SOEP-Core v32 - Documentation on biography and life history data

Goebel, Jan (Ed.); DIW Berlin / SOEP (Ed.) Research Report SOEP-Core v32 - Documentation on biography and life history data SOEP Survey Papers, No. 418 Provided in Cooperation with: German Institute for Economic Research (DIW Berlin) Suggested Citation: Goebel, Jan (Ed.); DIW Berlin / SOEP (Ed.) (2017) : SOEP-Core v32 - Documentation on biography and life history data, SOEP Survey Papers, No. 418, Deutsches Institut für Wirtschaftsforschung (DIW), Berlin This Version is available at: https://hdl.handle.net/10419/155351 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-sa/4.0/ SOEP Survey Papers Series D – Variable Descriptions and Coding The German Socio-Economic Panel study SOEP-Core v32 – Documentation on Biography and Life History Data 418 SOEP — The German Socio-Economic Panel study at DIW Berlin 2017 Jan Goebel (ed.) ███ Running since 1984, the German Socio-Economic Panel study (SOEP) is a wide-ranging representative longitudinal study of private households, located at the German Institute for Economic Research, DIW Berlin. The aim of the SOEP Survey Papers Series is to thoroughly document the survey’s data collection and data processing. The SOEP Survey Papers is comprised of the following series: Series A – Survey Instruments (Erhebungsinstrumente) Series B – Survey Reports (Methodenberichte) Series C – Data Documentation (Datendokumentationen) Series D – Variable Descriptions and Coding Series E – SOEPmonitors Series F – SOEP Newsletters Series G – General Issues and Teaching Materials The SOEP Survey Papers are available at http://www.diw.de/soepsurveypapers Editors: Dr. Jan Goebel, DIW Berlin Prof. Dr. Martin Kroh, DIW Berlin and Humboldt Universität Berlin Prof. Dr. Carsten Schröder, DIW Berlin and Freie Universität Berlin Prof. Dr. Jürgen Schupp, DIW Berlin and Freie Universität Berlin Please cite this paper as follows: Jan Goebel (ed.). 2017. SOEP-Core v32 – Documentation on Biography and Life History Data. SOEP Survey Papers 418: Series D. Berlin: DIW/SOEP This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License. © 2017 by SOEP ISSN: 2193-5580 (online) DIW Berlin German Socio-Economic Panel (SOEP) Mohrenstr. 58 10117 Berlin Germany [email protected] ███ SOEP-Core v32 – Documentation on Biography and Life History Data Jan Goebel (ed.) The files described in this documentation are part of a collection, which is released with doi:10.5684/soep.v32 ███ 1 General Introduction and Overview of available datasets ......................................... 6 2 Biographical Information in the Meta File PPFAD (Month of Birth, Year of Death, Immigration Variables, Living in East or West Germany in 1989) ............................. 19 2.1 The Month of Birth in the data set PPFAD .............................................................................................. 19 2.2 Construction of variables ........................................................................................................................ 20 2.3 Year of birth ............................................................................................................................................ 23 2.4 Year of death ........................................................................................................................................... 24 2.5 Immigration information ........................................................................................................................ 26 2.6 Living in East or West Germany in 1989 ................................................................................................. 35 3 MIGSPELL: The Migration-Biography (Samples M1 and M2) .................................... 37 3.1 Introduction to the new release of MIGSPELL ........................................................................................ 37 3.2 Summary description of MIGSPELL ......................................................................................................... 38 3.3 Overview: the structure of the migration biography questions ............................................................. 38 3.4 Description of the variables in MIGSPELL ............................................................................................... 41 3.4.1 Summary description of the changes in MIGSPELL.................................................................................... 41 3.4.2 Synopsis of the variables in MIGSPELL (systematic order) ........................................................................ 42 3.4.3 Levels and value labels of the categorical variables (alphabetical order) ................................................. 43 3.4.4 The date-variables in MIGSPELL ................................................................................................................. 48 3.5 The imputation of missing date values ................................................................................................... 49 3.5.1 General remarks ........................................................................................................................................ 49 3.5.2 Procedure for the imputation of the missing date values ......................................................................... 50 3.6 The integration of the migration biographies of the three waves 2013-2015 (bd, be, bf) ..................... 53 3.6.1 Structure Tables: $$p_mig-variables to MIGSPELL-variables for waves bd to bf ...................................... 55 3.6.2 Synopsis: Mapping of the migbiography variables of waves 2013-2015 to the MIGSPELL variables ........ 61 4 Activity Biography in the Files PBIOSPE and ARTKALEN ........................................... 65 5 BIOJOB: Detailed Information on First and Last Job ................................................ 74 5.1 Overview ................................................................................................................................................. 74 5.2 Structure and Contents of BIOJOB .......................................................................................................... 75 5.3 Steps of Coding ....................................................................................................................................... 94 6 The couple history files BIOCOUPLM and BIOCOUPLY, and marital history files BIOMARSM and BIOMARSY ................................................................................... 97 6.1 Sources of the couple and marital history .............................................................................................. 98 6.2 Construction of marital histories ............................................................................................................ 99 6.3 Construction of couple histories ........................................................................................................... 101 6.4 BIOCOUPLM: A monthly couple biography ........................................................................................... 105 6.5 BIOCOUPLY: An annual couple biography ............................................................................................. 108 6.6 BIOMARSM: A monthly marital history ................................................................................................ 111 ███ SOEP Survey Papers 418 3 SOEP v32 6.7 BIOMARSY: A annual marital biography ............................................................................................... 113 7 BIOBIRTH: A Data Set on the Birth Biography of Female Respondents ................... 116 7.1 Population and purpose of the data set BIOBIRTH ............................................................................... 116 7.2 Structure of the data set ....................................................................................................................... 117 7.3 Information basis of the birth biography .............................................................................................. 118 7.4 A new source of biographical information – the youth questionnaire ................................................. 120 7.5 The fertility histories of male respondents in BIOBIRTH ....................................................................... 120 7.6 Integration of “Familien in Deutschland” – FiD .................................................................................... 121 7.7 Identification process of the children in the SOEP data base ............................................................... 122 7.8 Identification of the children of parents with completed fertility histories ......................................... 124 7.9 Identification of the children for women who have no biography data/ not completed the biography questionnaire ........................................................................................................................................ 124 7.10 Updating BIOBIRTH ............................................................................................................................... 124 8 BIOTWIN: TWINS in the SOEP ............................................................................... 128 8.1 Population and contents of the data set BIOTWIN ............................................................................... 128 8.2 The twin survey of 2006 ........................................................................................................................ 129 8.3 Construction of variables in the data set BIOTWIN .............................................................................. 129 9 BIOSIB: Information on siblings in the SOEP .......................................................... 132 9.1 General description of the data set ...................................................................................................... 132 9.2 Sources of information on siblings in the SOEP .................................................................................... 132 9.3 Overview on the number of siblings in BIOSIB...................................................................................... 133 9.4 Organization of the data in BIOSIB ........................................................................................................ 133 10 BIOAGE01, BIOAGE03, BIOAGE06, BIOAGE08, BIOAGE10, BIOAGE12: Generated variables from the “Mother & Child”, “Parent”, and “Pupils” questionnaires ........ 140 10.1 Introduction .......................................................................................................................................... 140 10.2 Respondents in the ‘Bioagel’ Data Set .................................................................................................. 141 10.3 Topics and Variables ............................................................................................................................. 144 10.4 Generated Variables ............................................................................................................................. 145 11 BIOAGE17: The Youth Questionnaire .................................................................... 149 11.1 Genesis and Target Population of the Youth Questionnaire ................................................................ 149 11.2 Contents and Structure of the Data Set BIOAGE17 .............................................................................. 151 11.3 Special Features of Some Questions and Variables .............................................................................. 152 12 BIOSOC: Retrospective Data on Youth and Socialization ....................................... 161 12.1 Structure of the Data Set BIOSOC ......................................................................................................... 162 12.2 Special Features of Some Questions and Variables .............................................................................. 162 SOEP Survey Papers 418 4 SOEP v32 13 BIOPAREN: Biography Information for the Parents of SOEP-Respondents ............. 170 13.1 Short summary ...................................................................................................................................... 170 13.2 How biography information has been collected in the SOEP ............................................................... 170 13.3 How is BIOPAREN generated? ............................................................................................................... 173 13.4 What’s new in version v32? .................................................................................................................. 177 13.5 Complete list of variables in BIOPAREN ................................................................................................ 178 14 BIOIMMIG: Generated and Status Variables from SOEP for Foreigners and Migrants 213 14.1 Content ................................................................................................................................................. 213 14.2 Status Variables and Carrying Forth of Information ............................................................................. 213 14.3 Updating of Time-Dependent Information ........................................................................................... 215 14.4 Using this File ........................................................................................................................................ 216 14.5 Using BIOIMMIG as a Cross-Section ...................................................................................................... 216 14.6 Documentation of the Variables ........................................................................................................... 217 15 BIORESID: Variables on Occupancy and Second Residence .................................... 269 15.1 Sources of Variables .............................................................................................................................. 270 15.2 Population of Interest ........................................................................................................................... 270 15.3 Variable List of the Data Set BIORESID .................................................................................................. 273 15.4 Recent Changes in the Data Set ............................................................................................................ 273 16 BIOEDU: Data on educational participation and transitions .................................. 274 17 LIFESPELL: Information on the Preand Post-Survey History of SOEP-Respondents 277 ███ SOEP Survey Papers 418 5 SOEP v32 1 General Introduction and Overview of available datasets by Jan Goebel By compiling a comprehensive set of questions on the individual life history into userfriendly variables, the SOEP database provides users with a representative collection of biographical information for the entire German population.1 This covers information on the individual career path since the age of 15, on marital status and childhood biography, the first job, social background and migration history. The function of these data is, on the one hand, to make important background information available for analyses (e.g. information on fertility as an explanatory variable when analyzing labor market supply of women), and, on the other hand, to support self-contained analyses (e.g. on occupational careers or intergenerational transmission of education). In general, each respondent of the SOEP questionnaire (surveying age starts in the calendar year a person turns 17 years) will answer the biographical questions only once (retrospectively). In the beginning of the SOEP, this occurred within the framework of the first three waves (1984 to 1986). Due to the inevitable ‘mortality rate’ of the panel (refusal to participate, death, relocation abroad), this process unfortunately leads to missing biographical entries for persons who did not participate in all three waves. Because of this, since 1988 all biographical information (occupation, marital status, family, first job and social background) is, in principle, collected during the first interview for new respondents in existing sample households. It should be noted that - due to the costs involved and the increased response burden—the main objective of surveying the biographical information in the course of the very first interview is not applied to the first wave of new subsamples. For example, in sample C (East Germany, field work started in 1990) the biographical questionnaire was first collected in 1992. Consequently, the surveyed persons in sample C who left SOEP before 1992 or who refused to complete the biography questionnaire in 1992 have no biographical information included in the SOEP data. Starting with Sample J in 2011 we now enable member of new subsamples to fill in an integrated questionnaire, combining individual and biographical questions. This procedure allows us to collect biographical information during the first wave without increased response burden, however at the expense of a slightly different individual questionnaire when compared to “old” samples. In such a case, the effected variables will be set to “-5 Not included in this version of the questionnaire” for the entire subsample. Summing up, in principle most of the biographical information in the SOEP is collected by means of the so-called ‘Lebenslauf’ (`life history’) questionnaire. Although naming 1 A general introduction into the SOEP database can be found in our Desktop Companion (DTC) at http://about.paneldata.org/soep/dtc/ ███ SOEP Survey Papers 418 6 SOEP v32 conventions, positioning of questions and the scope of this questionnaire have been changed and revised several times (see below), it has been addressed once at each respondent throughout the SOEP. Since 2000, a separate youth questionnaire exists which contains youth-specific questions.2 A whole new series of age-triggered instruments for collecting biographical data was implemented in 2003. The target of the first of these questionnaires is to collect information about newborn children. It is aimed at their mothers of children aged up to 15 months. As a result, the SOEP has started to survey the development of children from the very beginning of their life and will provide users with a completely new type of data. In 2005, a follow-up questionnaire targeted at children aged 2 to 3 years was implemented. Again, the information was collected from the mothers. It contained questions on the child’s individual development and the mother’s specific experiences during this formative period of raising the child. There will be follow-up interviews to collect data about these children at specific ages which are typically associated with decisions relevant to their individual development. The respective questionnaire targeted at children aged 5 to 6 was implemented in 2008. A questionnaire targeted at children aged 7 to 8 years is used for the first time in 2010. This questionnaire will be answered by mothers and fathers (in contrast to earlier agetriggered instruments which were answered by mothers, only) and will therefore delivered in two separate files. A questionnaire targeted at children aged 9 to10 years is used for the first time in 2012. In 2014 we introduced a specific questionnaire for pupils aged 11 to 12 years. A chronological listing of the various changes related to the survey of biographically relevant information for the time period 1984 to 2015 can be found below. The differences in gathering information among and between the various sub-samples are reported with respect to ‘Timing’ (when respondents were asked), ‘Coverage’ (which parts of the biographical topics and single indicators were asked), and ‘Positioning’ of the biographical questions in the diverse survey instruments. 1984 The focus of the survey from samples A and B was the occupational biography. This information was collected (retrospectively) with the help of a ‘life-course calendar’ and covered the time period from the age of 15 up to the current age (or up until and including the maximum age of 65). The ‘calendar’ takes the form of a matrix with one column for every year of age and up to nine specifications of occupational activities (school, apprenticeship/training, military and community service, employed full time, employed part time, unemployed, househusband/wife, retired and other; question 62 in the standard Individual Questionnaire in Wave 1). 1985 The focus for samples A and B was on collecting marital and family status information in retrospect (questions 81-88 in the standard Individual 2 Due to survey-related reasons biographical information was not asked of first-time respondents aged 16 or 17 years until 1999. For this group of persons, much of the biographical information (i.e. on marital status and family information, occupation history since the age of 16 and social background) can generally be reconstructed using variables collected by means of the Individual Questionnaire, i.e. from the yearly ongoing survey. SOEP Survey Papers 418 7 SOEP v32 Table 1: Biographical data in SOEP Files in the SOEP Database Biography Sub-area Number of Question in the ‘Lebenslauf’ Questionnaire (2006) Comparable Questions in the Youth Questionnaire (2006) SOEP Target Population Analysis Unit Update Requirements (Source File for Update) Status: Available / Not Available (up to Wave BE) PPFAD Country of birth 2, 3 61, 62 All persons surveyed Individual No Available PPFAD Year of immigration 4 63 For persons not born in Germany Individual No Available MIGSPELL Migration biography migration modul in CAPI version Sample M only SPELL No Available BIOIMMIG Immigration biography 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 15a 64, 65, 66, 67, 68, 69, 70, 71 For persons not born in Germany Individual No Available PPFAD Living in East or West Germany in 1989 16 - All persons surveyed Individual No Available BIOPAREN Place of childhood; Life at childhood residence; grew up with parents, Living together with parents 17,17a, 19, 20 72, 73, 75, 76 All persons surveyed Individual No Available BIOSIB Number of brothers and sisters 18 74 All persons surveyed Individual Yes Available BIOPAREN Parents living region, year of birth, year of death, nationality, country of birth 21, 22, 23, 23a 77, 78, 79 All persons surveyed Individual Partly (year of death from PPFAD) Available ███ SOEP Survey Papers 418 14 SOEP v32 Files in the SOEP Database Biography Sub-area Number of Question in the ‘Lebenslauf’ Questionnaire (2006) Comparable Questions in the Youth Questionnaire (2006) SOEP Target Population Analysis Unit Update Requirements (Source File for Update) Status: Available / Not Available (up to Wave BE) BIOPAREN Religious affiliation of parents 28 85 All persons surveyed Individual No Available BIOSOC Parents took care about efforts at school 29 41 All persons surveyed Individual No Available BIOSOC Respondent’s last school marks 30 37 All persons surveyed Individual No Available BIOSOC Relationship to parents during youth 31 13 All persons surveyed Individual No Available BIOSOC Sport and activities during youth 32, 33, 34, 35 16, 21,22, 25 All persons surveyed Individual No Available PBIOSPE Occupational biography 36 - All persons surveyed Spell Yes ($P, $PKAL) Available BIOSOC Year and place of acquiring a school degree 37, 38, 41 27 All persons surveyed Individual No (although possible using $P) Available $PGEN Level of school degree 39, 40, 42 28 All persons surveyed Individual Yes ($P) Available BIOSOC Number of foreign classmates in last attended school class 43 45 All persons surveyed Individual No Available BIOSOC Target school degree 44, 45 29, 30 All persons surveyed Individual No Available ███ SOEP Survey Papers 418 15 SOEP v32 Files in the SOEP Database Biography Sub-area Number of Question in the ‘Lebenslauf’ Questionnaire (2006) Comparable Questions in the Youth Questionnaire (2006) SOEP Target Population Analysis Unit Update Requirements (Source File for Update) Status: Available / Not Available (up to Wave BE) BIOSOC Target vocational degree 53, 54 48, 49 All persons surveyed Individual No Available BIOJOB First job (age, occupational position, public sector, industry) 55, 56, 57, 58, 59, 60a, 60b - All persons surveyed Individual Yes, if person previously did not work ($P) Available BIOJOB Occupational changes 61 - All persons surveyed Individual Yes Available Available BIOJOB Last job (year, scope, public sector branch, occupational position) 62, 63, 64, 65, 66, 67 - All persons surveyed Individual Yes BIORESID Year since living personally in current apartment; second residence 68, 69 - All persons surveyed Individual No Available BIOBIRTH BIOBRTHM Births 70 - All women surveyed; since 2000 men, too Individual Yes ($P, $PBRUTTO, $KIND) Available BIOMARSY Family status (marriage biography) 71, 72 - All persons surveyed Spell Yes ($P, $PBRUTTO) Available BIOCUPLY Partnership biography partnership modul in CAPI version All persons surveyes SPELL Yes ($P, $BRUTTO) Available MIGSPELL Migration biography migration modul in CAPI version Sample M only SPELL No Available ███ SOEP Survey Papers 418 16 SOEP v32 Files in the SOEP Database Biography Sub-area Number of Question in the ‘Lebenslauf’ Questionnaire (2006) Comparable Questions in the Youth Questionnaire (2006) SOEP Target Population Analysis Unit Update Requirements (Source File for Update) Status: Available / Not Available (up to Wave BE) BIOAGE17 Youth Youth Questionnaire 16 and 17 year old respondents Individual No Available BIOAGEL Newborns Mother & Child Questionnaire Mothers of newborns Individual No Available BIOAGEL Infants Questionnaire on children aged 2 to 3 years Mothers Individual No Available BIOAGEL Preschooler Questionnaire on children aged 5 to 6 years Mothers Individual No Available BIOAGEL Elementary school Questionnaire on children between the age of seven and eight Parents Individual No Available BIOAGEL Elementary school Questionnaire on children aged 9 to 10 years Mothers Individual No Available BIOEDU Educational history All persons surveyed Individual Yes Available Lifespell Preand Post-Survey history Dropout studies All persons Spell $BRUTTO Available ███ SOEP Survey Papers 418 17 SOEP v32 SOEP Survey Papers 418 18 SOEP v32 2 Biographical Information in the Meta File PPFAD (Month of Birth, Year of Death, Immigration Variables, Living in East or West Germany in 1989)1 by Elisabeth Liebau, Christian Schmitt, and Diana Schacht The file PPFAD includes, among other more survey related variables like responding status, some most important demographical information for each person who has ever participated in SOEP in at least one wave. These are, on the one hand, longitudinally checked data on sex (variable SEX) and the date of birth (year of birth in variable GEBJAHR in 4-digits and month of birth in variable GEBMONAT), and, on the other hand, generated demographic variables on the year of death (TODJAHR and TODINFO), on the country of origin (GERMBORN and CORIGIN), on the year of the first immigration to Germany (IMMIYEAR), on the migration background (MIGBACK and MIGINFO), as well as on the geographic area a person lived in prior to German unification (LOC1989). In the following section, the construction of these generated variables will be explained briefly. 2.1 The Month of Birth in the data set PPFAD Introduction of variables From wave T onwards (2003) the data set PPFAD contains – in addition to the year of birth – the month of birth (GEBMONAT). This new variable is accompanied by the supporting variable GEBMOVAL which indicates the data source for the month of birth. GEBMONAT and GEBMOVAL can take the following characteristics: • GEBMONAT: Month of birth; 1 (January) to 12 (December) • GEBMOVAL: Month of birth— data-source 1 Generated 2 Info as stored in PPFAD 3 Info derived from data set $KIND 4 Info derived from data set SP (own response) 5 derived from data set $LELA (own response) 6 derived from BIOAGE01 (mother-child-questionnaire) (NEW with Wave W / Survey year 2006) 1 Based on earlier work of Joachim R. Frick, Olaf Groh-Samberg, and Florian Henkel. SOEP Survey Papers 418 19 SOEP v32 7 derived from Youth Questionnaire (own response) (NEW with Wave Z / Survey year 2009) The month of birth was asked in wave S individual questionnaire (SP). Furthermore, the month of birth was asked in the biography data set, starting with wave T ($LELA, file not available with the SOEP data distribution). Additionally the month of birth is recorded for all children within the file $KIND (starting with wave T). With wave W, an further source of information was introduced with a number of biographical questionnaires, including the mother-child questionnaire (filled in by mothers of newborns), and a number of additional biographical questionnaires, where parents report on their children around ages three, six and eight (BIOAGE01, BIOAGE03, BIOAGE06, BIOAGE08a, BIOAGE08b). All these biographical questionnaires record the year and month of the child. Information from the Youth Questionnaire (self-response, age 17) is also considered. All these sources of information are used to derive the month of birth, where valid information is used to replace missings in one or more of the mentioned sources. This procedure provides the relevant information for most of the current panel members. The information remains missing for persons who lack any of the above information, including temporary dropouts or people who exited in a previous wave, and before providing data in any of the sources mentioned above. For some of those persons, the month of birth could be reconstructed (this refers primarily to newborns for whom the month of moving into the household is considered as a proxy in case no other reliable information is available). This reconstruction remains an approximation and might differ from the true month of birth in individual cases. The variable GEBMOVAL displays an ordinal scaling of the level of reliability, where individual response on one’s own date of birth is given preference over derived information, and parent response is considered more reliable at younger ages of the child. 2.2 Construction of variables The month of birth is constructed in an hierarchical order from the files: • Generated (basis: $P, $PBRUTTO $KIND) • $KIND • SP • $LELA • BIOAGE01, BIOAGE03, BIOAGE06, BIOAGE08a, BIOAGE08b • Youth Questionnaire ($PAGE17) whereas the latter information overrides the former. ███ SOEP Survey Papers 418 20 SOEP v32 This means the generated information will only be utilized if no further, questionnaire based information for the month of birth is available. The generated month of birth could only be constructed for people who were born while their parents were members of the SOEP. The information was derived from two sources: • For newborn children the month of moving into the household was used as an approximation of the real month of birth (relevant file $PBRUTTO). • For parents who reported a birth in a certain month, a link to the child was established, assigning the month of birth to the child (relevant file $P). Several adjustments and tests of the generated data have been done which showed that – in the cases in which the generated data was also collected by SP, $LELA, $KIND, and $BIOAGE[n] – the data generation is almost always congruent with the collected data and therefore has proven to be reliable. Frequencies: Month of Birth and Month of Birth: Data Source (File: PPFAD / up to Wave BF) Table 1: GEBMONAT Month of Birth Frequency Percent Cumulative Percent Valid -5 Not Present in Version of Questionnaire 17,243 15.15 15.15 -3 Answer improbable 1 0 15.15 -1 No Answer 15,719 13.81 28.96 1 January 7,268 6.,38 35.34 2 February 6,648 5.84 41.18 3 March 7,363 6.47 47.65 4 April 6,624 5.82 53.47 5 May 6,941 6.1 59.56 6 June 6,553 5.76 65.32 7 July 6,971 6.12 71.44 8 August 6,772 5.95 77.39 9 September 6,857 6.02 83.42 10 October 6,563 5.77 89.18 11 November 6,045 5.31 94.49 12 December 6,272 5.51 100.00 Total 113,840 100.00 Source: SOEP v32, doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 21 SOEP v32 Table 2: GEBMOVAL Month Of Birth, Data Source Frequency Percent Cumulative Percent Valid -5 Not Present in Version of Questionnaire 17,243 15.15 15.15 -3 Not Valid 1 0 15.15 -1 No Answer 15,719 13.81 28.96 1 Generated from gebmonth (parents) 1,524 1.34 30.29 3 $KIND, Info from mother 11,437 10.05 40.34 4 Info From SP 26,769 23.51 63.86 5 Info From $LELA 28,471 25.01 88.87 6 Info From bioage[n] 7,294 6.41 95.27 7 Info from $PAGE17 5,382 4.73 100.00 Total 113,840 100.00 Source: SOEP v32, doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 22 SOEP v32 2.3 Year of birth (not generated) ███ SOEP Survey Papers 418 23 SOEP v32 Variable CORIGIN “Country of origin” Value Value Label Frequency Percent Cumulative Percent -2 does not apply 6,803 6.0 6.0 -1 no answer / don’t know 2,494 2.2 8.2 1 Germany 87,446 76.8 85.0 2 Turkey 2,625 2.3 87.3 … 183 Niger 1 0.0 99.8 222 Unspecified Eastern European country 155 0.1 100.0 333 Other unspecified foreign country 50 0.0 100.0 Total 113,840 100.0 Source: SOEP v32, doi: 10.5684/soep.v32. CORIGIN contains information on the country of birth for all persons who have ever been a part of a SOEP household (i.e., the population from PPFAD). Respondents who were born in Germany were assigned the code “1” (see GERMBORN). Persons who were not born in Germany were assigned another country of birth than Germany depending on the information given in the wave-specific individual questionnaires ($P or $PAUSL) or the variations of the “Biography/Life history” questionnaires (integrated biographical data files for Waves A to L in BIOLELA or life course information on first-time respondents since Wave M in $LELA), and from the additional questionnaire for 16-17-year-olds in use since 2000 ($JUGEND). For those respondents who were not born in Germany and whose country of birth could not be determined, additional indicators were used to minimize the number of missing values. These indicators were used in the following order: - Respondents’ country of origin was considered to be the same country as that of their first, second, or previous citizenship if this was not German (NATION$$ from PGEN, $P or $JUGEND, or $PNAT from $BRUTTO). - Since 1996, first-time respondents’ have been asked to state whether they are a member of a broader group of immigrants such as Ethnic Germans from Eastern Europe or whether they are EEA or EU citizens. Respondents citing Eastern Europe as their country of origin were coded “222 - Unspecified Eastern European country,” whereas respondents who stated that they were from an EU country or a precursor country were coded “444 – Unspecified country within EU”. - Respondents who stated being from another region than Germany (including East and West Germany before 1989) are coded “333” (see also variables GP10803 to JP108B03). - Mothers’ country of birth was considered to be the respondents’ most probable place of birth if the respondent was born before the mother immigrated to Germany. ███ SOEP Survey Papers 418 30 SOEP v32 - When sample D was launched, the country of birth was obtained from the address protocol (variable LPHERKFT). This information was used to fill in further missing values in CORIGIN. If the country of birth was still missing after this procedure, CORIGIN was coded “-1”. CORIGIN includes a few more missing values than GERMBORN due to cases in which it was not possible to determine a country of birth other than Germany. Permanent nonrespondents were assigned the missing value “-2”. A high share of these are sample G cases that were not asked to state their country of birth when the sample was launched (around 900 cases). For persons who were born in another country than Germany, IMMIYEAR designates the year of immigration to Germany. Variable IMMIYEAR “Year of immigration to Germany “ Value Value Label Frequency Percent Cumulative Percent -2 does not apply 94,249 82.8 82.8 -1 no answer / don’t know 3,364 3.0 85.8 1950 19 0.0 85.8 … 2015 24 0.0 100.0 Total 113,840 100.0 Source: SOEP v32, doi: 10.5684/soep.v32. IMMIYEAR contains information on the year of immigration to Germany for all persons who have ever been a part of a SOEP household (i.e., the population from PPFAD) and who were not born in Germany (see GERMBORN). The information on this variable was collected from the wave-specific individual questionnaires ($P or $PAUSL) or the variations of the “Biography/Life history” questionnaires (integrated biographical data files for Waves A to L in BIOLELA or life course information on first-time respondents since Wave M in $LELA), and from the additional questionnaire for 16-17-year-olds in use since 2000 ($JUGEND). Since sample M, information on all of a respondent’s stays in Germany is collected (up to 15 moves between countries, see MIGSPELL). For all cases in which a respondent had more than one stay in Germany, IMMIYEAR contains the respondent’s last year of immigration to Germany. For those respondents who were not born in Germany and whose year of immigration could not be determined, additional indicators were used to minimize the portion of missing values. These indicators were used in the following order: ███ SOEP Survey Papers 418 31 SOEP v32 32 - When a respondent entered the SOEP for the first time because he/she had just moved into the household from abroad (see $PZUG from $PBRUTTO), the household entry year was considered to be the same as the immigration year. - Mother’s year of immigration was used as a proxy for the respondent when the respondent was born before the mother immigrated to Germany. If the year of immigration was still missing after this procedure, IMMIYEAR was coded “-1”. Persons born in Germany were coded “-2” (see GERMBORN) as were permanent nonrespondents. A high share of the latter are sample G cases that were not asked to state their country of birth when the sample was launched (around 900 cases). Variables MIGBACK and MIGINFO The SOEP data comprises a sizeable number of immigrants to Germany and their descendants. The variable MIGBACK is useful in identifying the latter. It combines information on respondents’ country of birth (see GERMBORN) and parental information such as their place of birth and their citizenship. The information for this variable comes predominantly from PPFAD (GERMBORN), PGEN (NATION$$), and the relevant biographical data sets (BIOPAREN, BIOIMMIG). The variables were also updated using information from the wave-specific individual questionnaires ($P or $PAUSL), the variations of the “Biography/Life history” questionnaires (integrated biographical data files for Waves A to L in BIOLELA or life course information on first-time respondents since Wave M in $LELA), and the additional questionnaire for 16-17-year-olds in use since 2000 ($JUGEND). Table 5 lists information used for generating MIGBACK and MIGINFO. Table 5: Information used for MIGBACK and MIGINFO File Information/variable used PPFAD Born in Germany (GERMBORN) PGEN Citizenship (NATION$$) $P $P $P Acquisition of German citizenship (at birth/after) Dual citizenship Both parents born in Germany (Yes/No) $JUGEND $JUGEND $JUGEND $JUGEND $JUGEND Acquisition of German citizenship (at birth/after) Mother: Born in Germany Mother: Country of birth Father: Born in Germany Father: Country of birth BIOIMMIG Status group of migrant (BIIMGRP) BIOPAREN Mother: Country of origin (MORIGIN) BIOPAREN Father: Country of origin (VORIGIN) BIOPAREN Mother: Citizenship (MNAT) BIOPAREN Father: Citizenship (VNAT) $LELA $LELA $LELA Mother: Born in Germany Father: Born in Germany Mother: Country of birth ███ SOEP Survey Papers 418 32 SOEP v32 33 File Information/variable used $LELA $PBRUTTO AKIND/EKIND Father: Country of birth PNAT AK07A and EK03A Source: SOEP v32, doi: 10.5684/soep.v32. In the following sections, the variables MIGBACK and MIGINFO are described in detail. Special attention is given to the filtering function of GERMBORN for MIGBACK. Variable MIGBACK “Migration Background” Value Value Label Frequency Percent Cumulative Percent -1 no answer / don’t know 6,781 6.0 6.0 1 No migration background 72,619 63.8 69.8 2 Direct migration background 17,201 15.1 84.9 3 Indirect migration background 14,827 13.0 97.9 4 Migration background, not differentiated 2,412 2.1 100.0 Total 113,840 100.0 Source: SOEP v32, doi: 10.5684/soep.v32. Respondents were assigned to the MIGBACK categories based on country of birth (see GERMBORN): Being born in another country than Germany indicates, by definition, a direct migration background (2), while respondents born in Germany may have either no (1) or an indirect (3) migration background. Respondents whose parents had no migration background were assigned the code “1no migration background”, while respondents whose father or mother had a migration background were assigned the code “3 - indirect migration background”. However, parental information is not available for all respondents. Whenever this information was missing (see MIGINFO) but the respondent was born in Germany and when further indicators also suggested that there was no migration background (e.g. NATION$$), the respondent was considered having “1 - no migration background” (see MIGINFO). Since some of these respondents may be the descendants of immigrants, MIGBACK may slightly underestimate the number of persons having a “3 - indirect migration background”. Whenever information on a respondent’s country of birth was missing (see GERMBORN), the respondent’s first, second, and previous citizenships were taken into account. Having had non-German citizenship at some point in time was considered an indicator of a migration background. In a similar manner, BIIMGRP from BIOIMMIG was taken into account. If the migration background could not be differentiated further, however, respondents were assigned the code “4 – Migration background, not differentiable”. Given that citizenship is also ███ SOEP Survey Papers 418 33 SOEP v32 34 available for household members who never participated in an interview (e.g., $PNAT, AK07A, EK03A), MIGBACK contains fewer missing entries than, for example, GERMBORN. If the migration background is still missing after this procedure, MIGBACK is coded “-1”. Note that any updates in related variables may also lead to an update of the MIGBACK variable. For instance, a respondent who never stated his or her citizenship but later states having German citizenship will be classified as having a migration background of some form. This retrospective perspective may lead to updates of the migration background variable with every new wave. To provide the highest level of transparency possible, we include a variable for the sources used to create the migration background variable: MIGINFO. Variable MIGINFO “Information source of MIGBACK” Value Value Label Frequency Percent Cumulative Percent -1 no answer / don’t know 6,781 6.0 6.0 1 Direct information without parental information 27,728 24.4 30.3 2 Proxy information without parental information 1,294 1.1 31.5 3 Direct information with parental information 51,994 45.7 77.1 4 Proxy information with parental information 26,043 22.9 100.0 Total 113,840 100.0 SOEP v32, doi: 10.5684/soep.v32. MIGINFO can indicate the quality of information given in MIGBACK. MIGINFO provides information about the usage of proxy information in the generation process of MIGBACK due to missing values in respondents’ and their parents’ migration histories in the SOEP. Overall, MIGINFO can take on four different codes: either direct or proxy information is available on respondents, and either parental information is available or not. With proxy information, we are referring to information on the respondent reported either by the parents (AK07A or EK03A) or by the interviewer ($PNAT from $PBRUTTO or the respective Infratest information), whereas we consider information on the country of birth (GERMBORN) or information on a respondents’ citizenship (NATION$$ from PGEN, $P or $JUGEND) to be direct respondent information. Please note that information on GERMBORN has partially been derived from other indicators (see GERMBORN). Whenever this is the case, MIGINFO reports proxy and not direct information. ███ SOEP Survey Papers 418 34 SOEP v32 The parental information refers to any information on the migration background of the respondents’ mother or father or both. This includes information on the country of birth (see GERMBORN), country of citizenship (NATION$$ from PGEN, $P or $JUGEND), or proxy information on the migration history ($PNAT from $PBRUTTO or the respective Infratest information). MIGBACK information is considered to be particularly reliable for cases coded “3” on MIGINFO, in contrast to the other cases of missing parental information (1 and 2 in MIGINFO). The cases coded “4” in MIGINFO refer primarily to children in SOEP households whose information was reported by their parents. 2.6 Living in East or West Germany in 1989 The variable LOC1989 in the meta-file PPFAD provides information about the geographic area a person lived in prior to the German reunification, differentiating “East Germany (DDR incl. East Berlin”, “West Germany (Bundesrepublik Deutschland incl. West Berlin)”, and “abroad (Ausland)”. This information has been generated for all individuals in SOEP. Variable LOC1989 “Where did you live in 1989?” Codes 1 East Germany (German Democratic Republic [DDR] including East Berlin) 2 West Germany (Federal Republic of Germany [BRD] including West Berlin) 3 Abroad (Ausland) Missing Codes -2 does not apply; born after 1989 -1 not available After asking this information from all respondents in 2003 (variable TP121 in file TP), a corresponding question has been included in the biography questionnaire since wave U (2004) [Question 16 / variable UB16 in file ULELA, UJ58 in file UJUGEND] which will collect this time-independent information from all future first time respondents. For all respondents interviewed up until 2003, the following information was used as input to generate LOC1989: • Information on place and date of last school attendance [variables BSSCHEND and BSSCHWO in file BIOSOC / variables $B38 and $B3701 in file $LELA with $ starting in wave U, 2004], ███ SOEP Survey Papers 418 35 SOEP v32 • Sample affiliation [variable PSAMPLE in file PPFAD], • year moved in at current address [variable BRMOVEIN in file BIORESID / variable $B68 in file $LELA with $ starting in wave U, 2004], • sample region [variables $SAMPREG in file PPFAD], • year of first immigration to Germany [variable IMMIYEAR in file PPFAD] • In case of inconsistent information from these various sources, the data collected in 2003 via variable TP121 and the information from the biography questionnaire collected since 2004 is considered superior. Persons without any individual information and aged less than 18 years in 1989 were assigned parental information, if available. • The variable LOC1989 is completed for SOEP samples A through M2. Variable LOC1989 Where did you live in 1989? Freq. Percent Cum. [-2] trifft nicht zu 30,756 27.02 27.02 [-1] keine Angabe 12,471 10.95 37.97 [1] East Germany (DDR) incl. East Berlin 14,422 12.67 50.64 [2] West Germany (FRG) incl. West Berlin 47,839 42.02 92.66 [3] Abroad (Ausland) 8,352 7.34 100.00 Total 113,840 100.00 Source: POPULATION of PPFAD SOEP v32, doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 36 SOEP v32 3 MIGSPELL: The Migration-Biography (Samples M1 and M2) Integrated Version: Waves bd to bf by Klaudia Erhardt1 3.1 Introduction to the new release of MIGSPELL The previous release, v31, of MIGSPELL included the waves bd and be of the migrant survey within the SOEP. At that time, although the questions were changed in a significant way between 2013 and 2014, the MIGSPELL-generation changes were minor, because it was known that other changes and the inclusion of the new sample M2 would be incorporated in the next wave. Therefore, we postponed the decision on the ultimate structure of MIGSPELL until wave bf was ready to be included. With this release, v32 of MIGSPELL, the integration of waves bd to bf has been carried out. The integration entailed major modifications in the number and coding of the MIGSPELL variables, due to different operationalizations of the status at entry. But these amendments of the question design apply only to the migration biographies of respondents who were born abroad and their migrations to Germany, but not their recorded moves back to their birth country or to other countries. Questions related to German-born respondents as well as to stays abroad of foreign-born respondents remained unchanged over the waves. In addition, in the new release, an improved imputation procedure for the replacement of missing dates has been implemented, meaning that fewer spells will drop from analyses due to a missing start date. We also eliminated some bugs that had been noticed since the last release. So the end of the last spell of a migration biography is the interview month, as was always intended, but in the last release was a constant that served as a place holder in programming that had been forgotten to be replaced. In the following sections, the variables of MIGSPELL and their generation are described in detail. 1 Contact: [email protected] SOEP Survey Papers 418 37 SOEP v32 3.2 Summary description of MIGSPELL MIGSPELL is derived from the migration biographies, which are collected from each new respondent of the IAB-SOEP migration samples M1 and M2.2 The moves of the respondents were captured through a loop structure of the questionnaire (see Figure 1 and Figure 2 on pp. 3 - 4), with the number of loops limited to 15. The $$p_mig files of the SOEP-distribution hold these data in "wide" format: A proper set of variables for each potential loop has been laid out. For MIGSPELL, the original data has been transformed into spell format. Each migration that actually took place is represented by a spell. Additionally, a spell has been generated for the period from birth to first move (if there has been any), or from birth to interview date (if the respondent has not moved). This has the advantage that every respondent from the IAB-SOEP migration samples is represented in the MIGSPELL dataset, not only those who had moved to another country. Because two moves could be captured within one loop cycle, the maximum number of spells a person can have in MIGSPELL is 31. However, this is extremely rare.3 MIGSPELL contains data on the moves of foreign-born migrants as well as on the stays abroad of German-born respondents. The requirement was to capture only periods of at least three months, which was generally observed, but not always. 3.3 Overview: the structure of the migration biography questions Figure 1 and Figure 2 show two flow charts of the migration biography questions of the IABSOEP migration sample. Because compiling such flow charts is very labor-consuming, we provide them only for the first wave of sample M1. As mentioned, the questions to survey the stays abroad of foreign-born and German-born respondents remained unchanged for samples M1 and M2 over all waves. Concerning the moves to Germany of foreign-born respondents, the changes affected mainly the response alternatives (and subsequently, minor filter paths), but not the whole structure as such. For this, the flow charts are still helpful for all the waves to date. As you see from the "Exit"-lists in the charts, few respondents undergo more than two loop cycles, i.e. undertook more than 4 moves between countries. 2 See http://www.diw.de/en/diw_01.c.485464.en/iab_soep.html for a description and additional documenation of the IAB-SOEP migration sample M1 for waves bd and be (survey years 2013 and 2014). For sample M2 and the survey year 2015, documentation will follow. 3 For details on the transformation from wide data to spell data, see: Klaudia Erhardt. 2014. How to Generate Spell Data from Data in “Wide” Format. Based on the migration biographies of the IAB‐SOEP Migration Sample. SOEP Survey Papers 228: Series G. Berlin: DIW/SOEP SOEP Survey Papers 418 38 SOEP v32 Figure 1: Flow chart of the "Coming to Germany"-part of the migration biography SOEP Survey Papers 418 39 SOEP v32 ostatus Status at entry ("other" open answers) Generated from open answers to status-variables. Replaces variables ostatus, ostatusde and ostatusoc of former releases of migspell. Open answers are captured for the status variables related to stays abroad. For stays in Germany, open answers have been captured only in the first wave (bd). In former releases of migspell, coded open answers have been assigned directly to the variable status (and statusde, statusoc), if they fitted a level. In contrast, in this release of migspell, ostatus is designed to mirror if a level fits to a level of status1 or status2, but no direct assignment to status1 or status2 has been made. The first digit of the two-digit codes in ostatus indicates if the level fits variables status1 or status2, the second digit indicates which level of status1 or status2 a certain level of ostatus fits. Example: open answers with the meaning "Spouse, child, or family member" were coded 23 in ostatus, because the standardized answers with the same meaning are assigned to code 3 of status2. One-digit codes do not correspond to a level of status1 or status2 1 Visit of family or friends 2 Love attachment 3 Au pair, gap year spent on voluntary social work 4 Intern, trainee 5 Military service, soldier 11 German migrant from Eastern Europe 21 Labor force (not bd) 22 Labor force with job agreement at entry 23 Spouse, child, family member 24 Asylum seeker, refugee 25 Student, apprentice 26 Seeking for job 27 Tourist 28 With tourist visa 29 None of these / other start Start date of a spell in months from jan 1900 Generated variable from starty_imp and startmo_imp startmo Start month of an episode (original values) Source variables see section 1.6.1, Table 2 to Table 7 startmo_imp Start month of an episode (imputed values) Missing values in startmo are replaced with imputed values, where possible (see section 1.5) ███ SOEP Survey Papers 418 46 SOEP v32 starty Start year of an episode. (original values) Source variables see section 1.6.1, Table 2 to Table 7 starty_imp Start year of an episode (imputed values) Missing values in starty are replaced with imputed values, where possible (see section 1.5) status1 Legal background of entry Generated Variable. Mapping of source variables to levels of status1 see section 1.6.2, Table 8. status1 and status2 replace variables status, statusde, statusoc of former releases of migspell 1 German migrant from Eastern Europe 2 German citizen, grown-up outside Germany (since be) 3 EU-citizen (only be) 4 EUor EEZ-citizen with right to free movement (since bf) 5 EUor EEZ-citizen without right to free movement (since bf) 6 Other citizens status2 Status at entry Generated Variable. Mapping of source variables to levels of status2 see section 1.6.2, Table 8 status1 and status2 replace variables status, statusde, statusoc of former releases of migspell 1 Labor force (not bd) 2 Labor force with job agreement at entry 3 Spouse, child, family member 4 Asylum seeker, refugee 5 Student, trainee 6 Seeking for job 7 Tourist 8 With tourist visa 9 None of these / other staytime Duration of stay (months) Generated from imputed start and end variables: = end - start + 1 stype Spell type (marker for type of stay) Generated from variable move 1 Stay in Germany ███ SOEP Survey Papers 418 47 SOEP v32 48 2 Stay abroad tcountry Target country of the next move Coding according to SOEP-convention Source variables see section 1.6.1, Table 2 to Table 7 3.4.4 The date-variables in MIGSPELL With spell data, the information on start date and end date of each episode is crucial. The migration biographies constitute a special kind of spell data, without parallelities and gaps: Because the questionnaire did not allow for reporting more than one place of habitation at the same time, the spells of a person can not be parallel to each other. Further - as one necessarily has to stay at one place or another - the migration biographies have no true gaps, the stays are successive to each other. The participants were only asked at what time they moved to a certain country, but not at what time they left. Because of the successive character of the data at hand, the end of a spell could be derived from the start of the next spell. However, sometimes respondents were not able to recollect the exact dates of their moves, or they named dates that contradicted the time sequence of moves, which led to missing values in the start date variables. The respondents were not bullied into stating a date and time contradictions were not immediately clarified during the interview, so these kinds of missings were to be expected. Overall, 369 spells (=2.4%) have a missing value in the original startyear and/or the startmonth. In order to make those spells accessible for data analyses, the missing values were replaced by imputed values to the extent possible, resulting in only 9 spells with missing values in the imputed version of the startyear and/or startmonth. From the imputed versions of startyear and startmonth the start and end variables have been generated. They contain integers counting the month that have passed since January 1, 1900, (i.e. including January). This is in contrast to the SOEP standard, where the counting of time begins January 1, 1983, and also to the Stata standard, where the counting begins January 1, 1960. We had to advance the zero-point of the time scale, because otherwise the SOEP-convention on missing codes (which are integers < 0 and > -10) would have conflicted: The migration biographies may begin earlier than January 1983 or 1960, which would result in negative values if we had used the SOEP or Stata standard. In certain cases, missing and valid codes would then have become indistinguishable. ███ SOEP Survey Papers 418 48 SOEP v32 49 If you are using Stata for data analyses and want to benefit from the inbuilt Stata time and date functions, you can transform the start and end variables into the Stata standard by subtracting 271. But before doing so, you have to replace the SOEP missing values by values that are distinguishable from valid values, such as Stata missing codes: gen start_stata = start gen end_stata = end recode start_stata end_stata (-1 = .a) (-3 = .b) replace start_stata = start_stata - 721 replace end_stata = end_stata - 721 format start_stata end_stata %tmCCYY_mon 3.5 The imputation of missing date values 3.5.1 General remarks As already mentioned, the start and end variables have been generated from the imputed startyear and startmonth information. The original variables starty and startmo have not been touched, so that they can be used for analyses instead of the imputed versions. All replacements of values have been made in the variables starty_imp and startmo_imp. In this section we explain how the imputation was performed. Before presenting the applied rules and algorithms for the replacement of missing dates, the treatment of the special case "birth-date" is described. The birth-date is represented in starty and startmo of the first spell of a respondent, which relates to the episode from birth to the first move to another country. If there had been missing values in the birth-date (which was not the case), they would not have been subject to imputation. In the first spell, there is no floor for possible values, and therefore a span for the estimation is not to be determined. However, the birth-date was compared with the birth-date from ppfad. The birth-date is surveyed in each wave anew, and ppfad holds the newest answer of the respondent. If there was a difference between the birth-date from starty and startmo with the one from ppfad, starty_imp and starty_mon was updated with the newer information, and the flag variables f_startmo_imp and f_starty_imp, respectively, were set to 1. In one case, the difference was more than 28 years, which was regarded as an improbable value, and the update of starty_imp and startmo_imp was undone. ███ SOEP Survey Papers 418 49 SOEP v32 3.5.2 Procedure for the imputation of the missing date values In the next section, the imputation of startyear and startmonth for the missing dates is explained. The corresponding Stata syntax is too complex to be part of this documentation. In case of questions please contact the author via [email protected] 3.5.2.1 The imputation of missing values in the startyear In a first step, the consistency of the startyears' sequence is tested and set to -3 if it is inconsistent, i.e. the startyear is earlier than the startyear of the previous spell. After that, all spells with a missing value (-1 or -3) are flagged. Then a marker is generated for spells with directly succeeding missing startyears. For series of more than one spell with a missing startyear, rule 1 applies: Rule 1: If there are more than one missing startyear in directly succeeding spells, they will not be imputed. Instead, the startmonth is also set to missing, namely, set to the value of the startyear (which is either -1 or -3). Single spells with a missing value in startyear are treated as follows: Rule 2: The missing startyear is replaced with the startyear of the next spell minus 1. If the next spell is the last spell of a case, the missing startyear is replaced by the interview year minus 1. If the such imputed startyear is smaller than the startyear of the preceding spell, or if it is equal to the startyear of the preceding spell but the startmonth is smaller than the startmonth of the preceding spell (that is to say, if the imputed start time is earlier than the start time of the preceding spell), then the subtraction of 1 from the startyear of the next spell is undone, the imputed startyear equals the startyear of the next spell. With this procedure, in a rare constellation, the spell with a missing startyear conflicts with the time sequence of either the preceding or the following spell. For example: The spell with a missing startyear has a startmonth "May", the preceding spell starts June, 2011, the following spell starts February, 2012. To handle this kind of constellation, a second test of the sequence consistency is performed, resulting in a reset of the originally missing, now imputed startyear to missing. In practice, mostly the startmonth is also missing if the startyear is missing, so that the chance to meet this constellation is very small. Following rule 2, a spell with an imputed startyear lasts one year and 11 months maximally. ███ SOEP Survey Papers 418 50 SOEP v32 3.5.2.2 The imputation of missing values in the startmonth The imputation of missing startmonths is only done if the imputed startyear is not missing. Otherwise, there is no sense in keeping or imputing the startmonth information, and startmo_imp is set to the same missing code than starty_imp (either -1 or -3). So the frame for the imputation of missing startmonths is always the startyear of the spell in question. For the imputation of missing startmonths, rule 1 does not apply. More than one succeding missing startmonth may be imputed if the startyear is known, such that a lower and upper limit for the missing data can be established. Multiple succeeding spells with missing startmonth information are referred to as a "series" of spells with a missing startmonth. The principle of the imputation of a series of spells with missing startmonth is to portion the time between the last and the next established startmonth of the same year onto the spells with a missing startmonth that lay in between. It has to be distinguished between cases where a) all spells in a year have missing startmonths and b) at least one spell in a year has a valid startmonth. Considerations for case a) all spells of a year have missing startmonths: Rule 3: 12 months are divided between the number of succeeding spells with a missing startmonth in the same year. The algorithms are: a1) startmo_imp = 1 + round(c * span) a2) span = 12 / ctot + 1 with: c = running number of the spell with a missing startmonth (the succeeding spells with missing starmonth within a year numbered from 1 to n) ctot = total of succeeding spells with missing startmonth within a year EXAMPLE: startmonth before imputation: -3, -1, -1 span = 12 / 3+1 = 3 startmonth after imputation: 4, 7, 10 (1 + 1 * 3 / 1 + 2*3 / 1 + 3*3) NOTE: the algorithm means that only 11 spells with a missing startmonth can be placed within one year, although there is enough "place" for 12 spells, if each lasts 1 month. The reason for that is the "+ 1" in formula a1). Only by this, a single spell in a year with missing startmonth is assigned the desired result 7 instead of 6. ███ SOEP Survey Papers 418 51 SOEP v32 As the provisions for surveying the migration biographies say to capture only stays that lasted at least 3 months (which was not always met, however), the constellation of 12 spells with a missing startmonth in a year does not occur empirically. Considerations for case b) there is at least one spell with a valid startmonth in the same year: Rule 4 The adjacent spells with a valid startmonth in the same year form the lower and upper limits of the time span that can be divided between the number of succeeding spells with a missing startmonth in the same year. If a series of spells with a missing startmonth begin in the first or end in the last month in a year, the lower limit is 1, the upper limit is 12, respectively. Otherwise, the lower limit is the startmonth of the last spell before the series with missing startmonth, the upper limit is the startmonth of the next spell after the series with missing startmonth. The algorithms are: b1) startmo_imp = lol + round(c * span) b2) span = upl - lol + 1 / ctot + 1 with: lol = lower limit = startmonth of the last spell with valid startmonth before the series of spells with missing startmonth, or 1 respectively upl = upper limit startmonth of the next spell with valid startmonth before the series of spells with missing startmonth, or 12 respectively c and ctot as above. We now understand, that formulas a1) and a2) are only special cases of formulas b1) and b2). NOTE: The minimum permissible value for span is 1. In other words, between lower and upper limit there must be a gap of at least as many free time units (months) as there are spells with a missing startmonth in the series. EXAMPLE: startmonth before imputation: 2, -1, 3, 10 span = 10-2 / 2+1 = 8/3 = 2,66 imputed startmonths: 2 + round(1 * 2,66) = 5 2 + round(2 * 2,66) = 7 startmonth after imputation: 2, 5, 7, 10 ███ SOEP Survey Papers 418 52 SOEP v32 After the imputation of missing values in starty and startmo, the variables starty_imp and startmo_imp are either both missing or both non-missing, because we decided a) to impute all missing startmonths if starty was valid and b) not to keep a nonmissing startmonth if starty was missing. From the imputed variables starty_imp and startmo_imp the variable start has been generated by: gen start = cond(starty_imp > 0 & startmo_imp > 0, /// ((starty_imp-1900)* 12) + startmo_imp, starty_imp) meaning: if starty_imp and startmo_imp are not missing, start is calculated as: starty_imp - 1900 * 12) + startmo_imp otherwise it is equal to starty_imp (which is a missing code). The end of a spell is generated by: by persnr: gen end = cond(start[_n+1] > 0, start[_n+1]-1, start[_n+1]) meaning: end is calculated as the start of the next spell minus 1, provided that the start of the next spell is not missing and that the next spell belongs to the same case. 3.6 The integration of the migration biographies of the three waves 2013-2015 (bd, be, bf) The questionnaire for the survey of the migration biographies was different from wave to wave. The version for wave 2015 is expected to be a final version, but because legal regulations are involved, which may change in future, this is not set in stone. Regardless, the questionnaire for the new respondents of the M1 and M2 IAB-Soep-Migration samples has remained the same for the forthcoming wave bg, whereas the questionnaire for the M3 IABSoep-Migration sample (i.e. refugees) is very different, and it is neither sensible nor feasable to include the migration biographies of the M3 sample into MIGSPELL. An integration of different structures necessarily has to compromise. Consequently, some core variables of MIGSPELL - while mirroring the variables of the bdp_mig-file perfectly in the first wave - now have a more complex relationship to the $$p_mig files. For analyses and their correct interpretation you should be aware of the information that goes - and does not go - into the different categories of the MIGSPELL variables. E.g., some categories systematically receive data only from a single wave, or from a certain group in wave bd, but from another group in wave bf. ███ SOEP Survey Papers 418 53 SOEP v32 To allow for keeping track of the relationship between the variables of the $$p_mig files and the MIGSPELL file, we provide the single structure tables for German and abroad-born respondents of each wave (Table 2 to Table 7) and a synopsis of the levels of each variable of the migration biographies as a source for the core MIGSPELL variables (Table 8) in the following sections. ███ SOEP Survey Papers 418 54 SOEP v32 3.6.1 Structure Tables: $$p_mig-variables to MIGSPELL-variables for waves bd to bf Table 2: bdp integrated Synopsis 1: Stays abroad (for German-born migrants) Var## country starty startm status1 status2 jobpr lfgroup nmtype tcountry coverage 01 bdpm_l_0203 -2 tnz: DE integr. bdpm_l_0103 bdpm_l_0102 --- --- bdpm_l01_28 bdpm_l01_2902 all cases from here: repetition for each value of Var## 01 bdpm_l01_3001 bdpm_l01_3002 --- bdpm_l01_3101 bdpm_l01_3101 --- bdpm_l01_3201 bdpm_l01_3203 stay abroad 01 bdpm_l01_32a01 bdpm_l01_32a02 --- --- --- --- bdpm_l01_3301 bdpm_l01_3303 stay in DE 02 bdpm_l02_3001 bdpm_l02_3002 --- bdpm_l02_3101 bdpm_l02_3101 --- bdpm_l02_3201 bdpm_l02_3203 stay abroad 02 bdpm_l02_32a01 bdpm_l02_32a02 --- --- --- --- bdpm_l02_3301 bdpm_l02_3303 stay in DE etc. 15 bdpm_l15_3001 bdpm_l15_3002 --- bdpm_l15_3101 bdpm_l15_3101 --- bdpm_l15_3201 bdpm_l15_3203 stay abroad 15 bdpm_l15_32a01 bdpm_l15_32a02 --- --- --- --- bdpm_l15_3301 bdpm_l15_3303 stay in DE (up to int.date) Source: SOEP v32, doi: 10.5684/soep.v32 SOEP Survey Papers 418 55 SOEP v32 62 bdp_mig bep_mig bfp_mig migspell signification bepm_l_08 (2) bepm_l##_17 (2) bfpm_l_08 (2) bfpm_l##_19 (2) status1 : 2 German citizen, grownup outside Germany bepm_l_08 (3) bepm_l##_17 (3) status1 : 3 EU-citizen bfpm_l_08 (3) bfpm_l##_19 (3) status1 : 4 EUor EEZ-citizen with right to free movement bfpm_l_08 (4) bfpm_l##_19 (4) status1 : 5 EUor EEZ-citizen without right to free movement bfpm_l_08 (5) bfpm_l##_19 (5) status1 : 6 Other citizens status2 status2 : -6 status2 : -5 status2 : -3 status2 : -2 bepm_l_08 (-1) status2 : -1 bepm_l_0901 (2) bepm_l_10 (1) bepm_l##_1801 (2) bepm_l##_19 (1) bfpm_l_09 (1) bfpm_l_11 (1) bfpm_l##_20 (1) bfpm_l##_22 (1) status2 : 1 Labor force bdpm_l01_1801 (1) bdpm_l##_2501 (1) bdpm_l##_2201 (1) bdpm_l##_3101 (1) bepm_l##_1401 (1) bepm_l##_2501 (1) bfpm_l##_1601 (1) bfpm_l##_3701 (1) status2 : 2 Labor force with job agreement at entry bdpm_l01_1801 (3) bdpm_l##_2501 (3) bdpm_l##_2201 (2) bdpm_l##_3101 (2) bepm_l_08 (6) bepm_l_0901 (4) bepm_l##_17 (6) bepm_l##_1801 (4) bepm_l##_1401 (2) bepm_l##_2501 (2) bfpm_l_09 (4) bfpm_l_11 (4) bfpm_l##_20 (4) bfpm_l##_22 (4) bfpm_l##_1601 (2) bfpm_l##_3701 (2) status2 : 3 Spouse, child, family member bdpm_l01_1801 (4) bdpm_l##_2501 (4) bdpm_l##_2201 (3) bepm_l_08 (4) bepm_l##_17 (4) bepm_l##_1401 (3) bfpm_l_11 (5) bfpm_l##_22 (5) bfpm_l##_1601 (3) status2 : 4 Asylum seeker, refugee bdpm_l01_1801 (5) bdpm_l##_2501 (5) bdpm_l##_2201 (4) bdpm_l##_3101 (3) bepm_l_08 (7) bepm_l_0901 (3) bepm_l##_17 (7) bepm_l##_1801 (3) bepm_l##_1401 (4) bepm_l##_2501 (3) bfpm_l_09 (3) bfpm_l_11 (3) bfpm_l##_20 (3) bfpm_l##_22 (3) bfpm_l##_1601 (4) bfpm_l##_3701 (3)) status2 : 5 Student, trainee ███ SOEP Survey Papers 418 62 SOEP v32 bdp_mig bep_mig bfp_mig migspell signification bdpm_l01_1801 (6) bdpm_l##_2501 (6) bdpm_l##_2201 (5) bdpm_l##_3101 (4) bepm_l_0901 (1) bepm_l01_11 (8) bepm_l##_1801 (1) bepm_l##_20 (8) bepm_l##_1401 (5) bepm_l##_2501 (4) bfpm_l_09 (2) bfpm_l_11 (2) bfpm_l##_20 (2) bfpm_l##_22 (2) bfpm_l##_1601 (5) bfpm_l##_3701 (4) status2 : 6 Seeking for job bfpm_l_09 (5) bfpm_l_11 (6) bfpm_l##_20 (5) bfpm_l##_22 (6) status2 : 7 Tourist bepm_l_08 (8) bepm_l##_17 (8) status2 : 8 with tourist visum bdpm_l01_1801 (7) bdpm_l##_2501 (7) bdpm_l##_2201 (6) bdpm_l##_3101 (5) bepm_l_08 (9) bepm_l_0901 (5) bepm_l##_17 (9) bepm_l##_1801 (5) bepm_l##_1401 (6) bepm_l##_2501 (5) bfpm_l_09 (6) bfpm_l_11 (7) bfpm_l##_20 (6) bfpm_l##_22 (7) bfpm_l##_1601 (6) bfpm_l##_3701 (5) status2 : 9 None of these / other jobpr jobpr : -6 bdpm_l01_1801 (1,2,3,4,5,6,7 od. all levels?) bdpm_l##_2501 (1,2,3,4,5,6,7 od. all levels?) bdpm_l01_2201 (all levels?) bepm_l_0902 (-2,-1,1,2) bepm_l##_1802 (-2,-1,1,2) jobpr : -5 jobpr : -3 bfpm_l_13 (-2) bfpm_l##_24 (-2) jobpr : -2 bfpm_l_13 (-1) bfpm_l##_24 (-1) jobpr : -1 bdpm_l01_1801 (1) bdpm_l##_2501 (1) bdpm_l01_2201 (1) bdpm_l##_3101 (1) bepm_l_0902 (1) bepm_l##_1802 (1) bepm_l##_1401 (1) bepm_l##_2501 (1) bfpm_l##_1601 (1) bfpm_l##_3701 (1) jobpr : 1 Yes (undiff.) bfpm_l_13 (1) bfpm_l##_24 (1) jobpr : 2 Prospective job bfpm_l_13 (2) bfpm_l##_24 (2) jobpr : 3 Employment contract bfpm_l_13 (3) bfpm_l##_24 (3) jobpr : 4 Job as self-employed bepm_l_0902 (2) bepm_l##_1802 (2) bfpm_l_13 (4) bfpm_l##_24 (4) jobpr : 5 No ███ SOEP Survey Papers 418 63 SOEP v32 bdp_mig bep_mig bfp_mig migspell signification bfpm_l_13 (5) bfpm_l##_24 (5) jobpr : 6 Did not look for job bfpm_l_13 (6) bfpm_l##_24 (6) jobpr : 7 Does not apply, was a child lfgroup lfgroup : -6 lfgroup : -5 lfgroup : -3 lfgroup : -2 lfgroup : -1 bdpm_l01_19 (2) bdpm_l##_26 (2) bepm_l01_11 (2) bepm_l##_20 (2) bfpm_l_10 (1) bfpm_l_12 (1) bfpm_l##_21 (1) bfpm_l##_23 (1) lfgroup : 1 Seasonal worker, contract for work and labor bdpm_l01_19 (5) bdpm_l##_26 (5) bepm_l01_11 (5) bepm_l##_20 (5) bfpm_l_12 (2) bfpm_l##_23 (2) lfgroup : 2 Highly qualified and experts with special entry conditions bfpm_l_12 (3) bfpm_l##_23 (3) lfgroup : 3 Qualified labor force with priority check by the Fed. Work Agency bfpm_l_12 (4) bfpm_l##_23 (4) lfgroup : 4 Other labor force with priority check by the Fed. Work Agency bepm_l01_11 (6,7) bepm_l##_20 (6,7) bfpm_l_10 (3) bfpm_l_12 (5) bfpm_l##_21 (3) bfpm_l##_23 (5) lfgroup : 5 Trainee, au pair bdpm_l01_19 (1) bdpm_l##_26 (1) bepm_l01_11 (1) bepm_l##_20 (1) bfpm_l_10 (4) bfpm_l_12 (6) bfpm_l##_21 (4) bfpm_l##_23 (6) lfgroup : 6 Self-employed, entrepreneur bdpm_l01_19 (6) bdpm_l##_26 (6) bepm_l01_11 (9) bepm_l##_20 (9) bfpm_l_10 (5) bfpm_l_12 (7) bfpm_l##_21 (5) bfpm_l##_23 (7) lfgroup : 7 Other bdpm_l01_19 (3) bdpm_l##_26 (3) bepm_l01_11 (3) bepm_l##_20 (3) lfgroup : 8 Relocated to Germany by employer bdpm_l01_19 (4) bdpm_l##_26 (4) bepm_l01_11 (4) bepm_l##_20 (4) lfgroup : 9 Sent to Germany by company Source: SOEP v32, doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 64 SOEP v32 65 4 Activity Biography in the Files PBIOSPE and ARTKALEN by Paul Schmelzer and Maik Hamjediers1 The spell file PBIOSPE is based on the information on activity status over the life course, which is collected as a matrix from every respondent answering the Biography Questionnaire (Question 45 in 2015).2 The observations start at the age of 15 and end at the current age (up to age 65). This information on activity status covers only the period up to the time the biography is collected. To update the ongoing occupational career in PBIOSPE, information from the yearly Individual Questionnaire is also used. In this questionnaire, respondents are always asked their occupational status for every month of the previous year (Question 118 in 2015).3 Therefore, the information on activity status collected on a monthly basis in the yearly personal questionnaire and stored in the file ARTKALEN in spell format is aggregated into yearly values and combined with the information gathered from the Biography Questionnaire.4 In the following, the method of combining the data is described. There have been no changes how the data is generated since the previous version, distributed in 2015.5 But if you have been working with older versions of the dataset (versions distributed in 2008 and earlier) you should check the section at the end of the chapter, where you will find information on previous changes. But before we move on to the details, we provide a brief overview of the contents of PBIOSPE. Table 2 contains a list of all the variables in the dataset. The variables BEGIN and END indicate the beginning and the end of a spell. These variables are age entries. There are also variables that refer to calendar years: BEGINY and ENDY (Y stands for Year). The variable SPELLTYP contains information on the activity status during the spell, e.g., employed full-time or unemployed. The SPELLNR is a serial identifier of spells of each activity status of a given person. Missing information on the beginning or end of a spell causes what is known as censoring problems. There are two types of missing data. First, data can be missing on periods outside the observation window (before the age of 15 and after the age of 65). Second, data can be missing on years within the observation window due to item non-response in particular years or due to temporary drop-outs (the latter applies to calendar 1 Based on earlier work by Rainer Pischner, Henning Lohmann, Marco Giesselmann and Mila Staneva. 2 See Chapter 1 for general information on the collection of biography information. 3 For persons who were temporarily unavailable for interviewing, it is sometimes possible to fill in the gaps in their occupational status. If these persons fill out the additional questionnaire for temporary drop-outs later on, we can use the information collected there (see files $PLUECKE). 4 For more information, see Haisken-DeNew, John and Joachim R. Frick (2005): DTC - Desktop Companion to the German Socio-Economic Panel Study (SOEP), Chapter 3. 5 The only exception is the lack of information on short work hours (spelltyp ‘2’ in ARTKALEN), due to omitting this question in the questionnaire of 2015. ███ SOEP Survey Papers 418 65 SOEP v32 information only). In this case, we speak of “gaps.” There are nine different patterns (see Table 1). Table 1: Coding of the variable ZENSOR Right: Left: not censored censored missing censored before gap not censored 1 2 3 censored missing 4 5 6 censored after gap 7 8 9 Note: ‘(99) Gap’ spells are all markes as (-2) Source: SOEP v32, doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 66 SOEP v32 67 Table 2: Contents of PBIOSPE (variables) As mentioned above, PBIOSPE combines information collected in the biography questionnaire and the calendar matrix of the individual questionnaire. The two types of information are merged into PBIOSPE following a number of rules. First of all, it is important to acknowledge that the Biography Questionnaire Matrix as well as the Individual Questionnaire Matrix allow for multiple activity statuses for a given year or month. No concept of main activity is used. A common combination is, for instance, “housewife/- husband” and “working part-time”. There are a number of other plausible combinations, but also combinations that are less plausible. However, a list of valid combinations of activity Variable Description HHNR Original Household Number PERSNR Never Changing Person ID SPELLNR Serial Number Of The Spell Per Person SPELLTYP Type Of Spell BEGIN Age Spell Begins END Age Spell Ends BEGINY Year Spell Begins ENDY Year Spell Ends ZENSOR Censor Variable SPELLINF Spell Construction Information ERHEBJ Survey Year Biography Data KALYEAR First Observation Year Calendar BEGINB1 Age Spell Begins, 1st Initial Biography Spell ENDB1 Age Spell Ends, 1st Initial Biography Spell BEGINK1 Age Spell Begins, 1st Initial Calendar Spell ENDK1 Age Spell Ends, 1st Initial Calendar Spell BEGINYB1 Year Spell Begins, 1st Initial Biography Spell ENDYB1 Year Spell Ends, 1st Initial Biography Spell BEGINYK1 Year Spell Begins, 1st Initial Calendar Spell ENDYK1 Year Spell Ends, 1st Initial Calendar Spell BEGINB2 Age Spell Begins, 2nd Initial Biography Spell ENDB2 Age Spell Ends, 2nd Initial Biography Spell BEGINK2 Age Spell Begins, 2nd Initial Calendar Spell ENDK2 Age Spell Ends, 2nd Initial Calendar Spell BEGINYB2 Year Spell Begins, 2nd Initial Biography Spell ENDYB2 Year Spell Ends, 2nd Initial Biography Spell BEGINYK2 Year Spell Begins, 2nd Initial Calendar Spell ENDYK2 Year Spell Ends, 2nd Initial Calendar Spell BEGINB3 Age Spell Begins, 3rd Initial Biography Spell ENDB3 Age Spell Ends, 3rd Initial Biography Spell BEGINK3 Age Spell Begins, 3rd Initial Calendar Spell ENDK3 Age Spell Ends, 3rd Initial Calendar Spell BEGINYB3 Year Spell Begins, 3rd Initial Biography Spell ENDYB3 Year Spell Ends, 3rd Initial Biography Spell BEGINYK3 Year Spell Begins, 3rd Initial Calendar Spell ENDYK3 Year Spell Ends, 3rd Initial Calendar Spell BEGINK4 Year Spell Begins, 4th Initial Biography Spell ENDK4 Year Spell Ends, 4th Initial Biography Spell BEGINYK4 Year Spell Begins, 4th Initial Calendar Spell ENDYK4 Year Spell Ends, 4th Initial Calendar Spell ███ SOEP Survey Papers 418 67 SOEP v32 statuses defined according to legal or similar constructs would need to be based on very strong assumptions. In addition—in particular in case of the yearly matrix in the Biography Questionnaire—activities are reported that took place in a calendar year in consecutive months, which makes it impossible to exclude combinations of activities. Therefore, no data cleaning is performed at this stage. As a consequence, the data may contain information on more than one activity for a given point in time. This also defines the rules for aggregating the monthly ARTKALEN data into yearly values. Take, for example, a person who was in full-time employment from January to November 2007, and unemployed in December 2007. The exact months are recorded in the dataset ARTKALEN. In the aggregated data, which is merged with the yearly data from the Biography Questionnaire, you find the information that the person worked full-time and was also unemployed in the year 2007. There is a second level of aggregation of ARTKALEN information as the data on type of activity, which is recorded in the variable SPELLTYP is more detailed than in PBIOSPE. The respective information is aggregated as described in Table 3. Table 3: Aggregation of ARTKALEN spell information into PBIOSPE PBIOSPE ARTKALEN 1 School/University School, College (1) 2 Apprenticeship/Training Vocational Training (4), First Job Training, Apprenticeship (13), Continuing Education, Retraining (14) 3 Military/Civilian service Military, Community Service (9) 4 Full-time employed Full-Time Employment (1), Short Work Hrs (2) 5 Part-time employed Part-Time Employment (3), Second Job (11), Mini-job (up to 400 euros) (15) 6 Unemployed Unemployed (5) 7 House-Husband/Wife Housewife, Husband (10) 8 Retired Retired (6) 9 Other Maternity Leave (7), Other (12) 99 Gap Information on gaps in ARTKALEN is not used. Gaps are calculated on the basis of the merged dataset. Source: SOEP v32, doi: 10.5684/soep.v32 As stated above, the calendar information is used to update the biography information. However, there is also a certain overlap of the periods covered by the two types of data. This is shown in Table 4. It indicates, for persons included in PBIOSPE, the year in which the biography information was collected (variable ERHEBJ). This year is usually also the last year for which biography information is available.1 The table also shows the first year recorded in the calendar data (variable KALYEAR). 1 Please note that some biographies were collected in 2011 although they are part of Wave 27. This results from the fact that some members of Sample I were interviewed in early 2011 instead of 2010. ███ SOEP Survey Papers 418 68 SOEP v32 Table 4: Overlap between biography and calendar information First observation in ARTKALEN (compared to erhebj*) same year or later % earlier erhebj* 1 year % 2 years % 3 years % 4+ years % Total n 1984 0.1 100.0 0.0 0.0 0.0 11,001 1987 0.0 36.4 33.5 30.1 0.0 505 1988 0.0 100.0 0.0 0.0 0.0 164 1989 0.5 99.5 0.0 0.0 0.0 193 1990 0.0 100.0 0.0 0.0 0.0 180 1991 0.0 100.0 0.0 0.0 0.0 157 1992 0.0 8.4 3.6 88.0 0.0 3,930 1993 0.0 76.6 0.3 2.3 20.7 304 1994 0.2 98.3 0.3 0.2 1.0 918 1995 0.2 99.1 0.0 0.1 0.6 1,037 1996 0.2 97.9 0.0 0.0 1.9 480 1997 0.0 98.5 0.0 0.0 1.5 478 1998 0.7 98.1 0.0 0.2 1.0 415 1999 0.1 26.6 72.8 0.0 0.5 1,821 2000 0.0 90.2 0.9 7.7 1.3 235 2001 0.0 6.3 93.6 0.0 0.0 7,529 2002 0.2 48.1 0.4 39.0 12.4 526 2003 0.1 16.9 81.3 0.1 1.6 2,193 2004 0.0 68.8 4.2 20.1 6.9 432 2005 0.0 89.0 3.4 0.7 6.9 292 2006 0.0 92.2 4.2 0.0 3.7 217 2007 0.0 16.2 83.4 0.1 0.3 1,858 2008 0.0 68.9 2.9 26.9 1.3 309 2009 0.0 89.5 2.1 0.5 7.9 190 2010 4.1 79.2 16.7 0.0 0.0 7,887 2011 2.5 91.7 5.7 0.0 0.1 5,670 2012 3.9 95.6 0.3 0.2 0.1 2,205 2013 92.6 7.1 0.2 0.1 0.0 3,860 2014 38.1 61.0 0.7 0.0 0.2 423 2015 0.0 79.3 12.0 7.5 1.1 266 Total 7.7 59.8 24.7 7.3 0.6 55,675 Notes: *) Year of biography data collection (variable ERHEBJ). Source: SOEP v32 (PBIOSPE), doi: 10.5684/soep.v32 In the majority of cases (59.8 percent), the earliest calendar information is available for the year before the biography interview. This is the case for persons who answered the Biography Questionnaire in their first year as survey respondents. The calendar in the Individual ███ SOEP Survey Papers 418 69 SOEP v32 Questionnaire refers to the year before the survey. There are, however, changes over time. In 1998, it was decided that first-time respondents from new samples would not be given the Biography Questionnaire in the first wave but in the second in order to reduce the entry threshold for these new respondents. Consequently, for the majority of persons in years after new samples were integrated (1999, 2001, 2003, 2007, 2010/11 – Samples E to I), the earliest calendar information is available two years before the biography information was collected. This was once again changed in 2011 and in 2013: respondents from samples J and K were given both questionnaires in their first year in SOEP and 2013 the sample M was introduced via a special Biography Questionnaire about individual migration histories without surveying the calendar data. The same applies to first-time respondents who are members of an old sample (e.g., persons who moved into a panel household) - they answer the Biography Questionnaire at the time of their first interview. The pattern is quite stable for most years before 1999. Notable exceptions is the year 1992. This is explained by the integration of East Germany into the SOEP in 1990 (Sample C). The majority of the respondents in these samples answered just the Biography Questionnaire at the entrance into the SOEP. Another exception is the year 1987. In the years 1985 to 1987, the life course matrix was not part of any of the questionnaires. Therefore the respective biography information was only available for persons who were interviewed in 1984. In 1988, biographic information was also collected for persons who became respondents in 1985, 1986, and 1987 (for all years ERHEBJ=1987). In addition There are even some cases (0.6 percent = 307 cases) where the biography information was collected a long time after the person started to respond to the Individual Questionnaire (up to 31 years). These are respondents who failed to answer to the Biography Questionnaire at a given time and therefore the biography information was collected later. In these—albeit very rare—cases, there is substantial overlap between the periods covered by the calendar and biography information. Table 5: Sources of PBIOSPE spells n % % cum. biography only 211,450 51.9 51.9 calendar only 130,600 32.0 84.0 1 biography, 1 calendar spell 63,394 15.6 99.6 2+ biography, 1 calendar spell(s) 582 0.1 99.7 1 biography, 2+ calendar spell(s) 1,062 0.3 100.0 2+ biography, 2+ calendar spell(s) 35 0.0 100.0 Total 407,123 100.0 Source: SOEP v32 (PBIOSPE ), doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 70 SOEP v32 After merging the information from the Biography Questionnaire and ARTKALEN, the data is transformed into spells, whereby each spell is defined by the duration of a given status. A question that arises when merging the data is how to handle overlapping pieces of information. The basic principle is to assign a value of a given status in a given year if the status is recorded in the calendar or in the biography information or both. An example might help to illustrate this: the calendar records full-time employment for the years 2005 and 2007 while the biography records full-time employment for the period from 2000 up to 2006. The merged data from PBIOSPE contains a spell that begins in 2000 and ends in 2007. However, the initial information is restored by including additional variables, which allows for alternative ways of merging the data (see below). The variables SPELLINF, ERHEBJ, and KALYEAR contain general information on the sources of the information captured in a given spell. Table 5 shows that the majority of spells are based on biography information only (51.9 percent). Slightly less than one-third of all spells (31.7 percent) are not observed in the Biography Questionnaire but only in the calendar data. The remainder of spells contain information from biography as well as calendar data. Usually these spells combine one period observed in the Biography Questionnaire with a period observed in the calendar. Only 0.4 percent of the spells combine more than one period in any of the two sources (SPELLINF=4, 5 or 6). The variables BEGINB1-ENDYK4 document the initial information from the two different sources and are probably not of interest to the majority of users. However, on the basis of these variables, users are able to fully separate the Biography data from the aggregated ARTKALEN data. This is advisable if you want to use the more detailed ARTKALEN information and combine it with the yearly information from PBIOSPE for earlier years only. The variable names indicate the “source” of the original information utilized (B: Biography - Questionnaire or K: calendar information from the yearly survey). As an example, we discuss one of the spells that combines information on more than one period from any of the two sources. The spell number 4 of person 9205 starts in 1983 and ends in 1994 (SPELLTYP=4: full-time employment). As the variable SPELLINF (=5) shows, this a spell that combines one period from the biography data with two periods from the calendar data. According to the biography data, the person worked full-time from 1983 (BEGINYB1) until 1992 (ENDYB1). There is overlapping information from the calendar data available from 1986 onwards (KALYEAR). According to these data, the person worked full-time from 1986 (BEGINYK1) to 1990 (ENDYK1) and from 1993 (BEGINYK2) to 1994 (ENDYK2). During the years 1991 and 1992, no full-time employment is recorded in the calendar data, which contradicts the information from the biography data. ███ SOEP Survey Papers 418 71 SOEP v32 greater than 27, is it replaced by the PBIOSPE data. (95% of these cases have an AGEFJOB below 27.) f) If we observe item non response concerning AGEFJOB and NOJOB, but spell information is available, the missing value is replaced by the corresponding PBIOSPE spell data. g) If even the ‘What did you do since you were 15’ question had not been answered, there still was a chance to extract similar information out of the PBIOSPE-file by considering the question ‘What did you do every month last year’. h) If we still had no valid information, the value of AGEFJOB was left out of the dataset. i) Due to the fact that PBIOSPE information are collected only until the end of the year preceding the actual wave, for respondents without first job information from both the biography questionnaire and PBIOSPE we further look for a first job using information from the current wave individual questionnaire. YOUTH-respondents j) For respondents who are regularly employed, information is taken from the Youth Questionnaire; AGEFJOB is coded as year of questioning minus year of birth minus one (only if the respondent does not state that he/she is still in school, etc.). k) If we additionally observe a spell starting before the respondent answers the Youth Questionnaire, information from PBIOSPE is used if the respondent does not state in the current questionnaire that he/she is still in school, etc. l) If respondents answer that they have no regular employment but provide an employment spell starting after the time of the first interview, information from $P (for details see m) is taken if available (only if the respondent does not state that he/she is still in school, etc.). m) For respondents with inconsistent first job information (simultaneous employment and school attendance/apprenticeship, differing job info in Youth Questionnaire and PBIOSPE) the question ‘Are you currently engaged in paid employment?’ asked in the Individual Questionnaire turned out to be the most reliable source of information. If a respondent states to be fullor part-time employed in a wave subsequent to the youth interview, AGEFJOB info is derived from the latest information of that kind. n) If people do not answer at least one of the questions ‘Do you currently earn money?” and ‘Do you earn money as an apprentice, full-time worker or part-time-worker?” but have an employment spell, like in m) the earliest $P information is taken if available (only if the respondent does not state that he/she is still in school, etc.). ███ SOEP Survey Papers 418 78 SOEP v32 o) If information from the Youth and the Individual Questionnaire (including PBIOSPE) are inconsistent concerning AGEFJOB, then the variable is set to missing. p) Due to the fact that PBIOSPE information are collected only until the end of the year preceding the actual wave, for respondents without first job information from both the Youth Questionnaire and PBIOSPE we further look for a first job using information from the current Individual Questionnaire. The pointer variable AGEINFO provides the coding information described above. Value labels of AGEINFO indicating the source of information are: (1) LELA-files (case (a) above) (2) PBIOSPE if AGEFJOB<15, but spell begin > 15 (c) (3) PBIOSPE if ‘not worked’ at interview but later spell begin (b) (4) PBIOSPE if ‘not worked’ at interview but earlier spell begin (d) (5) PBIOSPE if AGEFJOB>27 and earlier spell begin (e) (6) implausible information therefore set missing (h) (7) PBIOSPE if ‘not worked’-question and AGEFJOB not answered, but ‘what done at 15’-question answered (f) (8) PBIOSPE if ‘not worked’-question, AGEFJOB and ‘what done at 15’-question not answered, but ‘what done last year’-question answered (g) (9) completely missing (10) SP if no info from bio interview and PBIOSPE but employment in current Individual Questionnaire (i) (11) info drawn from Youth Questionnaire(j) (12) info drawn from PBIOSPE for persons who state in the Youth Questionnaire to be regularly employed and additionally have an employment spell starting earlier (k) (13) info drawn from $P for persons who state in the Youth Questionnaire not to earn money relating to an employment/job or to earn money but relating to a part-time job or a practical training, and have a subsequent employment spell (l) (14) info drawn from $P for persons with inconsistent first job information from the Youth Questionnaire or PBIOSPE, but valid employment information from an Individual Questionnaire subsequent to the biography interview (m) (15) info drawn from $P for persons with item non response in one of the questions ‘Do you already earn money from jobs?’ or ‘Do you earn that money as a trainee, fulltime or part-time employee?’ and with info in PBIOSPE (n) SOEP Survey Papers 418 79 SOEP v32 (16) completely missing (17) set to missing because of inconsistent information (o) (18) info drawn out of UP, the last wave of the SOEP (p) For more than 50% of the cases with AGEINFO = 3, 7, or 8 (AGEINFO=7 or 8 only if information collected after biography interview) it is possible to extract information from the regular questionnaires. For respondents with AGEINFO=10 or 11, information referring to the variables OCCFJOB, FJBLUE, FJWHITE, FJSELFE, FJSEFSIZ, FJCIVS, REQEDUC and CIVILSFJ are taken from the Individual Questionnaire (same year as of youth interview). While for respondents having AGEINFO=10 this approach is intuitive, for the persons having AGEINFO=11 we act on the assumption that the job declared in the respective Individual Questionnaire is still the first job of that person. This assumption seems plausible due to the low age of all persons responding to the YOUTH Questionnaire. In the YOUTH Questionnaire there is no question on the first job. But we can follow up their professional career by the statements given in the activity calendar in the subsequent waves. This can lead to problems if these youths report student jobs. For that reason we decided to take information from the question “Are you currently engaged in paid employment?” asked in the Individual Questionnaires of subsequent waves as the relevant source of information for this group of respondents. The earliest information of that kind determines the variable AGEFJOB. Some respondents have very low values with respect to AGEFJOB. Most of these jobs turn out to be low-skilled and starting before 1970. The respective persons are either blue collar workers (mostly unskilled) or self-employed (mostly helping in family business). We think these characteristics suggest that these specifications are valid. EINSTIEG_ARTK/EINSTIEG_ARTK_INFO The variable EINSTIEG_ARTK (by Marco Giesselmann and Mila Staneva) provides the year of available survey information related to the entry into the working force. It is primarily based on information found in the spell dataset ARTKALEN and generally founded on a different conceptualization of job entry than AGEFJOB. EINSTIEG_ARTK_INFO is a pointer variable indicating the source of the information. There are three main reasons why two seemingly redundant variables like AGEFJOB and EINSTIEG_ARTK are both included in BIOJOB. a) EINSTIEG is based on a more clear and consistent definition of what a first labor market entry is. Here the first labor market entry is conceptualized as the entry in the first job after the completion of (secondary and tertiary) education and apprenticeship. SOEP Survey Papers 418 80 SOEP v32 AGEFJOB, though, captures labor market entries at very different stages of the educational and employment biography. One reason for this is that it largely relies on a self-assessment of what a labor market entry is. In the Biography Interview all respondents are asked when they first started to work and this leads to very diverse self-reported labor market entries ranging from the first side-job in high school to the first full-time stable employment matching the own professional field. b) As described above for people who have never been employed at the time point of the Biography Interview the very first observed labor market entry from the spell-data PBIOSPE is used for the generation of the “age at first job”-variable. This also leads to inconsistent labor market entries since this generating strategy often captures student side-jobs. c) Additionally, EINSTIEG refers to the earliest yearly measurement after the transition. Although this year is, due to panel structure and interview date, not necessarily to the year of entry, only this strategy allows a clear assignment of covariates from the yearly measurement to the labor market entry. Among the several plausible concepts and operationalization of labor market entries with SOEP Data (among them AGEFOB), we consequently hold this indicator particularly suited for scholars who want to study the impact of labor market institutions on early career outcomes. By not being focused on first full-time or standard employment, but also regarding shifts into atypical employment as labour market entry, employment biographies spanned from this entry point capture the uncertainties and instabilities associated with the early career phase. At the same time, side-jobs or apprenticeships are explicitly assigned to the educational phase and excluded from the concept of labor market entry. A detailed description of the generation process of EINSTIEG_ARTK is given in the respective documentation file Introduction to the Variable EINSTIEG. EINSTIEG_PBIO/EINSTIEG_PBIO_INFO The variable EINSTIEG_PBIO provides the year of the entry into the working force. It is primarily based on information found in the spell dataset PBIOSPE and founded on the same concept as EINSTIEG_ARTK. EINSTIEG_PBIO_INFO is a pointer variable indicating the source of the information. The motivation behind another operationalization of job entry is straight-forward: Using the algorithm of EINSTIEG_ARTK only the job market entry-years of less than 7500 respondents can be reconstructed. This low number is explained by the fact that to be identified by our algorithm the beginning of the job (and the end of the educational) biography of a SOEP-participant has to be part of the ARTKALEN-dataset. This is only the case if one became part of the survey in late youth or early adolescence and did not leave the SOEP Survey Papers 418 81 SOEP v32 82 sample before a first employment could be observed, so only for a fraction of first jobs as defined by EINSTIEG_ARTK dates can be estimated. To offer a compromise between the problems of AGEFJOB and the few observations reconstructed by EINSTIEG_ARTK a third variable was created: EINSTIEG_PBIO, which instead of using information from ARTKALEN employs the dataset PBIOSPE, which includes spell-data gathered from the retrospective activity calendar which is part of the biography questionnaire. For understandable practical reasons though this data is just available on a yearly basis and not a monthly one like it is the case with ARTKALEN. The implied loss of granularity induces a potential higher risk of misclassifications compared to EINSTIEG_ARTK while enabling us to reconstruct job entries for a vastly higher amount of respondents, namely almost everyone who ever filled out the biography questionnaire. Still the potential use of EINSTIEG_PBIO compared to EINSTIEG_ARTK is much more restricted as again for most identified first jobs which fall in the time frame before the person became part of the sample and whose information deviates from agefjob there is just no further information available at all. A more detailed description of the generation process of EINSTIEG_PBIO is given in the respective documentation file Introduction to the Variable EINSTIEG. NOJOB The underlying question for the variable NOJOB is ‘I have never been employed up to this date’. This variable has the label ‘never been employed until the date of the interview” (1). If NOJOB has a missing value, in general there should exist AGEFJOB information, for special cases, see above. Due to the lack of a comparable question in the Youth Questionnaire, respondents of this questionnaire are given the value (1) as long as no consistent AGEFJOB information is available. STILLFJ This variable is based on the question ‘Are you still employed in the same job and at the same place?’. It applies only to LELA respondents who do not state ‘I have never been gainfully employed’ and whose biography interview was after 2000. Value labels: (1) Yes (2) No ███ SOEP Survey Papers 418 82 SOEP v32 83 FULLTIME The FULLTIME-variable is used to indicate, whether the first job of a person was a full-time or a part-time job. The value labels are (0) part-time job or marginal employment (1) full-time job. This variable is generated out of the file PBIOSPE for all respondents. For persons with first job information stemming from the Biography Questionnaires, FULLTIME possibly does not refer to the declared first job if PBIOSPE does not contain the respective job spell (i.e. due to item non response or incomplete answering of the activity biography within the Biography Questionnaire). OCCFJOB The variable OCCFJOB provides information on the occupational position at the first job. Due to different versions of the questionnaires in the GSOEP´s different samples we face some difficulties. Table 1 gives an overview. ███ SOEP Survey Papers 418 83 SOEP v32 Table 1: Number of Possible Values for Occupational Classifications in the First Job Farmers (not selfemployed) Blue Collar Workers Self-employed White Collar Workers Civil Servants Sample A, B (84-95) - 5 5 5 4 Sample C (90-95) 4 5 5 4 4 Sample D (94/95) 4 5 5 4 4 Sample A,B,C,D (96) - 3 4 3 4 Sample A,B,C,D (97-99), E (99) - 3 4 4 4 Sample A,B,C,D,E (00) - 3 6 4 4 Sample A,B,C,D,E,F (01) - 3 10 4 4 Sample A,B,C,D, E,F (02) - 5 10 6 4 Sample A,B,C,D, E,F,G(06),H(06), I(10),J(11),K(12) - 5 10 6 4 Source: SOEP v32 (PBIOSPE), doi: 10.5684/soep.v32 Facing these differences we decided to standardise the occupational classification. Only four types of occupational status were taken into account: blue collar workers, white collar workers, civil servants, and self-employed. The group ‘Farmers’ is included in the blue collar worker group. The potential value labels for OCCFJOB are: (1) blue collar worker (2) self-employed (3) white collar worker (4) civil servant Further details are provided by the variables FJBLUE (for blue collar workers), FJSELFE (self-employed), FJWHITE (white collar workers), and FJCIVS (civil servants). Table 2 shows the number of possible values. Table 2: Number of Possible Values for the subcategories of the variable OCCFJOB FJBLUE FJSELFE FJWHITE FJCIVS Sample A,B,C,D, E,F,G,H,I,J,K(12) 9 4 7 4 Source: SOEP v32 (PBIOSPE), doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 84 SOEP v32 Due to the fact that the PBIOSPE-file is used for the coding of AGEFJOB in certain cases (see above) there is less information on OCCFJOB than on AGEFJOB. FJBLUE The FJBLUE variable provides detailed information on the first occupational status if the person was a blue collar worker. Certain value labels are only given for certain samples, because of the already mentioned differences in the questionnaires. The following value labels are assigned: (10) unand semiskilled farmers (sample C/D) (11) unskilled worker (12) semiskilled worker (20) skilled worker (30) farmers (sample C/D) being foreman or master craftsman (31) foreman (sample A/B) (32) foreman (sample C/D) (40) master craftsman (41) farmers (sample C/D) in middle and higher management FJSELFE/FJSEFSIZ The FJSELFE variable provides detailed information on the first occupational status if the person was self-employed. FJSEFSIZ gives the number of employees in the respondent’s firm. Again there are differences due to the different versions of questionnaires. The following value labels are assigned: (10) independent farmer (20) free lances, self employed academics (30) other self employed workers (40) helping within family business FJSEFSIZ has the following value labels: (10) number of employees ≤ 9 (all subsamples (see exceptions for samples C/D), up until wave M) (11) no co-workers (all subsamples, from wave R on) ███ SOEP Survey Papers 418 85 SOEP v32 86 (12) number of co-workers 1-9 (all subsamples, from wave N on) (20) number of employees > 9 (all subsamples (see exceptions for samples C/D)) (30) number of employees ≤ 10 (sample C (waves I to L) / D (waves K to L), only if info drawn from biography questionnaire) (40) number of employees > 10 (sample C (waves I to L) / D (waves K to L) , only if info drawn from biography questionnaire) FJWHITE FJWHITE gives detailed information on persons, who were first employed as white collar workers. The subvalues of unskilled labour without degree (21), or with degree (22) are, due to uncomparable values in the LELA-files, only drawn from the $P-Files. (Beginning with BIOJOB 2004). Potential value labels: (10) industrial foreman (20) employee / unskilled labour (21) same as (20), but without degree (22) same as (20), but with degree (30) employee / skilled labour (40) employee / professional labour (50) employee / managerial labour FJCIVS FJCIVS provides detailed information on first employment as a public servant. The following value labels occur: (10) low level civil servant (20) middle level civil servant (30) high level civil servant (40) executive civil servant ███ SOEP Survey Papers 418 86 SOEP v32 ISCO88, STBA EGP, ISEI, MPS, SIOPS These variables – job classifications and different prestige scores – concerning in each case the first job but are not generated within this file and therefore they are not described within this documentation. REQEDUC REQEDUC provides information about the required education for the first job. This information has been asked in the Biography Questionnaire for the first time in the year 2001, but comparable information are gathered by the Individual Questionnaire in all waves. For all respondents having their first job subsequent to their biography interview, information is drawn out of the generated file $PGEN. Neither respective variables in $P nor those in $PGEN provide full information for all waves. In both data sources no differentiation is made between vocational college degree and university degree. As $PGEN info is equally coded in all waves, it is preferred to $P info. Potential value labels: (10) no training (20) completed vocational training (30) vocational college or university degree (31) vocational college degree (32) university degree CIVILSFJ CIVILSFJ indicates if the first job was assigned to the civil service or not. This information has been asked in the 2001 Biography Questionnaire for the first time For respondents having their first job subsequent to their biography interview, information is drawn out of the generated file $PGEN where this information is provided since the first wave in 1984. The following value labels occur: (1) Yes (2) No ███ SOEP Survey Papers 418 87 SOEP v32 (5) accredited professional school (6) technical or professional college (7) university General remark: Some persons answered more than once the Biography Questionnaire (but this occurs very rarely). The data-set BIOJOB contains only information from one Biography Questionnaire, in most cases the earlier one. 5.3 Steps of Coding 1. Creating a dataset using the data concerning all aspects of the job biography (working force entry, position, etc.) drawn from BIOLELA, MLELA, NLELA, OLELA, PLELA, QLELA, RLELA, SLELA, TLELA, ULELA, VLELA, WLELA, XLELA, YLELA, ZLELA, BALELA, BBLELA, BCLELA (internal DIW files with biographical information up to wave BB), QJUGEND, RJUGEND, SJUGEND, TJUGEND, UJUGEND, VJUGEND, WJUGEND, XJUGEND, YJUGEND, ZJUGEND, BAJUGEND, BBJUGEND, BCJUGEND (internal DIW youth biography files), QP, RP, SP, TP, UP, VP, WP, XP, YP, ZP, BAP, BBP, BCP (needed for consistency checks with respect to the youth biography files). 2. Using the PBIOSPE-data to retrieve spell information during the first occupation. 3. Using PPFAD for personal data (year of birth, sex, sample). 4. Using several files containing generated information about job classification (ISCO), prestige scores and industry sector classification (NACE) concerning the first job. 5. Combining all data concerning the employment biography into a new data file BIOJOB, where priority is set as mentioned above. 6. Coding of AGEFJOB. (for details, see above) 7. Setting the pointer variable AGEINFO indicating the source of the information of AGEFJOB. (for details, see above) 8. Excluding one value for respondents, who stated to have two occupational positions in their first job. Exclusion based on consistency checks. 9. Assignment of the variable OCCFJOB, with respect to the different versions of the questionnaire. Possible value labels: FJBLUE, FJSELFE, FJWHITE, FJCIVS. ███ SOEP Survey Papers 418 94 SOEP v32 10. Definition and assignment of new value-labels for the sub-category FJBLUE, nine labels possible, for details see above. 11. Definition and assignment of new value-labels for the sub-category FJSELFE, four labels possible, for details see above. 12. Definition of the variable FJSEFSIZ, indicating the numbers of employees. 13. Definition and assignment of new value-labels for the sub-category FWHITE, seven labels possible, for details see above. 14. Definition and assignment of new value-labels for the sub-category FJCIVS, four labels possible, for details see above. 15. Coding of the variables REQEDUC and CIVILSFJ. 16. Coding of the variables INTEDUC1 to INTEDUC4. 17. Computing the age at the most recent change of occupation if necessary. 18. Check of consistency: Does information about the age at the most recent change of occupation make sense? If inconsistencies appear, the value is set to a missing value. 19. Assignment of value labels for the variables specifying the last job: 20. Definition and assignment of value labels of the variable CURREMPL indicating if a respondent is gainfully employed at the time of the biography interview. 21. Specification of the year of last employment (YEARLAST). 22. Coding of the variables SCOPELJ and CIVILSLJ. 23. Excluding one value for respondents, who stated to have two occupational positions in their last job. Exclusion based on consistency checks. 24. Assignment of the variable OCCLJOB, with respect to the different versions of the questionnaire. Possible value labels: LJBLUE, LJSELFE, LJWHITE, LJCIVS. 25. Definition and assignment of new value-labels for the sub-category LJBLUE, nine labels possible, for details see above. 26. Definition and assignment of new value-labels for the sub-category LJSELFE, four labels possible, for details see above. 27. Definition of the variable LJSEFSIZ, indicating the numbers of employees. 28. Definition and assignment of new value-labels for the sub-category LJWHITE, seven labels possible, for details see above. 29. Definition and assignment of new value-labels for the sub-category LJCIVS, four labels possible, for details see above. SOEP Survey Papers 418 95 SOEP v32 30. Collecting of job information for people with AGEINFO = 3, 7 or 8, if possible. 31. Collecting of job information for people with AGEINFO = 12, 14 or 16, if possible. 32. Coding of the variable FULLTIME. 33. Definition of missing values for all variables. 34. Hand-editing of inconsistencies between different variables. 35. Final listing 36. Definition and assignment of new value-labels for the sub-category LJCIVS, four labels possible, for details see above. 37. Collecting of job information for people with AGEINFO = 3, 7 or 8, if possible. 38. Collecting of job information for people with AGEINFO = 12, 14 or 16, if possible. 39. Coding of the variable FULLTIME. 40. Coding of the EINSTIEG variables 41. Definition of missing values for all variables. 42. Hand-editing of inconsistencies between different variables. 43. Final listing SOEP Survey Papers 418 96 SOEP v32 Table 1: Overview of the datasets Provided type of history Time unit Starting time Sample BIOMARSM Marital Histories Month Entry into SOEP All adult SOEP-participants BIOMARSY Marital Histories Annual Year of birth All adult SOEP-participants BIOCOUPLM Relationship Histories Month Entry into SOEP All adult SOEP-participants BIOCOUPLY Relationship Histories Annual Year of birth Adult respondents answered the biography questionnaire after wave 27 or which were observed since the age of 17 6 The couple history files BIOCOUPLM and BIOCOUPLY, and marital history files BIOMARSM and BIOMARSY by Maik Hamjediers and Paul Schmelzer1 With the BIOMARSM/Y and BIOCOUPLM/Y the SOEP provides consistent and continuous marital and partnership histories for nearly all adult respondents. Whereas BIOMARSM and BIOCOUPLM just build on the prospective information at the time of each interview, BIOMARSY and BIOCOUPLY – containing also retrospective data – provide complete marital histories of respondents, starting at the year of their birth. But since until wave 27 no questions on a respondents’ couple history were asked, BIOCOUPLY includes only those respondents who have answered the biography questionnaire in wave 28 or later (and those who have been observed since the age of 17). Thus, all four datasets contain different information and therefore Table 1 gives an overview on the most important differences. Source: SOEP v32, doi: 10.5684/soep.v32 Note that the marital status in the $PGEN data files, stored as $FAMSTD, is derived from BIOCOUPLM and BIOMARSM at the time of the interview. Although the partner indicator PARTZ$ supplied in the $PGEN data files is considered in the generating process of BIOCOUPLM and BIOMARSM, due to different generating processes, it might not entirely 1 This documentation is a new version of previous SOEP documentations for the same files and has benefited from the work made by previous generators. For readability reasons, we do not specifically cite and specify text that has been used directly from the older SOEP documents. SOEP Survey Papers 418 97 SOEP v32 match with $FAMSTD. Furthermore, consistency checks between waves were done as well. That way, changes in $FAMSTD between data distributions but also in the couple and marital history datasets are possible for former waves. This documentation proceeds with a short description of the sources for the generating process and of the editing process of constructing logical and consistent marital and couple histories. These steps are important to understand the description of the four data files following in the next parts. 6.1 Sources of the couple and marital history For the construction of individual marital and relationship histories we gathered information 1. on current marital and couple status conducted in the personal questionnaire $P (and $PAUSL and $PLUECKE); 2. on monthly information on events, that may have occurred since the last personal interview, which are also stored in $P (and $PLUECKE); 3. on the generated partner pointer PARTZ$$ of the generated dataset $PGEN, which links current partners living in the same household; 4. on the marital and relationship biography from the biographical questionnaire $LELA (and BIOLELA). The personal questionnaire comprises a question on the marital status at the month of interview (Question 147 in 2015), whereby information on registered partnerships is included only since 2011. Furthermore it is asked for the current couple status, which entails whether someone has a partner and if so whether they are living together (Question 148 and 149 in 2015). For immigrants we also used information on the marital status derived from the foreigner questionnaire stored until 1995 in $PAUSL and for temporary drop outs we replaced missing information with data from a short version of the personal questionnaire, which information is stored in $PLUECKE. Moreover at least once the majority of SOEP participants answers the biography instead of the personal questionnaire, and for those waves the current marital and couple status is stored in $LELA. This hereby out of the four sources gathered information on the current marital couple status is used to generate spells, beginning at the time of the interview and reaching to the next interview in which a change in the status is reported. For those interviewed again it is also asked for any changes that occurred during the last year (Question 173 in 2015).2 This monthly information on the events ‘moved in together’, ‘marriage’, ‘divorce’, ‘separation’, ‘death of partner’ and since 2011 ‘starting a new 2 Due to the fact that events were collected retrospectively from first of January of the last calendar year until the month of interview, events in the beginning of a year could have been reported twice in two consecutive waves. Taking variations up to twelve months for the same event into account, the timing of the events were corrected. SOEP Survey Papers 418 98 SOEP v32 relationship’ that may have occurred since the last personal interview supplements the previous generated marital and couple status spells In addition, we added the generated partner pointer, which was just used as another source of. These three kind of information – events, the current status and the partner indicator – were used to generate the monthly datasets BIOMARSM and BIOCOUPLM. For the annual files, we additionally resorted on the retrospective information of the biographical questionnaire, which nearly all respondents just answer one time while participating in the SOEP. Since until 2010 it was just asked for the last three marriages (Question 72 in 2010), for all respondents having used the old version it was only possible to generate marital and no couple biographies from their year of birth on. In 2011, design of the retrospective questionnaire collecting information on couples changed notably. It now contains information on up to three previous marriages, registered same-sex partnerships or long-term relationships, defined as lasting for at least six months (Question 83 in 2015). Hence, up to four relationships, are possible to mention.3 Therefore for all respondents who have used the biography questionnaire after 2010 we generated the annual data BIOCOUPLY, whereas all responses to any biographical questionnaire were used to generate BIOMARSY. Naturally, the information drawn from the biographical questionnaires are proceeded by the responses to the personal questionnaire for the following waves. This is done by extending the monthly spelldata from BIOCOUPLM and BIOMARSY to an annual format; thus for this time, also shorter relationships were possible to mention, for example via the monthly information on the events mentioned above. 6.2 Construction of marital histories Information on marital history mainly stems from respondents’ retrospective reports on their own history. Thus, no other benchmark on the substance of these reports exists. This led to inconsistencies – for example, overlapping of reported marriages or impossible changes between legal states. In contrary to the relationship histories, where odd patterns of the reported histories occurred but were also possible for the most part, rules for logical histories of the legal marital status had to be laid down in the generation process: 1. Every individual marital history has to start with the state ‘unmarried’. We did not allow a person to be married before age 16. 2. From ‘unmarried’, one can only change to ‘married’ or into ‘living in a registered same-sex partnership'. 3 Please note, that also between 2011 and 2014 the questions of the biography questionnaire were changed. From 2011 to 2013 it was also possible to mention a fourth previous relationship, thus the questions for that relationship was limited compared to the question about the other relationships. Furthermore, the biography questionnaire of the National Survey of Families (FiD-Sample) left out the option of reg. same-sex partnerships. SOEP Survey Papers 418 99 SOEP v32 3. There is no possible return to ‘unmarried’ once a person was ever ‘married’.4 The only possible change from ‘married’ is to ‘divorced’, ‘widowed’ or ‘divorced or widowed’. 4. The only possible change from ‘divorced’ or ‘widowed’ is to ‘married’. 5. Reported separations of married spouses are taken into account as well. This led to the new category ‘married, separated’ which is coherent with the BIOCOUPL datasets and contains the time from a separation until a divorce or death of the respondent’s spouse. Yet, this rule was not applied for move-outs and later returns into the household of the same partner. If another marriage with a new partner is reported without a divorce or a death of the previous partner, after the ‘married, separated’- spell an episode of the category ‘divorced or widowed’ is also added and the beginning and end of those spells as well as the censoring were adjusted (Figure 1). Algorithms rearrange and correct original data to produce logically consistent marital histories following these five rules. Thus, we inserted an obligatory first ‘unmarried’ spell starting with birth and ending with the first marriage. Possible contradictions in the data were checked and edited as well, i.e.: a) contradictions between responses given in the same year in the different, previously described sources; b) illogical sequences between years. While most of the original data was used, we edited some spells, e.g. because of the notion of “romantic wedding”. This idea assumes that respondents often define their marital status in subjective or even affective terms rather than referring to the legal marital status. For instance, we frequently observe unmarried couples both reporting to be married for one year, and then returning to define themselves as singles (this phenomenon we call “romantic marriage”). Another rather emotional defined marital status are changes between the reported state of being divorced and being married, but not living together. This might stand for an insecure period after a break-up of a marriage and the changes between the states are corrected until 4 Except for the annulations of a marriage. However, given that this event is extremely rare – in particular compared to the many returns to ‘unmarried’ that we find in the data – we did not consider the possibility of annulations. Figure 1: Example for treatment of separated marriages in BIOMARSM/Y Respondents information Added spells persnr spellnr spelltyp begin end divorce persnr spellnr spelltyp begin end divorce 1 1 married 10 50 0 < 1 1 married 10 50 0 1 2 married, separated 50 -1 -1 1 2 married 100 150 0 1 3 divorced or widowed -1 100 0 1 4 married 100 150 0 SOEP Survey Papers 418 100 SOEP v32 any clear indication for an actual divorce is reported. A third frequent pattern refers to divorced respondents who report to be unmarried after they started a new relationship, since they may want to avoid the new partner to come to know about the former marriage. Since this kind of change of the legal marital status is impossible, the status of being divorced is assigned till the next reported marriage. If responses on marital status alternate between ‘divorced’ and ‘widowed’ without reporting another marriage in between, the information reported most often was used to correct these contradictions. Sequences alternating short-termed between ‘divorced’ / ‘unmarried’ and ‘married’ are replaced by ‘married’, as long as a new partner, a longer consistent sequence or the report of an event does not confirm this possibly new marriage. If the very end of a reported sequence indicates an ‘unmarried’ spell after being married and without further information, the marital history ends with a ‘divorced or widowed’, indicating that the reason for the end of a marriage is not known. Likewise, reported weddings during still existing marriages are ignored mostly. In cases where information of both partners on their joint marriage is available, contradictions in dates were not dissolved and the original information is not replaced. You can identify these spells by sorting the couples in the BIOCOUPLM dataset using the variable COUPID and decide which information you find more reliable. Completeness is a further criterion for construction of marital histories in the sense that the spell system is a closed system of spells starting from birth or the entrance into the SOEP going to the last year of sample membership. Due to item as well as partial unit non-response (i.e., a person of a SOEP households refuses to give a personal interview) and due to inconsistent information, ‘gap’ spells are introduced as another category of SPELLTYP on its own. Gap spells can occur at any place in the spell system, i.e., there are no restriction rules like the ones above. If information on marital status is missing for more than two years we inserted gap spells indicating that we have no knowledge of what happened during these periods. Likewise, missing retrospective biography information is indicated by an inserted gap spell. Also, if three finished marriages were reported in the biography questionnaire and a fourth current marriage via the next marital status, another gap is inserted. Missing information due to item non-response in the life course questionnaire may also affect the dates of the beginning or end of a spell. 6.3 Construction of couple histories As stated above, most of reported odd patterns in the couple histories may be possible and hence, no verified decision between measurement error and uncommon reality can be made. For that reason, whenever possible, couple histories are left as they were stated. Only in rare cases, restrictions, corrections on orderings of events or changes of declared years are SOEP Survey Papers 418 101 SOEP v32 conducted (read the following paragraphs for details). Further corrections to smooth out irregularities are thus left to the user. Firstly, the following default rules were obeyed to obtain logical and consistent histories if no other information forced to do otherwise: 1. Every individual couple history starts at the year of their birth with the state ‘single’. Because legal marriages are not possible before the age of 16, we restricted the age of marriage to be at least 16. Thus, we did not restrict age within a relationship, that is, unmarried relationships are allowed to start anytime. 2. Every spell set for a certain couple starts with the state ‘coupled, partner not in household’. One exception exists: if respondents report a year of moving together that lies before the start of their relationship, this specific couple history starts with ‘coupled, partner in household’. Note that in this case the information when this couple moved together is not available in BIOCOUPLM/Y anymore. Analogical the date of moving out is also lost if it was reported to come after the end of a relationship. Both dates you can still look it up in the original source. 3. If there is no evidence to the contrary, it is assumed that married couples live together and moved together before marriage. That is: for married couples their specific couple history starts with ‘coupled, partner not in household’ and is followed by a spell ‘coupled, partner in household’ before their marriage spell ‘married, spouse in household’ starts. Thus, if a couple moved together in the same year they married, a spell ‘coupled, partner in household’ is included anyway. The information whether spells needed to be inserted in order to this rule is stored in the variable REMARK (see Figure 2). Note that dates of becoming a couple and moving together or whether they moved together at all are not known in that case and were set equal to the time of marriage. SOEP Survey Papers 418 102 SOEP v32 4. Since it was possible to mention another relationship in the questionnaire, it is assumed that periods between those relationships mentioned were ‘single’ states and thus assigned accordingly. Note that this relates to long-term relationships (longer than six months) reported via the biography questionnaire only. In contrast to those, relationships derived from reported changes in the familiar situation in the personal questionnaire may also be shorter than six month. 5. Any (formerly) married couple that is not an active relationship anymore, i.e. married couple which is separated but not yet divorced, ends with a spell ‘married-separated’ (see Figure 3). As long as they are not divorced (or the partner hasn’t died) yet, the end date of those separated spells is the same as their last interview year. Contrary to this, the separation spell is not added if marriage clearly ended with a divorce or the death of the respondent’s partner and no previous break-up was reported. Be aware, that because the question for a separation does not differentiate between a move-out and a clear ending of a relationship, there is some uncertainty whether a marriage is still active, but not in a joint household or the relationship ended and the spouses are n o t d i v o r c e d Figure 2: Example for additional spells before marriages Respondents information Added spells persnr spellnr spelltyp begin end remark persnr spellnr spelltyp begin end remark 1 1 single 10 50 1 < 1 1 single 10 50 1 1 2 coupled, not living together 50 50 2 1 2 married, living together 50 100 1 1 3 coupled, living together 50 50 2 1 4 married, living together 50 100 1 Figure 3: Example for treatment of separated marriages in BIOCOUPLM/Y Respondents information Added spells persnr spellnr spelltyp begin end divorce persnr spellnr spelltyp begin end divorce 1 1 married, living together 10 50 0 < 1 1 married, living together 10 50 0 1 2 married, separated 50 150 -1 1 2 coupled, not living together 100 150 0 1 3 single 50 100 0 1 4 coupled, not living together 100 150 0 SOEP Survey Papers 418 103 SOEP v32 with other spells that contain the actual couple status(es) over the entire separation episode and is therefore redundant in terms of completeness of couple histories. The additionally assigned codes ‘implausible’, ‘unknown’ and ‘unit nonresponse’ indicate implausibility or a lack of information for the respective period. Variable SPELLNR is a chronological index number for each individual’s spell during the observation period. Due to the fact that the spells’ duration is measured in months via the personal questionnaire, it is important to note that an individual may encounter several events in the same year. In this case the variable SPELLNR allows the user to order spells with respect to the respondent’s life course.The variables BEGINY and ENDY provide the years in which a spell begins and ends, whereas the variables BEGIN and END indicate respondent’s age for users’ convenience. In BIOCOUPLY, spell systems for each individual always start with the respondent’s birth. The SPELLTYP of the first spell per definition is ‘single’. As SPELLTYP does contain missing values, so do BEGIN(Y) and END(Y), indicating that the exact year of change in the couple status is not known. Missing dates indicate that the year was either not reported (-1) (or especially for ‘married, separated’-spells the event of divorce is not reported) or that the reported year of change is implausible (-3), i.e. contradictory to other information. In order to differentiate the reasons for missing information the user can utilize the variables REMARK and CENSOR. Consistent with BIOCOUPLM the data file BIOCOUPLY also provides the indicator variables PDEATH and DIVORCE. PDEATH indicates whether a respective spell ends with the death of a person’s partner, but is not restricted to married persons, thus does not only refer to widowhood. The states of widowhood can easily be derived from BIOCOUPLY by using the spells of marriage in SPELLTYP and checking for the death of the respective spouse in PDEATH (also BIOMARSM/Y contains the marital status and therefore information about widowhood). PDEATH of single spells are assigned a ‘(-2) does not apply’. DIVORCE works in a similar fashion, indicating whether the last marriage spell, that is the separated spell, ended in divorce. Hence, if it did not end in divorce it is coded as a still ongoing marriage. In this case END is updated by the year of the last interview. Variable REMARK provides information on whether we had to edit or supplement original information given by respondents in order to construct consistent couple biographies. Spells in BIOCOUPLY are marked as ‘edited’ (in contrast to ‘original’) if the editing process involved substitution of reported information because of inconsistence with responses in previous or following interviews or information reported by a partner. Furthermore we have added spells, e.g. between two marriages because two marriages with separate persons at the same time are not legal (see later section for more details on editing); these spells are marked as ‘added spells’. Furthermore, ‘first spells’ and ‘gap spells’ are marked separately, whereas ‘added’ or ‘edited’ first spells are marked as ‘first spells’. For BIOCOUPLY we also generated a variable SOURCE, which contains information about whether the respective spell SOEP Survey Papers 418 110 SOEP v32 was generated on basis of answers to the biography questionnaire or / and of responses to the personal questionnaire. The variable CENSOR indicates whether a spell is left-censored, rightcensored or censored on both sides (see Table 2 above within the explanations on BIOCOUPLM). The coding furthermore provides information on the reasons of censoring. In principle, spells might be censored if they precede or follow a gap spell or if BEGIN or END is missing. The last spell for each person is marked as right-censored if a person is still in the SOEP and the current relationship status is open (‘last spell’). 6.6 BIOMARSM: A monthly marital history Spells in data file BIOMARSM contain prospectively collected information on marital biographies starting with the information reported in the first personal interview. The data file comprises eleven variables: the case and individual identifiers HHNR and PERSNR as well as nine spell specific variables. Variable SPELLTYP documents marital status with the possible categories ‘unmarried’, ‘married’, ‘divorced’, ‘widowed’ and ‘divorced or widowed’. Once married, a later spell ‘not married’ is not assigned anymore. Note that we renamed the known SOEP code ‘(1) single’ to ‘not married’. This is necessary to indicate that it is possible the respondent might have a partner anyway. If you are interested in this information, we recommend using BIOCOUPLM instead of BIOMARSM. Take also into account that we do not differentiate between marriages and registered same-sex partnerships for the SPELLTYP categories (3) to (6). SPELLTYP has one additional category ‘divorced or widowed’ which indicates that a marriage definitely ended though we do not know whether via divorce or death of the spouse. This may be due to missing information from the biographical questionnaires or due to a respondent’s frequent shifts between both categories without ever reporting the death of the partner or divorce as an event. A sixth state is ‘gap’ indicating a lack of reliable data for this period. SOEP Survey Papers 418 111 SOEP v32 Variable SPELLNR is a chronological index number for each individual’s spells during the observation period. Variables BEGIN and END indicate the month in which a marital spell starts and ends. Monthly histories start with a value of 1 in January 1983 and ranges until the current margin in month 384, i.e., December 2014. In principle, the month in which a spell starts is the very same month the previous spell ends in. Please take into account that the months of BEGIN and END are not imputed when the exact month of the change in the familiar situation is not reported. Instead, in those cases the time of the interview is left as the BEGIN or END date when a change is observed in the data. To distinguish between these two cases we introduced a new variable called EVENTS, containing the information whether the exact month of the BEGIN or END of a spell is reported or not. It is important to note that a ‘first spell’ in BIOMARSM is not the very first spell of a person, but the first observed marital status since the person is taking part in the SOEP. Accordingly, the first spell in BIOMARSM is almost always left-censored. Variable CENSOR informs about whether a spell is leftor right-censored and if so, why. Most spells in BIOMARSM are in fact censored. In order to provide the user with detailed information on the nature of censorship we distinguished ‘left’, ‘right’, and combined ‘leftand right-censored spells’ with respect to the reason for censoring: ‘first spell’ or ‘last spell’, ‘spell ends with death’, spell Variables of BIOMARSM HHNR Identifier of original sample household PERSNR Personal identifier SPELLNR Consecutive spell number (chronological order) SPELLTYP Marital status (1) Unmarried (2) Married (3) Divorced / reg. same-sex partnership annulled (4) Widowed / reg. same-sex partnership deceased (5) Divorced or widowed / reg. same-sex partnership annulled or deceased (6) Married, separated / reg. same-sex partnership, separated (7) Living in reg. same-sex partnership (9) Gap BEGIN Month when spell begins [1=Jan 1983 to 396=Dec 2015] END Month when spell ends [1=Jan 1983 to 384=Dec 2014] BEGINY Year spell begins [1983 to 2015; -3=implausible; -1=missing] ENDY Year spell begins [1983 to 2015; -3=implausible; -1=missing] CENSOR Censoring information [0 to 14] (see explanation above, Table 2) EVENTS Month information: exact month of spell begin or end known? REMARK Further spell information (1) Original spell (2) Edited spell (3) Added spell (4) Gap Spell (5) First Spell SOEP Survey Papers 418 112 SOEP v32 ‘precedes’ or ‘succeeds a gap’ (see Table 2 above within the explanations on BIOCOUPLM). Of course, ‘death’ and ‘last spell’ are not mutually exclusive, thus we overwrite the latter with the former reason for being right-censored if the last interview is in the year of death or precedes it. Variable REMARK provides information on whether we had to edit or supplement original information provided by respondents in order to construct consistent couple biographies. Spells in BIOMARSM are marked as ‘edited’ if the editing process involved substitution of reported information because of inconsistence with responses in previous or following interviews or information reported by a partner. Furthermore we have added spells, e.g. between two marriages because two marriages with separate persons at the same time are not legal (see later section for more details on editing); these spells are marked as ‘added spells’. Finally ‘first spells’ and ‘gap spells’ are marked separately, whereas also ‘added’ or ‘edited’ first spells are marked as ‘first spells’. 6.7 BIOMARSY: A annual marital biography Data file BIOMARSY supplements BIOMARSM with retrospectively collected information on the marital history since a respondent’s year of birth. Whereas the marital history in BIOMARSM is measured in months, BIOMARSY depicts the marital biography on an annual basis. In contrast to BIOCOUPLY the BIOMARSY data set contains also the respondents to the biography questionnaire before wave 28. Please note, that until wave 28 the biography questionnaire just asked for three previous marriages and therefore the number of reported marriages in BIOMARSY is limited. The BIOMARSY file comprises eleven variables. The individual and household identifiers HHNR and PERSNR as well as SPELLTYP are basically the same in all data sets. Once married, a later spell ‘not married’ is not assigned anymore. Note again that we renamed the known SOEP code ‘(1) single’ to ‘not married’ and take into account that we do not differentiate between marriages and reg. same-sex partnerships for the SPELLTYP categories (3) to (6). This is to indicate that it is possible the respondent might have a partner anyway. Like BIOMARSM, it is important to notice that SPELLTYP has one additional category ‘divorced or widowed’ which indicates that a marriage definitely ended, though we do not know whether via divorce or death of the spouse. This may be due to missing information from the biographical questionnaires or due to a respondent’s frequent shifts between both categories without ever reporting the death of the partner or divorce as an event. SOEP Survey Papers 418 113 SOEP v32 114 Regarding the fact that duration of spells is measured in years it is important to notice that an individual may encounter several events in the same year. In this case the variable SPELLNR allows the user to order the spells with respect to a respondent’s life course. The variables BEGINY and ENDY provide the years in which a spell begins and ends, while the variables BEGIN and END indicate for users’ convenience the respective age of the respondent. The spell system for each individual in BIOMARSY always starts with the birth of the respondents. We thus created a first spell for each individual ever interviewed in the SOEP starting in the year of birth and continuing at least until the year in which a person turns age 16. The SPELLTYP of the first spell per definition is ‘unmarried’. Even if a respondent reported an earlier marriage in the biography questionnaire we restricted its beginning to age 16, however, we marked the beginning of this spell as ‘implausible’. There are some missing values (-1 and -3) in BEGINY as well as in ENDY (resp. BEGIN and END) indicating that we do not know the exact year of a change in the marital status. This can have two reasons. First, it may simply indicate that the respondent did not report the year in which a marriage began or ended. In order to differentiate the reasons for missing information the user may utilize the variables REMARK and CENSOR. Second, within single case corrections some dates are set ‘implausible (-3)’ if overlapping of marriages appear unsolvable or contradictions with the reported year and i.e. the year of death exist. Variables of BIOMARSY HHNR Identifier of original sample household PERSNR Personal identifier SPELLNR Consecutive spell number (chronological order) SPELLTYP Marital status (1) Unmarried (2) Married (3) Divorced / reg. same-sex partnership annulled (4) Widowed / reg. same-sex partnership deceased (5) Divorced or widowed / reg. same-sex partnership annulled or deceased (6) Married, separated / reg. same-sex partnership, separated (7) Living in reg. same-sex partnership (9) Gap BEGIN Age of respondent when spell begins [-3=implausible; -1=missing] END Age of respondent when spell ends [-3=implausible; -1=missing] BEGINY Year when spell begins ENDY Year when spell ends CENSOR Censoring information [0 to 14] (see explanation above, Table 2) SOURCE Source of information (1) derived only from biography questionnaire (2) derived from biography and personal questionnaire (3) derived only from personal questionnaire REMARK Further spell information (1) Original spell (2) Edited spell (3) Added spell (4) Gap Spell (5) First Spell ███ SOEP Survey Papers 418 114 SOEP v32 115 REMARK indicates whether a spell was ‘edited’ or ‘added’ in the same way as in BIOMARSM (see above). Variable CENSOR indicates if a spell is left or right censored or censored on both tails. The coding of CENSOR (see Table 2 above) provides also information about the reasons for censoring. In addition to what was said before about gap spells in BIOMARSM, gaps or missing values in BEGIN or END may appear in BIOMARSY if a respondent reported a terminated first marriage and the beginning of a second marriage, but did not report the reason for and/or the year of the end of the first marriage. The last spell of each person is marked as right censored, irrespective that the person died, quit the SOEP or is still married currently. ███ SOEP Survey Papers 418 115 SOEP v32 7 BIOBIRTH: A Data Set on the Birth Biography of Female Respondents25 by Christian Schmitt 7.1 Population and purpose of the data set BIOBIRTH The file BIOBIRTH provides information on fertility histories of adult respondents in the SOEP. Until 2014 (version 30, wave BD) the data was stored in two separate files: BIOBIRTH containing female fertility histories, and BIOBRTHM providing male fertility histories. It is important to note that the latter file only records the male fertility histories for respondents who entered the SOEP in 2001 or later (for more details see below). Since 2015 (version 31, wave BE) all fertility histories of new respondents as well as the old and continuously updated data of the fertility histories collected in BIOBIRTH (until 2014) and BIOBRTHM (until 2014) are stored in a single file - BIOBIRTH (this naming is identical to the previous fertility histories of women). The variable SEX (distinguishing male and female respondents) is added to the BIOBIRTH dataset since wave BE. Moreover, the file BIOBIRTH is also supplemented with the fertility histories of the “Familien in Deutschland” (FiD) panel survey, which was integrated into the SOEP data-base in 2015. Records from the FiD subsamples can be distinguished by inspecting the variable BIOVALID. For more details on integration and extension procedures see below. Fertility histories in BIOBIRTH provide information on every woman (as well as every man with a panel entry since 2001) who has ever provided at least one successful SOEP interview. Note that the data is right censored for respondents who left the panel early in their fertile lifephase and it is left censored for persons who entered the SOEP at higher ages without ever filling in a biographical questionnaire. The variable BIOVALID provides information on whether individual level information is based merely on information derived from household composition and family relations, or on biographical questionnaire data. The variables EINTRITT and AUSTRITT in ppfad (panel entry and exit) provide information on censoring (left-censoring can be ignored if a biographical questionnaire exists for a given person) For each of the mentioned adult respondents BIOBIRTH documents the fertility history. The annual update focuses on including new information on becoming a biological parent as based on data collected with the individual or the biographical questionnaire, respectively. Furthermore adults who have been interviewed for the first time but who have not yet provided information on their fertility histories are included. The latter case applies to either new adult household members, or teenagers who have reached the required minimum for a participation in the personal questionnaire (16 years). BIOBIRTH constitutes an accumulative data set, in which the entire birth biography of all SOEP respondents is presented. 25 Information on female birth biographies was designed with reference to earlier works by Joachim R. Frick. SOEP Survey Papers 418 116 SOEP v32 7.2 Structure of the data set BIOBIRTH covers the following information: (1) Person identifier (PERSNR), respondent’s year of birth, status information on the origin of the included data, number of children derived from fertility histories, total number of children derived from fertility histories and subsequent consideration of changes in household structure up to the last date of interview. (2) A sequence of 15 variables relating to 1 out of 15 children, including child’s person identifier (KIDPNR[nn], provided the child could be identified within the SOEP’s household structure), child’s sex, child’s year of birth, child’s month of birth. BIOBIRTH contains the following variables for all adult women (and men since 2001): • HHNR Invariable number of the original household • PERSNR Invariable personal identifier of the respondent • SEX Respondent’s sex • GEBJAHR Respondent’s year of birth • BIOVALID Status of the birth biography: (Please mind: The codes in the variable BIOVALID have been extended in 2015 (wave BF)). 10: no birth biographical entries 20: youth biography questionnaire, no fertility histories on children 30: birth biography questionnaire, no children in fertility history 31: birth biography questionnaire, one or more children in fertility history 40: FiD: no birth biography, no data on children 41: FiD: no birth biography, information on existing children in FiD parent/couple questionnaire 50: FiD: birth biography questionnaire, no children in fertility history 51: FiD: birth biography questionnaire, one or more children in fertility history. • BIOYEAR Survey year when birth biography questionnaire was completed (1985ff.). Code “-2” is assigned if the respondent never completed a birth biography questionnaire. ███ SOEP Survey Papers 418 117 SOEP v32 • BIOAGE Age of the woman at the time of the birth biography survey. Code “-2” is assigned if the respondent never completed a birth biography questionnaire. • SUMKIDS Total number of children born (more precisely: total number of children identifiable within SOEP by merging all available data up to the time of the last observation (SUMKIDS=BIOKIDS+subsequent births during panel participation). • BIOKIDS Total number of children identified in the birth biography. Code “-2” is assigned if the respondent never completed a birth biography questionnaire. • KIDGEB[nn] Year of birth of the child [nn] (for the first child up to the fifteenth child). • KIDSEX[nn] Sex of the child [nn] (for the first child up to the 15th child). • KIDPNR[nn] Personal number of the child [nn] (for the first child up to the 15th child), given it is identifiable in the SOEP. • KIDMON[nn] Month of birth of a child [nn] (for the first child up to the 15th child). With respect to the variables KIDGEB[nn], KIDSEX[nn], KIDPNR[nn], and KIDMON[nn] identical missing codes apply: The code “-2“ is assigned if there’s no [nn]th child identified for a mother/father. The code “-1” applies if an [nn]th child can be identified but the information on the birth year and/or sex and/or personal identifier is unavailable, or if it could not be identified. For every respondent a maximum of up to 15 children are considered. The sequence of children within BIOBIRTH is recorded with regards to the birth order in terms of age of the children. The order ranks from the oldest child specified under KIDPNR01 to the youngest child. If the age is missing it is listed in the first record (KIDPNR01), and in subsequent records following KIDPNR01 if more than one child’s personal identifier remains missing. 7.3 Information basis of the birth biography The main basis of the individual fertility history considered in BIOBIRTH is the information collected with the biography questionnaire26, in which the number, birth year and sex of the biological children for every adult respondent are collected. For adults with information on 26 The information collected over the course of the biography survey for every adult contains the number of children, the year of birth, the sex, residence status of the child, and, if applicable the year of death of the biological child. The biography data is stored in wave specific files ($LELA), which are not provided with the SOEP distribution. SOEP Survey Papers 418 118 SOEP v32 children stemming from the biography questionnaire the BIOVALID code “31” is assigned. Women who completed this questionnaire, but did not report any biological children receive the code “30”. In correspondence to this, the samples of the FiD panel integrated into the SOEP since 2015 (wave BE) contribute the codes 50 and 51 for respondents with biography questionnaires with, or without children, respectively. Codes 10 and 40 correspond accordingly for SOEP and FiD data. 41 is a qualitatively new code for parents in the FiD who did not fill in the birth biography but who have provided reliable information on their status as a parent in other sources of FiD (this relates to specific questions in the FiD parent and couple questionnaire). A minority of respondents did not provide any information on fertility histories by filling in the biographical questionnaire for several reasons27. For these cases the variable BIOVALID is assigned the code 10 (original SOEP samples), and 40 (new FiD subsamples), respectively. This group is subject to a risk of underestimating the total number of births, particularly since births prior to the entry into the SOEP cannot be identified unless current household structure provides substantive evidence, usually reflected by parent-child co-residence. Respondents without a valid fertility history (codes 10 and 40) can be distinguished in three major groups: • Respondents who were 16 years of age at the time of the first interview. In most cases these respondents participate in the biography survey at a later date. Thus, the parent-child relationship recorded earlier in BIOBIRTH (as based on household structure) can be verified and supplemented with data on their fertility histories at a later date. • Respondents who were at about 30 years of age or younger at the time of first interview. In this sub-population, children are not yet adults and still reside in the parental home in most cases. Since information from the biographical questionnaire is missing, a final distinction in social and biological children is difficult particular for records considered in earlier SOEP waves when the intra-household relationships ($STELL) were less refined, compared to the recent setup. Hence these older records have a higher likelihood of misspecifying a social as a biological parent-child relationship. • Respondents who were well over 30 years of age at the time of the first interview. In these cases some of the children are likely to already have left the parental home, and therefore are no longer part of the survey population. For that reason, the number of biological children might be underestimated in this group of respondents to a larger extent as compared to younger women. 27 Beside the reason ‘refusal’, the collection date of the life history biographies differ among SOEP sub-samples. SOEP Survey Papers 418 119 SOEP v32 Source: SOEP v32, doi: 10.5684/soep.v32 BIOYEAR Year of Biography Survey Year of Biography Survey Frequency Percent Cumulative Percent -2 13.096 17,93 17,93 1985 6.618 9,06 27 1986 83 0,11 27,11 1987 104 0,14 27,25 1988 220 0,3 27,55 1989 211 0,29 27,84 1990 202 0,28 28,12 1991 161 0,22 28,34 1992 2.635 3,61 31,95 1993 230 0,31 32,26 1994 592 0,81 33,07 1995 528 0,72 33,8 1996 255 0,35 34,14 1997 231 0,32 34,46 1998 203 0,28 34,74 1999 1.033 1,41 36,15 2000 361 0,49 36,65 2001 9.429 12,91 49,56 2002 899 1,23 50,79 2003 2.691 3,68 54,47 2004 822 1,13 55,6 2005 667 0,91 56,51 2006 530 0,73 57,24 2007 2.576 3,53 60,77 2008 594 0,81 61,58 2009 438 0,6 62,18 2010 9.597 13,14 75,32 2011 6.908 9,46 84,78 2012 3.227 4,42 89,2 2013 651 0,89 90,09 2014 4.606 6,31 96,4 2015 2.630 3,6 100 Total 73028 100,0 Source: SOEP v32, doi: 10.5684/soep.v32 ███ SOEP Survey Papers 418 126 SOEP v32 SUMKIDS Total Number of Births Total Number of Children Born Frequency Percent Cumulative Percent 0 26310 36,03 36,03 1 13921 19,06 55,09 2 19720 27 82,09 3 8782 12,03 94,12 4 2790 3,82 97,94 5 888 1,22 99,16 6 328 0,45 99,6 7 138 0,19 99,79 8 79 0,11 99,9 9 28 0,04 99,94 10 23 0,03 99,97 11 11 0,02 99,99 12 7 0,01 100 16 2 0 100 17 1 0 100 Total 73028 100.00 Source: SOEP v32, doi: 10.5684/soep.v32 BIOKIDS Number of Births from Biography Births from Birth Biography Questionnaire Frequency Percent Cumulative Percent -2 13096 17,93 17,93 0 22340 30,59 48,52 1 11293 15,46 63,99 2 15589 21,35 85,33 3 7169 9,82 95,15 4 2283 3,13 98,28 5 714 0,98 99,26 6 282 0,39 99,64 7 129 0,18 99,82 8 71 0,1 99,92 9 32 0,04 99,96 10 16 0,02 99,98 11 7 0,01 99,99 12 4 0,01 100 16 2 0 100 17 1 0 100 Total 73028 100.00 Source: SOEP v32, doi: 10.5684/soep.v32 SOEP Survey Papers 418 127 SOEP v32 8 BIOTWIN: TWINS in the SOEP by Christian Schmitt 8.1 Population and contents of the data set BIOTWIN The file BIOTWIN contains all twins that were ever identified within the SOEP. To be classified as a twin, a person is required to: • have exactly the same age as his or her sibling (year & month of birth), • have a relationship to the head of the household that indicates that he or her and a second persons are siblings, and • have the same mother (as far as a pointer to the mother is available). Furthermore, it is not only twins that are recorded in the BIOTWIN data set, but also triplets or quadruple siblings. The following variables are stored within the BIOTWIN data set: • HHNR Invariable number of the original household. • PERSNR Invariable personal identifier of the first sibling. • PNRTWIN Invariable personal identifier of the second sibling, the twin. • PNRTRIP Invariable personal identifier of the third sibling. • PNRQUAD Invariable personal identifier of the fourth sibling. • PNRMOTH Pointer to the personal identifier of the mother of the twin-group. • BIOMONOZ Monozygotic group? Information if the group is monozygotic. • INFSOURC Source of information from which the status of being a twin is derived The central variable PERSNR is assigned to the sibling with the lowest personal identifier in the twin group. The PNRTWIN and – in rare cases if available – PNRTRIP or PNRQUAD contain the personal identifier of second, and third or fourth sibling in the group. This means that every case in the data set consists of a group of twins (or triplets or quadruplets). The code “-2” is assigned to PNRTRIP and/or PNRQUAD if a third or fourth twin sibling doesn’t exist. PERSNR and PNRTWIN however should always contain valid codes. The variable PNRMOTH provides the link to the mother of the group and is derived from the data sets $KIND (reference to this $KIND was discontinued in with wave Z / 2009) and/or BIOBIRTH. ███ SOEP Survey Papers 418 128 SOEP v32 8.2 The twin survey of 2006 In 2006, a questionnaire was distributed among all households with potential twin groups, identified up till then. The aim was to validate that none of these twins had been identified by mistake. The variables INFOTWIN and BIOMONOZ contain new information which was derived from this survey. The result of the survey could widely validate the selection of the twin population, contained in the BIOTWIN data set of the SOEP. More than 80% of households with potential twins as of 2006 could be contacted and were interviewed in the twin survey. Among these only 3 groups of twins turned out to be identified erroneously (those false positives were removed from the BIOTWIN data set). Thus the algorithms of identifying twins within the SOEP could prove to be widely reliable. Additional information that was collected with the twin survey contributed to identifying a number of mothers of twins, for whom the mother-child-link was missing previously. Furthermore the twin survey provided additional information on monozygotic respectively dizygotic twins. The variable BIOMONOZ was extended, in order to reflect this additional information (see below for more details). 8.3 Construction of variables in the data set BIOTWIN The variable BIOMONOZ28 indicates if the group is monozygotic. If the information could be validated in the twin-survey in 2006 the code is set to 1 for monozygotic twins and 2 for dizygotic twins. If the information on being monoor dizygotic twins could not be validated in the twin survey, which was carried out in 2006, the code is set to 0 if the sex of all the siblings is identical, and this group thus might be monozygotic. Please pay attention to the fact that the labels and values of the variable BIOMONOZ from wave W onwards are not consistent with values and labels from previous waves. The variable INFOTWIN is introduced with wave W and provides information on the source from which the status of being a member of a twin group is derived from and whether this information could be validated in the twin-survey in 2006. INFOTWIN can take the following characteristics: 1 Generated up to 2006 – basis: household co-residence, identical parent, year & month of birth – not validated by in the twin survey 2006 2 Possible Twin or Triplet – Information not revisable in twin survey 2006 3 Possible Twin or Triplet –Answer refused in twin survey 2006 28 This variable existed before wave W but was restructured to reflect the additional information which became available with the 2006 twin questionnaire. ███ SOEP Survey Papers 418 129 SOEP v32 4 Twin or Triplet – Information validated by twin survey 2006 5 Twin or Triplet – New since 2006 (congruent year & month of birth) 6 Twin or Triplet – New since 2006 (congruent year of birth / missing month of birth) The selection of twins within the SOEP, which compiles the data set BIOTWIN, is based on the either the month of birth, or an identical year of birth. Priority is given to congruent months of birth, as a woman might – in rare cases – give birth at two different times in a year. Hence the month of birth plays a central role in identifying potential twin-groups. According to that logic people with a) valid month of birth information or b) identical month of birth, or c) with an identical year of birth and missing data on the month of birth among both siblings are classified as twins. In a second step, the relationship of these potential twins to the head of household is scanned ($STELL). If the relationship of both persons assures that they are siblings, then they are assumed to be twins. In a third step the pointer to the mother is checked for both siblings with focus on the files $kind / BIOBIRTH. If this maternal link is identical for both siblings, it is transferred into the variable PNRMOTH. An overview of central information in the file BIOTWIN (Version 2015 / Wave BF) Table 1: Siblings in BIOTWIN29 Sibling type n Valid Mother Pointers Twins 1428 1278 Triple 50 50 Quadruple 4 4 Source: SOEP v32, doi: 10.5684/soep.v32 29 Please note: sibling groups contribute observations for each individual of the twin-pair/triplet, that is – a pair of twins would provide two entries to the data set, one for each twin. SOEP Survey Papers 418 130 SOEP v32 Table 2: BIOMONOZ Gender Combination of Siblings Frequency Percent Cumulative Percent Valid (-1) No Answer 2 0,14 0,14 0 Possibly Identical Twins 721 50,49 50,63 1 Definitely Identical Twins 46 3,22 53,85 2 Definitely Fraternal Twins 659 46,15 100 Total 1428 100 Source: SOEP v32, doi: 10.5684/soep.v32 Table 3: INFOTWIN Twin Status: Source of Information Frequency Percent Cumulative Percent Valid 1 Generated - not in twin survey 2006 45 3,15 3,15 2 Twinsurvey 2006 (answer not verified) 73 5,11 8,26 3 Twinsurvey 2006 (answer refused) 2 0,14 8,4 4 Twinsurvey 2006 (answer validated) 196 13,73 22,13 5 Gen. since 2007 (basis: year of birth & month ) 717 50,21 72,34 6 Gen. since 2007 (basis: year of birth / month miss) 395 27,66 100 Total 1428 100 Source: SOEP v32, doi: 10.5684/soep.v32 SOEP Survey Papers 418 131 SOEP v32 9 BIOSIB: Information on siblings in the SOEP by Josephine Kraft and Daniel D. Schnitzlein 9.1 General description of the data set BIOSIB provides information on siblings living within the SOEP households. The data set contains the person numbers of all siblings in an observed family. It includes information on their sex, their year of birth, the number of siblings, the individual’s position within the birth order, and on the relationship between the observed siblings. 9.2 Sources of information on siblings in the SOEP Information on siblings in the SOEP is available from three sources: • First, the respondents are asked about their siblings in the biography questionnaire and the youth questionnaire. This information (for example have or ever had siblings yes/no, number of sisters, number of brothers) is stored in the file BIOPAREN (for detailed information please see the chapter on BIOPAREN). • Second, in the years 1991, 1996, 2001, 2006, 2011 and 2013 all respondents were asked about their family relations. Among other questions on their family, the individuals were asked about their siblings (like above, have or ever had siblings yes/no, number of sisters, number of brothers) outside the SOEP household in 1991, 1996 and 2001. From 2006 on they were asked about all siblings within and outside the household. The information from these questions is stored in the $$p-files. • Third, siblings within the SOEP households are observed directly. The aim of this file is to provide the never changing person ID of the siblings of each respondent as far as they can be identified in the SOEP. With this information it is possible to use the whole range of personal or household information for each sibling to carry out detailed sibling analyses. This file therefore adds to the information in BIOTWIN (for details see chapter on BIOTWIN), which provides the person IDs for all twins in the SOEP. ███ SOEP Survey Papers 418 132 SOEP v32 9.3 Overview on the number of siblings in BIOSIB Figure 1 gives an overview on the information stored in BIOSIB. The data set contains 13,716 families with 34,484 individuals that have at least one sibling identified in the SOEP. 17,437 of them have one sibling, 10,584 have two siblings, 3,987 have three siblings, 1,398 have four siblings and 1,040 have five or more siblings identified in the data. As is apparent from Figure 1 in all cases most of the siblings are identified by having the same mother and the same father (for more details about the identification of siblings and the identification of biological siblings please see the description to SIBDEF1-SIBDEF11 on the subsequent pages). Figure 1: Number of persons with information on their siblings, by number of siblings Source: own calculation, based on SOEPv32, doi: 10.5684/soep.v32 9.4 Organization of the data in BIOSIB Each row in the dataset represents one individual for which at least one sibling could be identified. Therefore a family with three siblings appears three times in BIOSIB, one time for each child. The person IDs of the siblings are ordered by birth order starting with the oldest sibling. ███ SOEP Survey Papers 418 133 SOEP v32 List of variables HHNR Original Household Number PERSNR Never Changing Person ID SIBPNR1 – SIBPNR11 Person Number of 1st – 11th Sibling SIBDEF1 – SIBDEF11 Sibling Relation to 1st – 11th Sibling FAMCOUNT Family Counter POS_SIB Position in the birth order NUM_SIB Number of observed siblings in the SOEP SEX Gender of Individual GEBJAHR Year of Birth of Individual SEXSIB1 – SEXSIB11 Gender of 1st – 11th Sibling GEBSIB1 – GEBSIB11 Year of Birth of 1st – 11th Sibling The variables HHNR, PERSNR, SEX, GEBJAHR, SEXSIB1-SEXSIB11 and GEBSIB1GEBSIB11 are generated from the information stored in PPFAD. The newly generated variables SIBPNR1-SIBPNR11, SIBDEF1-SIBDEF11, FAMCOUNT, POS_SIB, and NUM_SIB are described on the next pages. SOEP Survey Papers 418 134 SOEP v32 Variable SIBPNR1 – SIBPNR11 Label: Person Number of 1st-11th Sibling Values: (-1) No answer (-2) Does not apply (-3) Answer improbable Description: The variables provide the never changing person IDs for the siblings of the individual identified by PERSNR. The sibling relationship is generated from the parent information in BIOBIRTH, BIOBRTHM and BIOPAREN (for detailed information on these files please see the relevant chapters above). Two persons are defined as siblings if they report both, the same mother and father, only the same mother, or only the same father. This information on the sibling relationship is stored in SIBDEF1-SIBDEF11. In the case of inconsistent information on parents in BIOBIRTH and BIOPAREN, BIOPAREN was assigned the lowest priority. Please note, that BIOPAREN uses a social definition of parenthood based on cohabitation. In contrast, BIOSIB contains both biological (BIOBIRTH/BIOBRTHM) and social siblings with a higher priority on biological relations. SOEP Survey Papers 418 135 SOEP v32 Bioage06: “Your child between the ages of five and six”, Children Aged 5-6 years The questionnaire is given to all mothers whose child turns six in the current survey year. In the exceptional case that a mother cannot complete the questionnaire, the father responds. Bioage08a and Bioage08b: “Parent” Questionnaire, children aged 7-8 years The questionnaire is given to both parents of children turning eight in the current survey year. Data of parent 1, which is usually the mother, can be found in bioage08a. Data of parent 2, which is usually the father, can be found in bioage08b. Bioage10: “Your child between the ages of nine and ten”, Children Aged 9-10 years (only FiD: Bioage10b) The questionnaire is given to all mothers (FiD: and fathers) whose child turns ten in the current survey year. In the exceptional case that a mother cannot complete the questionnaire, the father responds. Data of parent 1, which is usually the mother, can be found in bioage10a. Data of parent 2 (FiD only), which is usually the father, can be found in bioage10b. Bioage12: “Pupils between the ages of eleven and twelve”, Children Aged 11-12 years The questionnaire is given to all children themselves who turn twelve in the current survey year. Number of Children and Twins in the Bioage Data Sets The data set has grown over the years and now contains data on 12,661 children. At the moment (2015), there are 139 children for whom information has been provided through all of the “Mother & Child”, “Parent”, and “Pupils” questionnaires (bioage01, bioage03, bioage06, bioage08a/b, bioage10, and bioage12). Data are available for 234 children from six questionnaires, 694 children from five questionnaires, 1,405 children from four questionnaires, for 2,697 children from three questionnaires, and for 3,140 children from two questionnaires. For 4,352 children data are available from one questionnaire. An overview of the number of children in each data set is given in Table 3. SOEP Survey Papers 418 142 SOEP v32 Table 10: Number of respondents to the “Mother & Child”, “Parent”, and “Pupils” questionnaires Bioage File Survey Year Total 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 $01 318 247 246 234 205 185 196 1503 353 378 349 444 312 4970 $02 (FiD) 787 647 568 187 2189 $03 257 222 237 246 186 1061 914 745 701 454 386 5152 $06 237 210 687 696 612 858 825 657 4782 $08a 646 703 715 647 629 779 4119 $08b 409 490 500 439 399 556 2793 $10 403 510 699 686 656 616 3570 $10b (FiD) 242 310 291 301 1144 §12 606 608 1214 Total 318 247 503 456 442 668 592 5738 4623 4508 4168 4013 3914 30190 Source: SOEP v32, doi: 10.5684/soep.v32 Table 4 gives an overview of the number of children per family in the specific age range. Note that there might be more children in the household than indicated here: these children are not in the age range of the “Mother & Child”, “Parent”, and “Pupils” questionnaires and therefore not covered by the bioagel data set. ███ SOEP Survey Papers 418 143 SOEP v32 Table 11: Number of children in the family in the specific age range Bioage File Number of Children in the Family 1 Child 2 Children 3 Children 4 Children 5 Children 6 Children Bioage01 2909 815 113 16 2 3 Bioage02 (FiD only) 1671 232 18 Bioage03 3332 858 92 16 3 2 Bioage06 2922 811 71 5 1 Bioage08a 2607 624 84 3 Bioage08b 1772 418 59 2 Bioage10 2460 469 56 1 Bioage10b (FiD only) 785 169 7 Bioage12 1082 63 2 Source: SOEP v32, doi: 10.5684/soep.v32 The data also includes twins. Bioage01 contains 109 pairs, bioage02 57 pairs, bioage03 106 pairs, bioage06 64 pairs, bioage08a 73 pairs, bioage08b 45 pairs, bioage10 67 pairs, bioage10b 24 pairs, and bioage12 19 pairs of twins. 10.3 Topics and Variables The bioage data contain information regarding • Pregnancy and childbirth • Child health • Childcare situation • Changes in living circumstances since birth of child • Child’s abilities • Parenting experiences • Expectations about the child’s success in school • Educational goals and aspirations of the parents • Educational behavior of the parents • Self-conception of the role of parents ███ SOEP Survey Papers 418 144 SOEP v32 The rules used to generate the variables from the questionnaires are consistent over the various bioage questionnaires. In the new integrated bioage long data set, the data is presented in “long” format, i.e. this dataset contains information from bioage01, bioage03, bioage06, bioage08a and bioage08b, bioage10 as well as bioage12. In addition, all variables from the FiD-Study are included as well: if there was a corresponding SOEP Variable, the data was combined into the SOEP variable; if there was no corresponding SOEP Variable or if the data was coded differently, the information is provided in separate variables with the suffix _fid. A few variables are generated by combining the information from two or more variables. The present documentation provides detailed information on variables generated using information from other files. An overview of the specific variables and the rules by which they were generated is given in section 10.4 “Generated Variables”. 10.4 Generated Variables This section provides additional information on variables that have been generated from combinations of variables from other datasets than the parent questionnaires. AGE Variable label “Child's age in months” Variable format 2-digit integer Comment: This variable provides the child’s age in months as a combination of month of birth and interview month. As the exact day of birth remains unknown, information is only an approximation and may vary by one month. Note that the information concerning year and month of birth from the “Mother & Child” and “Parent” questionnaires proved to be partially inconsistent. Therefore, as of the beginning of 2012 the child’s age in months is computed using birth information from ppfad. For further information, refer to the documentation on ppfad. ███ SOEP Survey Papers 418 145 SOEP v32 PREGY Variable label “Mother: pregnant at interview in survey year X” Variable format 4-digit integer Bioage File 01 Comment This variable is based on information from the previous year’s individual questionnaire provided by the mother on her pregnancy status at the time of the interview. If pregnancy was reported (or was unknown) and a child was born, the year in which the interview took place is contained in BCPREGY. Hence, this information is available only for those women in the sample for at least two years with a completed individual interview in the first year and a completed “Mother & Child” questionnaire in the second year. Please note that some mothers are not aware that they are pregnant in the early stages of pregnancy. The time of observation starts in survey year 2003 with the 2002 birth cohort. SOEP Survey Papers 418 146 SOEP v32 PREGMO Variable label “Mother: pregnancy month at interview” Variable format 2-digit integer Bioage File 01 Comment This variable is based on the exact month of birth (BCPREGMO), the duration of childbearing in weeks (BCSSW) and the interview month of the previous year’s personal interview. Hence, this information is available only for those women in the sample for at least two years. As the exact day of birth is unknown, this variable remains a close approximation. PREBEG Variable label “Spell Begin Pregnancy (Month, 01.83=1)” Variable format 3-digit integer Bioage File 01 Comment The variable BCPREBEG contains information on the beginning of pregnancy (i.e., the month of conception). Information is given in the regular SOEP spell format: values start with 1 for January 1983 (e.g., the earliest spell in bioage01, survey year 2010, is 304, which equals April 2008). The variable is based on the exact month of birth (BCPREGMO) and the duration of pregnancy in weeks (BCSSW). Accordingly, information is available only for women who completed the “Mother & Child Questionnaire” and for whom the duration of the pregnancy is known. Note that the month of conception may vary by one month as the exact date of birth remains unknown. ███ SOEP Survey Papers 418 147 SOEP v32 PREEND Variable label “Spell End Pregnancy, Birth (Month, 01.83=1)” Variable format 3-digit integer Bioage File 01 Comment The variable BCPREEND contains information on the end of pregnancy (i.e. the month of birth). Information is given in the regular SOEP spell format: values start with 1 for January 1983 (e.g., earliest spell in bioage01, survey year 2010, is 304, which equals April 2008). This variable is based on the exact month of birth (BCPREGMO) and the duration of pregnancy in weeks (BCSSW). SEX Variable label “Gender of child” Variable format 1-digit integer Comment The sex of the child is not asked for in the “Mother & Child” and “Parent” questionnaires. Information on this variable stems from the ppfad. SEXRESP Variable label “Sex of respondent (parent)” Variable format 1-digit integer Bioage File 08a, 08b Comment This variable tells whether the mother or father answered the respective questionnaire. The information for this variable comes from the ppfad file. SOEP Survey Papers 418 148 SOEP v32 149 11 BIOAGE17: The Youth Questionnaire1 by Marco Giesselmann, Mila Staneva and Tabea Naujoks2 A special group of first time respondents are young persons living in a panel household, who reach the surveying age of 17 years. From this specific group of panel entrants, we are able to obtain some more detailed information on youth and socialisation than from other new sample members. At the same time, certain life-course dimensions (as the partnershipor employment biography) have not yet developed in 17 year-olds. With regard to these specifics, the standard biography questionnaire is not appropriate to this group. Thus, we use an independent questionnaire for this special group of first time respondents: the Youth Questionnaire. This instrument is used since the year 2000 and can be understood as an alternative version of the Biography Questionnaire, collecting more comprehensive information on relationships with parents, leisure-time activities, and past achievements in school, as well as on personality characteristics. In addition, there are numerous prospective questions about educational plans and plans for further training, as well as questions about expectations for future career and family. A number of statements regarding specific circumstances—including the expectations for the future mentioned above—are directly related to the time at which the questionnaire was completed. However, they provide a multifaceted background for long-term analyses since these young people will continue to be interviewed in subsequent years like other SOEP respondents. The Youth Questionnaire also contains retrospective questions, for example, at what age the teenager started his or her first job or first music lessons, what recommendations he or she received regarding choice of secondary school level, and which grades he or she repeated. 11.1 Genesis and Target Population of the Youth Questionnaire The Youth Questionnaire is aimed at youths who have reached the surveying age of 17 years3 and are therefore being interviewed for the first time. This questionnaire takes the place of the supplementary Biography Questionnaire, since the latter does not apply to the young people’s family or career situations. As a rule, information on social origin can be obtained from the parents’ Individual Questionnaire, in case the youth lives together with the respective parent. If the teenager does not live with either parent, the Youth Questionnaire collects information on the missing parent(s). Young people who immigrated to Germany are also given the 1 In earlier SOEP-data releases BIOAGE17 was called BIOYOUTH. 2 Replaces earlier versions by Henning Lohmann and Sven Witzke, Jürgen Schupp, and Michaela Frühling, Thorsten Schneider, and Bettina Isengard. 3 More precisely, this refers to youths who live in an already existing panel household and are or will turn 17 years old in the year of the survey. They are therefore 16 or 17 years old at the time of the interview. █████ SOEP Survey Papers 418 149 SOEP v32 150 standard questions on immigration from the supplementary Biography Questionnaire. This guarantees that all important information collected in the Biography Questionnaire is also available on these young people. A preliminary version of the Youth Questionnaire was tested in 2000 in samples A-E on individuals born in 1983. An expanded and revised questionnaire entered the field one year later, in 2001, for all samples (A-F). In samples A-E, young people born in 1984 were surveyed, and in sample F, those born in the years 1982 to 1984. With the expansion of the number of birth cohorts, entries for the birth year 1983 are also collected for sample F (data previously existed only for samples A-E), which also creates a clear increase in the number of entries. In the following years, also the youths from additional samples have been interviewed. In 2014, Data from the SOEP-related FID-study from the years 2010 to 2014 has been integrated in the SOEP (Sample L). Therefore, the number of cases in BIOAGE 17 has increased with SOEP Version 31 retrospectively for the years 2010 to 2013, compared with previous versions. You might want to use the psample-variable as filter, if you want to exclude these cases and reproduce your old sample. For an overview of the target population in each survey year, see Table 1. In total, we have gathered interview data from 6,6641 analysable observations up to the present. Table 1: Target Population for the Youth Questionnaire by year, sample and age Suvey year Sample frequency A-E F G H I J K L M 2000 17 yrs 232 2001 17 yrs 17-19 yrs 618 2002 17 yrs 17 yrs 352 2003 17 yrs 17 yrs 17 yrs 365 2004 17 yrs 17 yrs 17 yrs 373 2005 17 yrs 17 yrs 17 yrs 368 2006 17 yrs 17 yrs 17 yrs 307 2007 17 yrs 17 yrs 17 yrs 17 yrs 346 2008 17 yrs 17 yrs 17 yrs 17 yrs 261 2009 17 yrs 17 yrs 17 yrs 17 yrs 243 2010 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 404 2011 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 531 2012 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 537 2013 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 567 2014 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 577 2015 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 17 yrs 560 Status: up to wave BF (2015) Source: SOEP v32, doi: 10.5684/soep.v32 █████ SOEP Survey Papers 418 150 SOEP v32 151 In 2006, a new questionnaire on cognitive potential was introduced. Like the Youth Questionnaire, this instrument is aimed at youths who have reached the surveying age of 17 years. In order to keep the Interview at an acceptable length, the standard Individual Questionnaire is now left out for respondents of the Youth Questionnaire. The 2006 data on cognitive potentials was provided for secondary analysis in 2009 (dataset COGDJ). 11.2 Contents and Structure of the Data Set BIOAGE17 From a technical perspective, four different types of questions are asked in the Youth Questionnaire: A) Questions used to complete certain biographical files (BIOIMMIG, BIOPAREN). These questions are identical to questions in the standard Biography Interview. This applies to the topic blocks ‘Origin’ (questions 60 to 71) and ‘Childhood and Parents’ House’ (questions 7287). The corresponding variables are not included in BIOAGE17, but combined with biographical information from non-youth new entrants in the files BIOIMMIG and BIOPAREN. B) Questions that are similar to items in the standard Biography Interview, but go further into detail. This applies to the topic blocks ‘Relationships’ (questions 12-14), ‘Free time and Sport’ (questions 15-25) and ‘Education and Career plans’ (questions 26-55). These variables are stored in BIOAGE17. Corresponding Variables obtained from other new sample members (with a standard Biography Interview) are included in the dataset BIOSOC. Depending on the complexity and scope of the analysis, the user might want to combine corresponding data from BIOAGE17 and BIOSOC in order to access all panel members. C) Questions that specifically relate to young persons and therefore have no equivalent in the standard Biography Interview. This applies to the topic blocks ‘Residence’ (questions 1-3), ‘Jobs and Money’ (questions 4-11), ‘Future’ (question 59) and ‘Attitudes and Opinions’ (questions 86-87). These Variables are stored in BIOAGE17 and have no equivalent for other panel entrants in BIOSOC. D) Since 2006, selected time-variant questions from the unanswered regular individual questionnaire (which is not handed out to first time panel respondents from existing panel households, see 1.3) are added to the Youth Questionnaire. This refers to the questions 56 to 58, 91, 92 and the topic block ‘personality’ (questions 93 to 101). This data is not included in BIOAGE174, but stored in an additional dataset $PAGE17. 4 The first ten items in question 92 are still stored in BIOAGE17, for details see 13.3. █████ SOEP Survey Papers 418 151 SOEP v32 254 BIGOBACK BI: Rueckkehr Heimat (ab 1994) BI: Go back home? BIO Question: Q17 Comment: The question BIGOBACK (using BIOLELA) asks whether one intends to return home to the native country whereas BISTAY (using $PAUSL) asks wether one intends to stay in Germany. The wording and the answer possibilities are different in both questions. Further, there is no particular reason to believe that the two variables even are consistent. Starting 2001, this is not defined for new entrants. German: Planen Sie selbst, in Ihr Herkunftsland wieder zurückzukehren? "[1] Ja, ganz sicher" "[2] Ja, wahrscheinlich" "[3] Eher unwahrscheinlich" "[4] Nein, sicher nicht" English: Are you planning to go back to live in your native country? "[1] Yes, certainly" "[2] Yes, probably" "[3] Probably not" "[4] No, Certainly not" See also: BIGOBACK, BISTAY, BISTAYY Year File Variable 1984-93 A-J BIOLELA n/a 1994 K BIOLELA P230Z 1995 L BIOLELA P230Z 1996 M MLELA MB230Z 1997 N NLELA NB230Z 1998 O OLELA OB230Z 1999 P PLELA PB230Z 2000 Q QLELA QB230Z 2001 -- $LELA n/a 2000 Q QJUGEND QJ69 2001 -- $JUGEND n/a █████ SOEP Survey Papers 418 254 SOEP v32 255 BISTAY BI: Wunsch in D zu bleiben BI: Desire to Stay in Germany BIO Question: Q17 Comment: The question BIGOBACK (using BIOLELA) asks whether one intends to return home to the native country whereas BISTAY (using $PAUSL) asks wether one intends to stay in Germany. The wording and the answer possibilities are different in both questions. Further, there is no particular reason to believe that the two variables even are consistent. This variable is not defined for youths answering the $JUGEND biography questionnaire. German: Wie lange wollen Sie in Deutschland bleiben? "[1] Kehre innerhalb eines Jahres zurück" "[2] Einige Jahre und zwar..." "[3] Für immer in D bleiben" English: How long would you like to stay in Germany? "[1] Go back within 12 months" "[2] Several years, specifically..." "[3] Always stay in Germany" See also: BIGOBACK, BISTAY, BISTAYY █████ SOEP Survey Papers 418 255 SOEP v32 256 Year File Variable 1984 A APAUSL AP67A01 1985 B BPAUSL BP96A01 1986 C CPAUSL CP87A01 1987 D DPAUSL DP89A01 1988 E EPAUSL EP77A01 1989 F FPAUSL FP99A01 1990 G GPAUSL GP96A01 1991 H HPAUSL HP99A01 1992 I IPAUSL IP99A01 1993 J JPAUSL JP99A01 1994 K KPAUSL KP96A01 1995 L LPAUSL LP107A01 1996 M MP MP101A01 MP100A 1997 N NP NP109A01 NP108A 1998 O OP OP11401 OP113 1999 P PP PP12601 PP125 2000 Q QP QP13401 QP133 2001 R RP RP12701 RP126 2002 S SP SP12601 SP125 2003 T TP TP13301 TP132 2004 U UP UP13501 UP134 2005 V VP VP14601 VP145 2006 W WP WP13601 WP135 2007 X XP XP14601 XP145 2008 Y YP YP14501 YP144 2009 Z ZP ZP14401 ZP143 2010 BA BAP BAP14601 BAP145 2011 BB BBP BBP14701 BBP146 2012 BC 2013 BD 2014 BE 2015 BF BCP BDP BEP BFP n/a BDP15001 BDP149 n/a BFP16401 BFP163 2013 BD 2014 BE 2015 BF BDP_MIG BEP_MIG BFP_MIG BDPM_P_3701BDPM_P_36 n/a BFPM_P_5201 BFPM_P_51 2000 -- $JUGEND n/a █████ SOEP Survey Papers 418 256 SOEP v32 257 BISTAYY BI: Dauer des geplanten Aufenthalts BI: Years Desired to Stay in Germany BIO Question: Q17 Comment: This variable is not defined for youths answering the $JUGEND biography questionnaire. German: Wie lange wollen Sie in Deutschland bleiben? Einige Jahre und zwar... English: How long would you like to stay in Germany? Several years, specifically... See also: BIGOBACK, BISTAY, BISTAYY █████ SOEP Survey Papers 418 257 SOEP v32 258 Year File Variable 1984 A APAUSL AP67A02 1985 B BPAUSL BP96A02 1986 C CPAUSL CP87A02 1987 D DPAUSL DP89A02 1988 E EPAUSL EP77A02 1989 F FPAUSL FP99A02 1990 G GPAUSL GP96A02 1991 H HPAUSL HP99A02 1992 I IPAUSL IP99A02 1993 J JPAUSL JP99A02 1994 K KPAUSL KP96A02 1995 L LPAUSL LP107A02 1996 M MP MP101A02 1997 N NP NP109A02 1998 O OP OP11402 1999 P PP PP12602 2000 Q QP QP13402 2001 R RP RP12702 2002 S SP SP12602 2003 T TP TP13302 2004 U UP UP13502 2005 V VP VP14602 2006 W WP WP13602 2007 X XP XP14602 2008 Y YP YP14502 2009 Z ZP ZP14402 2010 BA BAP BAP14602 2011 BB BBP BBP14702 2012 BC 2013 BD BCP BDP n/a BDP15002 2014 BE 2015 BF BEP BFP n/a BFP16402 2013 BD 2014 BE 2015 BF BDP_MIG BEP_MIG BFP_MIG BDPM_P_3702 n/a BFPM_P_5202 2000 -- $JUGEND n/a █████ SOEP Survey Papers 418 258 SOEP v32 259 BISCGER BI: In Dt. Schule besucht? BI: Attended School in Germany BIO Question: Q18 Comment: This question asks only if one has ever attended a (primary/secondary) school in Germany, but does not ask whether one received a certificate/diploma, such as the generated variable $PSBIL in the file $PGEN. Since 2014 and in the IAB-SOEP Migrationsample the question is about the last attended school. German: Haben Sie in Deutschland eine Schule besucht? "[1] Ja" "[2] Nein" English: Did you attend school in Germany? "[1] Yes" "[2] No" See also: BISCGER, BISCGRAD, BISCGERC, BISCGC, BISCGCF, BISCGCFN █████ SOEP Survey Papers 418 259 SOEP v32 260 Year File Variable 1984 A APAUSL AP06A01 1985 B BPAUSL BP100A01 1986 C CPAUSL CP100B01 1987 D DPAUSL DP97A01 1988 E EPAUSL EP90A01 1989 F FPAUSL FP107A 1990 G GPAUSL GP107A 1991 H HPAUSL HP107A 1992 I IPAUSL IP107A 1993 J JPAUSL JP107A 1984-93 A-J BIOLELA B46A 1994 K BIOLELA P280Z 1995 L BIOLELA P280Z 1996 M MLELA MB280Z 1997 N NLELA NB280Z 1998 O OLELA OB280Z 1999 P PLELA PB280Z 2000 Q QLELA QB280Z 2001 R RLELA RB280Z 2002 S SLELA SB11 2003 T TLELA TB11 2004 U ULELA UB11 2005 V VLELA VB11 2006 W WLELA WB11 2007 X XLELA XB11 2008 Y YLELA YB11 2009 Z ZLELA ZB11 2010-2012 B$ B$LELA B$B11 2013 BD 2014 BE 2015 BF BDLELA BELELA BFLELA BDB12 BEB45 BFB47 2013 BD 2014 BE 2015 BF BDP_MIG BEP_MIG BFP_MIG BDPM_L_70 BEPM_L_61 BFPM_L_136 2000 Q QJUGEND QJ63 2001 R RJUGEND RJ65 2002 S SJUGEND SJ65 2003 T TJUGEND TJ65 2004 U UJUGEND UJ65 2005 V VJUGEND VJ65 2006 -- $JUGEND n/a █████ SOEP Survey Papers 418 260 SOEP v32 261 BISCGRAD BI: In welche Klasse in dt. Schule BI: Which Grade School BIO Question: Q19 Comment: The question here is not on the highest schooling achieved, but rather what was the grade or class when one first came to Germany. German: In welche Klasse sind Sie in Deutschland in die Schule gekommen? English: Which class/grade did you attend when you came to Germany? See also: BISCGER, BISCGRAD, BISCGERC, BISCGC, BISCGCF, BISCGCFN █████ SOEP Survey Papers 418 261 SOEP v32 262 Year File Variable 1984 A APAUSL n/a 1985 B BPAUSL n/a 1986 C CPAUSL n/a 1987 D DPAUSL n/a 1988 E EPAUSL n/a 1989 F FPAUSL FP108A 1990 G GPAUSL GP108A 1991 H HPAUSL HP108A 1992 I IPAUSL IP108A 1993 J JPAUSL JP108A 1984-93 A-J BIOLELA B47A 1994 K BIOLELA P290Z 1995 L BIOLELA P290Z 1996 M MLELA MB290Z 1997 N NLELA NB290Z 1998 O OLELA OB290Z 1999 P PLELA PB290Z 2000 Q QLELA QB290Z 2001 R RLELA RB290Z 2002 S SLELA SB12 2003 T TLELA TB12 2004 U ULELA UB12 2005 V VLELA VB12 2006 W WLELA WB12 2007 X XLELA XB12 2008 Y YLELA YB12 2009 Z ZLELA ZB12 2010-2012 B$ B$LELA B$B12 2013 BD Since 2014 BE BDLELA BELELA BDB13 n/a 2000 Q QJUGEND QJ64 2001 R RJUGEND RJ6601 2002 S SJUGEND SJ6601 2003 T TJUGEND TJ6601 2004 U UJUGEND UJ6601 2005 V VJUGEND VJ6601 2006 -- $JUGEND n/a █████ SOEP Survey Papers 418 262 SOEP v32 263 BISCGERC BI: Besuch spezieller Vorbereitung BI: Attended Special Foreigner Prep Class BIO Question: Q20 German: Haben Sie vorher eine spezielle Vorbereitungsklasse für Ausländer in Deutschland besucht? "[1] Ja" "[2] Nein" English: Did you attend a special preparation class for foreigners in Germany? "[1] Yes" "[2] No" See also: BISCGER, BISCGRAD, BISCGERC, BISCGC, BISCGCF, BISCGCFN █████ SOEP Survey Papers 418 263 SOEP v32 270 From which residence do you usually go to work?4 From this one From the other one Not applicable Since 2014, Western 15.1 Sources of Variables The information for the years 1994 and 1995 stem from the file BIOLELA. Information for later years are taken from the wave-specific data sets $LELA. In principle, SOEP respondents answer the Biography Questionnaire only once, so every person has only one record with wave-specific information in BIORESID. For fieldworkrelated reasons, very few people have answered the Biography Questionnaire twice. For these, the first interview is taken as relevant for BIORESID. Further cases are dropped if their information stems from an interview completed before 1994. 15.2 Population of Interest The BIORESID dataset as of wave 2015 contains information on 46,457 individuals, stemming from samples A-M. The data set is supplemented every year by new respondents filling in the supplementary Biography Questionnaire. █████ SOEP Survey Papers 418 270 SOEP v32 271 Table 1: Survey Year in BIORESID Survey Year n 1994 993 1995 1,075 1996 471 1997 480 1998 415 1999 2,039 2000 243 2001 8,816 2002 508 2003 2,319 2004 449 2005 299 2006 223 2007 2,232 2008 336 2009 196 2010 9,845 2011 6,860 2012 2,993 2013 398 2014 277 2015 5,050 Total 46,457 Status: up to wave BF (2015) Source: SOEP v32, doi: 10.5684/soep.v32 █████ SOEP Survey Papers 418 271 SOEP v32 272 Table 2: Samples in BIORESID Source: SOEP v32, doi: 10.5684/soep.v32 The information in BIORESID is treated as time-invariant. Although, in principle, it is possible to update the information on occupancy for some individuals on the basis of more recent information, we abstain from doing so for selectivity reasons. Sample n A Germans (West) 2,131 B Foreigners (West) 738 C Germans (East) 1,430 D Immigrants 1984-93 1,399 E Supplement 1998 1,871 F Innovation 2000 9,844 G High Income 2002 2,262 H Supplement 2006 2,198 I Incentivation 2009 1,870 J Supplement 2011 5,409 K Supplement 2012 2,579 L Family Types 9,956 M Migration Sample 4,770 █████ SOEP Survey Papers 418 272 SOEP v32 273 15.3 Variable List of the Data Set BIORESID Table 3: Description of the Data Set BIORESID Variable Name Content of the Variable Entries for Surveyed Person HHNR Original household number (invariant) HHNRAKT Current wave HH number (wave of biography interview) PERSNR Never changing person ID SYEAR Survey year Occupancy BRMOVEIN Year person moved in current dwelling Second Residence BRSECHOM Having a second residence BRSECREG Region of second residence BRSECUSE Use of second residence BRSECWOR Second residence at place of work Specification of Interview Situation BRINTA Type of interview INTID Identifier of the interviewer 15.4 Recent Changes in the Data Set In 2012 the Biography Questionnaire was integrated in the Individual Questionnaire. This revised questionnaire version was used on the new sample K and did not contain the question about the use of the second residence. As a result respondents from sample K who have a second residence are coded with “-5” on the variable BRSECUSE. In 2013 the new sample M was interviewed with a special version of the Biography Questionnaire which does not contain the questions on occupancy. Therefore, in 2013 and 2014 sample M is not part of the BIORESID dataset. In 2015 the sample M was interviewed on occupancy and is now a part of the BIORESID dataset. In 2014, Data from the SOEPrelated FID-study from the years 2010 to 2014 has been integrated in the SOEP (Sample L). Therefore, the number of cases in BIORESID has increased with SOEP Version 31 retrospectively for the years 2010 to 2013, compared with previous versions. █████ SOEP Survey Papers 418 273 SOEP v32 274 16 BIOEDU: Data on educational participation and transitions by Henning Lohmann and Sven Witzke The Socio-Economic Panel Study (SOEP) contains a broad range of variables which cover early child education and care, educational participation, educational degrees and other related topics. However, the respective questions are included in different questionnaires (e.g., personal questionnaire, household questionnaire, youth questionnaire) and the variables are not always in a format which is suited for longitudinal analyses. For instance, transitions such as school enrolment or entry into tertiary education are not documented in a single variable but can only be reconstructed by comparing the status in a wave t with the status in a wave t+1 (e.g., a transition into tertiary education took place if a person was not in university in wave t but is in university in wave t+1). Generating such variables is time-consuming and prone to errors. It is the aim of the BIOEDU dataset to provide ready-made variables on educational transitions and related topics in order to support analyses in a longitudinal perspective. The BIOEDU dataset is primarily based on prospectively collected information. Therefore, it contains most information for those persons who have been part of the survey population at the time when they have attended school or other educational institutions. In total the dataset contains information on 90,734 persons. This is the part of the SOEP sample for which we have observed an educational transition and/or an educational degree. For the larger part of this group we have observed an educational degree only (n=65,016). These are persons who have not been a part of the sample at the time when they participated in education or experienced educational transitions. The smaller part of the sample is more interesting for longitudinal analyse of educational participation. These are persons who lived in a survey household at the time of educational participation.1 Depending on the age of the individual the dataset contains variables on: • early child education and care (ECEC) • entry into primary school • transition to secondary school • first exit from secondary school • secondary school attendance after first exit from school • first entry into and exit from vocational training 1 Accordingly the first group is much older than the second group. At the time of the first observation in the sample the first group is on average 45 years old while the second group has an average age below 9 years. █████ SOEP Survey Papers 418 274 SOEP v32 275 • vocational training participation after first • first entry into and exit from tertiary education • tertiary education participation after first exit • highest ever obtained educational degrees and last observed educational participation The SOEP as a general household panel study is not specifically directed at the analysis of educational life courses. Nevertheless, right from the beginning of the panel in 1984 the survey instruments contained questions on the educational attainment of the respondents (aged 17 and older) and children younger than 17 years living in survey households. After more than 30 years of survey duration these data provide a precious source for the reconstruction of educational life courses. In the following we describe how we use these data to reconstruct educational transitions starting before school enrolment and up to postsecondary education. The reconstruction of transitions is primarily based on yearly information on educational participation (i.e. entry and exit reconstructed from changes in participation). For later transitions there is some more information as explicit questions on the end of general school, vocational training and tertiary education are part of the questionnaires (changes during the year prior to the survey, only for persons aged 17+ years, exception: already obtained degrees before age 17 in youth questionnaire). One remark on the variable naming conventions: The variable names always begin with “be” which stands for “biography education” (in analogy to other biography datasets). The third and fourth letter denote the type of transition or similar. For instance, t0 stands for variables on the first and t1 for the last year in child care. Variables on starting school contain a t2 and so on (up to t8= exit from tertiary education). Variables containing an x as the third letter contain information on the last observed year in education or on the highest educational degrees ever obtained (x4, x6, x8). Using this dataset you should keep in mind that most of the information covered by the dataset is not directly asked in the SOEP questionnaires but has been derived from the combination of several variables. In the process of reconstruction assumptions have been made which we try to describe as detailed as possible in our exhaustive documentation of the dataset (see below). The more these assumptions are based on additional knowledge, e.g. provided by strict institutional regulations, the better for the reconstruction of the transitions. The dataset covers transitions starting in early childhood up to tertiary education. For a part of the sample only one of these transitions or episodes is observed, for others the whole sequence from elementary education until the exit from tertiary education. The variable beinfo █████ SOEP Survey Papers 418 275 SOEP v32 276 provides an overview on the frequencies of these different patterns. In total the dataset contains information on more than 90,000 persons. This is the part of the SOEP sample for which we have observed an educational transition and/or an educational degree. For 597 cases we have full information (pattern 811111111). We have provided a number of variables where we documented the process of data generation and the sources where the data stem from (betXinfo, variables with suffixes _s or _g). You could use these variables as indicators of the degree of uncertainty in the process of the reconstruction of educational transitions. The less the variables could be reconstructed just using the basic algorithm (e.g., bet2info<>”0000|0|0000”), the higher is the degree of uncertainty. The same applies to long durations between an observed exit and the observation of a matching educational degree (e.g., a high value in bet6cert_g). It is certainly advisable to check if certain deviations in the process of data generation “explain” substantial results. E.g., if children living in households where interviewed in August (this information is provided in betXinfo) have a much higher propensity of starting school late (bet2agemo), this might just be a data artefact because it is difficult to decide if the information the household provided referred to the school year which just started in August or to the school year which just ended at the time of the interview. In general, you should expect that there are no such systematic measurement errors in the reconstructed variables. But if you want to have a closer look on potential biases you could use the respective variables which document the data generation process. This documentation describes a beta version of the dataset (v32_0.1). If you have comments or encounter while using the dataset, please let us know. This is just a brief introduction to the dataset. Way more detailed information (especially concerning the algorithms used to reconstruct information) is provided in the following publication which can be easily found on the DIW website. It is highly recommended for people interested in working with BIOEDU to have a look at it. Lohmann, Henning / Witzke, Sven (2011): BIOEDU (beta version): Biographical data on educational participation and transitions in the German Socio-Economic Panel Study (SOEP), DIW Data Documentation 58, Berlin. █████ SOEP Survey Papers 418 276 SOEP v32 277 17 LIFESPELL: Information on the Preand Post-Survey History of SOEP-Respondents by Martin Kroh and Hannes Kröger Prospective panel surveys typically face the problem that no information is available on units of analysis after respondents have left the survey. The SOEP team therefore regularly conducts drop-out studies to identify the whereabouts of attritors. These studies draw on official register data and allow us to determine whether a person is still living in Germany, is deceased, or has moved abroad since the last SOEP interview. The information is combined in a spell file LIFESPELL. This dataset reports all available information on the preand the post-survey history of all persons who have ever been a member of a SOEP household. The LIFESPELL file lends itself particularly to mortality research, migration research, and nonresponse research. It extends the period under investigation from the last SOEP interview to the last drop-out study and thus reduces the problem of selective observational probabilities of units of analysis. For users less familiar with spell files, we also provide a STATA code for converting a spell file into a long format file (person x year) at the end of this chapter. The file includes spells for every person ever living in a SOEP household. Each spell indicates one of the following states: • Living abroad • Living in Germany • Living in Germany + part of a SOEP HH1 • Deceased • No information about status The information comprised by the LIFESPELL file includes the following: • Year of birth 1 This code does not distinguish between active and inactive respondents, such as children and temporary nonrespondents. As long as interviewers can contact the household (even though an interview might not take place), we have information on the vital status of respondents. Therefore, all HH members are coded as being in the SOEP until the HH finally drops out of the SOEP and is no longer followed up by the interviewer. * indicates that the fact is only necessary for subgroups of the population █████ SOEP Survey Papers 418 277 SOEP v32 278 • Year of immigration* • Year of entry into SOEP • Year of exit out of SOEP • Year of emigration* • Year of death * indicates that the fact is only necessary for subgroups of the population The year of birth, and year of immigration are self-reported (retrospective) information from personal interviews (e.g., p-files). The year of entry and exit are taken from gross information reported by the interviewer (e.g., pbrutto-files). The year of emigration and the year of death can either come from related persons in the household (e.g., deceased-files), the interviewer (e.g., pbrutto-files), or, finally, from drop-out studies (i.e., population registers). The following register-based drop-out studies have been conducted in the past: a) 1992 drop-out study The study provides information about the dates of death or emigration, or one can infer that a person was still living in Germany until 1992. b) 2001 drop-out study The study provides information about the dates of death or emigration, or one can infer that a person was still living in Germany until 2001. Additionally, the study allows us to analyze regional mobility within Germany. It includes HH drop-outs from 1985-1998. c) 2006 drop-out study This study is not a register study, but a survey among nonrespondents. Attritors were contacted by mail and responded by mail. It allows us to include the year of death for those whose letters returned by post with the mark “deceased” or whose letters were answered by relatives or friends. Those who answered themselves can also be assumed to still have been living in Germany in 2006. No information is available for cases where there was no response. No information about emigration is given. The study includes drop-outs from 20012004. █████ SOEP Survey Papers 418 278 SOEP v32 279 d) 2008 drop-out study The latest of the studies provides information about the year of death and those still living in Germany in 2008. Emigration is captured insufficiently. It includes drop-outs from 1985-2006. █████ SOEP Survey Papers 418 279 SOEP v32