Clear in its core, blurred in the outer contours: culturally normative conceptions of the family in Germany

Over the course of several decades and generations, we have observed a complex set of changes in private living arrangements (Charles et al. 2008; Lesthaeghe 1995) and in the patterns of family trajectories (Huinink 2013; Brückner and Mayer 2005), with considerable cross-national variation in the pathways and in the timing of these changes (Duranton et al. 2009; Reher 2004). Such processes have often been interpreted as a consequence of increasing individual freedom to pursue subjective needs and preferences, referring, e.g. to the concepts of individualisation (Beck 1992) or, in demography, to the ‘second demographic transition’ (Lesthaeghe 1995). Accordingly, subsequent changes in living arrangements as well as cross-country and individual differences have often been explained from a utilitarian economic perspective, referring to individual resources and constraints (Becker 1993 [1981]) and their interaction with welfare states and their institutional settings (Esping-Andersen 1990). Often, such interpretations gave the impression that these were the only remaining relevant influential contexts for action in contemporary late-modern societies.

Recent research continues to reveal evidence that this ‘post-modern’ paradigmatic-theoretic frame is too one-sided and that changes in the ways in which family lives are forged are much less radical. Behavioural patterns still follow, to a large degree, religious belief systems, norms and social expectations of individuals’ social networks (Craig and Powell 2012; Vaaler et al. 2009), with family networks having a particularly strong influence on family-related biographic decisions (Moscatelli and Bramanti 2017). Family practices are also often guided by stereotypes and behavioural scripts (Eldén 2012). These again are likely to transmit unconscious culturally normative conceptions: People have culture-specific normative ideas of how family lives and family trajectories should look. These ideas seem to be shaped by the societies’ predominant religious orientation (Voicu et al. 2009) and history (Kalmijn 2007; Reher 2004). In terms of how they affect behaviour, such conceptions interact with individual resources and institutional contexts (Bernardi et al. 2008). And they set boundaries against radical social change and political control of behaviour. Several empirical findings reveal that culturally normative conceptions remain firmly in place, even if their influence may in some ways appear more subtle today than it used to until the mid-twentieth century. Examples include the high fertility rate that persists despite low institutional support for childcare in English-speaking countries (McDonald and Moyle 2010), the persistently asymmetric gender-based division of labour despite the levelling of educational status between men and women (Lewis et al. 2008), the strong preference for two children despite one child offering a higher efficiency of investment in the ‘quality of children’ (Sobotka and Beaujouan 2014) or the strong orientation towards the nuclear family despite the supposed individual freedom of arranging private living (Huinink 2014; Charles et al. 2008).

This raises the question as to which culturally normative conceptions of the family are prevalent today. Research has shown that there have been changes in the ideas of what family relations and family organisation are expected to look like, i.e. relationships are now expected to be less authoritarian (du Bois-Reymond 1995), and the division of responsibilities is expected to be more egalitarian (Bühlmann et al. 2010). But aside from a two-child norm that prevails in Europe, defining two as a ‘normal’ number of children (Sobotka and Beaujouan 2014), there is little knowledge regarding culturally normative conceptions of which living arrangements people envision when they speak of ‘family’. This question is addressed below, based on a multi-method approach with data for Germany.

As our analyses are of an exploratory nature, they do not test hypotheses or one particular theory. Nevertheless, they are theory-driven in the sense that there are a range of theoretical approaches which are inspiring our research question and which provide orientation for our operationalisations and interpretations. These are outlined below.

The phenomena we are interested in are part of what contemporary sociology of knowledge calls ‘common-sense knowledge’ or ‘knowledge in everyday life’ (Berger and Luckmann 1991 [1966]). This approach can inform us about the nature of the studied culturally normative conceptions, their emergence, the mechanisms of their reproduction and their impact on social life. Berger and Luckmann consider society to be a dialectical process between objective and subjective reality. By ‘subjective reality’, Berger and Luckmann mean everything that is considered to be knowledge of reality and as such inter-subjectively shared in a given society: everything that is ‘taken for granted as reality by the ordinary members of society in the subjectively meaningful conduct of their lives. It is a world that originates in their thoughts and actions, and is maintained as real by these’ (Berger and Luckmann 1991 [1966]: 33), which means: It is socially (re-)constructed. Despite remaining undefined by Berger and Luckmann, we can take the term ‘culture’ as a synonym for ‘subjective reality’ in this understanding. This cultural universe of shared meanings, rules, habits and convictions enables people to successfully act and interact with each other in society. It is constructed in processes of externalisation, objectivation and internalisation. Despite its socially constructed nature, human culture is taken for granted in the same way as physical nature.

The phenomena and mechanisms Berger and Luckmann are describing obviously present a challenge for empirical research, since they are not directly observably and they occur to some extent unconsciously. Consequently, theoretical models instructing research, in particular quantitative research, are rare. One of the few examples is the concept of ‘cultural life scripts’ (Janssen and Rubin 2011; Berntsen and Rubin 2002). This concept assumes that people have internalised scripts of how to live a ‘normal’ life course with ideal ages for specific biographical events. In relation to these scripts, people aim to be ‘on time’ or feel themselves to be ‘off time’ if their own life trajectory deviates from the script.

A more comprehensive, non-issue-specific approach for capturing the influence of cultural knowledge is provided by the concept of ‘leitbild’ (Lück et al. 2017; Diabaté and Lück 2014). A leitbild (German for ‘guiding image’, plural: leitbilder) is a ‘bundle of collectively shared and visually imagined conceptions of normality – with “normality” implying that something is personally desired, socially expected, and/or presumably very widespread (i.e. common and self-evident)’ (Lück et al. 2017: 66). The concept integrates the three mechanisms of attitudes, social norms and frames or routines, which are assumed to be active simultaneously, and it formulates implications for empirical research. These include the implication that a leitbild is too complex to be measured by one survey item, and should rather be operationalised based on principal component analysis or similar multivariate analyses. And they include the implication that the elements of such an operationalisation should not only consist of what respondents personally consider desirable, but also of what they feel is expected of them and considered as desirable by the society surrounding them (Lück et al. 2017). According to Berger and Luckmann, leitbilder are each small segments of common-sense knowledge. Nevertheless, they remain complex conceptions of one particular field of everyday life, such as the family.

Before their theoretical elaboration, leitbilder had been incorporated by Birgit Pfau-Effinger into her concept of ‘gender’ or ‘care arrangements’ (Pfau-Effinger 1996, 2004). This concept focuses on the issue of dividing paid work and care work between mother, father and the state. In this context, it assumes that there are nation-specific arrangements of leitbilder or ‘cultural models’ (Pfau-Effinger 2004: 382) that define how this division ought to be organised. Together with accordingly designed institutions, these leitbilder shape individual behaviour – a process that corresponds to the interaction between objective and subjective reality, as described by Berger and Luckmann (1991 [1966]). Based on macro level research, Pfau-Effinger identifies several leitbilder for care arrangements in Western Europe, such as the ‘housewife model (of the male breadwinner marriage)’, the ‘part-time carer model (of the male breadwinner marriage)’ or the ‘dual breadwinner model with partner-shared childcare’ (Pfau-Effinger 2004: 383f.).

After the Second World War, like most societies in Europe, West Germany was strongly oriented towards the nuclear family and the complementary division of labour within the couple, known as the ‘male breadwinner model’. The emphasis on the importance of the family was particularly strong, given the experience of indoctrination by the fascist regime and the hope that children could be protected from ideologisation if the family was maintained as a strong institution (Szynka 2016). This rationale was supported by the influence of the churches, in particular, the Roman Catholic Church. It resulted in the principle of subsidiarity, according to which the family received – and receives – strong institutional support, without the government being legitimately justified in interfering with the family’s private sphere and the socialisation of children, unless absolutely necessary.

This principle is mirrored by many institutional arrangements, such as the (West) German constitution, which states that ‘marriage and family shall be particularly protected by the public order’ (Deutscher Bundestag 2017: article 6). Unlike most other European countries, Germany has a ministry exclusively for family affairs. Until 2007, parental leave included financial support for a three-year break for a parent (usually the mother) to take care of the child. Also until recently, the West German infrastructure for public childcare was limited to kindergarten serving only children aged 3–6, with restricted and inflexible opening hours. School hours typically ended at noon, and mothers were expected to prepare lunch. There are, however, many forms of direct payment, e.g. child allowances (Kindergeld) of roughly 200 € per month and child. Supporting the family as an institution also implied – and partly still implies – supporting marriage and the ‘male breadwinner’ arrangement of distributing paid and unpaid work. Married couples in a male breadwinner constellation are supported financially by generous tax breaks (Ehegattensplitting) as well as by free health insurance for non-employed spouses and under-age children.

This cultural and institutional context led to the fact that the change of family and gender roles proceeded more slowly in West Germany than in economically comparable countries in the North-West of Europe, although there were profound changes in Germany too from the late 1960s onwards (Dorbritz 2008). In theoretical frameworks for European international comparison, West Germany therefore tends to be placed in conservative categories, along with Southern European states: In Esping-Andersen’s (1990) typology of welfare regimes, it belongs to the ‘conservative-corporatist’ regime, and in the categorisation of Lewis and Ostner (1994), it resembles a ‘strong male breadwinner state’.

In the meantime, East Germany developed a quite different family conception. Here the influence of the church has traditionally been weak – in particular the Roman Catholic church plays a minor role (Klüsener and Goldstein 2016); and the socialist era further established a secular society. Also, East Germany was unable to supply its economy’s demand for labour through immigration, as West Germany did, and therefore was highly dependent on women – including mothers – engaging in the labour market. Accordingly, the full-time working mother was established as a role model, supported by a public all-day care for children from an early age on (Kinderkrippe, Kindertagesstätte) (Leitner et al. 2008). And while this was true in a similar way in several Eastern European socialist countries, East Germany also developed highly egalitarian gender attitudes (Bauernschuster and Rainer 2012) as well as pluralised family forms (Dorbritz 2008). Births out of wedlock have been and continue to be much more frequent in the East than in the West. East German women become mothers more often and at younger ages than West German women do (Dorbritz 2008).

There is a continuity of differences between East and West Germany (Goldstein and Kreyenfeld 2011; Dorbritz 2008). Nevertheless, the two parts have also converged somewhat since 1990. Particularly, West Germany has been subject to profound political and culturally normative change. Persistently low fertility brought attention to unresolved tensions in the work–life balance, and put pressure on policymakers to resolve them. Since reunification, West Germany has been confronted with the liberal family conception of East Germany as well as East German mothers’ strong orientation towards full-time employment and short child-related career breaks. Accordingly, several institutional reforms have been initiated. In 2007, a new regulation for paternal leave came into force (Elterngeld & Elternzeit), encouraging shorter leaves and male participation by paying for 12 months plus 2 extra months if both parents share the leave. The payment is tied to the last income (in general 67%), reducing the opportunity costs of leave for people with a high income. Since August 2013, parents have been entitled to a place in public childcare for children aged 1 and 2. These reforms do encourage new care arrangements. In Pfau-Effinger’s typology of care arrangements, West Germany has developed from a ‘housewife model (of the male breadwinner marriage)’ towards a ‘part-time carer model’ (Pfau-Effinger 2004, 2005), whereas a ‘dual breadwinner model with external childcare’ has rather been the norm in East Germany for several decades (Pfau-Effinger 2005). However, political conflicts regarding the right family conception and the right family policies persist until today. And it is unclear to what extent the political reforms are mirrored in a (preceding or subsequent) change in cultural conceptions. So far, little research on family conceptions has been carried out. Existing studies, such as the one by Mühling et al. (2006), show for West Germany that cultural conceptions shifted towards greater gender equality; however, when it comes to parental roles, they are persistently traditional and differences between generations are not very pronounced (Mühling et al.2006: 71).

In this paper, we use a multi-method design, combining quantitative and qualitative analyses. We proceed in three steps: First we identify conceptions of the family in society as well as their prevalence. This is based on quantitative data. Then, we enhance these understandings by adding further aspects that characterise or constitute a family, such as the quality of relationships among family members, based on qualitative data. Finally, we examine the degree to which these conceptions are rooted in particular social groups or strata, again using quantitative data and multivariate analyses.

The quantitative data are mainly drawn from the ‘Family Leitbild’ survey (FLB 2012) conducted in Germany in 2012, which was inspired by the theoretical leitbild concept described above. A sample of n = 5000 participants, representative of the population aged 20–39, was interviewed on the phone using the CATI method in interviews of roughly 30 minutes. Weights were constructed for correcting moderate non-response biases and to compensate for disproportionality in the use of landline and mobile phone numbers. The survey was conducted by the German Federal Institute for Population Research (BiB). Interviews were carried out by the polling institute TNS Infratest (Lück et al. 2013).

Partly quantitative but predominantly qualitative data are taken from the study ‘Family in Pictures’ (FiB 2015), again inspired by the leitbild concept and again carried out by the Federal Institute for Population Research (BiB) in Germany. In this study, n = 101 participants with German citizenship of 16 years or older were selected using a snowball system, following pre-defined quotas designed to ensure an approximate balance of women and men, of East and West Germans as well as of five stages in the family biography: living together with parents, living as a single in one’s own household, living with a steady life partner but without children, living with one’s own children in a household, being a parent in an ‘empty nest’ stage. The youngest participant was 16, the oldest 79 years old. Each participant drew a picture of what, in his or her subjective view, constituted a ‘proper’ family, and was then asked about the conception underlying their drawing in a guided telephone interview, as described in more detail below (Lück et al. 2018).

Due to the focus on German residents aged 20–39, the age structure of the survey sample is more homogeneous and younger than in the qualitative sample. This may also lead us to expect that the cultural conceptions reported in the survey will be somewhat more homogeneous and liberal than in the latter case. However, in the qualitative sample too a majority of roughly two thirds of participants are relatively young adults in their 20s, 30s or 40s. Both samples have a gender bias, over-representing women, which is corrected by a weighting factor in the survey sample.

The quantitative analyses are based on one question and a set of items in which the individual respondent’s conception of a family was measured by proposing seven different living arrangements and asking for each of them whether or not the respondent would consider it a family:

• a couple, woman and man, married, with children

• a couple, woman and man, not married, with children

• a gay or lesbian couple with their own children

• a woman with a child and a new partner, not married

• a woman with a child, without a partner

• a couple, woman and man, married, no children

• a couple, woman and man, not married, no children.

Respondents were able to answer ‘yes, this is a family’ or ‘no, this is not a family’. With fewer than 1% of responses missing for each item, evidently these items were properly understood and very well received by the respondents. The constellations represented by the items vary across several dimensions, i.e. partnership, institutionalisation through marriage and parenthood. Therefore, certain patterns of response across all seven items can reveal a principle which the respondent in question followed, indicating a particular understanding of what ‘family’ in general means to this respondent. Since the plausibility of the assumption that a certain conception of the family underlies a pattern of response often depends on a precise combination of all seven answers, the operationalisation will rely on such exact combinations – and not on indicators generated by cluster analysis or similar multivariate analyses.

In the (predominantly) qualitative study ‘Family in Pictures’, the participants were sent a set of documents and materials by post. These included, among other things, six coloured pencils as well as a blank sheet of paper with the instruction: ‘In your imagination, what constitutes a ‘proper’ family? Please draw it!’.

After the research team received a drawing by post, a group of researchers formulated an ad hoc interpretation of the family conception underlying the drawing. The participant was then contacted by phone for a subsequent interview. In this semi-structured telephone interview, the participant was asked first of all to describe her or his basic thoughts and intentions when reading the instruction and starting to draw. The answer to this question helped to decide whether or not the drawing did represent a subjective abstract conception of a family at all (or, for example, merely the participant’s own family of origin). Then detailed questions followed regarding the meaning of each detail of the drawing, so that the participant was prompted to self-interpret her or his own drawing.

The interpretation of the participant’s conception of a family was based on two foundations: Inasmuch as the participant’s self-interpretation was considered authentic, this self-interpretation was adopted, since it was able to reveal insights no other source could have offered. Inasmuch as the self-interpretation seemed to be biased and used for ex post censorship of politically incorrect aspects, the researcher’s ad hoc interpretation was adopted. As a final result, cultural conceptions of the family were described, based on subjective conceptions that had been mentioned repeatedly across all participants.

Both studies aim to measure collectively shared subjective conceptions of the family, based on the aforementioned concept of the leitbild (Lück et al. 2017) as a common theoretical foundation. However, these are not just different techniques for measuring the exact same phenomenon. They have slightly different foci. Firstly, the instruction in the study ‘Family in Pictures’ encourages participants to draw one rather than several pictures of a family, so that the full range of possible variations is not necessarily captured. Secondly, the instruction asks for a picture of a ‘proper’ family, which may prompt respondents to describe the core of a family conception rather than the edges. As a consequence, we expect both operationalisations to reveal valid measurements of family conceptions in Germany, albeit with an emphasis on the core of those conceptions in the qualitative study and on their diversity in the survey. Comparison of the two datasets may give an idea of the range in which conceptions of the family are identifiable.

Below, we combine the data from the studies described above, aiming to utilise each of their strengths and compensate for their weaknesses, to collect as much evidence as possible for describing the culturally normative conceptions of the family in Germany.

This implies that the nuclear family as described by Talcott Parsons (1955) in the mid-twentieth century still represents the core of the family conceptions. It is far from being the only living arrangement considered a family, which may be a result of a cultural change during recent decades; but it does appear to be the only living arrangement for which it remains beyond question that it deserves to be called a family. All or almost all contemporary conceptions of the family may incorporate many living arrangements beyond the nuclear family, but there are none omitting the nuclear family. Family conceptions may have broadened their outer contours in comparison to what they presumably looked like five decades ago, but their core has not shifted.

The criterion that seems to be the most relevant for including a living arrangement in the conception of what constitutes a family is parenthood. In addition to married and unmarried heterosexual couples with children, also all other living arrangements with children receive particularly high ratios of agreement (between 82.4% and 88.1%). So, the existence of any parent–children relationship is sufficient for considering a living arrangement a family.

The fact that not only married, but also unmarried heterosexual couples with their own children receive an agreement of nearly 100% seems to imply, at first sight, that marriage is a criterion that has lost its relevance for considering a living arrangement a family. That, however, is only partially true. Being married is not a necessary condition, but it can be a sufficient condition for considering a couple a family, in the sense that it substitutes for the criterion of parenthood. Twice as many respondents agree that a childless but married heterosexual couple is a family than is the case for unmarried couples. This may mean that marriage is often interpreted as a precursor of a first birth or at least as a sign of intending to have children. It also may mean that, in the eyes of many Germans, the institutionalisation of the couple bond through legal marriage changes the quality of the relationship as such.

Among the three items representing ‘unconventional’ living arrangements with children – ‘a gay or lesbian couple with their own children’, ‘a woman with a child and a new partner, not married’ and ‘a woman with a child, without a partner’ – there are marginal differences in agreement. These indicate that the presence of two adults having a role as social parent for the child is also a criterion of some importance: The single mother does not fulfil this criterion and receives slightly less agreement; the same-sex couple does, and is declared a family more often. The item ‘a woman with a child and a new partner’ could be understood either as a stepfamily in which the new partner takes on the parental role, or as a single mother who has a partner with no interest in the child. Accordingly, its evaluation has an in-between position. The difference between same-sex couples and heterosexual couples does not seem to make any significant difference.

A last remarkable finding is that even the item with the least agreement is considered a family by a third of the respondents. So, despite the fact that the nuclear family still marks the core understanding of what constitutes a family and the lowest common denominator, it seems that the vast majority of personal conceptions of the family are very inclusive and mirror the increase in heterogeneity of family forms over the course of the last decades. However, in order to reveal what shape these understandings take, we need to take into account the patterns of response across all seven items.

In searching for meaningful combinations of answers across the seven items, presented in Figure 1, we combine two strategies: a ‘deductive’ one and an inductive one. The ‘deductive’ strategy implies formulating conceptions of the family that we expect to exist, then explicating which combination of responses across the seven items would represent each understanding and, in a third step, ascertaining the prevalence of these patterns in the data. Our expectations in this respect are not derived from theory, as should be the case in deductive research in a strict sense; they are merely based on the theoretical interpretations of the findings presented in the previous section. Nevertheless, the opportunity to define in advance how a conception of the family that we expect to exist should translate into a pattern of response, does increase the likelihood that our interpretation of this pattern is accurate, should analyses reveal that it exists.

Nevertheless, as our analyses are exploratory and there is little empirical knowledge on which they can rely, it is likely that the data reveal different patterns of response than expected. Consequently, we enhance the ‘deductive’ strategy by adding an inductive one. We do not restrict ourselves to ascertaining the prevalence of the expected patterns, but we ascertain all possible combinations of responses, then identify those that are frequent and, in a third step, try to identify underlying family conceptions also for those that we had not foreseen.

Which family conceptions do we expect to identify? Our interpretations of the findings presented in the previous section suggest that parenthood, marriage, a combination of parenthood and marriage, as well as the co-presence of a child and two parents may be relevant criteria for considering a living arrangement a family. The findings also suggest that there may be an understanding that any living arrangement can be called a family, co-existing alongside the conception that the term ‘family’ only includes nuclear families in the sense of married heterosexual couples with their own children, or possibly also unmarried couples with their own children. Consequently, we expect to find the seven conceptions of what constitutes a family presented in Table 1 as an equivalent for a set of hypotheses.

Table 1 also shows the combination of responses that we would expect each family conception to lead to and that can be used as an operationalisation, thus identifying the underlying conception. In most cases, it is possible to define exactly one combination of responses that a family conception should result in. There is one exception: People with the ‘child-/two-parent-centred’ understanding who believe that living arrangements with children and two parents are families may come to different decisions regarding the item (d) ‘a woman with a child and a new partner, not married’. The new male partner of the mother may be interpreted as a caring stepfather, or he may be interpreted as a person with no strong interest in the child, a fact we cannot measure in the data. Accordingly, this one understanding can lead to two patterns of response, either including or excluding item (d).

Below we investigate the data by looking for the prevalence of the aforementioned expected patterns of response as well as by looking for any frequent combination of responses that may reflect a family conception we have not yet been aware of. For answering seven items with ‘yes, this is a family’ or ‘no, this is not a family’, there are 128 possible combinations, which implies a statistically expected ratio of 0.8% for each combination. Out of the 128 possible combinations, 56 are observable in the data. However, most of these combinations are rare. So, even if they should represent conceptions of the family held by individual respondents, it is unlikely that they represent relevant cultural conceptions in German society. If we reduce the list of patterns to those that were realised by at least 1% of the weighted sample (approx. 50 respondents) then 16 patterns remain, covering 93.4% of all interviewees. If we reduce the list to those combinations that were realised by at least 2% of the weighted sample (approx. 100 respondents), then nine patterns of response remain, covering 85.2%. These nine patterns seem relevant enough to interpret them as potential manifestations of underlying conceptions of the family. They are presented in Table 2, together with those understandings which we expected to find (cf. Table 1) but which do not fulfil the criterion of representing at least 2% of the sample. Altogether, the conceptions presented in Table 2 represent 88.0% of the sample.

In comparison to our previously defined expected patterns as presented in Table 1, the results are in some cases surprising: While the three most liberal understandings – the ‘inclusive’, the ‘child or marriage-centred’ and the ‘child-centred’ family conception – appear to be very frequent, the three most selective understandings are hardly found at all. Only 0.3% of respondents evaluate the items according to the ‘marriage-centred’ principle that a marriage defines a family. Only 0.4% support an ‘exclusive’ understanding according to which only a married heterosexual couple with their own children can be considered a family. If we add up respondents with an ‘exclusive’ understanding and those with a ‘semi-exclusive’ conception who also consider unmarried heterosexual couples with their own children as families, we reach a ratio of 1.4%. This is still below the statistically expected ratio (of 1.6%) any two combinations should have if answers were distributed completely evenly.

These findings suggest that the narrow understandings that only nuclear families or only marriages qualify as family are not present to any significant degree. However, the results of the qualitative study ‘Family in Pictures’ that will be discussed later on in more detail, put this finding into perspective: Here the ‘exclusive’ conception that the family is first of all a nuclear family is much more present. Roughly one in four participants draws or at least mentions it. We can therefore summarise by saying that young adults in Germany refuse to restrict their idea of a family to married heterosexual couples with their own children, but as a culturally established common core of individual understandings of what constitutes a family, the nuclear family is certainly present (see Section 5.3).

The three most liberal conceptions we expected to exist in society (cf. Table 1), by contrast, exist as individual conceptions and represent a highly significant share of the population: Around 19% agree that any living arrangement with children can be called a family (‘child-centred’). Another 23% agree that any living arrangement with children or with a married couple should be called a family (‘child- or marriage-centred’). Another 25% even share the widest understanding possible that any living arrangement with two or more people qualifies as a family (‘inclusive’). Remarkably, the prevalence increases with the degree of inclusiveness of the family conceptions. Together these three make up for two thirds (66.8%) of the population aged 20–39, and can be regarded as the dominant personal understandings of what constitutes a family. No other combination of responses is nearly as frequent as these three.

The ‘child-/two-parent-centred’ conception is more demanding than those described above. It is the most exclusive one that we observed in our data in significant numbers. Taking both patterns of response that represent this conception (based on the two possible interpretations of item (d) ‘a woman with a child and a new partner, not married’, as described above), a total of 6.2% share this logic. The majority (5.0%) includes item (d), interpreting the ‘new partner’ as a social stepfather; a minority (1.1%) does not.

Aside from these expected conceptions of the family, the findings reveal five further combinations of responses that are quite frequent although sometimes difficult to interpret:

• We find a ‘couple-centred’ conception (2.7%) that follows the principle that all living arrangements with a couple are families.

• Another pattern could be called the ‘complex’ conception (2.4%) as it represents the somewhat elaborate understanding that all living arrangements with children and two parents, as well as childless marriages, are families.

• The ‘heteronormative’ conception (2.2%) seems to reflect the idea that living arrangements with children plus childless marriages are families – except those involving a same-sex couple.

• The pattern we called ‘community-centred’ conception (2.0%) appears to have the underlying principle that living arrangements in which all members have close intimate ties to each other are families. It only excludes the ‘woman with a child and a new partner’, presumably because in this living arrangement there is one relationship – the relationship between the child and the new partner – that one would not expect to be intimate.

• The last pattern to be mentioned is the most frequent one (4.0%) although it does not reflect any consistent principle that we are able to recognise: Presumably, it does not represent a conception of the family in its own right, but rather a common inconsistent deviation from one or several patterns.

Further patterns of response, particularly those with ratios above 1%, have also been investigated for underlying principles; but we were unable to identify any further consistent comprehensible principle. For the remaining 12% of response combinations, we assume that these represent the absence of any consolidated conception of the family.

Aside from conceptions regarding which composition families normally have or ideally should have, participants of the study ‘Family in Pictures’ also emphasise other aspects that characterise or constitute a family, such as the quality of relationships among family members, the family practices in regard to shared time, and the housing conditions in which a family lives.

Independently of the question concerning which constellation of people constitutes a family, participants of the study ‘Family in Pictures’ consider several social criteria to be important in describing what a ‘proper’ family looks like, as the qualitative interviews reveal. To fully comprehend conceptions of the family, we must also take these criteria into account.

One issue frequently addressed by the participants is the quality of relations among family members. For many participants, it is more important how people are related to each other socially and emotionally than their biological family relationship. Family requires close emotional ties and a feeling of togetherness. It implies solidarity and readiness to support one another. Thus, family provides feelings of security, satisfaction and happiness. In the drawings, the quality of relationships is very often symbolised by family members holding each other’s hands as well as by smiling faces, a sun that symbolises happiness or by a heart that stands for love.

The family relationships are explicitly or implicitly linked to the family functions that participants expect a ‘proper’ family to provide. One frequently mentioned function is mutual help and support, another one is care for children or other needy members. Care is mostly thought of as care for children, and is barely distinguished from the functions of bringing up and socialising children. Sometimes also reproduction is mentioned as a function that constitutes families.

These functions are crucial in the sense that they justify the constellations people consider as families. Emphasising the function of mutual support, for example, typically is combined with an ‘inclusive’ conception that all constellations of people living together are families. The ‘child-centred’ conception that a family is a living arrangement with children is typically derived from the understanding that a family is supposed to raise, socialise and care for children. The ‘child-/two-parent-centred’ conception, requiring not only children but also two parents, is based on the argument that ‘it is crucial for educating children that there is not only one contact person’. In their rationale for including the extended family network, participants refer to the network’s role in providing support for educating children. Even for considering a pet a family member, participants provide functional arguments: ‘The pet is supposed to provide guidance for taking on responsibility’. The rationale for a ‘heteronormative’ conception of the family, excluding same-sex couples, may refer to the family’s function of reproduction or by presuming that complementary gender roles, including a male breadwinner and a female caretaker, are important for families.

Aside from family functions and the quality of family relationships, some participants mention the biographical time limitation as characteristic of families. In this perspective, the family is perceived as a biographical phase. This understanding is typically combined with an emphasis on the parent–child relationship, and argues that the family only exists as long as the children are young, live at home and are taken care of by the parents. When children are grown up, move out and become independent, the family of heritage stops existing. On the other hand, other participants emphasise that the family continues to exist after the children have moved out – typically combining this view with the idea of families being networks of mutual support.

A further social criterion for family corresponds to the theoretical approach of family practices (Morgan 1999). On this understanding, the family is constituted by shared time and shared activities. Only when family members actively spend time with each other, when they interact or communicate, does family come alive as a social relationship. Nevertheless, not every interaction or communication between human beings is considered as constituting family, but only the interaction between family members. This implies that there also needs to be an independent conception of who qualifies as a family member in principle. Also various understandings of the family’s functions can be combined with the idea of family being constituted by actively shared time.

A last set of criteria that can be found in the data involves the living and housing conditions of families. Many participants think of the family as a unit that shares a common household which is presumably considered a precondition for the provision of care work or for other family functions. Several people envision the family as ideally living in a lone standing house, either in the countryside or with a garden. For the house, the provision of protection may be an underlying idea. For house, garden and countryside, it is argued that this is a setting which allows children to grow up peacefully and be sheltered. House and garden are typically combined with a rather restrictive conception of the family in terms of its composition. Finally, participants also associate a ‘proper’ family with a certain level of wealth, providing financial security.

In a final step, we wish to disaggregate the prevalence of the various family conceptions by the segments of society in which they are particularly frequent or uncommon. In order to do so, we turn back to our quantitative data source and test the influence of socio-economic and socio-demographic characteristics on the respondents’ family conceptions in a multinomial logistic regression analysis (Table 3). This method is common and seems suitable to us for identifying the determinants of the various values of a nominal dependent variable, such as the respondents’ various conceptions of family. The model presented (in Table 3) is the result of a series of variations and improvements.

The dependent variable distinguishes four conceptions of the family, for which case numbers allow a multivariate analysis, plus one residual category summarising all other patterns of response. The most liberal and most frequent ‘inclusive’ understanding that every living arrangement is a family serves as a reference category. The other three conceptions distinguished are: (1) child- or marriage-centred (living arrangements with children or a marriage constitute a family), (2) child-centred (living arrangements with children constitute a family) and (3) a combination of child-/two-parent-centred, semi-exclusive and exclusive (living arrangements with a couple and children constitute a family). A further category has not been distinguished due to low case numbers.

The overall model fit is barely acceptable, but remains unsatisfactory for causal analyses (Nagelkerke’s R² = .109). Forgoing the residual category (o) in the dependent variable, it would be reduced even further to R² = .089. This indicates that the various family conceptions are generally observable in all parts of society; their prevalence does not vary much by social position. Nevertheless, all understandings can be characterised based on a number of significant explanatory variables that indicate in which strata they are rooted in particular.

The ‘inclusive’ understanding that all living arrangements constitute a family is somewhat more difficult to characterise, based on the presented regression analysis, since it is the reference category. It is characterised in relation to and in reversion of all other categories. Accordingly, this conception is particularly common among women socialised in East Germany. It is somewhat more frequent among older people and among people who have a partner, but are not married. And it is characteristic of non-religious people with no tertiary education.

The ‘child- or marriage-centred’ understanding that living arrangements with children or a marriage constitute a family is particularly common among married people, which is likely to be causally related to the fact that marriage is a criterion for defining the family. It is also typical of West Germans with higher secondary or tertiary education. And it is somewhat more common among younger people, men and people without a partner.

The ‘child-centred’ understanding that living arrangements with children constitute a family is also particularly typical of university-educated West Germans, and in that respect is similar to the previous conception. Furthermore, it is common among parents, which is again likely to be linked to the emphasis on parenthood as a criterion for defining family. It is somewhat more frequent among men, young people and among people living in urban centres.

The combination of ‘child-/two-parent-centred’, ‘semi-exclusive’ and ‘exclusive’ conceptions, indicating that living arrangements with a (married) couple and (their own) children constitute a family, is characteristic of men, religious people and Roman Catholics. It is more frequent in West Germany and somewhat more common among people with tertiary education.

Having a migrant background does not have a significant influence on an individual’s conception of the family, except that people with such a background do not tend to fall into the residual category of ‘all other conceptions’. This may be due to the heterogeneity of people with a migrant background, including among others 40 respondents with Turkish, 39 with Polish and 33 with Italian citizenship. Household income and community size also play a minor role.

In our paper, we set out to show which culturally normative conceptions of family exist in Germany. To achieve this goal, we used a multi-method design, combining qualitative and quantitative analyses. This approach turns out to deliver major benefits. The survey data fail to capture several crucial aspects of conceptions of the family, particularly the role of grandparents and family functions, both of which were not addressed by the standardised survey questions. The qualitative data, by itself, would not be able to give a realistic idea of the prevalence of the various family conceptions or their association with social groups. In combination, both approaches are much stronger and reveal a quite comprehensive result. Nevertheless, this paper only scratches the surface of a very under-researched area. Further studies are called for addressing other societies, further aspects of family conceptions, and perhaps even other cultural conceptions aside from family lives. Also, the methodological tools need to be researched and developed further. Data on culturally normative aspects, as studied here, are hard to capture due to their partly unconscious nature and due to the biases we need to assume, given that answers in standardised interviews may follow social expectations. These limitations have to be kept in mind also when interpreting the findings presented here.

Our findings suggest that among young adults in Germany, there are a number of quite diverse conceptions of the family. Three particularly inclusive personal conceptions of the family, however, stand out as dominant and represent about two thirds of the population. These are the ‘inclusive’ conceptions that any living arrangement with two or more people qualifies as a family, the ‘child-centred’ conception that any living arrangement with children can be called a family as well as the ‘child- or marriage-centred’ conception that any living arrangement with children or with a married couple should be called a family. The most exclusive family conception we observe with a relevant prevalence (about 6%) is the ‘child-/two-parent-centred’ conception that living arrangements with children and two parents are families.

These conceptions are relatively evenly spread across the young adult population (aged 20–39) of Germany, represented by the survey data, and only weakly correlated with socio-economic characteristics. Among the tendencies we do observe, it is apparent that those with a more restrictive conception of the family who only consider living arrangements involving a (married) couple with (their own) children to be a family, tend to be male, religious and Roman Catholic. One main difference between people oriented towards the three rather inclusive conceptions lies in their own family background, with people tending to consider their own situation as constituting a family: At least, having children is a strong predictor for sharing a ‘child-centred’ conception, and being married is the strongest predictor for sharing a ‘child or marriage-centred’ conception of the family.

In contrast to the quite liberal and inclusive appearance of conceptions of the family gained from the survey data, findings from the study ‘Family in Pictures’ reveal a somewhat different picture. Here, the most exclusive conceptions, particularly the idea of the family as a heterosexual married couple with their own children, appear to be very present. The main differences between the methodological approaches are that the instruction to draw a picture induced participants to pick one family form rather than a universe of possible forms and that it asks for the ‘proper’ family instead of ‘a’ family. Accordingly, the study ‘Family in Pictures’ rather reveals a culturally established common core of individual family conceptions, whereas the pattern of response across seven survey items rather reflects the boundaries of the personal conceptions of family. The core of today’s very liberal family conceptions is indeed still marked by the nuclear family in the sense that this is the only living arrangement for which it remains unquestioned that it deserves to be called a family. It is the core which all other conceptions are constructed around, even if they are broad and very inclusive and presented as a counter concept to it. Given the domination of the nuclear family in the drawings of a ‘proper’ family, we assume that it still serves as a culturally normative conception that provides orientation and a reference, even if partly in a negative way.

The conceptions concerning the (‘proper’) constellation of a family are closely related to various social aspects characterising the family, and in particular to the question of which function(s) families have and ought to fulfil: For example, emphasising the provision of mutual support and solidarity among family members as a family’s main function fits together well with an ‘inclusive’ understanding. Seeing the upbringing, socialisation and care of children as main functions goes together well with a ‘child-centred’ family understanding. Emphasising the necessity of two contact persons for educating children properly may provide a rationale for more exclusive conceptions, narrowing the universe down to living arrangements with two parents. Considering biological reproduction as a core function is likely to lead to an exclusion of same-sex couples.

Also other social criteria associated with a ‘proper’ family are closely linked to the preferred view of which functions the family is responsible for. That family members feel emotionally closely attached to each other, for example, is both a precondition and an outcome of their responsibility for taking care of each other if needed. That people feel protected and sheltered by their family is part of this function as such. Perceiving the family as an institution for providing care work is likely to induce the idea that the family is tied to a limited biographical phase in which children are young, live in their parents’ home and need care. The ideas that a family should live in a house in the countryside have a pet or be framed by a network of extended family are further possible consequences of this perception, inasmuch as these criteria are considered relevant for fulfilling the function of raising children responsibly.

Against the background of the – limited – research on culturally normative family conceptions, these findings seem plausible and are able to fill research gaps. We recognise the cultural change that Pfau-Effinger (2005) describes and that Germany appears to have undergone during recent decades, considering the political and institutional changes described in Section 3. This is reflected in the prevalence of the very inclusive conceptions we see in the quantitative data. Nevertheless, none of these inclusive conceptions are yet able to describe a new well-established societal consensus. Therefore, we also observe an ambiguity and contrariness between these new conceptions on the one hand, and the exclusive understanding of family as a nuclear family on the other hand, which remains in place, offers orientation and allows institutional arrangements such as the tax breaks for male breadwinner couples to persist. The ambiguity is mirrored particularly in the explanations of drawings of nuclear families in the qualitative data.

All in all, the culturally normative family conceptions are diverse. They all share a common culturally well-established idea that family in its prototypical form is a nuclear family, consisting of a heterosexual married couple with their own children, linked to a number of functions, forms of organisation, qualities of relationship and settings. They mostly are very generous and inclusive extensions of this core conception, including a large number of other living arrangements and forgoing some of the family’s ‘traditional’ functions. Precisely which living arrangements are included varies, however. In that sense, the common cultural conception of the family is very clear in its core and blurred in its outer contours. For young people drafting their life plans, today’s conceptions of what constitutes a family leave a lot of freedom for diversity. However, for those who come without firm personal preferences, they still propose the nuclear family as a frame of reference and as the safe beaten track.

No potential conflict of interest was reported by the authors.

Detlev Lück is a Senior Researcher at the Federal Institute for Population Research (BiB) in Wiesbaden, Germany.

Kerstin Ruckdeschel is a Senior Researcher at the Federal Institute for Population Research (BiB) in Wiesbaden, Germany.