Sexual and Gender Minority (SGM) Research Meets Household Panel Surveys: Research Potentials of the German Socio-Economic Panel and Its Boost ...

Page created by Clifton Page
 
CONTINUE READING
Sexual and Gender Minority (SGM) Research Meets Household Panel Surveys: Research Potentials of the German Socio-Economic Panel and Its Boost ...
European Sociological Review, 2021, 1–15
                                                                                                     doi: 10.1093/esr/jcab050
                                                                                                                   Data Brief

Sexual and Gender Minority (SGM) Research

                                                                                                                                                         Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Meets Household Panel Surveys: Research
Potentials of the German Socio-Economic Panel
and Its Boost Sample of SGM Households
Mirjam M. Fischer 1,*, Martin Kroh2,3, Lisa De Vries3, David Kasprowski2,
Simon Kühne3, David Richter3 and Zaza Zindel3
1
 Institute of Sociology and Social Psychology, University of Cologne, Albertus-Magnus-Platz, 50923 Koeln,
Germany, 2Faculty of Sociology, University of Bielefeld, Universitätsstraße 25, 33615 Bielefeld, Germany
and 3German Institute for Economic Research, German Socio-Economic Panel, Anton-Wilhelm-Amo-
Straße 58, 10117 Berlin, Germany
*Corresponding author. Email: m.fischer@uni-koeln.de
Submitted June 2021; accepted August 2021

Abstract
There are numerous challenges to studying structural inequality in sexual and gender minority (SGM)
populations, from the difficulty of obtaining a representative sample to issues comparing data across
populations. This data brief illustrates how the largest household panel survey in Germany, the
German Socio-Economic Panel (SOEP), and its recent nationwide boost sample of SGM households,
Sample Q, have addressed these challenges. We discuss the case of the SOEP and its boost sample to
illustrate the untapped potentials that lie in other national household panels for research into the lived
experiences of SGM people, couples, and families. Based on the SOEP example, we show the poten-
tial for cross-sectional comparative, retrospective, and longitudinal life course research using the
detailed information on the extended families (parents, children, siblings, and relatives) of SGM
respondents in household panel surveys.

Background                                                                   This data brief illustrates how the largest household
Rapid changes in the legal and social environment                            panel study in Germany, the German Socio-Economic
around sexual and gender minority (SGM) people, cou-                         Panel (SOEP), and its recent nationwide boost sample of
ples, and families in many western societies are increas-                    SGM households, Sample Q, have addressed these
ing the demand for reliable, representative data to study                    challenges.
their lived experiences. There are numerous challenges                          First, in the past decades, SGM research has relied on
to studying structural inequality in SGM populations,                        convenience samples, in which participants are typically
from the difficulty of obtaining a representative sample                     recruited through SGM-specific venues, media outlets,
to issues comparing data across national populations.                        or social media (Kühne and Zindel, 2020). This reflects

C The Author(s) 2021. Published by Oxford University Press.
V
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial License (https://creativecommons.org/
licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
For commercial re-use, please contact journals.permissions@oup.com
Sexual and Gender Minority (SGM) Research Meets Household Panel Surveys: Research Potentials of the German Socio-Economic Panel and Its Boost ...
2                                                                   European Sociological Review, 2021, Vol. 00, No. 0

the classic challenge of sampling SGM populations: the         generally lies at about 1 per cent of all couples in the
lack of a complete list of all target population members.      data (e.g. Lengerer and Bohr, 2019), and compared to
In fact, such a list would not be desirable, as it would in-   census data, samples in general and small samples in
fringe on the privacy of a vulnerable group in society. A      particular come with increased uncertainty regarding
variety of strategies have been developed in recent years      population inference. Also, SGM-specific topics have
to facilitate population inference beyond convenience          primarily been addressed up to now by convenience

                                                                                                                             Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
samples (Valfort, 2017; OECD, 2019). Section 2 of this         sample studies in the context of health or experiences of
article describes a strategy for recruiting a probability-     discrimination (Pöge et al., 2020). The absence of SGM-
based sample of SGM households (SOEP Sample Q)                 specific topics in household surveys limits the possibil-
based on a large-scale screening that draws on a nation-       ities to generalize prevalence and effects, and do not
wide dual frame of landline and mobile phone numbers.          allow investigation of a wide range of mechanisms
    The second challenge lies in creating a sampling           underlying individual life-course differences between
frame. When pursuing the collection of data that are not       SGM and non-SGM people. This is a missed opportun-
biased by convenience recruitment strategies, social sci-      ity: despite some limitations, household surveys offer
entists have been turning increasingly to official popula-     potential for population inference and comparability,
tion registers, census data, and, where possible, to           both between groups in a single country and cross-
existing social surveys with household grids, which            nationally. The SOEP boost sample of SGM households
allow for the identification of cohabiting same-sex cou-       supplements an existing probability-based sample of res-
ples (Fischer, 2016). In some countries, it is possible to     idents in Germany, thereby addressing the issue of an
identify the entire population of same-sex couples in reg-     overall low number of observations. In Section 4, we de-
istered partnerships and marriages (e.g. Sweden; Kolk          scribe the potential for cross-sectional comparison with-
and Andersson, 2018) or the population of cohabiting           in the population of residents in Germany, for instance,
same-sex couples (e.g. The Netherlands; Steinmetz and          comparative analysis of SGM and cis-heterosexual peo-
Fischer, 2019). These data have been used to study life        ple across topics covered in the SOEP. In this section, we
transitions, such as divorce and parenthood (Kolk and          also describe SGM-specific topics surveyed by SOEP,
Andersson, 2018), economic outcomes (Aldén et al.,            such as minority stress and resilience.
2015; Antecol et al., 2008), and academic achievement              A fourth challenge lies in the fact that samples of the
of children with same-sex parents (Kabátek and Perales,       target population of SGM people do not lend themselves
2021). Yet, the identification of same-sex couples in          to explaining the social inequalities affecting this group,
these data remains error-prone (DeMaio et al., 2013).          since the data lack a comparison group of cis-
By definition, it is limited to people in partnerships and     heterosexual respondents. The study of social inequal-
a multiplicity of gender identities within relationships       ities rests on comparative research designs, and without
remains invisible (Naylor, 2020). While these strategies       a suitable comparison group, the direction and magni-
are appropriate for addressing some research questions         tude of unequally distributed resources and power can-
pertaining to partnered people, we argue that there is         not be quantified. The SOEP addresses both these issues
untapped potential in representative household surveys         by surveying SGM and non-SGM respondents with
for the field of SGM research beyond partnerships. In          identical questionnaires and by using an additional
Section 3 of this article, we discuss alternative              SGM-specific survey module with the target group.
approaches to identifying SGM households in existing               A fifth and final challenge to research on structural
household panel surveys, in particular, the approach           inequality in SGM populations arises from the fact that
taken by the SOEP in introducing direct questions on           longitudinal data on SGM populations are rare, particu-
sexual orientation and gender identity into the survey.        larly probability-based longitudinal data. This is a major
We also examine the number of interviews with SGM              problem, as key life transitions remain largely unexam-
people in the SOEP over time, before and after adding          ined. In their review of a decade of SGM family research
the boost Sample Q.                                            between 2010 and 2020, Reczek (2020) identified the
    A third challenge arises from the limited range of re-     lack of integration of SGM experiences across the life
search topics that can be addressed with census and            course, from childhood to old age, as one of the primary
register data from official statistical agencies.              gaps in existing SGM family research. Panel data also
Academically driven household surveys offer an alterna-        offer other advantages over cross-sectional data, in add-
tive, and have a key advantage of covering a broad range       ition to improving our fragmented understanding of so-
of topics. They also have disadvantages: the number of         cial forces in the lives of SGM people to date (Halaby,
observations of same-sex couples in these studies              2004; Brüderl et al. 2019). Lazarsfeld and Fiske (1938)
Sexual and Gender Minority (SGM) Research Meets Household Panel Surveys: Research Potentials of the German Socio-Economic Panel and Its Boost ...
European Sociological Review, 2021, Vol. 00, No. 0                                                                    3

noted in their classic article on panel studies that panel    information about sexual orientation could only be
data allow for analysis not only of short-term individual     inferred from same-sex partnerships and information on
change, but also of rare but critical life events (see also   marital status.
Elder et al., 2003). Studying these effects across people         In 2019, the SOEP began work to expand the exist-
who are affected by them and who are not may function         ing Samples A–P through a randomized boost sample of
as a ‘substitute for the use of a control group’              SGM households, Sample Q. Recruitment was done in

                                                                                                                           Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
(Lazarsfeld and Fiske, 1938, p. 597), thus improving          two phases. In the first phase, the SOEP screened a large
causal inferences from observational data.1 Also, infor-      probability-based sample of the general population
mation on the household contexts and relationships be-        using computer-assisted telephone interviewing (CATI),
tween members of a family in household panel surveys          whereby 70 per cent were contacted at their landline
provide unique perspectives on linked life courses            numbers and 30 per cent at their mobile phone numbers.
(Bengtson et al., 2005); in Germany, there is only one        By taking a dual-frame approach (e.g. Buskirk and Best,
study using the PAIRFAM panel to study intergenera-           2012), we are able to infer sampling probabilities based
tional ties of about one hundred people who currently         on the mixture of mobile and landline sub-samples.
are or ever were in a same-sex relationship (Hank and         Moreover, these sampling probabilities and resulting de-
Salzburger, 2015). In Section 5 of this article, we de-       sign weights were augmented by estimated non-response
scribe the potentials of SOEP data for life course re-        weights to compensate for refusals in subsequent field-
search on SGM people. In it, we report the number of          work phases (for details, de Vries et al., 2021).
interviews by cohort and period, the number of inter-             Of the overall total of 74,998 households that were
views by person, the number of critical life events           screened nationwide, 53,497 completed the screening
observed by prospective as well as retrospective survey       interview (see Table 1). SGM people were identified by
data, and finally, the number of family members inter-        asking whether they identified as heterosexual (attracted
viewed in the SOEP.                                           to the ‘other’ gender), lesbian/gay (attracted to the same
    In light of the challenges discussed here, the German     gender), bisexual (attracted to ‘both’ genders), or an-
Socio-Economic Panel (SOEP) and its boost Sample Q            other sexual orientation. The items were expanded by
of SGM households address a number of key challenges          providing a write-in option for respondents who chose
in the current data infrastructure. In Germany, the           ‘another sexual orientation’, which resulted in answers
SOEP and its SGM boost sample comprise the first data         such as queer, pansexual, asexual, polysexual, or demis-
source of this kind. Due to their unique characteristics      exual. All respondents who did not select ‘heterosexual’
and deliberate inclusion in a general household panel,        were screened in as part of the target group. Gender was
the SOEP data are likely to be of interest to the inter-      surveyed using an internationally established two-step
national SGM research community.                              method. In the first step, respondents indicate whether
                                                              they were assigned female or male at birth (in their ori-
                                                              ginal birth certificate). In the second step, respondents
Study Design                                                  report their current gender identity, which may be fe-
The SOEP is a nationwide, probability-based panel sur-        male, male, transgender, or none of these (again with a
vey of private households in Germany with a large num-        write-in option). People who identified as transgender or
ber of respondents (currently over 30,000 people in over      none of these genders were screened in, as well as people
20,000 households surveyed every year). The SOEP has          whose current gender identity did not align with the gen-
been running since 1984. All household members are            der they were assigned at birth. Besides the sexual orien-
surveyed on various life domains on an annual basis.          tation and gender identity (SOGI) items, the interview
The majority of the SOEP interviews are conducted by          included a number of other questions, such as family
professional interviewers in computer-supported, per-         situation, household composition, and a range of gen-
sonal interviews. The panel is based on a number of           eral opinion items.
probability samples (Samples A–P) of the general                  Based on the SOGI questions, 2,824 households were
German population as well as different boost samples of       identified as belonging to the target population of SGM
specific subpopulations such as migrants (for an over-        households. If people were identified as members of the
view, see Siegers et al., 2021). As a consequence, SGM        target population, they were invited to take part in the
respondents have always been part of the SOEP due to          SOEP survey in the second phase. The final gross boost
the random sample of the entire population of                 sample of households with complete contact informa-
Germany. However, the SOEP did not include a direct           tion and willingness to participate in the main survey
question on sexual orientation until 2016. Before that,       consisted of 835 households. Overall, 477 households
4                                                                                             European Sociological Review, 2021, Vol. 00, No. 0

Table 1. Results of the random telephone screening and response for 2019 SOEP boost Sample QTotal screening
interviews

                                                  N             Per cent of total           Per cent of valid           Per cent of identified   Per cent of gross
                                                                screening interviews        screening interviews        target group             sample Q

74,998                                               100

                                                                                                                                                                       Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Non-response to SOGI items                        21,501        28.7
Completed screening interviews                    53,497        71.3                        100
  Not in SGM target group                         50,673                                     94.7
In SGM target group                                2,824                                      5.3                       100
  Refusal to participate in SOEP                   1,731                                                                 61.3
  No contact information provided                     70                                                                  2.5
  Assumed false positive screenings                  188                                                                  6.7
Final gross boost                                    835                                                                 29.6                    100
   sample Q
  Not eligible                                         17                                                                                          2.0
  Refusal                                             170                                                                                         20.3
  Other not completed interviews                      171                                                                                         20.6
Completed interviews                                  477                                                                                         57.1

    Bold numbers correspond to the entire samples after screening, selecting the target group and the final gross boost sample.
    Source: Gross telephone screening Sample Q.

Table 2. Sexual orientation and gender identity in the 2019 SOEP boost Sample Q

                                                         Sample Q total                                                            SGM only

                                    N               % un-weighted                 % weighted                 N              % un-weighted              % weighted

Sexual orientation
  Heterosexual                     127                   22.52                       26.49                   0                    0.00                   0.00
  Lesbian/gay                      231                   40.96                       33.32                 231                    53.60                  45.98
  Bisexual                         182                   32.27                       34.41                 182                    42.23                  47.48
  Other                             17                   3.01                        4.67                   17                    3.94                   6.44
  No information                    7                    1.24                        1.10                    1                    0.23                   0.09
Gender identity
  Male                             302                   53.55                       53.89                 240                    55.68                  56.55
  Female                           253                   44.86                       44.77                 182                    42.23                  41.61
  Diverse                           8                    1.42                        1.16                   8                     1.86                   1.60
  No information                    1                    0.18                        0.18                   1                     0.23                   0.25

    Note: The column Sample Q total refers to the overall sample; the column SGM only refers to respondents who reported their sexual orientation to be lesbian/gay,
bisexual or other, and/or their gender identity to be transgender or other.
   Source: SOEP v36 2019.

were successfully interviewed, which corresponds to a                                  identified as heterosexual. These respondents also did
response rate of 57 per cent. The fieldwork period of the                              not report a minoritized gender identity nor did they re-
main data collection was from April to November 2019.                                  port a history of transitioning genders (i.e. they are cis-
    Tables 2 and 3 report some basic descriptive statistics                            gender heterosexual in all but one case). The right col-
for Sample Q in 2019. Table 2 gives an overview of the                                 umn shows the distribution of sexual and gender iden-
distribution of sexual and gender identities in the boost                              tity for those respondents who reported a minoritized
Sample Q. The left columns show this distribution for                                  sexual and/or gender identity (SGM only). Table 3
all respondents in Sample Q, both raw and weighted.                                    shows socio-demographic information for Sample Q,
Despite the initial screening on SOGI items, 22 per cent                               again for the entire boost sample and SGM people only.
of respondents indicated in the main interview that they                               Around 40 per cent of respondents in the boost sample
European Sociological Review, 2021, Vol. 00, No. 0                                                                                                               5

Table 3. Socio-demographics in the 2019 SOEP boost Sample Q

                                                                Sample Q total                                                   SGM only

                                             N             % un-weighted            % weighted              N            % un-weighted               % weighted

Age (in years)

                                                                                                                                                                      Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
18–24                                       71                  12.59                  17.07                55                 12.76                    18.01
25–34                                       108                 19.15                  22.60                91                 21.11                    25.58
35–44                                       92                  16.31                  16.41                83                 19.26                    19.49
45–54                                       116                 20.57                  16.74                90                 20.88                    16.17
55–64                                       84                  14.89                  11.26                64                 14.85                    10.91
65þ                                         93                  16.49                  15.93                48                 11.14                    9.84
Education
No degree/in education                       8                  1.42                    1.57                3                   0.70                    0.80
Lower secondary degree                      71                  12.59                  12.42               42                   9.74                    9.13
Middle secondary degree                     160                 28.37                  27.87               122                 28.31                    27.43
Advanced technical degree                   48                  8.51                    8.17               37                   8.58                    8.72
Higher secondary degree                     243                 43.09                  42.12               205                 47.56                    47.68
Other                                       20                  3.55                    5.20               12                   2.78                    4.02
No information                              14                  2.48                    2.65               10                   2.32                    2.23
Partnership
No                                          229                 40.60                  42.57               177                 41.07                    43.28
Yes                                         333                 59.04                  57.13               252                 58.47                    56.31
No information                               2                  0.35                    0.30                2                   0.46                    0.41
Children under 16 in Household
No                                          507                 89.89                  85.54               391                 90.72                    85.96
Yes                                         57                  10.11                  14.46               40                   9.28                    14.04

   Note: The column Sample Q total refers to the overall sample; the column SGM only refers to respondents who reported their sexual orientation to be lesbian/gay,
bisexual or other, and/or their gender identity to be transgender or other.
   Source: SOEP v36 2019.

are unpartnered, which illustrates the necessity to imple-                          reported sexual orientation only in recent years, if at all.
ment measures beyond partnership inference.                                         For instance, UKHLS and HILDA introduced measures
                                                                                    of sexual orientation in 2012 and SOEP in 2016 (Uhrig,
                                                                                    2013; Wooden, 2014; Kühne et al., 2019). In 2016,
SGM Respondents in the SOEP, 1984–2019                                              respondents were asked whether they identified as het-
Existing national household panel surveys, such as the                              erosexual (that is, attracted to the other gender), lesbian/
Panel Study of Income Dynamics in the United States of                              gay (that is, attracted to the same gender), bisexual (that
America (PSID), the Socio-Economic Panel in Germany                                 is, attracted to both genders), or another sexual orienta-
(SOEP, see Goebel et al., 2018), the Household, Income                              tion. Pretesting of the measure showed that people
and Labour Dynamics in Australia (HILDA) survey, the                                needed the explanations in parentheses to understand
Swiss Household Panel (SHP), and Understanding                                      the identity labels, but unfortunately, this led to the use
Society in the United Kingdom (UKHLS) include sub-                                  of terminology that reinforces outdated binary notions
stantial numbers of SGM respondents by design. The                                  of gender. The SOEP did not survey gender-diverse pop-
limitation of these longitudinal data on the national                               ulations prior to 2019. In the 2019 survey, trans*
population lies in their possibilities for identifying SGM                          respondents could be identified in the boost sample
respondents.                                                                        only; the core questionnaire included a binary measure
   Partnership-inferred sexual orientation based on                                 for gender. As of 2021, gender will be measured in a
household grids misses a substantial part of the general                            non-binary way, and respondents will be asked about
sexual minority population, and it glosses over any gen-                            any changes that may have been made to their birth cer-
der diversity within these partnerships. While many                                 tificate over the life course. At present, the number of
household panel surveys have been going on for decades,                             trans* observations that can be identified in the panel is
some of them started to introduce measures of self-                                 very small (N ¼ 38).
6                                                                        European Sociological Review, 2021, Vol. 00, No. 0

    Considering that sexual and gender diversity remain             information from which sexual orientation could be
stigmatized in contemporary societies, the disclosure of            inferred (bars in the middle). This remains the only
minoritized sexualities and genders is considered sensi-            source of information for any respondent who attired
tive, and prior research has shown evidence of mode                 prior to 2016. In more recent panel waves, less than 10
effects on the reporting of sexual orientation (Kühne              per cent of adult respondents reported information on
et al., 2019). Not every survey mode is equally suited to           cohabitation or civil partnerships. In contrast, the num-

                                                                                                                                Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
obtaining valid SOGI data. The SOEP is conducted by                 ber of respondents who self-reported their sexual iden-
computer-assisted personal interviewing. For sensitive              tity has increased over time to more than 80 per cent in
questions on topics around SOGI, interviewers turn the              2019 (bars on the outer left).2 For about 10 to 20 per
laptop toward the respondents and let them answer the               cent of adult respondents, we do not have any informa-
respective questions in privacy. This strategy is aimed at          tion on sexual orientation for any of these three meas-
minimizing social desirability bias.                                ures (bars on the outer right).
    Figure 1 shows the distribution of different sexual                 Based on all sources of information on sexual orien-
orientation information (self-identification, same-sex              tation and gender identity combined, the SOEP panel in
partnership, same-sex civil partnership) for the respond-           2019 includes 7,762 interviews of 1,522 adult respond-
ents in SOEP in each survey year. In this figure, we con-           ents between the years 1984 and 2019. Each of these
sider sexual orientation to be time-invariant. Before self-         respondents either cohabitated with a same-sex partner
reported sexual orientation was introduced in the survey            and/or self-identified as non-cis-heterosexual. The num-
in 2016, particularly in the 1980s and 1990s, cohabiting            ber of available interviews that can be ascribed to SGM
partnership data and legal civil partnerships (legally rec-         respondents increases after 2000 with the legal introduc-
ognized since 2001) were the primary sources of                     tion of same-sex civil partnerships and almost doubles

Figure 1. Information used to classify adult respondents by sexual orientation

Source: SOEP v36 2019.
European Sociological Review, 2021, Vol. 00, No. 0                                                                       7

                                                                                                                              Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Figure 2a. Number of Interviews with SGM Respondents by Panel Wave (Total 7,762).
Source: SOEP v36 2019.

to a total of 1,077 annual person-interviews with SGM           Modifying such questions involves a low implementa-
respondents in 2019 with the introduction of Sample Q           tion cost and is crucial to avoid alienation of SGM
(Figure 2a).                                                    respondents. The SOEP has opted to add the category of
                                                                social mother or social father to the existing option of
                                                                biological mother or biological father. Households com-
Questionnaire Content
                                                                plete an additional questionnaire when a child was born
The coverage of topics in the SOEP is broad, ranging            into the household since the previous interview. This
from employment, income, and education to health,               questionnaire used to be aimed at the biological mother
well-being, social relationships, fertility, and attitudes.3    of the child, and the questions pertained to the circum-
Respondents in the SGM boost Sample Q completed the             stances of the birth. As of 2021, this questionnaire first
same questionnaire as the core SOEP sample in 2019. In          asks about the relationship to the new child (i.e. bio-
addition, the SGM boost Sample Q also completed an              logical, social or adoptive parenthood, or other relation-
SGM-specific module. Therefore, SOEP combines a                 ship). If the respondent completing the questionnaire
wide range of general topics (and their comparability           gave birth to the child, there is another set of items ask-
with non-SGM respondents) with SGM-specific topics.             ing about the circumstances of conception (i.e. whether
Table 4 in the appendix gives an overview of topics that        medical assistance was provided in the form of hormo-
overlap between the questionnaire of the boost Sample           nal therapy or IVF treatments). These minor alterations
Q and all other SOEP Samples A–P as well as the SGM-            make the questionnaire inclusive to SGM respondents at
specific questions.                                             minimal cost.
   When integrating SGM samples into existing house-
hold surveys, it must be considered that surveys tend to
use gendered language and gendered concepts at various          Life Course Research Potentials
points in their questionnaires. Questions pertaining to         The extended period of observation of SGM respondents
parenthood may use terms like ‘mother’ and ‘father’ and         in SOEP holds unique analytical potential for research
presume a biological relationship with the child.               across the life course and across generations
8                                                                                         European Sociological Review, 2021, Vol. 00, No. 0

                                                                                                                                               Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Figure 2b. Number of Interviews of Respondents with Different Sexual Orientations (N ¼ 1,522 SGM respondents and N ¼ 99,509
all other respondents).
Source: SOEP v36 2019.

Table 5. Number of SGM Respondents by age and cohort

                                                                              Age groups

Cohorts                   17–19              20–29             30–39              40–49          50–59        60–69        70þ         Total

1914–1939                                                                            8           42             68         112           230
1940–1949                                                         6                 69           159           272         211           717
1950–1959                                      2                  34               130           276           221                       663
1960–1969                   18                221                378               710           537                                   1,864
1970–1979                   33                211                678               631                                                 1,553
1980–1989                   92                650                684                                                                   1,426
1990–2001                  506                803                                                                                      1,309
Total                      649               1,887             1,780              1,548          1,014         561         323         7,762

    Note: The total of 7,762 observations refers to 1,522 unique SGM adult respondents.
    Source: SOEP v36 2019.

(Mannheim, 1928; Ryder, 1985). The 7,762 interviews                                    Section 3 already documents the number of inter-
on SGM respondents cover cohorts between 1914 and                                   views with SGM respondents from a perspective of
2001, with age at the time of interview ranging from 17                             repeated cross-sections. Owing to the long period of ob-
to 88 years (see Table 5).                                                          servation in the SOEP, some respondents have been
European Sociological Review, 2021, Vol. 00, No. 0                                                                       9

Table 6. Family-of-origin characteristics reported by SGM respondents (N ¼ 972), SOEP 1984–2019

                                                                        SGM boost      SOEP Samples      Total     Outcome
                                                                         Sample Q
                                                                                            A–P

Parental socio-demographic information

                                                                                                                              Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
  Birth year father                                                         377             498          875        90.0%
  Birth year mother                                                         383             522          905        93.1%
  School education (father)                                                 368             465          833        85.7%
  School education (mother)                                                 381             498          879        90.4%
  Professional education (father)                                           387             491          878        90.3%
  Professional education (mother)                                           376             517          893        91.9%
  Religious affiliation (father)                                            387             429          816        84.0%
  Religious affiliation (mother)                                            388             471          859        88.4%
  Nationality (father)                                                      383             499          882        90.1%
  Nationality (mother)                                                      394             516          910        93.6%
  Country of origin (father)                                                 60             411          471        48.5%
  Country of origin (mother)                                                 56             428          484        49.8%
  Father’s current location                                                 401             514          915        94.1%
  Mother’s current location                                                 413             529          942        96.9%
Occupational and class classifications
  Current occupational classification KldB10 (father)                       335             434          769        79.1%
  Current occupational classification KldB10 (mother)                       266             361          627        64.5%
  ISEI (father)                                                             335             432          767        78.9%
  ISEI (mother)                                                             266             360          626        64.4%
  ISCO (father)                                                             335             432          767        78.9%
  ISCO (mother)                                                             266             360          626        64.4%
  Treiman’s standard int. Occupational prestige score (ISCO; father)        335             432          767        78.9%
  Treiman’s standard int. Occupational prestige score (ISCO; mother)        266             360          626        64.4%
  EGP Class scheme (father)                                                 270             417          687        70.7%
  EGP Class scheme (mother)                                                 240             357          597        61.4%
  EGP Class scheme (combined, at least one parent information)              347             482          829        85.3%
Parental information during adolescence (at the age of 15 years)
  Main residence during childhood                                           433             531          964        99.2%
  Still living at residence of childhood                                    433             514          947        97.4%
  Years cohabiting with parent(s)                                           429             503          932        96.0%
  Father’s occupational status at age 15                                    380             467          847        87.1%
  Mother’s occupational status at age 15                                    387             481          868        89.3%
  Fight with father at age 15                                               434             434          868        89.3%
  Fight with mother at age 15                                               434             436          870        89.5%
Siblings
  Information on siblings general                                           431             528          959        98.7%
  Having at least one sibling                                               369             463          832        86.8%
  At least one sister                                                        1              103          104        12.5%
  At least one brother                                                       1              300          301        36.2%
  Twins                                                                      0               10          10         0.01%

  Source: SOEP v36 2019.

interviewed in more than 30 different panel waves so far          comparable across SGM and cis-heterosexual respond-
(Figure 2b). While the peak in the number of interviews           ents, with a mean of 8 and 5 panel waves, respectively.
with SGM respondents results from the boost sample,                  In the SOEP data, a number of critical life events can
the number of panel waves per respondent appears                  be observed in SGM respondents. For instance, there are
10                                                                European Sociological Review, 2021, Vol. 00, No. 0

447 unemployment spells in 345 SGM respondents.              time. Initial cross-sectional reports based on these data
Moreover, we can trace a total of 847 births to SGM          confirm that there is considerable inequality between
respondents, and we observe 692 partnerships during          SGM and non-SGM populations in different life
their participation in SOEP. Beyond partnerships, there      domains (De Vries et al., 2020; Kasprowski et al.,
are other family relationships that can be analysed. For     2021). These descriptive reports are suggestive of the po-
instance, 274 respondents in the SOEP are mothers of         tential of these data for studying people with minori-

                                                                                                                          Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
SGM (adult) children; 229 respondents are fathers of         tized sexual and gender identities.
SGM (adult) children. In addition, a total of 447 SOEP           The SOEP project fills an important gap because sur-
respondents are siblings of an SGM respondent. Hence,        vey respondents provide their sexual and gender identity
the SOEP data offer unique analytical potential for the      directly, hence including singles and non-cohabiting
study of social ties in families and partnerships over the   couples and a suitable comparison group within the
life course. Information on parents, siblings, children,     same dataset. The boost sample ensures a large enough
and partners can be obtained not only from interviews        group size to make meaningful comparisons with the
with the respective people in the same household but         non-SGM population. The survey covers general topics
also from biographical information reported by               such as work, family, and health as well as population-
respondents, for instance, when they were 15 years of        specific topics such as openness about sexual orienta-
age. Table 6 shows the available biographical informa-       tion, minority stress, and experiences of sexual stigma
tion on parents and siblings of SGM respondents.             and discrimination. The inclusion of the SGM sample
                                                             into the general panel means that there will be longitu-
                                                             dinal data that are unique and direly needed. Overall,
Conclusions
                                                             data with all of these features are exceptionally rare. In
Using the example of SOEP and its boost sample of            light of this, the SOEP data may be of interest to the
SGM respondents, we have illustrated the untapped            larger field of LGB studies in Europe, beyond the scope
potentials of household panels for research into SGM         of Germany alone.
populations. The SOEP has taken advantage of the op-             The data are available to employees and students at
portunity to maximize its panel structure to make empir-     universities and other research institutes for non-
ical longitudinal inferences with sufficient statistical     commercial scientific research, and are provided based
confidence in service of a rigorous SGM Sociology. Due       on a data use contract (www.diw.de/en/soep).
to its probability-based sampling, the SOEP already
included SGM people even before the introduction of
the SGM boost sample. In 2016, respondents were asked                              Endnotes
about their sexual orientation, making lesbian, gay, and     1   Lazarsfeld and Fiske (1938) also note that the survey
bisexual panellists visible in the core SOEP samples (A–         experience of respondents in longitudinal surveys
P). This created analytical possibilities through the use        may increase the reliability of measurement (for sup-
of retrospective existing data on these respondents. The         portive empirical evidence, see Kroh et al., 2016).
latest boost sample of SGM households (Sample Q) is              The main disadvantages of panel data are the possi-
integrated into the larger panel as of 2019 and offers           bility of selection bias due to attrition as well as
sufficient supplementary observations to obtain unique           panel conditioning, that is, the reactivity of
probability-based, longitudinal survey data with annual          responses to previous waves of the panel (Lazarsfeld
interviews of SGM respondents as the panel progresses.           and Fiske, 1938).
    We have illustrated the unique potential of house-       2   Besides the raw data on indicators of sexual orienta-
hold panel surveys to provide detailed accounts of fam-          tion, the SOEP team provides two generated sum-
ily constellations, not just of couples and SGM parents          mary indicators of sexual orientation: sexor is
but of all family relationships (siblings, parents, rela-        categorical and time-invariant and distinguishes be-
tives). These data make it possible to conduct SGM fam-          tween (probably) lesbian/gay and bisexual respond-
ily research in context of wider family relationships, a         ents on the one hand and (probably) heterosexual
key avenue for future research outlined by Reczek                respondents on the other hand. The second variable,
(2020). The identification and inclusion of SGM people           sexorinfo, distinguishes the source of information
in panel surveys take advantage of this potential. We            that was used for this classification. Researchers
provided examples of de-gendering of survey items that           who wish to restrict their samples to respondents
may alienate SGM respondents. They were minor and                who self-reported their sexual orientation can do so
inexpensive, and they maximize comparability across              with the sexorinfo variable.
European Sociological Review, 2021, Vol. 00, No. 0                                                                                   11

3 Many of these questions are surveyed annually;                        Weekly Report (36/2020). Berlin: DIW. https://doi.org/10.
  others are part rotating modules. We recommend the                    18723/diw_dwr:2020-36-1.
                                                                      DeMaio, T. J., Bates, N. and O’Connell, M. (2013). Exploring
  online SOEP Companion to anyone interested in
                                                                        measurement error issues in reporting of same-sex couples.
  using these data (http://companion.soep.de/). It con-
                                                                        Public Opinion Quarterly, 77, 145–158.
  tains an overview of all topics covered, variables,                 Elder, G. H., Johnson, M. K. and Crosnoe, R. (2003). The emer-
  technical documents, fieldwork reports, and prac-                     gence and development of life course theory. In Mortimer, J.

                                                                                                                                           Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
  tical tips on how to get started.                                     T. and Shanahan, M. J. (Eds.), Handbook of the Life Course.
                                                                        Boston, MA: Springer, pp. 3–19. https://doi.org/10.1007/978-
                                                                        0-306-48247-2_1.
Funding                                                               Fischer, M. M. (2016). Identifying same-sex couples in
This data collection was funded by the Federal Ministry of              cross-national survey data: a comparison of same-sex couples’
Education and Research, Berlin, Germany (BMBF) [ref.                    demographic and socio-economic traits in six European coun-
01UW1803A, 01UW1803B] and it is part of a larger research               tries. In Meuleman, R., Kraaykamp, G. and Wittenberg, M.
project into Sexual and Gender Diversity in Germany (GesMin)            (Eds.),    Nederland       in    Context:      Verschillen    en
[ref. 01UW2002A, 01UW2002B]. Both projects are conducted                Overeenkomsten. Den Haag: DANS Symposium.
in collaboration between the German Institute for Economic            Goebel, J. et al. (2018). The German Socio-Economic Panel
Research and the University of Bielefeld. The DFG funded the            (SOEP). Journal of Economics and Statistics, 239, 345–360.
research network Sexual Orientation and Gender Identity in            Halaby, C. N. (2004). Panel models in sociological research: the-
Germany (SOGI-GER) Bundling Interdisciplinary Expertise to              ory into practice. Annual Review of Sociology, 30, 507–544.
stimulate interdisciplinary scientific collaboration and research     Hank, K. and Salzburger, V. (2015). Gay and lesbian adults’ re-
based on the SOEP-LGB data [ref. FI 2490/1-1].                          lationship with parents in Germany. Journal of Marriage and
                                                                        Family, 77, 866–876.
                                                                      Kabátek, J. and Perales, F. (2021). Academic achievement of
References                                                              children in same- and different-sex-parented families: a
Aldén, L. et al. (2015). Effect of registered partnership on labor     population-level analysis of linked administrative data from
  earnings and fertility for same-sex couples: evidence from            the Netherlands. Demography, 58, 393–418.
  Swedish register data. Demography, 52, 1243–1268.                   Kasprowski, D. et al. (2021). LGBTQI people in Germany face
Antecol, H., Jong, A. and Steinberger, M. (2008). The sexual            staggering health disparities LGBTQI people in Germany face
                                                                        staggering health disparities. In DIW Weekly Report
  orientation wage gap: the role of occupational sorting and
                                                                        (5/2021). Berlin: DIW. https://doi.org/10.18723/diw_
  human capital. Industrial and Labor Relations Review, 61,
                                                                        dwr:2021-5-1.
  518–543.
                                                                      Kolk, M. and Andersson, G. (2018). Two decades of same-sex
Bengtson, V. L., Elder, G. H. Jr. and Putney, N. M. (2005). The
                                                                        marriage in Sweden: a demographic account of developments
  lifecourse perspective on aging: linked lives, timing, and his-
                                                                        in marriage, childbearing and divorce. Social Policy and
  tory. In Johnson M. L. (Ed.), The Cambridge Handbook of
                                                                        Family Dynamics in Europe, 2, 1–37.
  Age and Ageing. New York, NY: Cambridge University Press,
                                                                      Kroh, M., Winter, F. and Schupp, J. (2016). Using person-fit
  pp. 493–509.
                                                                        measures to assess the impact of panel conditioning on reli-
Brüderl, J., Kratz, F. and Bauer, G. (2019). Life course research
                                                                        ability. Public Opinion Quarterly, 80, 914–942.
  with panel data: an analysis of the reproduction of social in-      Kühne, S., Kroh, M. and Richter, D. (2019). Comparing
  equality. Advances in Life Course Research, 41. https://doi.          self-reported and partnership-inferred sexual orientation in
  org/10.1016/j.alcr.2018.09.003.                                       household surveys. Journal of Official Statistics, 35, 777–805.
de Vries, L. et al. (2021). SOEP-Core-2019: Design, nonres-           Kühne, S. and Zindel, Z. (2020). Using Facebook and Instagram
  ponse, and weighting in the Sample Q (queer) (No. 940).               to recruit web survey participants: a step-by-step guide and
  SOEP Survey Papers.                                                   application. Survey Methods: Insights from the Field, Special
Buskirk, T., and Jonathan B. (2012). “Venn Diagrams,                    Issue, 1–14.
  Probability 101, and Sampling Weights Computed from Dual-           Lazarsfeld, P. and Fiske, M. (1938). The “panel” as a new tool
  Frame Telephone RDD Designs.” JSM Proceedings, Survey                 for measuring opinion. Public Opinion Quarterly, 2,
  Research Methods Section, American Statistical Association,           596–612.
  3696–3710.                                                          Lengerer, A. and Bohr, J. (2019). Is there an increase in same-sex
de Vries, L. et al. (2020). LGBTQI people on the labor market:          couples in Germany? Theoretical considerations and empirical
  highly educated, frequently discriminated against. In DIW             findings. Zeitschrift Fur Soziologie, 48, 136–157.
12                                                                        European Sociological Review, 2021, Vol. 00, No. 0

Mannheim, K. (1928). Das Problem der Generationen. Kölner             Until 2019). SOEP survey papers 960: Series C. Available
  Vierteljahreshefte Für Soziologie, 7, 157–184.                      from: https://www.diw.de/surveypapers.
Naylor, L. A. (2020). Counting an invisible class of citizens: the   Steinmetz, S. and Fischer, M. M. (2019). Surveying persons in
  LGBT population and the U.S. Census. Public Integrity, 22,           same-sex relationships in a probabilistic way—an example
  54–72.                                                               from the Netherlands. Journal of Official Statistics, 35,
OECD. (2019). Society at a Glance 2019: OECD Social                    753–776.
  Indicators. Paris: OECD Publishing.                                Uhrig, S. C. N. (2013). An examination of poverty and sexual

                                                                                                                                      Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Pöge, K. et al. (2020). Die gesundheitliche Lage von lesbischen,      orientation in the UK (No. 2014-02). University of Essex,
  schwulen, bisexuellen sowie trans- und intergeschlechtlichen         Colchester: Institute for Social and Economic Research work-
  Menschen [The health situation of lesbian, gay, bisexual, trans      ing paper series.
  and inter people]. Journal of Health Monitoring, 5, 1–120.         Valfort, M.-A. (2017). LGBTI in OECD countries: a review. In
Reczek, C. (2020). Sexual- and gender-minority families: a 2010        OECD Social, Employment and Migration Working Papers,
  to 2020 decade in review. Journal of Marriage and Family, 82,        Vol. No. 198 (Issue June). pp. 185. https://doi.org/10.1787/
  300–325.                                                             d5d49711-en.
Ryder, N. B. (1985). The Cohort as a concept in the study of so-     Wooden, M. (2014). The Measurement of Sexual Identity in
  cial change. In Mason, W. M., Fienberg, S. E. (Eds.), Cohort         Wave 12 of the HILDA Survey (and associations with mental
  Analysis in Social Research. New York, NY: Springer. https://        health and earnings). HILDA PROJECT DISCUSSION
  doi.org/10.1007/978-1-4613-8536-3_2.                                 PAPER SERIES, University of Melbourne. Retrieved from
Siegers, R., Steinhauer, H. W. and Dührsen, L. (2021). Data           https://melbourneinstitute.unimelb.edu.au/assets/documents/
  Documentation: Documentation of Sample Sizes and Panel               hilda-bibliography/hilda-discussion-papers/hdps114.pdf.
  Attrition in the German Socio-Economic Panel (SOEP) (1984

Mirjam Fischer is a postdoctoral researcher in
Sociology at the University of Cologne. Her re-                      Lisa de Vries is a doctoral researcher at the Faculty
search interests include inequlities in (mental)                     of Sociology and a member of the Bielefeld
health, well-being, social networks, and family ties                 Graduate School in History and Sociology at
between people with minoritized sexual and/or gen-                   Bielefeld University. Current research interests com-
der identities and the cis-heterosexual majority, and                prise discrimination and inequality in the labor
sampling and measurement issues related to                           market, career paths of sexual and gender minor-
LGBTQI* populations.                                                 ities, and LGBTQI*-parent families.

Martin Kroh is a Professor of Quantitative                           David Kasprowski is a doctoral researcher in the
Methods of Empirical Social Research at the                          Socio-Economic Panel Research Infrasturcture at
Faculty of Sociology of Bielefeld University and re-                 the German Institute for Economic Research in
search fellow at the Socio-Economic Panel at the                     Berlin. He is a member at the Berlin Graduate
German Institute for Economic Research (DIW                          School for Social Sciences at the Humboldt
Berlin). His research interests include the link be-                 University. Current research interests comprise sex-
tween social and political inequality. He also works                 ual orientation and gender diversity as stratifying
on political attitudes and behaviors in the context                  factors, particularly of education and labor market
of families and generations.                                         outcomes.
European Sociological Review, 2021, Vol. 00, No. 0                                                         13

Simon Kuehne is a Professor of Applied Social Data     researcher and survey manager at DIW Berlin. He is
Science at the University of Bielefeld. His research   interested in the development of emotions, well-
interests include survey methodology, social media,    being, and life satisfaction across the adult life-span
and dimensions of social inequality such as gender,    and the influence of life-events on the development
sexuality, or health. His work has been published in   of life satisfaction and well-being.

                                                                                                                 Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
journals, such as Proceedings of the National
Academy of Sciences (PNAS), Sociological Methods       Zaza Zindel is a doctoral researcher at the Faculty
& Research, and Journal of Official Statistics.        of Sociology and a member of the Bielefeld
                                                       Graduate School in History and Sociology at
David Richter is a Professor of Survey Research at     Bielefeld University. Her research interests include
the Department of Educational Science and              survey research and methodology, social media re-
Psychology at the Freie Universität Berlin and a      search, discrimination, and social inequality.
14                                                                      European Sociological Review, 2021, Vol. 00, No. 0

Appendix

Table 4. Questionnaire content of SOEP SGM boost Sample Q

                                            Overlap with SOEP Samples A–P                   Unique to the boost Sample Q

                                                                                                                                  Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Identification of sexual orientation   Sexual orientation (self-identifying),
                                         same-sex partner, marital status:
                                         registered civil partnership
Sexual orientation                                                                    Disclosure of sexual orientation,
                                                                                        identification with sexual orientation,
                                                                                        sexual orientation, and the society
Identification of gender identity                                                     Gender, birth gender (registered)
Gender identity                                                                       Disclosure of trans* identity,
                                                                                        identification with trans* identity,
                                                                                        trans* identity and the society
Discrimination                         Experienced discrimination on the basis        Areas of discrimination
                                         of 10 different characteristics
Family situation                       Marital status, relationship status,
                                         partner in same household, financial
                                         support, changes in family situation
Children                               Detailed information about every child in
                                         the household depending on age:
                                         parenting goals, parenting role,
                                         parenting style, big five, Strengths and
                                         Difficulties Questionnaire (SDQ),
                                         temperament, Vineland Adaptive
                                         Behaviour Scales, child care, school,
                                         pregnancy and birth, health
Partner                                Detailed information about partner in
                                         household
Household                              Housing situation, neighbourhood,
                                         detailed information about every
                                         Person in the Household, household
                                         composition, household income
Childbirth preferences                 Desire to have children                        Realization, barriers, contact to donor
Situation in the last year             Paid work, maternity leave/statutory
                                         parental leave, statutory period of
                                         care, officially registered unemployed,
                                         education or training, finish education,
                                         left a job
Employment situation                   Employment status, occupational status,
                                         job search, occupational change, job
                                         position, sector of business/industry,
                                         type of employment contract, type of
                                         company, income, sources of income,
                                         working time, extra hours, duty stroke,
                                         side jobs, volunteer jobs, bonuses or
                                         extra pay, retirement/pension
Asset                                  Residential properties, real properties,
                                         building loan contract, financial
                                         investments, companies, vehicles,
                                         tangible properties, liabilities,
                                         inheritance

                                                                                                                   (continued)
European Sociological Review, 2021, Vol. 00, No. 0                                                             15

Table 4. (Continued)
                                        Overlap with SOEP Samples A–P           Unique to the boost Sample Q

Health and illness                 Current health, sleeping hours, diseases,
                                     severely disabled, partially incapable
                                     of work, trips to the doctor, hospital
                                     stay, sick leaves, health insurance

                                                                                                                    Downloaded from https://academic.oup.com/esr/advance-article/doi/10.1093/esr/jcab050/6400293 by guest on 29 December 2021
Timing schedule                    Everyday Life, leisure time
Well-being
  Cognitive well-being             General life satisfaction, satisfaction in
                                     11 further domains
  Affective well-being             Affective well-being, emotional well-
                                     being
Risk aversion                      Risk aversion in general
Attitudes and opinions             Big five, justice, attitudes towards
                                     LGBT*s, political interest, political
                                     opinion, worries, social networks
Biography/socio-demographics       Origin, nationality, migration, stays
                                     abroad, occupation, childhood, educa-
                                     tion, parents, grandparents, siblings,
                                     children
Migration history, religion        Citizenship, residency status, language
                                     skills, contact to Germans and non-
                                     Germans, attachment to place of resi-
                                     dence, contacts abroad, religion
You can also read