Zeitschrift für Bevölkerungswissenschaft 33 (2008) 3-4:409–432 DOI 10.1007/s12523-009-0022-5 W I S S E N S C H A F T L I C H E A B H A N D L U N G – B E G U TA C H T E T
Measuring spatial mobility with the German Microcensus: The case of German return migrants* Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer Abstract: In the context of far-reaching policy reforms on immigration and integration in Germany, there has been a growing interest in better statistical and scientific information about the underlying migration and integration processes. The article argues that, in addition to the necessary efforts to supplement existing statistical systems and set up new surveys, there is a chance to provide more in-depth information through more exhaustive analyses of data already available. The main aim of this contribution is to test whether the information on spatial mobility recorded in the German Microcensus since the early 1980s is a reliable, valid source for migration studies. In addition to a discussion on general methodological features of this official annual population survey, the paper compares estimates of the international migration of German citizens based on the Microcensus with reference data drawn from German population registers. Differentiating the results of other recent contributions on this topic, the analyses show that migration scholars should make more use of the question on “place of residence twelve months ago” when estimating migrants’ socio-economic characteristics and motives. It argues that the German Microcensus on its own, as well as in combination with other similar official population surveys from other countries, provides a valuable data source for nationally- and internationally-comparative migration studies. Keywords: Mobility · Migration · Microcensus · Population Register · Germany
*
We thank Walter Becker, Monika Kübler, Constanze Quiatkowski and Martin Conrad of the Federal Statistical Office for regular help with data of the German migration and population statistics, and Klaus-Jürgen Duschek as well as Robert Herter-Eschweiler for helpful explanations on the intricacies of the Microcensus. Furthermore, thanks are due to Reinhard Pollak for his interest in this project and continuous advice with survey methodologies.
Andreas Ette ( ) · Philip Graze · Lenore Sauer Federal Institute for Population Research, Friedrich-Ebert-Allee 4, 65185 Wiesbaden, Germany E-Mail:
[email protected],
[email protected],
[email protected] URL: http://www.bib-demographie.de Rainer Unger Universität Bremen, Centre for Social Policy Research, Health Economy, Health Policy and Outcomes Research, Barkhof , Parkallee 39, 28209 Bremen, Germany E-Mail:
[email protected], URL: http://www.zes.uni-bremen.de
410
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
Die Erfassung räumlicher Mobilität im Mikrozensus: Das Beispiel deutscher Rückkehrer aus dem Ausland Zusammenfassung: Im Kontext weitreichender politischer Reformen zur Einwanderungs- und Integrationspolitik in Deutschland stieg in den vergangenen Jahren das Interesse an hochwertigen statistischen und wissenschaftlichen Daten und Informationen über die zugrundeliegenden Migrations- und Integrationsprozesse. Neben den notwendigen Ergänzungen bereits existierender statistischer Erfassungssysteme und der Entwicklung neuer Umfragen zu diesem Themenbereich lassen sich aber auch durch vertiefende Analysen bereits vorhandener Datenquellen neue Erkenntnisse erzielen. Im deutschen Mikrozensus werden Informationen zur räumlichen Mobilität bereits seit den frühen 1980er Jahren erfasst. Die Untersuchung der Validität und der Zuverlässigkeit der in der Migrationsforschung bisher nur unzureichend genutzten Datenquelle ist das Hauptanliegen dieses Beitrags. In einem ersten Schritt werden generelle methodologische Charakteristika des Mikrozensus diskutiert. Darauf aufbauend werden auf dem Mikrozensus basierende Schätzungen zur internationalen Migration deutscher Staatsbürger mit Referenzinformationen der deutschen Einwohnermeldeämter verglichen. Die Ergebnisse des Beitrags differenzieren bisherige Untersuchungen zum gleichen Thema und zeigen, dass bei der Analyse der sozio-ökonomischen Charakteristika und Motive von Migranten zukünftig mehr auf die Auswertung der Frage nach dem „Wohnort vor zwölf Monaten“ zurückgegriffen werden sollte. Weiterhin zeigen die Ergebnisse, dass der Mikrozensus für sich wie auch in Kombination mit ähnlichen offiziellen Bevölkerungsumfragen anderer Länder eine wertvolle Datenquelle für nationale als auch international vergleichende Migrationsstudien darstellt. Schlagwörter: Mobilität · Migration · Mikrozensus · Bevölkerungsregister · Deutschland 1
Introduction
Germany’s immigration and integration policies have experienced profound changes in recent years. It is only ten years ago that the then-new Red-Green coalition government started to place this policy area on a new legal foundation: It introduced a fundamental reform of citizenship and prepared Germany’s first Immigration Act. Furthermore, these years have seen changes in the public perception of immigration and integration. Previously, Germany was officially declared a “non-immigration country”, although the empirical reality has told a different story for more than three decades (Thränhardt 1995). The recent changes in both public policy and in public discourse are best understood as a “process of normalisation”, resulting in greater interest in science-based political management of migration and integration processes in Germany (Bommes 2006). The introduction of the Migrationsbericht (migration report) as a yearly reporting system on immigration, published by the Federal Office for Migration and Refugees, and the soon-to-be-established integration indicator report, published by the Federal Government
Measuring spatial mobility with the German Microcensus
411
Commissioner for Migration, Refugees and Integration, are good examples of these developments. So far, official statistics are not well equipped to scientifically monitor and analyse immigration and integration, and longstanding calls by scholars for modifications (e.g. Diehl/Haug 2003; ifo Schnelldienst 2006) have only recently led government commissions to press for changes in the statistical systems (e.g. Sachverständigenrat für Zuwanderung und Integration 2004; Unabhängige Kommission Zuwanderung 2001). Recent years’ efforts to introduce new information to long established statistical systems and the financing of new surveys to fill gaps in the information are therefore highly welcome. The article however argues that, in addition to these improvements, there is a need to make better use of the data that are already available. In particular, it argues that the German Microcensus – which so far was only used for analyses of migrant stocks and their integration processes – also includes information on migration flows, which are rarely used for scientific analyses (for an exception see Diehl/Grobecker 2006). The aim of this paper is to discuss how the Microcensus conceptualizes spatial mobility, and to assess whether it provides a reliable, valid data source for scientific analyses of spatial mobility in general and international migration in particular. The politics of migration make Germany a special case with respect to the official measurement of immigration. Nevertheless, problems of statistical registration of migration are not by any means only a German concern. Rather, the measurement of spatial mobility is a general challenge, and although largely different systems have been developed, they all seem to encounter difficulties in correctly assessing the phenomenon at stake (Boyle et al. 1998). It therefore comes as virtually no surprise that not only scholars in Germany complain about this situation. Compared to countries with no population registers, one could even argue that Germany’s statistical system provides a relatively precise, comprehensive picture of spatial mobility. The local residents’ registration offices (Einwohnermeldeämter) collect information on the number of persons moving in and out of every municipality in Germany. Since registration is obligatory for any person residing longer than two weeks at a given place, migration is directly observable within this system, and is published separately as the German migration statistics (Wanderungsstatistik). Countries which do not have population registers have to make use of statistics on visas and residence permits, their decennial censuses, as well as their yearly official population surveys (e.g. the Current Population Survey in the USA) to provide information on migration flows. In Germany, the Microcensus constitutes a yearly official population survey comparable to that in countries such as the USA, but so far it is not used to estimate spatial mobility. Instead, Germany’s statistical system relies almost exclusively on its population registers, even though they only provide aggregated data with a small number of basic demographic characteristics (sex, age, nationality, marital status, place of origin and destination). The article argues that the German Microcensus is a potential additional data source to measure spatial mobility, including individual-level data on the socio-economic characteristics of migrants which could complement population register data. Spatial mobility, and international migration in particular, is a rather rare social event, with only a small share of the population actually changing their place of residence within a certain period. Because even in large-scale social surveys the case numbers are presumably very small, the article’s main focus is methodological. It tests whether mobility analyses of single migrant groups can be carried out on the basis of the German Micro-
412
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
census. The article proceeds along the following structure: It firstly presents the German Microcensus and its conceptualization of spatial mobility. Secondly, it discusses certain methodological features of this survey and how these potentially affect the reliability and accuracy of estimating spatial mobility, and thirdly it compares the volume and structure of migrants calculated on the basis of the Microcensus with information based on the population register to test for the validity of this data source. To do so, the paper focuses not on spatial mobility in general, but on the international migration of German citizens immigrating to Germany from abroad. There are at least two reasons for selecting this particular group of migrants: First of all, the article attempts not only to test whether general information on spatial mobility can be obtained, but also to show whether analyses of particular groups of migrants are possible. Furthermore, recent years have seen an increasing interest on the part of scholars and politicians alike in Germans’ international mobility. So far, governments have usually been more interested in managing and measuring the migration of foreigners, and not of their own citizens. In consequence, there is generally less information available about the mobility of German own citizens, and it seems an interesting test and proving ground whether the Microcensus is able to shed some light on this particular group of migrants (Nowok et al. 2006: 206). In the article’s conclusion, the results are used to formulate recommendations to increase the usefulness of the Microcensus for measuring spatial mobility. 2
The German Microcensus and its concept of spatial mobility
The German Microcensus is an official representative population survey containing structural population data of Germany. It was introduced in West Germany in 1957, whereas it has been carried out in East Germany since 1991. It is a multipurpose survey providing statistical information on the economic and social situation of the population, on families and their living conditions, as well as on employment, the labour market and the education system. Its original intention was to provide updates between two population censuses. It is therefore carried out on a yearly basis, with a one-percent sample of the whole population living in Germany. This includes private households, as well as persons living in collective households, but does not include members of foreign armed forces or foreign diplomats and their family members. It results in a yearly sample of about 700,000 persons. The content of the Microcensus and its questionnaire is regulated by federal law. It changes regularly and adapts the survey to new demands by the administration, as well as to new scientific interests. This includes changes to the content of variables, the inclusion of entirely new questions or changes to its wording, as well as to the implementation and basic methodological foundations of the survey. Major changes concerning the content of the Microcensus were introduced between 1995 and 1996, whereas a fundamental change concerning its survey design was introduced between 2004 and 2005. This included in particular the introduction of a continuous survey carried out throughout the whole year, whereas all the interviews were previously conducted during a particular reference week in spring.
Measuring spatial mobility with the German Microcensus
Tab. 1:
413
Questions in the German Microcensus used to measure spatial mobility (Microcensus questionnaire 2006) Fragen im Mikrozensus zur Erfassung räumlicher Mobilität (Fragebogen des Mikrozensus 2006) 1st Person
2nd Person
3rd Person
4th Person
go to 116 ĸ Yes ………………
1
No ………………. go to 116 ĸ No answer ………
5th Person
1
1
1
1
8
8
8
8
8
9
9
9
9
9
1
1
1
1
1
go to 115c ĸ No ………………
8
8
8
8
8
go to 116 ĸ No answer ………
9
9
9
9
9
Voluntary
114 Were you living in the same place of residence twelve months ago?
Voluntary
Voluntary
115 Was your former place of residence in the Federal Republic of Germany? Yes ………………
115a In which state was your former place of residence? Please note the appropriate number of the list of states …………………………..
Voluntary
………………………. ………………………. ……………………….
…………………….
115b In which administrative district / which region was your former place of residence? Please note the appropriate number of the list of administrative districts / regions …..
………………………. ………………………. ……………………….
Voluntary
………….……….
115c In which country was your former place of residence? Please note the appropriate number of the list of countries …………………………..
………………………. ………………………. ……………………….
Source: German Federal Statistical Office, translation by the authors
……………………
414
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
With regard to spatial mobility, the Microcensus started in 1983 to ask about the situation of the interviewed person one year before the current survey. This included questions about economic activity, occupational status and economic sector, as well as region and country of residence (Statistisches Bundesamt 1988). Table 1 shows the wording and routing of the respective questions on spatial mobility as implemented in the German Microcensus in 2006. On the basis of the questionnaire, the Microcensus defines spatial mobility as a change in place of residence between the date of the current survey and twelve months previously. This retrospective conceptualization is identical to questions contained in many other population censuses and surveys which ask for the place of residence five years or only three months ago. The “twelve months ago” question is a compromise between periods being long enough to register certain forms of individual changes, without being too long to reduce errors of correct memory (cf. Ellis/Wright 1998). Compared to the Microcensus in general, the question on spatial mobility is subject to two restrictions: Whereas participation in the Microcensus is obligatory, answers to the questions on the situation one year ago are voluntary. The second restriction concerns the sampling fraction. Whereas the Microcensus in general includes one percent of the population in its survey, the questions on the situation one year ago were asked to only 0.45 % in the years up to 2004, and have only covered the whole sample since 2005. The reason for both restrictions is the European Union Labour Force Survey (EU LFS), which originally introduced this set of questions into the Microcensus but demanded lower statistical precision than the German legislature. 3
Methodological characteristics of the German Microcensus affecting the statistical quality of spatial mobility estimations
The main aim of this paper is to assess the validity of estimations of spatial mobility, based on the German Microcensus by comparing it with reference data drawn from the population register. Before this, the focus will be on different technical and methodological aspects of the Microcensus which affect its statistical quality and help to understand differences in the results between the two data sources. Four aspects are discussed in particular: (1) the impact of small domain sizes, (2) proxy interviews, (3) sample rotation patterns and (4) item non-response. 3.1
Small domain sizes
A general problem for migration scholars concerns the available data sources. Official statistics on spatial mobility in general, and on international migration in particular, commonly provide aggregated data produced in the context of administrative procedures only. Although these sources provide important information about the basic structure and development of migration, they do not provide an adequate basis to explain individual decision-making on migration (cf. Schnell et al. 2005: 253; Robinson 1950). Because of the very slight incidence of migration events taking place, only few sources exist containing individual-level data on international migration. Therefore it is of little surprise
Measuring spatial mobility with the German Microcensus
415
that survey data is mainly concerned with migrant stocks, leaving migration flows and their underlying individual decision-making outside their realm of interest. Three different strategies are available to sample migration decisions in surveys. These include the retrospective collection of migration histories (e.g. Courgeau 1985; Wagner 1989), the oversampling of the population of interest by conducting short screening interviews (e.g. Kalter 1997; Schneider/Meil 2008), as well as conducting panel studies (Schupp et al. 2008). Compared to these three strategies and their rather costly data collection procedures, it is the main argument of this paper that the German Microcensus can be used to estimate spatial mobility because its comparatively large sample size allows information to be collected about migration decisions which are representative of the population as a whole. Despite the large sample size, the numbers of respondents indicating spatial mobility are rather small in comparison with the sampled population as a whole. The question therefore arises as to whether these small domains have a negative impact on the accuracy and reliability of spatial mobility estimates. Table 2 presents the unweighted results for the average number of respondents indicating spatial mobility in general, and international migration in particular, for 1996-2004 and 2005-2007.1 The annual average shows that approximately 19,600 individuals in the period 1996-2004 and 43,600 in 2005-2007 indicate that they have changed place of residence during the last twelve months, resulting in a mobility rate for the population as a whole of 7.1 % in the former and 6.7 % in the latter period. There are obvious differences between German citizens and persons of foreign nationality: Between 12.9 % and 15.2 % of the latter group change their place of residence within one year. Focusing on those persons only who immigrated to Germany from abroad, the number of cases is certainly smaller. Here, the annual average shows that approximately 870 individuals in the former period and 1,690 individuals in the more recent period are recorded, resulting in a total mobility rate of approximately 0.3 %. Nevertheless, in each single year of the survey the case numbers have been clearly above 300 German citizens, indicating that they have been immigrating to Germany from abroad, resulting in more than 5,200 cases for the whole twelve-year period. The results in Table 2 show that the Microcensus involves a sizeable number of spatially mobile individuals in general, and German international migrants in particular. Nevertheless, the accuracy of estimates depends on the size of the sample in relation to the domain size, and case numbers are still very small in the example of spatial mobility compared to the overall sample. The underlying problem has been studied intensively in recent years, and is commonly referred to as small area or small domain size problems. It is concerned with the consequences of very infrequent characteristics, which could
1
The change in the survey design of the Microcensus between 2004 and 2005 from the concept of a reference week to a continuous survey carried out throughout the year affected the sample size of the mobility questions. Whereas the questions on spatial mobility during 1996-2004 were part of a sub-sample of approximately 0.45 % of the population, the new design and the data quality requirements resulted in the complete 1.0 % sample being addressed with the mobility questions from 2005 onwards. Consequently, the absolute number of cases is higher for 2005-2007 as against the nine previous years.
416
Tab. 2:
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
Sample sizes of spatially mobile persons in the German Microcensus by citizenship [annual averages for the period 1996-2004 and 2005-2007, unweighted results] Stichprobenumfang räumlich mobiler Personen im Mikrozensus nach Staatsbürgerschaft [jährliche Durchschnitte für die Jahre 1996-2004 und 2005-2007, ungewichtete Ergebnisse]
Same place of residence twelve months before this survey
1996-2004 German Foreign
2005-2007 German Foreign
1 Yes 8 No 9 No answer
241,638 17,053 10,747
14,095 2,528 1,066
569,029 38,182 27,155
36,305 5,388 3,039
0 Not applicable (persons aged under one) Total
2,323 271,760
232 17,917
4,990 639,355
227 44,956
Former place of residence within Germany
1996-2004 German Foreign
1 Yes 8 No 9 No answer 0 Not applicable (no change in place of residence; persons aged under one) Total
2005-2007 German Foreign
15,680 346 1,028
1,814 526 185
36,742 594 847
4,139 1,093 153
254,708 271,764
15,390 17,919
601,173 639,359
39,568 44,958
Source: German Microcensus, 1996-2007
substantially increase the standard error (cf. Ghosh/Rao 1994; Pfeffermann 2002). These studies generally argue that, having only a small sample for a given domain, the only possible solution to the estimation problem is to borrow information from other related data sets. Here, we could either refer to data measured for similar characteristics (e.g. spatial mobility in general instead of international migration), or to data measured for the characteristics of interest on previous occasions. In consequence, the following analyses are not based on single years, but on the whole period from 1996 to 2007, and follow one of the general guidelines for small domain estimations (for a similar approach see Dustmann/Weiss 2007). 3.2
Proxy Interviews
Another aspect affecting the accuracy of estimations concerns the reliability of answers given. Recent years have seen a number of studies on the data quality of answers to questions in the Microcensus. One of the starting points has been formed by studies on the economic activity of respondents (e.g. Schupp et al. 1999; Köhne-Finster/Lingnau 2008).
Measuring spatial mobility with the German Microcensus
417
These studies argue that different results on economic activity in the German Microcensus compared with other surveys are due to non-sampling errors. In particular they refer to the practice of proxy interviews. Although the Microcensus attempts personal interviewing, proxy interviews – where the interviewed person in the household provides information about other persons in the household – are allowed. An average of 27 % of all items of information in the Microcensus are based on proxy information. In the case of economic activity it is rather likely that in particular low paid and temporary activity is underreported when parents are asked about the employment of their children compared to asking them directly. There have also been some studies on the internal reliability of spatial mobility in recent years. These studies have been produced in the process of establishing a Microcensus Panel data set, and show that approximately 15 % of the answers given between 1996 and 1999 are implausible (Herter-Eschweiler 2003: 221). Analyses of the distribution of proxy interviews show their high proportion for persons under the age of 30 (cf. Zühlke 2008). In consequence, it is highly plausible to expect that these discrepancies are at least partly due to proxy interviewing and to different assessments between parents and their children about leaving the parental home. 3.3
Sample rotation patterns
A third aspect which is likely to affect the accuracy of estimates is the particular sample plan of the Microcensus which is based on four equally-sized 0.25 % samples of the population. Each of these four quarters remains in the Microcensus for four years, with one being exchanged every year. This partial rotation procedure leads to each household being asked in four consecutive years. The rate of overlap between each year’s survey is 75 %. Table 3 shows the sample plan of the Microcensus for 1996 to 2007. Each row shows in which survey the respective rotation quarter is interviewed for the first, second, third or fourth time, and when it subsequently drops out of the Microcensus. The columns list the year of the survey and which rotation quarter forms part of each year’s survey. One would expect that the question “What was your place of residence one year ago?” and the particular sample plan of the Microcensus leads to 75 % of the sample never being able to give a positive response, as they have already been in the sample and by definition have the same place of residence for more than one year. This would mean that the population likely to provide a positive answer to the question of spatial mobility is always restricted to the rotation quarter participating in the Microcensus for the first time (see Martí/Ródenas 2007 for such an argument). Although this reasoning sounds plausible at first sight, Figure 1 shows a rather balanced rate of spatially mobile persons in seven different rotation quarters in each of the four different waves. This overall similar pattern of spatial mobility in the different waves of the Microcensus is due to the fact that it is not individual persons and households, but sampling districts, which constitute the basic units of the Microcensus. On average, each sampling district covers nine apartments or houses which remain in the sample for four years, regardless of changes in the persons living there in the meantime. As the figure shows, the chance for spatial mobility is generally the same in all sampling districts, ranging from 4.5 to 7.6 % regardless of the district being part of the Microcensus for the first, second, third or fourth time. It is remarkable that the rotation quarter participating in the Microcensus for the first time even shows a rate of
418
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
Tab. 3:
Sample plan of the German Microcensus and the sequence of survey waves, 1996-2007 Stichprobenplan des Mikrozensus und Abfolge der Erhebungswellen, 1996-2007
Rotation quarter*
1996
3/3
4th w
4/3
3 w
1/4
2
w
3 w
4 w
2/4
1st w
2nd w
3rd w
3/4 4/4 1/5 2/5 3/5 4/5
rd
nd
1997
1998
1999
2000
2001
2002
2004
2005
2006
2007
th
4 w rd
st
1 w
th
nd
2
st
w
1 w
4th w rd
3 w nd
2
st
w
1 w
th
4 w rd
3 w nd
2
st
w
1 w
th
4 w rd
3 w nd
2
th
4 w rd
th
w
3 w
4 w
1st w
2nd w
3rd w
st
1 w
1/6 2/6 3/6 4/6 1/7 *
2003
nd
2
4th w rd
th
w
3 w
4 w
1st w
2nd w
3rd w
st
1 w
nd
2
st
w
1 w
4th w rd
3 w nd
2
th
4 w rd
w
3 w
1st w
2nd w st
1 w
Each single rotation quarter in the Microcensus is defined by the number of the quarter (number in front of the slash) and the specific sample from which it is drawn (number behind the slash). For more information on the survey design of the Microcensus see for example Herter-Eschweiler (2003: 221).
Source: Handbuch Mikrozensus-Panel 1996-1999, own illustration
spatial mobility being on average 1.7 percentage points below the rate in the other waves. It is difficult to provide an explanation for this puzzling result, and studies analysing the impact of rotation quarters on survey outcomes are scarce. Nevertheless, it is rather likely that these comparatively low mobility rates are due to the actual procedures and practices of the interviewers in the field. In many cases the same interviewer is responsible for the same sample district over all four years. In those interviews taken in districts which are part of the Microcensus for the first time, the interviewers have to ask the question on spatial mobility to get an answer. In districts participating for the second, third or fourth time, however, this interview procedure is likely to change if the interviewers are sure that this household did not live in the apartment twelve months before. For efficiency reasons
Measuring spatial mobility with the German Microcensus
Fig. 1: Abb. 1:
419
Share of spatially mobile persons by rotation quarter and wave of survey [unweighted results] Anteil räumlich mobiler Personen nach Rotationsviertel und Erhebungswelle [ungewichtete Ergebnisse]
Percent 8
7
6
5
4 3/4
4/4
1/5
2/5
3/5
4/5
1/6
2/6
3 1st Wave
2nd Wave
3rd Wave
4th Wave
Source: German Microcensus, 1997-2007
it is possible that the interviewer records answers without actually asking the question (cf. Berke 2009). These answers recorded by the interviewers themselves are probably those with greater reliability compared to those answers in the first wave, when the interviewer needs of necessity to rely on the respondents’ answers. Consequently, the underreporting of spatial mobility in the first wave is probably a result of the higher degree of item nonresponse and its potential bias towards persons who are spatially mobile. It is particularly this bias in the item non-response pattern which will form the focus of the next section. 3.4
Item non-response
A final problem for accurate estimations is related to item non-response, understood as missing answers to some questions of a survey. The quality of the German Microcensus generally benefits from the fact that participation and answering the questions of the survey are regulated by federal law, resulting in very low rates (< 10 %) of missing answers to single questions. The issue of item non-response received great public and academic attention within the context of the 1987 census and the debates on data protection related to it. In response, the 1980s and ‘90s saw a number of surveys analysing the consequences of having voluntary and obligatory questions (e.g. Esser et al. 1989; Emmerling/Riede
420
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
1994), arguing that voluntary questions result in serious information losses. In particular, item non-response is correlated with the type of interview, particular questions of the questionnaire (e.g. on education or commuting, etc.) and socio-economic characteristics of the respondent such as economic activity, occupation and age. It is precisely this correlation between non-response and particular characteristics of the respondent which generates bias in estimations. There are various indications in the literature that migration is correlated to non-response. Based on these discussions, Table 4 provides an impression of the pattern of item non-response concerning the question of spatial mobility. The table shows the most relevant variables, including methodological aspects of the Microcensus – the method of the interview and the wave of the survey – as well as socio-economic characteristics of the person interviewed – sex, age, marital status, size of household, education, economic activity, place of residence and citizenship. Overall, the results show that item non-response is not a serious issue, with only 4.3 % of all respondents providing no answer to the questions on migration. Behind this overall picture there are obvious correlations between particular methodological aspects of the Microcensus, as well as the socio-economic characteristics of the interviewed persons which introduce bias into the estimations. Concerning the personal aspects of the interviewed persons, the analysis confirms earlier results that were discussed above. There is a negative relationship between non-response and age and the size of the household, with young persons under the age of 40, as well as single person households, showing the highest non-response rates. This pattern is also confirmed by the finding on marital status, with singles being more inclined to refuse to provide answers to voluntary questions compared to married, widowed or divorced persons. Furthermore, non-response rates are skewed towards the economically active and educated persons, showing that inactive persons, as well as those with a lower level of education, have lower rates of non-response. Finally, there is also an East-West pattern, with interviewed persons in Western Germany having substantially higher non-response rates compared to Eastern Germany. In addition to these socio-economic characteristics of the respondents, the survey design of the Microcensus adds additional explanations to the non-response pattern, with the lowest non-response rates being found by persons being interviewed face-to-face. Instead, almost one person in five refuses to provide an answer when participating in the Microcensus with self-completion questionnaires. Finally, the rotation quarters discussed in the section above also have an influence on item non-response, with the highest non-response rate in the first wave and subsequently decreasing values. This is best explained by findings in other panel studies which show that interviewers and respondents build up a relationship of trust over time, resulting in better response rates on the side of the interviewed person. Similar to the aspects discussed before, the findings on non-response also show that the Microcensus is likely to underestimate spatial mobility because it is particularly those groups of the population with the highest probability of migration which show the highest non-response rates.
Measuring spatial mobility with the German Microcensus
Tab. 4:
421
Selectivity of item non-response concerning spatial mobility by different demographic and survey methodological characteristics Selektivität des Items Non-Response bei Fragen zur räumlichen Mobilität nach verschiedenen sozio-demographischen und erhebungstechnischen Charakteristika
Method of interview Face-to-face Self-completion Wave of survey 1st wave 2nd wave 3rd wave 4th wave Sex Male Female Age <20 20-39 40-59 60+ Marital status single married widowed / divorced Size of household 1 person 2 persons 3+ persons Collective households Education Inadequately completed general education General elementary education Intermediate general qualification General maturity certificate Activity employed unemployed inactive Place of Residence West Germany (not incl. Berlin) East Germany (not incl. Berlin) Berlin Citizenship German Foreign Total
Source: German Microcensus, 1996-2007
Non-response [in Percent]
N [in 1,000]
1.9 18.5
69,855 11,679
5.6 4.3 4.1 3.6
8,031 32,647 32,716 8,119
4.4 4.2
39,796 41,738
4.5 5.3 4.4 2.9
16,127 22,194 23,045 20,168
5.0 3.8 3.9
30,947 39,906 10,675
4.5 4.0 4.3 9.5
13,641 25,429 41,660 806
5.1 2.5 3.5 4.6
1,830 30,484 17,733 13,651
4.6 4.8 3.9
36,492 3,779 41,265
4.4 1.8 12.3
64,521 13,637 3,377
4.1 6.1 4.3
74,424 7,111 81,534
422
4
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
International migration of German citizens: Comparing results based on Microcensus and population register data
The former part has discussed different characteristics of the Microcensus likely to influence the accuracy of estimates of spatial mobility. Below, the paper empirically analyses the reliability and validity of estimates based on the German Microcensus with different data sources (see also Diekmann 2004). In particular it compares results on internationally mobile Germans based on information drawn from the Microcensus and German population registers. Although the group of German citizens immigrating to Germany from abroad is a particular group of migrants only, the results are directly transferable to intra-regional migration, as well as to international migrants with non-German citizenship. Before the results of the comparison between both data sources are presented, the most basic characteristics of population registers and how they are executed in Germany are presented in order to facilitate a better understanding of potential differences between the two types of data. 4.1
Estimating international migration with population registers
Population registers are generally seen as one of the most important data systems to provide information about spatial mobility in general, and about international migration in particular. For example, they provide information on international migration in ten of the EU-15 Member States. Most generally, population registers are built up from a base consisting of an inventory of the inhabitants of an area, modified continuously by current information on births, deaths, changes of residence, etc. (cf. Bilsborrow et al. 1997: 75f.), and information on migration is therefore a by-product of these administrative procedures. In Germany, spatial mobility is recorded in the population registers by registration and deregistration forms of the local resident registration offices. Measuring migration on the basis of this administrative source implies at least three problems: Firstly, population registers in most countries are based on events with no reference to individual persons. In consequence, every change of address which is registered at the residents’ registration office is counted as a separate event of spatial mobility. If a person moves several times a year, this individual is counted several times, resulting in proportionally higher rates of migration compared to the conceptualization applied in the Microcensus. The second problem concerns the fact that this information is based on an administrative source, with the consequence that spatial mobility is only recorded in cases where individuals register and deregister at the local residents’ registration offices. Temporary and seasonal migration in particular is often not registered, potentially resulting in comparatively lower migration rates compared to the concept applied in the Microcensus (e.g. Schriewer/Rodes 2008; Statistisches Bundesamt 2007). The third shortcoming of population register-based statistics is the small number of socio-economic information about the migrants. In Germany, in addition to the overall number of in- and out-migration, only information about citizenship, sex, age, country/region of origin, country/region of destination, marital status and religious affiliation are recorded. No information is registered about education, economic activity, etc. (Lederer 2004: 111f.).
Measuring spatial mobility with the German Microcensus
4.2
423
Estimating the volume of international migration by German citizens
Analysing the reliability and validity of estimates of spatial mobility in the German Microcensus, a first comparison focuses on the overall volume of German international migrants (Fig. 2). Based on the information drawn from the population register for 1996 to 2007, more than two million German citizens have migrated to Germany. There is an obvious trend in this overall number of decreasing immigration flows of German citizens migrating to Germany from other countries. Whereas this migration group consisted of more than 200,000 persons during the 1990s, it was only about 100,000 during the last two years. This development is mainly caused by the changes in the immigration of ethnic Germans (Aussiedler). The largest immigration of Ethnic Germans took place between 1989 and 1990. The numbers have been decreasing since then due to legal changes, as well as to the falling numbers of potentially immigrating ethnic migrants fulfilling the legal preconditions. Fewer than 8,000 immigrated in 2006, compared to almost 400,000 in 1990 and still 177,000 in 1996 (cf. Bundesministerium des Innern 2007). Compared to this large migration volume, with an annual average of 177,000 German migrants moving to Germany between 1996 and 2007, the volume of international migration by German citizens measured on the basis of the German Microcensus is comparatively small. Based on the results of the Microcensus, only slightly more than one million people migrated during this period. In conclusion, the Microcensus leads to a serious under-representation of spatial mobility. The discussion of methodological aspects of the Microcensus in the preceding section helps explain this finding. The practice of proxy interviews, the particular sample rotation pattern, and the issue of non-response, all help to explain the low estimates of international spatial mobility in the Microcensus. The differences in the operationalization of spatial mobility in both data sources provide additional explanations for this finding. Firstly, several movements during one year result in higher migration rates in the register, whereas the Microcensus would not register these moves in-between. The greater the interval of this reference period, the greater the underreporting of migration in the survey. Secondly, registers report migration at the time of their happening, whilst surveys only ask those respondents who have survived. Persons who have left the country between the two time periods cannot be registered. Furthermore, the retrospective measurement of mobility potentially results in underreporting because it is not remembered correctly at the time of the interview. A final explanation of the discrepancies is due to the different data collection strategies in the two data sources. All ethnic Germans are recorded in the population register at the reception centre where they first arrive in Germany. The data collection strategy of the Microcensus is very different to this administrative procedure used by the population register. Theoretically, ethnic Germans residing in a reception centre have a chance to be part of the sampling frame of the Microcensus because it includes reception centres and other types of collective households. Nevertheless, the high concentration of reception centres for ethnic Germans, with only one remaining since 2000, makes it rather unlikely that these centres and ethnic Germans respectively are adequately represented in the Microcensus. As a result, in particular those ethnic Germans who stay in reception centres for longer periods of time have a reduced chance of being adequately represented in the Microcensus.
424
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
Fig. 2: Abb. 2:
Return Migration of German Citizens: Comparing the volume of migrants based on Microcensus and Population Register data, 1996-2007 Rückwanderung deutscher Staatsbürger: Vergleich der Anzahl der Migranten nach Mikrozensus und Einwohnermelderegister, 1996-2007
in 1,000 300 Population Register Microcensus 250
200
150
100
50
0 1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
Source: Own calculations based on data from the German Microcensus and the German Migration Statistics (both German Federal Statistical Office), 1996-2007
In addition to the overall lower rate of spatial mobility as measured by the Microcensus in comparison to the population register, there are two additional aspects worth mentioning. Firstly, the development trend of falling numbers of German immigrants – obvious in population register data – is also observable in Microcensus data. In consequence, the under-representation of spatial mobility is likely to be due to the conceptualization of spatial mobility and a constant under-representation of mobile persons in the sampling frame, and not caused by a high level of volatility in the sampling procedure. A second aspect concerns the difference between the 1996 to 2004 period compared to the last three years (2005-2007). The two periods are based on different survey designs of the Microcensus. What is remarkable is the fact that the degree of under-representation is considerably smaller in the latter period. Whereas during the whole previous period, less than half of the movements have been accounted for with the Microcensus (48 %), the overlap has now increased to almost two-thirds (66 %). It shows that international migration of German citizens is subject to seasonal influences, probably due to educational exchange programs, as well as to seasonal retirement migration, and provides additional backing for the argument that the proportionally lower migration rates are not caused by volatility
Measuring spatial mobility with the German Microcensus
425
in the sampling procedure, but by constant characteristics of the Microcensus which one can account for in its interpretations. 4.3
Estimating the selectivity of international migration by German citizens
The results in the preceding part suggest that the sampling procedure misrepresents certain groups of mobile persons, but in a somewhat constant manner. In particular, the practice of proxy interviews, the sample rotation patterns and the non-response bias, combined with differences relating to the data collection strategy and the operationalization of spatial mobility, are the main explanatory factors accounting for the lower migration rates in the Microcensus compared to population register information. The following part focuses on the main demographic characteristics of German citizens immigrating from abroad in order to assess how these differences between the two data sources impact the socio-economic information about this group of migrants. In a first step, estimates of the age and sex composition of German international migrants are compared between the two data sources. Figure 3 shows that there is an overall match between the Microcensus and the population register, although the Microcensus estimates that slightly more German women compared to men are internationally mobile, whereas the population registers show a reverse ratio. The difference between the two data sources is below 2.5 percentage points in all age groups except among 21- to 30-year-old women, where the Microcensus displays a share that is 4.8 percentage points higher – almost 50 % more than that in the population registers. The results for men are also rather large compared to other age groups, although not as different as for women. It is reasonable to expect that the lower values calculated on the basis of the population registers are not necessarily the more precise estimates in this case. The group of 21- to 30-year-olds is highly mobile in the context of education and training. Exchange programs offer highly regulated programs, mostly resulting in stays of six or twelve months abroad with increasing shares of students participating (Deutscher Akademischer Austausch Dienst 2006). Because these movements are planned as temporary from the outset, it is likely that most students do not deregister in Germany when leaving and reregister when returning home. Overall, however, both data sources show largely similar sex and age structures, which provides credibility for the validity of both data sources. Countries of origin are compared in a second analysis. The results in Figure 4 show that the largest group of German international migrants during 1996 to 2007 immigrated from the category “other countries”, with 63 % on the basis of the population register and 62 % on the basis of the Microcensus. This large group certainly consists mainly of ethnic Germans immigrating from Eastern European and Asian countries. Overall, the figure shows that the match between the two data sources on countries of origin is rather high. Even disaggregating the data along single countries such as Switzerland or the United States produces rather similar estimates – mostly below one percentage point difference – on shares of German citizens immigrating from these countries. A final comparison of the aptitude of Microcensus data to measure spatial mobility concerns the area of destination, the particular the German state to which German international migrants move. The results of this distribution show a very poor match overall between the two data sources. The differences are around four percentage points in most
426
Fig. 3: Abb. 3:
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
Return Migration of German Citizens: Comparing age and sex distributions based on Microcensus and Population Register data for 1996-2007 Rückwanderung deutscher Staatsbürger: Vergleich der Alters- und Geschlechtsstruktur nach Mikrozensus und Einwohnermelderegister für die Jahre 1996-2007 Age in Years
Male
Female
71+
61-70
51-60
41-50
31-40
21-30
11-20
0-10
20 Percent
15
10
5
0
Microcensus
0
5
10
15
20 Percent
Population Register
Source: Own calculations based on data from the German Microcensus and the German Migration Statistics (both German Federal Statistical Office), 1996-2007
Measuring spatial mobility with the German Microcensus
Fig. 4: Abb. 4:
427
Return Migration of German Citizens: Comparing distributions of countries of origin based on Microcensus and Population Register data for 1996-2007 Rückwanderung deutscher Staatsbürger: Vergleich der Verteilung nach Herkunftsländern basierend auf Mikrozensus und Einwohnermelderegister für die Jahre 1996-2007
Percent 70 Microcensus
Population Register
60 50 40 30 20 10 0 United Kingdom, Ireland
France
Belgium, Sweden, Spain, Netherlands, Denmark, Greece, Luxemburg Finland Portugal, Italy
Austria Switzerland
Czech Republic, Hungary, Poland, Slovakia
USA
Other Countries
Source: Own calculations based on data from the German Microcensus and the German Migration Statistics (both German Federal Statistical Office), 1996-2007
cases, but rise to more than 25 percentage points in the case of Lower Saxony. This poor match, is however again best explained by the difficulties of the Microcensus to include the reception centres for ethnic Germans in its sample frame because there has only been one centre remaining since 2000, namely in Lower Saxony. Excluding Lower Saxony from these calculations results in a rather good match between the two data sources, similar to the two analyses presented above, with differences between the population register and the Microcensus being generally below two percentage points (see Fig. 5 for these results).2 Overall, the comparison between the two data sources and their estimates of the volume and structure of German citizens immigrating from abroad shows that these sour-
2
Hesse remains the only exception, which is explained by adjustments to the population registers in 2004 and 2005 in this Land, resulting in substantially higher international migration figures for these two years.
428
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
Fig. 5: Abb. 5:
Return Migration of German Citizens: Comparing average distributions of regions of destination based on Microcensus and Population Register data for 1996-2007 Rückwanderung deutscher Staatsbürger: Vergleich der Verteilung nach Zielregionen basierend auf Mikrozensus und Einwohnermelderegister für die Jahre 1996-2007
Percent 30 Microcensus
Population Register
25
20
15
10
5
0 BadenWürttemberg
Bavaria
Berlin
Brandenburg, Bremen, Mecklenburg- Hamburg, SchleswigWestern Holstein Pomerania, Saxony, SaxonyAnhalt, Thuringia
Hesse
North-Rhine Westphalia
RhinelandPalatinate, Saarland
Source: Own calculations based on data from the German Microcensus and the German Migration Statistics (both German Federal Statistical Office), 1996-2007
ces present largely different volumes of international migration, explained by the different conceptualizations of spatial mobility, the survey design and the sample frame of the Microcensus, as well as the influence of the item non-response bias. On the other hand, the comparison has shown a good match concerning the demographic characteristics of those migrants. Furthermore, the analysis has shown that we can explain and consequently account for the differences between the two data sources, making the Microcensus a very useful data source for migration studies because it allows rather detailed studies of the characteristics of spatially mobile persons that would be inconceivable in studies based on population registers only.
Measuring spatial mobility with the German Microcensus
5
429
Conclusions and recommendations
The aim of this contribution was to assess whether the German Microcensus can successfully be used for analyses of spatial mobility. In addition to some methodological considerations, the article tested the validity of this survey using the population register information as a source of reference. In particular, it focused on the international migration of German citizens between 1996 and 2007, a type of migration which has received increasing attention in recent years in the public and political debates without scholars being able to provide in-depth analyses of this group of migrants. In the context of this paper, this particular and rather small group functioned as a testing and proving ground for assessing the validity of spatial mobility information contained in the Microcensus. The results show that, on the one hand, the Microcensus underestimates the overall volume of international migration over the past twelve years by almost 50 percent. This is caused by a rather different conceptualization of migration compared to the population register, the focus on a particular reference week, the bias introduced by the practice of proxy interviews, the sample rotation pattern, item non-response distributions and the high concentration of reception centres for ethnic Germans which are not likely to be adequately represented in the sample frame. On the other hand, the comparison of the selectivity of these migration flows shows a very good match between the two data sources. If one analyses the distribution of countries of origin and regions of destination, as well as sex and age structures of these migrants, the results show that the Microcensus is even able to correct for some shortcomings in the data collection system of the population registers. In their analysis of the European Union Labour Force Survey (EU LFS), which originally caused the Microcensus to include those questions on spatial mobility in its questionnaire, Marti and Ródenas (2007: 113) propose to use this information only for analyses of migration stocks. They concluded from their analysis of the overlap of population register data on the volume of international migration compared with the estimates provided by the EU LFS that survey-based information largely underestimates the real flows, and is therefore of little assistance for analyses of international migration. Based on the results of this paper, this conclusion needs to be qualified. The bias introduced by item non-response can be corrected in statistical analyses, and an accurate estimation of the structure of basic demographic characteristics shows that the Microcensus is usefully applied to provide a more complete picture of the socio-economic profile of migrants and their individual motives for changing place of residence. With its interest in the potential and shortcomings of the German Microcensus, this article is among other papers with a similar interest but focusing on other thematic aspects where the Microcensus potentially provides helpful information (cf. Schupp et al. 1999; Lengerer et al. 2007). This growing awareness of scholars for the Microcensus is mainly due to the increasing efforts to increase its accessibility by publishing yearly scientific use files. Analyses of spatial mobility with the scientific use files are however rather restricted because the variables on the particular regions and countries of origin are published in aggregated format, which certainly reduces its potential. To improve the Microcensus as a potential data source for mobility analyses, these regulations for the establishment of scientific use files should be reconsidered. In the long run, one should also think about changing the status of these questions from a voluntary to an obligatory part of the que-
430
Andreas Ette, Rainer Unger, Philip Graze and Lenore Sauer
stionnaire. This would be a rather cost-efficient change with considerable potential for better spatial mobility analyses in Germany. Despite these recommendations, the inclusion of these questions on spatial mobility since the early 1980s has already made the Microcensus a rich reservoir for migration scholars of which insufficient use has so far been made. It should be regularly used for intra-regional as well as international migration in Germany by providing socio-economic characteristics of migrants and variables to analyse migration decisions. In this respect, the Microcensus has a rather large potential because there are few alternative data sources which provide this information. Furthermore, it is of additional assistance because the questions on spatial mobility are comparable to the conceptualization of migration in other official annual population and Labour Force Surveys. It therefore provides a largely compatible data source which also has considerable potential for internationallycomparative migration studies. 6
Bibliography
Berke, Paul, 2009: Ergebnisse einer nordrhein-westfälischen Zusatzbefragung der Interviewerinnen und Interviewer im Mikrozensus zu ihren Erfahrungen in der Feldarbeit. In: Statistische Analysen und Studien NRW 55: 15-35 Bilsborrow, Richard E. et al., 1997: International Migration Statistics. Guidelines for Improving Data Collection Systems. Geneva: International Labour Office Bommes, Michael, 2006: Einleitung: Migrations- und Integrationspolitik in Deutschland zwischen institutioneller Anpassung und Abwehr. In: Bommes, Michael; Schiffauer, Werner (eds.): Migrationsreport 2006. Fakten – Analysen – Perspektiven. Frankfurt/New York: Campus Boyle, Paul et al., 1998: Exploring Contemporary Migration. Harlow: Longman Bundesministerium des Innern, 2007: Migrationsbericht 2006 des Bundesamtes für Migration und Flüchtlinge, commissioned by the Federal Government. Nürnberg: Bundesamt für Migration und Flüchtlinge Courgeau, Daniel, 1985: Interaction between spatial mobility, family and career life-cycle: A French Survey. In: European Sociological Review 1: 139-162 Deutscher Akademischer Austausch Dienst, (ed.) 2006: Wissenschaft weltoffen 2006. Daten und Fakten zur Internationalität von Studium und Forschung in Deutschland. Bielefeld: W. Bertelsmann Verlag Diehl, Claudia; Grobecker, Claire, 2006: Neuzuwanderer in Deutschland, Ergebnisse des Mikrozensus 2000 bis 2003. In: Wirtschaft und Statistik 11/2006: 1-12 Diehl, Claudia; Haug, Sonja, 2003: Assessing migration and integration in an immigration hesitant country: the sources and problems of data in Germany. In: Studi Emigrazione/Migration Studies 40,152: 747-771 Diekmann, Andreas, 2004: Empirische Sozialforschung. Grundlagen, Methoden, Anwendungen. Reinbek: Rowohlt Dustmann, Christian; Weiss, Yoram, 2007: Return Migration: Theory and Empirical Evidence from the UK. In: British Journal of Industrial Relations 45,2: 236-256 Ellis, Mark; Wright, Richard, 1998: When Immigrants are not Migrants: Counting Arrivals of the Foreign Born using the U.S. Census. In: International Migration Review 32,1: 127-144 Emmerling, Dieter; Riede, Thomas, 1994: Zur Freiwilligkeit in der Auskunftserteilung im Mikrozensus. In: Wirtschaft und Statistik 6/1994: 435-449 Esser, Hartmut et al., 1989: Mikrozensus im Wandel. Untersuchungen und Empfehlungen zur inhaltlichen und methodischen Gestaltung. Stuttgart
Measuring spatial mobility with the German Microcensus
431
Ghosh, Malay; Rao, J. N. K., 1994: Small Area Estimation: An Appraisal. In: Statistical Science 9,1: 55-93 Herter-Eschweiler, Robert, 2003: Längsschnittdaten aus dem Mikrozensus: Basis für neue Analysemöglichkeiten. Bonn: Statistisches Bundesamt ifo Schnelldienst, (ed.) 2006: Zuwanderung nach Deutschland – wie zuverlässig ist die Statistik? München: Institut für Wirtschaftsforschung Kalter, Frank, 1997: Wohnortwechsel in Deutschland. Ein Beitrag zur Migrationstheorie und zur empirischen Anwendung von Rational-Choice-Modellen. Opladen: Leske + Budrich Köhne-Finster, Sabine; Lingnau, Andreas, 2008: Untersuchung der Datenqualität erwerbsstatistischer Angaben im Mikrozensus. In: Wirtschaft und Statistik12/2008: 1067-1088 Lederer, Harald, 2004: Indikatoren der Migration: Zur Messung des Umfangs und der Arten von Migration in Deutschland unter besonderer Berücksichtigung des Ehegatten- und Familiennachzugs sowie der illegalen Migration. Bamberg: europäisches forum für migrationsstudien Lengerer, Andrea et al., 2007: Familiensoziologische Analysepotenziale des Mikrozensus. In: Zeitschrift für Familienforschung 19,2: 186-209 Martí, Mónica; Ródenas, Carmen, 2007: Migration Estimation Based on the Labour Force Survey: An EU-15 Perspective. In: International Migration Review 41,1: 101-126 Nowok, Beata et al., 2006: Statistics on International Migration Flows. In: Poulain, Michel et al. (eds.): THESIM. Towards Harmonised European Statistics on International Migration. Louvain: Presses Universitaires de Louvain Pfeffermann, Danny, 2002: Small Area Estimation-New Developments and Directions. In: International Statistical Review 70,1: 125-143 Robinson,William S., 1950: Ecological Correlations and the Behavior of Individuals. In: American Sociological Review 15,3: 351-357 Sachverständigenrat für Zuwanderung und Integration, 2004: Migration und Integration – Erfahrungen nutzen, Neues wagen. Jahresgutachten des Sachverständigenrates für Zuwanderung und Integration. Nürnberg: Sachverständigenrat für Zuwanderung und Integration Schneider, Norbert F.; Meil, Gerardo, (eds.) 2008: Mobile Living Across Europe. Relevance and Diversity of Job-Related Spatial Mobility in Six European Countries. Opladen, Farmington Hills: Barbara Budrich Schnell, Rainer et al., 2005: Methoden der empirischen Sozialforschung. München: Oldenbourg Schriewer, Klaus; Rodes, Joaquín, 2008: Die offizielle und die verborgene europäische Wohlstandsmobilität. In: IMIS-Beiträge 33: 85-103 Schupp, Jürgen et al., 1999: Zur Erhebungsproblematik geringfügiger Beschäftigung: Ein Strukturvergleich des Mikrozensus mit dem Sozio-oekonomischen Panel und dem europäischen Haushaltspanel. In: Lüttinger, Paul (ed.): Sozialstrukturanalysen mit dem Mikrozensus. Mannheim: Zentrum für Umfragen, Methoden und Analysen (ZUMA) Schupp, Jürgen et al., 2008: Leben außerhalb Deutschlands – Eine Machbarkeitsstudie zur Realisierung von Auslandsbefragungen auf Basis des Sozio-oekonomischen Panels (SOEP). In: SOEP papers on Multidisciplinary Panel Data Research 120, August 2008 Statistisches Bundesamt, 1988: Arbeitsunterlagen zum Mikrozensus. Das Erhebungsprogramm des Mikrozensus seit 1957. Wiesbaden: Statistisches Bundesamt Statistisches Bundesamt, 2007: Qualitätsbericht Wanderungsstatistik. Wiesbaden: Statistisches Bundesamt Thränhardt, Dietrich, 1995: Germany: an undeclared immigration country. In: New Community 21,1: 19-36 Unabhängige Kommission Zuwanderung, 2001: Zuwanderung gestalten, Integration fördern. Berlin: Unabhängige Kommission Zuwanderung Wagner, Michael, 1989: Räumliche Mobilität im Lebenslauf. Stuttgart: Ferdinand Enke Zühlke, Sylvia, 2008: Auswirkungen von Proxy-Interviews auf die Datenqualität des Mikrozensus. In: Statistische Analysen und Studien NRW 53: 3-10
