house building activity, the selection is regularly updated.

How many households are questioned per year?

In the 2013 Microcensus roughly 336.200 households and respectively 683000 individuals were surveyed. Extrapolated this corresponds to 80611000 individuals.

For how long does the household data remain in the sample?

In the Microcensus a quarter of all households in the sample get replaced every year. Consequently, every household stays in the sample for 4 years.

How is the survey carried out?

The Microcensus survey is generally conducted face-to-face by an interviewer. This method is the fastest and involves the least effort for the interviewee. However, the respondents are also granted the opportunity to report the information in written form.

Who gets surveyed?

Interviewees are all persons of a household. Under certain circumstances, external information from other household members is allowed.

Are the respondents obligated to answer?

The questions of the Microcensus consists of a fixed basic and ammendatory programmes. On top of that there are ammendatory programmes which are conducted on a 4 year basis. There are certain parts that are not obligatory.

Which topics are surveyed?

The annual basic content of the Microcensus includes personal information (age, sex, citizenship etc.), information regarding connections between family and household as well as characteristics about main and secondary domiciles, employment, job search, unemployment, income etc. In the annual ammendatory programme additional questions are asked concerning employment, personal information and professional and general education. Within the 4 year ammendatory programmes information about commuting, the living situation, health insurance, health in general and disability are surveyed.

How complete are the respondent´s answers?

The response rate in the Microcensus is regularly 97 percent. 3 percent fall out mostly because they were unavailable. These so-called “known losses” are compensated for by a special procedure. The “non-response” quota, in questions which are optional, vary depending on the matter.h.

For which timeframe is the survey conducted?

The information, especially about employment, applies to the defined reference week. Usually it is the last week of April without any holiday.

According to the Microcensus, who is considered as employed?

Employed persons are all those who are 15 and older and who have worked at least one hour in the reference week (“labour force concept”).

How are the results illustrated?

The results of the Microcensus are offered for representation units “household – family- persons” in the form of a sum total. Besides federal results, there are also diverse results for individual states available. A deeper, more regional classification is restricted; therefore results can only be made available for large verification groups.

Where are the results published?

Currently the results are published for the press via official press releases. On top of that, the Federal Statistical Office regularly publishes Microcensus results in the so-called “Fachserien”: Fachserie 1, row 4.1.1: Position and development of employment; row 4.1.2: profession, education and working conditions of employees; Fachserie 1, row 3: households and families; Fachserie 13, row 1: health insured and annuity insurances as well as Fachserie 12, row: S.3: questions concerning health (irregular). State results are published by the states´ statistical offices. Comments on the results and methods can be found in the magazine “ Wirtschaft und Statistik”.

Who provides information?

Information can be received from the Federal Statistical Office, as well as from the staff of the Microcensus departments of the state offices.

Access to Data

Who has access to Microcensus data?

According to §16(6) of the Bundesstatistikgesetz, the Microcensus´ Scientific Use Files can only be delivered to independent domestic research institutes. It is a precondition to compose a license agreement with the Federal Statistical Office and you are also sworn to secrecy of all persons who work with the data. This special obligation usually applies to civil servants (university); otherwise it has to be carried out by the Federal Statistical Office.

The Microcensus guidelines restrict the usage of data to Germany.

If I do not belong to the subgroup, which has access to the Microcensus´ Scientific Use File, what alternatives of data usage are there?

  1. There is the possibility to fall back on publications. Detailed tables to various areas of the Microcensus can be found in the different "Fachseries" at the Federal Statistical Office.
  • Fachserie 1, Reihe 3: Haushalte und Familien
  • Fachserie 1, Reihe 4.1.1: Stand und Entwicklung der Erwerbstätigkeit
  • Fachserie 1, Reihe 4.1.2: Beruf, Ausbildung und Arbeitsbedingungen der Erwerbstätigen

  1. The Federal Statistical Office also carries out special evaluations on reimbursement of expenses. Contact:

    Statistisches Bundesamt
    Zweigstelle Bonn
    Gruppe F2 Bevölkerung, Mikrozensus, Wohnen und Migration
    Postfach 170377
    (Graurheindorfer Str. 198)
    53029 Bonn-Germany

    E-Mail: mikrozensus(at)destatis(dot)de

Is it possible to use the Microcensus´ Scientific Use Files for writing a thesis or dissertation?

The personal wish to use the data for a thesis or dissertation does not automatically fulfil the legal conditions. The Microcensus´ Scientific Use File can only be obtained by domestic universities and other research institutes with the responsibility to do independent research. If the purpose of the thesis or dissertation is part of department research etc. and the university applied for the data, it can be used within the context of university research. Therefore, the integration into the regular university research is important. For more information see: http://www.forschungsdatenzentrum.de/nutzung/fdz_nutzungsbedingungen.pdf

The statistical offices are offering the possibility to download anonymized data of the Microcensus as the so-called Campus File (so far available 1976, 1998, 2002 and Microcensus-Panel 1996-99 (updated: February 2012)). This can be easily accessed and used by students.

See below:
http://www.forschungsdatenzentrum.de/campus-file.asp

Where and how can the Microcensus data be ordered?

The available Microcencus data as a Scientific Use File can be ordered at the Research Data Centres of the Federal Statistical Office and the statistical offices of the Länder.

Research Data Centre

of the statistical offices of the Länder
- Geschäftsstelle -
c/o Information und Technik NRW
Postfach 101105
40002 Düsseldorf
forschungsdatenzentrum@it.nrw.de

of the Federal Statistical Office
- Forschungsdatenzentrum -
Gustav-Stresemann-Ring 11
65180 Wiesbaden
forschungsdatenzentrum@destatis.de

See: Request Form

If you are interested in the Microcensus or you are already a user, you can contact the GML and receive following consulting services:

  • Advice on to what extent the research plan can be realized with the data from the Microcensus
  • Advice on drawing up an application form (research plan)
  • Advice on technical questions (Technical equipment, data protection etc.)
  • Advice on questions about evaluation and analyses.

Which data protection measures are taken regarding the usage of the Microcensus Scientific Use File?

General precautions:

  1. Contractional obligations of the beneficiary of factual anonymized data;

  • appropriate technical and organizational measures for controlling the data usage:
  • Contractual penalty when attempting to re-identify
  • Usage boundaries to research purposes
  • No data forwarding to third party
  • Deletion or rather return of the data after conclusion of the research project
  • Handling of data extracts or duplicates as the original file
  • No investigation regarding the local implementation of the sample plans

  1. Secrecy of the local implementation of the sample plans through official statistics
  2. Systemless arrangement of the data

Anonymization measures regarding the Microcensus (basic file):

  1. Regional data only highly oversimplified available (merely state and structure of the residential area or rather oversimplified municipality size)
    • No single municipality with less than 500,000 inhabitants must be identifiable
    • A jurisdiction, which consists of several communities, may not have less than 400,000 inhabitants in any state.
  2. No nationality or group of nationalities with less than 50,000 people may be identifiable
  3. Oversimplifying features from all other variables – if necessary – so that in the univariate marginal distribution each acknowledged characteristic value for the Federal Republic of Germany comprises at least 5,000 cases.
  4. Only a 70-percent sample of the orignial data is delivered.

Sources of Information

Where can I find information about the mailinglist?

The microdata mailing list provides a discussion and information forum to official microdata. By E-Mail problems can be discussed and questions to official statistics can be answered.

You can subscribe here: The mailinglist for microdata and more mailinglists.

Is there an overview of publications and further documents of GML which are downloadable?

On the GML webpages under Downloads and Publications there are more documents to download.

Is there a catalogue of the publications which were generated based on the Microcensus Scientific Use File?

Publications based on the Microcensus Scientific Use File can be found here.

Is there information on more official microdata (besides the Microcensus) and their access possibilities?

Other official microdata is documented in MISSY systematically:

  • AES (Adult Education Survey)
  • EU-LFS (European Union Labour Force Survey)
  • CIS (Community Innovation Survey)
  • EU-SILC (European Union Statistics on Income and Living Conditions)
  • SES (Structure of Earnings Survey)

Further data is not documented systematically, but documentation (variable lists, questionnaires, setups, information about data access) is availabe for:


The website of the German Microdata Lab covers


An overview over further microdata which is available for the (scientific) public in Germany, but is not documented by the German Microdata Lab can be found here. For informationen regarding the access to official microdata click here.

The Survey Program and Topics of the Microcensus

Which topics does the Microcensus focus on? Does the Microcensus feature a consistens survey program?

The Microcensus questionnaire program consists of a consistent basic program which releases socio-economic basic information providing the current monitoring of the labour market. Addressing all households this basic program includes information on:

  • all persons of a household and how the household is made up,
  • occupation, unemployment, job search, economically non-active population,
  • attendance at kindergarten, school or university,
  • education and vocational training,
  • sources of livelihood, income level, compulsory pension insurance,
  • the compulsory and private health insurance and
  • the insurance for old-age care.

In addition to the basic program and its features, which are surveyed annually, there are also supplementary programmes. These programmes give information on vocational and further training, current and previous occupation and labour force participation. Furthermore, there are additional programmes with features focussing on different topics. These are not surveyed annually. From 2005 onwards the survey also includes ad-hoc modules of the Labour Force Survey.

Which additional surveys can be found in which Microcensus year?

The following tables give an overview of the different supplementary and additional programmes, their sampling fractions and years of survey.

How does the exceptional sampling fraction of 0.45 percent emerge in questions of the Microscensus and its supplementary and additional programmes?

Until 2004, the samples for the Microcensus supplementary and additional programmes are drawn with a sampling fraction of 0.5% or 1%. In order to meet the requirements of accuracy defined by Eurostat concerning the Microcensus survey program and the integrated EU Labour Force Survey, the sampling fractions differ in their size on county level (NUTS II): 0.4, 0.6, 0.8 or even 1.0%. For federal average this makes a sampling fraction of 0.45% on an average. These requirements of accuracy are defined in the Verordnung (EG) Nr. 577/98 des Rates vom 9.3.1998 zur Durchführung einer Stichprobenerhebung über Arbeitskräfte in der Gemeinschaft, Artikel 3; Repräsentatitivität der Stichprobe, as follows: (1) “As to a group of unemployed, representing 5% of the working age population, the relative standard error of estimated annual averages (or spring averages in case of an annual survey in spring) may only constitute on NUTS II level at the most 8% of the mentioned population group.

Thematical Questions

Employment

Are there any ISCO-Codes in the Microcensus in order to create prestige scales?

Until Microcensus 1995 the occupation is only coded in the German' Classification of occupation' (KLdB). Starting with the microcencus 2012 the KldB (2010 version) is used. As in the 1996 Microcensus, the occupation is also mapped by the ISCO 88 COM.

For distinctions between ISCO 88 and ISCO 88 (COM) see below: ISCO 88 (COM) - the European Union variant of ISCO and conversion technique. Starting with the microcencus 2011 ISCO (2008 version ISCCO-08) is used.

What occupation-related social science scales can be implemented using occupational data from the Microcensus?

As a GML service offering, the implementation of a number of social science instruments using Microcensus data can be found on the website. This service offering includes:

Which boundaries have to be set in order to compare the different types of employees in the individual Microcensuses?

A comparable demarcation is accomplished by:

Erwerbstätige EF34=1-4,10-12
Erwerbslose EF34=5-7
Nichterwerbspersonen EF34=8-9

Erwerbstyp (z.B. im MZ ab 2005: EF29)

Erwerbstätige EF29=1
Erwerbslose (einschließlich arbeitsuchende Nichterwerbspersonen) EF29=2-3
Nichterwerbspersonen EF29=4

Familiy/Household

Why are there no persons with the parameter value 3 (F3, married couple without any single children, but with single grandchildren) with the variable `family type´ from the Microcensus 1996 onwards?

With the variable `FAMILY TYPE` (Microcensus 1996-2004: EF575; from 2005 onwards: EF707) the family type F3 of the official family statistic is neither declared in the Scientific Use File, nor is it in the original material of the Federal Statistical Office. The types F2 and F3 are combined into one parameter value and are recoded with a "2".

In which years of the Microcensus are analyses on a household level possible?

With all the Scientific Use Files, analyses on a household level are possible.
See below:

For which Microcensus years, is it possible to identify the youngest child of the familiy?

As for all Scientific Use Files household and family specific analyses are possible. It is also generally possible to identify the youngest child of the family.
See below:

Which persons of the household were interviewed?

For all persons in the household, inquiries are conducted. Extrinsical information (proxy interviews) regarding different household members are under certain conditions permissible. Since the Microcensus 1999, it has been inquiried if the respondent was directly asked for personal information or if it was extrinsical information.

MZ 1999-2004 (Survey: Type of participation = EF51)

Person participated herself/himself: EF51=1
Proxy interview: EF51=2

MZ ab 2005 (Survey: Type of participation = EF17)

Person participated herself/himself: EF17=1
Proxy interview: EF17=2

Migration

In what language are migrants interviewed?

As for now (August 2006), questionnaires are only available in German.

How are educational degrees, which are achieved in foreign countries, assigned in the Microcensus?

For the categorization of degrees and vocational qualifications which were achieved in native countries, there is no list of equivalents to German degrees, because there is no formal rule of classification.

What information on migrant background is available in the Microcensus?

From 2005 onwards, information on migrant background is available in the Microcensus. The migrant background types, and the corresponding nationalities, are contained in the Microcensus Scientific Use Files from the year 2008 onwards (2008: mig, migs; 2009: EF2001, EF2008). For the years 2005 to 2007, they can be created retrospectively; see programs under Tools.

Income

Is the variable ´household income´ (standardized) in the Microcensus from 1996 onwards‚ based on the question for household income on the fill in form?

The variable, household net income (after standardization) (Microcencus 1996-2004; from 2005 onwards: EF707), includes basically the references from the directly asked question regarding the household income (question 126 in the fill in form 1+E). In the case of an Item-Nonresponse at the question regarding the household income, an estimate of the household income is conducted by the Federal Statistical Office, on basis of available individual references regarding the personal net income. At the same time in a first step, the grouped and censored individual references of " head income numbers" (mid-point of class; with the assumption of an equal distribution within the income bracket) are recoded. The sum of these income numbers are, in a second step, recoded into the corresponding income bracket. With this procedure, all, within one household, available income references are being used, regardless of the possibility that an individual person might have not given an income reference.

Regional Analysis with the Microcensus

Is the interpretation of data on a regional scale possible when using the German Microcensus?

The Microcensus’ Basic Scientific Use File, provides rather detailed information on professions, branches of economic activity, citizenship, etc. Information on geographical areas, however, are – due to data confidentiality reasons – limited. Only information on the German federal states (Bundeslaender) and the classification of community size (Gemeindegroessenklasse) are available.

In contrast the Mikrocensus’ Regional Scientific Use File (2000), provides less detailed information on socioeconomic and socio-demographic characteristics but rather detailed information on geographical areas. The most detailed regional areas are the ‘Mikrozensus-Kreisregionen’ (MZKR; district areas with at least 100,000 inhabitants), which are based on administrative local units (Kreise and siedlungsstrukturelle Kreistypen).
See here.

See Herter-Eschweiler, R. (2019): Der Mikrozensus und die Möglichkeiten seiner Regionalisierung, S. 205-225 in: Arbeitsgruppe Regionale Standards (Hrsg.) 2019: Regionale Standards Ausgabe 2019. GESIS Schriftenreihe Band 23. Köln: GESIS.
See here.

Is it true that East and West Germany can only be distinguished from one another through the corresponding federal state? That is, East and West Berlin can not be distinguished?

No, West- and East-Berlin can be differentiated by means of the variables "federal state" (EF1) and "Gemeindegrößenklasse" (community size).

The variable "Gemeindegrößenklasse" (community size) is in microcensus:

1991-1995: EF8
1996-2004: EF708
from 2005: EF563

Methodical Questions

Weighting and projection in the microcensus

Are there any specific weighting variables for the microcensus scientific use files?

The weighting and projection variables available for the scientific use files correspond to the respective variables of the original microcensus file.

What is the purpose of the microcensus weighting and projection variables?

Fundamentally, the microcensus is a 1% sample of German households. As with any other sample, this sample has inevitable random sampling errors. Moreover, as in any other survey, systematic attrition occurs when the microcensus is carried out (the share of household attrition is relatively low – 2.5-3% – due to the obligation that applies to provide information for the microcensus). The microcensus weighting variables serve to compensate for this systematic dropout as well as for the random sampling error. They also serve to adjust data to the population estimates. In addition to that, the weighting variables designed for the subsamples of each microcensus also contain a projection factor which effects a projection of a 70% microcensus sample (the scientific use file).

Why do the frequency figures of the scientific use files differ from the published microcensus results, despite weighting and projection?

These differences are owed to the fact that the scientific use file only represents a 70%-sample of households and apartments drawn from the original microcensus. It is thus compiled on the basis of an inherent sampling error. No weighting variables are designated to correct this sampling error. Weighting the scientific use file with the respective weighting variables from the 1%-original file leaves the random sampling error untouched that was responsible for this error in the first place.

Further Links

Significance tests using microcensus results

Can microcensus results be tested for their significance? Are conventional significance tests and inferences admissible for analyses with the microcensus?

Conventional significance tests and inferences generally require an unlimited random sample. The microcensus, however, is based upon a cluster sample, not a random sample. Cluster samples are characterized by an increased variance of the population estimates compared to a random sample. The accurate calculation of cluster-sample population estimates requires information about the stratification and clustering of the sample. Based upon this information, available for the 1973-1987 and from 1996 onwards microcensus, the German Microdata Lab provides tools to calculate the standard errors for totals (absolute frequencies), means, and regression estimates (only in German):

Zur Berechnung des Stichprobenfehlers im Mikrozensus

For further information see: GESIS-Methodenbericht 2009/02

Miscellaneous

What is the response rate of the Microcensus? Why are cases of "not stated" in the Microcensus, where disclosure is usually obligatory?

  1. Due to the obligation of selected households to disclose, household participation is approximately 97 percent; i.e. the Unit-Nonresponse (households not reached or not available etc.) is marginal compared to many other surveys.
  2. Several questions from the Microcensus are not subject to disclosure, e.g. questions on the year of marriage and education and advanced training of people older than 51 years etc. These questions were not answered in up to 10 percent of the cases (Item-Nonresponse). In the data set descriptions (key lists) it is noted for which questions and variables disclosure is required.
  3. Also by questions where disclosure is required, non-response is possible since for example sime respondents do not give information regarding their income. The statistical offices attempt to convince the respondents that their information is necessary. However, the interviewers are not always successful. Item-Nonresponse is especially high by written surveys than by traditional face-to-face interviews. Information regarding the type of interview is listed in Microcensus´ Scientific Use File beginning in 1996 with the variable "EF9 Bogenart". Further information on non-response can be found in the following publications:
  • Emmerling, D./Riede, T. (1994): Zur Freiwilligkeit in der Auskunftserteilung im Mikrozensus. Wirtschaft und Statistik (6): 435-449.
    See here
  • Lüttinger, P./Riede, T. (1997): Der Mikrozensus: amtliche Daten für die Sozialforschung. ZUMA-Nachrichten 41: 19-43. [.pdf]
  • Riede, T./Emmerling, D. (1994): Analysen zur Freiwilligkeit der Auskunftserteilung im Mikrozensus. Wirtschaft und Statistik (9): 733-742.
    See here
  • Afentakis, A.; Bihler, W. (2005): Das Hochrechnungsverfahren beim unterjährigen Mikrozensus ab 2005 Wirtschaft und Statistik (10): 1039-1048. [.pdf]

Are the interviews referred to as “Blaise field interview” (Blaise-Feldbefragung) in the variable “survey instrument” (Erhebungsinstrument; 1999–2004 Microcensuses) face-to-face interviews or self-administered written questionnaires?

Blaise field interviews are interviews administered with a laptop by an interviewer who enters the respondent’s answers directly into an electronic questionnaire. In the 1996–2004 Microcensuses (EF9), only a small proportion of the interviews were administered with a laptop. However, since the 2005 Microcensus (EF7), laptops have been universally used for data collection.

See also:
Thematic Structure:
See here

The following sources provide information on the use of laptops in household surveys:

  • Riede, T. & Dorn, V. (1991): “Zur Einsetzbarkeit von Laptops in Haushaltsbefragungen in der Bundesrepublik Deutschland” [The usability of laptops in household surveys in the Federal Republic of Germany]. Issue 20 of the series Ausgewählte Arbeitsunterlagen zur Bundesstatistik. Wiesbaden: Statistisches Bundesamt.
  • Helbig, Ina (1996): “Fünf Jahre Mikrozensus - Erfahrungen in Sachsen” [Five years of the Microcensus – Experiences in Saxony] Statistik in Sachsen (3): 1-5.
  • Helbig, Ina (1997): “Laptopeinsatz im Mikrozensus 1996 in Sachsen” [The use of laptops in the 1996 Microcensus in Saxony]. Statistik in Sachsen (1): 35.

On the universal use of laptops from the 2005 Microcensus onwards:

  • Lotze, S.; Breiholz, H. (2002a): Zum neuen Erhebungsdesign des Mikrozensus. Teil 1. Wirtschaft und Statistik (5): 359-366.
    See here
  • Lotze, S.; Breiholz, H. (2002b): Zum neuen Erhebungsdesign des Mikrozensus. Teil 2. Teil 2. Wirtschaft und Statistik (6): 454-459.
    See here
  • Statistisches Bundesamt (2019):
    See here

What are Jahresüberhänge and how do I handle them?

Since 2006, there have been so-called Jahresüberhänge (singular: Jahresüberhang) in every Microcensus except the 2007 survey. Jahresüberhänge are households or persons who could not be interviewed during the survey period – usually because they could not be reached – and who were therefore not interviewed until the following year. Such Jahresüberhänge are a consequence of the quarterly surveys that have been conducted since 2005. Respondents who cannot be reached in the selected quarter (usually the last quarter) are sometimes not interviewed until the first quarter of the following year. As a result of this delay, it can happen that some households are interviewed twice in one year. The information about whether a person or a household was interviewed in the following year – that is, whether a person or a household is a Jahresüberhang – can be found in variable EF5b. In the case of all Microcensuses since 2006, except the 2007 survey, thought must be given to how the Überhänge from the previous year should be handled.

When a household is interviewed twice in one year – once as scheduled and once to catch up on the previous year’s interview (Jahresüberhang) – it has the same household number each time. This can lead to problems when creating variables at household level by aggregating person-level records (so-called satzübergreifendes Arbeiten), for example “number of persons in the household”. In such cases, it is important to take Jahresüberhänge into account. When carrying out their own delimitations at the level of households, living arrangements, or families, researchers should use the following sorting order to ensure an unequivocal identification of the units: sampling district (EF3), reporting quarter (EF12), Jahresüberhang (EF5b), household number (EF4), and, if necessary, the number of the living arrangement in the household (EF27) or the number of the family in the household (EF25). Otherwise, duplications can occur. The following example illustrates the problem. here

However, if one wishes to replicate publications of the statistical offices, Jahresüberhang households and persons must be included. This inclusion is meaningful because it can be assumed that the distribution structures of the Jahresüberhang households, who are, as a rule, difficult to reach, are similar over time and that they differ from those of other households. During the extrapolation process, information on Jahresüberhänge (EF5b) can be disregarded.

Comparative Analysis with the Microcensus

Is it possible to compare the labor force survey and microcensus?

Principally, it is possible to compare the labor force survey and microcensus. In Germany, the EU labor force survey is carried out as a subsample of the microcensus. Since the 1996 microcensus, employment is surveyed according to the labor force concept using four guiding questions on employment. Based upon ILO standards, the population is categorized as employed civilians, unemployed civilians, and those not available for the labor market (children, retirees, etc.). In accordance with the ILO international standards the microcensus defined any person as unemployed if he/she was not employed for at least one hour in the one-week reference period, but actively sought work. While the ILO definition of unemployment, however, also emphasizes the criterion of labor-market availability, this criterion was not considered in the Federal Statistical Office’s publications on employment which were based upon the microcensuses until 1996. Labor-market availability has only been recorded separately since 1996. The corresponding variable for implementing the ILO definition, however, is included in the microcensus files from 1985.

Further information:

How do the individual microcensus samples have to be weighted to allow a truly comparative analysis?

Microcensus: scientific use files

SUF 1973

The scientific use files from 1973 correspond to a microcensus sample of 70%.

The adjustment of the microcensus results to the ongoing population estimates was not – as in more recent microcensuses – accomplished by weighting, but by duplicating and deleting cases (individual level) (EF76). The figures which served as the benchmark for the microcensus adjustment are based on the population estimates, organized by Germans/immigrants, sex, and administrative district. If, for instance, the designation by these attributes records an undercoverage in the MC sample compared to the benchmark of the population estimate with regard to one designated cohort, i.e. if the sample value amounts to less then 1% of the benchmark, sample cases are duplicated until the sample value corresponds precisely to the benchmark. Conversely, cases are deleted if an overcoverage of cases in a designated cohort is discerned (sample size more than 1% of the corresponding benchmark figure). The duplication and deletion procedure is described in detail in Krug/Nourney: Wirtschafts- und Sozialstatistik, 2. Auflage. München: Oldenbourg, 1987 (pp. 145-146). However, from the 3rd edition onward only brief mention is made of this procedure.

SUF 1989ff.

The microcensus scientific use files from 1989 onward also correspond to a 70% subsample.

For the data from 1989-95 there is merely an expansion factor at the household level (MC 1989: EF257U4; resulting from an adjustment of the reference value to the actual value). The adjustment of the microcensus results to the continuous population recording takes place at the individual level, so that for the individual there is no corresponding expansion factor available in the scientific use files 1989-1995. (Test analyses, however, have shown that the use of a household expansion factor in analyses at an individual level does not lead to substantially different results than when using the correct expansion factor.) From the microcensus 1996 onwards, both the household expansion factor and the expansion factor for individuals is available.

Comparative analyses

Because the above-described microcensuses, on the one hand, consist of aggregated as well as individual data and, on the other hand, are based upon various different sampling fractions and adjustment procedures, the conduct of evaluations is recommended as follows:

  • analysis in the case of population estimates if published or "official" results are obtained (approximately)

See below:

and the pages "Zur Abgrenzung der Bevölkerungsbegriffe und zur Hochrechnung im Mikrozensus SUF...", for example for 1973 und 2009.

For an expansion to the total population (100%) the weighted results have to multiplied by 100.

  • Analysis in the case of statistical model estimation

When estimating statistical models we recommend unweighted analysis.

To ensure that data for the scientific use files of the MC 1973-1987 remain fully unweighted (i.e. with no adjustment to the ongoing population estimates and without consideration of the SUF-sampling fraction of 70%), and that a comparative analysis is made for an unweighted estimate of MC data since 1989 onwards, the recommendation is to treat the “original ratios” of surveyed persons included in the MC data 1973-1987 as separate cases. This can be done (e.g. after separating the resident population: ef58=1, ef58=2, ef58=4) by limiting the analysis to the corresponding “original ratios” (for example microcensus1982: ef76=0, ef38>=0&ef38<=1).

The following essay contains further details on expansion factors when estimating total, proportionate and mean values as well as on the adjustment to the population estimates: Rendtel, U./Schimpl-Neimanns, B., 2001: Die Berechnung der Varianz von Populationsschätzern im Scientific Use File des Mikrozensus ab 1996. ZUMA-Nachrichten 48: 85-116.

More information can be found here.

0.45% subsample MC 1996-2004

When using the 0.45% subsample, the special extrapolation procedure with the variable EF755 (microcensus since 1996) needs to be considered. It is described by the Federal Statistical Office as follows: “(…) For the determination of an adjustment factor for the features of a 0.45% subsample, the results of a 1% sample and its subsample within an administrative district are projected with the calculated adjustment factor for the 1% sample, with differentiations between the employed, the unemployed, and those outside of the labor force combined with the variables Germans/immigrants as well as sex. The adjustment factor for the 0.45 % sample [see variable EF755] results from the proportion of the output from the 1% sample (= reference value) and the output from the 0.45% subsample (= actual value) (...)"

Also see:

Hochrechnungsfaktoren

Technical Problems

How are the raw data processed with SPSS?

On our website you can find all Microcensus SPSS-, STATA- und SAS-setups for creating an System File which are available as Scientific Use Files.

The SPSS-Setup can be used for the following tasks:

  • Importation of the raw data as it is provided by the federal Statistical Office
  • Transforming alphanumeric variables into numeric variables
  • Replacing missing values (blanks)
  • Recoding und Declaring the missing values
  • Adding Variable Labels
  • Adding Value Labels
  • Storing the final SPSS-Systemfile

General information about the Microcensus as well as about the processing of raw data with SPSS, can be found in the Readme-file of the respective SETUP.

From Microcensus 2007 onwards the research data centers of the statistical offices also provide system files for the most frequently used programs SPSS, SAS and Stata that were created with these setups.

How can I transform an SPSS-System File into a STATA-System File?

With the programm Stat/Transfer, it is possible to reformat an System File into another format. The specifics of the program (for example, the handling of missing values​​, conventions for representing long labels, etc) have to be taken into account.

With newer versions of SPSS Stata data files can also be created directly: SAVE TRANSLATE [...] \ TYPE = STATA.