Within-Track Differentiation and Leisure Activities as Strategies of Distinction

The socially inclusive educational expansion during the last few decades in Germany diluted the previous “elite” status of the Gymnasium and its certificate, the Abitur. In the absence of explicit “elite” institutions, unlike the UK, the USA or France, the question remains how social privilege is intergenerationally transferred in times of educational expansion? Several studies on this topic focused on distinction either through horizontal educational differentiation or via distinctive leisure activities. Building on Bourdieu’s theory of distinction and Lareau’s study on child-rearing practices, we argue that it is the set of different strategies of distinction that serves as a mechanism to reproduce privilege. Hence, we ask the question how parents’ and grandparents’ educational background influences the probability of applying a whole set of distinctive strategies. To test our assumptions, we use data from the youth questionnaire of the German Socio-Economic panel and analyse the effect of educational background on the application of distinctive educational strategies and distinctive leisure practices and the combination of both domains. The results indicate three main findings: first, having parents with an academic education increases the probability of applying any of the distinctive strategies under consideration. Second, having academically educated grandparents increases the probability of combining distinctive strategies within one domain. Third, the data do not support the assumption that the combination of curricular and leisure practices is particularly socially stratified.


Introduction
The second half of the twentieth century was characterized by a comprehensive educational expansion in all western societies (Breen and Jonsson 2005). Overall, this expansion resulted in a general educational upgrading in western societies, making an upper secondary degree (e.g., A levels) the standard school-leaving certificate. These trends led to a change in the intergenerational transmission of educational privilege from parents to children. Theoretical and empirical research on this topic focuses on the one hand on the acquisition of educational certificates (Raftery and Hout 1993), such as university degrees (Lörz and Schindler 2011) or doctoral degrees (Bachsleitner et al. 2018). On the other hand, it focuses on qualitative inequalities in K stratified curricula that allow for differentiation within the same level of education (Lucas 2001(Lucas , 2009. Investments in horizontally differentiating distinctive tracks as well as in higher educational degrees are supposed to maintain educational privilege, which results in better chances in the labor market. Looking at the top of the educational and occupational hierarchy, national elites in countries such as the UK, the US, or France commonly share a biography, which includes attendance at a small set of private secondary schools. These institutions share some common features, such as a classical academic curriculum or specific extracurricular activities (e.g., Reeves et al. 2017), and they form privileged learning environments (Dumont et al. 2013) enhancing the accumulation of knowledge, as attendance at these institutions is socially selective (Jerrim et al. 2015). Additionally, the focus on distinctive curricular content and extracurricular activities provides the formation of a typical upper-class habitus, which is rewarded in the labor market (Bourdieu 1996;Jackson 2009).
In a field-experimental study, Jackson (2009) shows that the combination of several elite signals, e.g., attending a private school and having elite interests, is more effective in terms of labor market returns than signaling a privileged status via one single signal only. Building on this finding as well as on Bourdieu's (1984) theory of distinction and Lareau's (2003) study on child-rearing practices as a theoretical foundation, we argue that it is not enough to look at school distinction and distinctive leisure practices separately (see, for example, Sawert and Gerhards 2019), but that privileged families combine different kinds of distinctive strategies to signal that they grew up in a family rich with cultural capital.
First, educationally privileged families make use of horizontal differentiation within the upper secondary system by choosing distinctive schools, tracks, and subjects. Second, these curricular strategies are combined with distinctive leisure practices, such as playing a classical instrument or doing distinctive sports. Based on 20 qualitative interviews, Sawert (2021) finds that it is precisely this combination of different strategies that distinguishes highly privileged families from less privileged families. The latter often allow their children to attend a good school or to participate in a distinctive sport, but to a lesser extent following a holistic lifestyle oriented toward distinction.
We test this finding using the representative data of the German Socio-Economic Panel (GSOEP) to analyze the social selectiveness of different means of upper-class distinction, which is elaborated in the following theoretical part. Specifically, we test whether applying more than one distinctive practice and combining education with leisure practices is socially more selective than focusing only on one distinctive practice. Additionally, we argue that it is not only the parents' education that affects the orientation toward distinctive practices but also the grandparents' education. Whereas the influence of the parents' educational level on the vertical and horizontal educational decision has been the focus of sociological research for decades (e.g., Reimer and Pollak 2010;Breen and Jonsson 2005;Blossfeld et al. 2015), authors have started to investigate the effect of grandparents' educational levels on the educational biographies of their grandchildren quite recently (e.g., Sheppard and Monden 2018). Sawert and Gerhards (2019) demonstrate that grandparents' educational background not only affects the educational success of their grandchildren but that it also has a direct effect on practicing highbrow leisure activities. Thus, we operationalize the educational background of the respondents using the information on parents' and grandparents' educational certificates.
In the following, we first develop the theoretical argument on distinction in school (Sect. 2.1) and out-of-school contexts (Sect. 2.2), as well as on the combination of both ways of distinction. In Sect. 3, we present the data basis used, as well as the operationalization of the variables on the basis of which we empirically test our assumptions. In Sect. 4, we present the results, first with a focus on the influence of educational origin on distinctive educational pathways in the school context (Sect. 4.1), and then on distinction in the out-of-school context (Sect. 4.2). In Sect. 4.3, we test the main argument of the article, that it is especially the combination of both distinction strategies that is socially selective. In Sect. 4.4, we subject our results to some robustness analyses before moving on to a final discussion of our findings in Sect. 5.
Our results partly support the argument to take a more comprehensive look at the combination of strategies of distinction. We find that educationally privileged households are more likely to perform at least one strategy of distinction and that having academically educated grandparents increases this likelihood substantially. However, the main argument that educationally privileged households combine educational and leisure strategies is only partly supported.

Theoretical Frame and State of the Art
Educational certificates are the main allocation principle for occupation chances in today's western societies (e.g., Blossfeld 1989;Solga 2005;Hillmert and Mayer 2004). Higher educational certificates are associated with lower risks of becoming unemployed (Reinberg and Hummel 2007) and positively affect the chances of obtaining a privileged occupational position (Müller 1998;Hillmert 2001;Klein 2011). A century ago, it was standard practice to leave school in Germany after the ninth grade (Volks-/Hauptschulabschluss). Educational reforms, already starting at the end of the nineteenth century (Sawert 2018) but accelerating in the 1960s, were aimed at a general educational upgrade of the overall population. The purpose of these reforms was to increase the number of potential students and to reduce social inequalities in access to higher educational tracks (Dahrendorf 1965;Becker 2007;Maaz 2006). These reforms were effective in increasing the number of pupils who finish school with the Abitur (equivalent to A levels), making it the most common school-leaving certificate in Germany since 2011 (Authoring Group Educational Reporting 2018). Whereas the traditional three-track structure in Germany closely linked school tracks with school-leaving certificates, this structure was loosened and alternative ways to the Abitur via non-Gymnasium school tracks were strengthened. Some German states further differentiated their non-Gymnasium tracks; the majority, however, reduced their school system mainly to two tracks; one school track beside the Gymnasium that is non-academic but opens up the option of attaining the Abitur too (Becker et al. 2016). Within the course of the quantitative expansion of upper secondary education, previously excluded, educationally less privileged social K groups have been included in upper secondary education. 1 Although educationally less privileged families tend more toward choosing vocational tracks to the Abitur than educationally privileged families (Maaz 2006;Schindler 2014), the inclusion of lower socio-economic strata diluted the social selectiveness and exclusivity of the Gymnasium Abitur to a large degree. In this context, authors (e.g., Blossfeld et al. 2015) argue that educational inequalities shifted from attaining an upper secondary degree toward tertiary educational differences, e.g., attending university.
Research focusing on these vertical differences in the acquisition of certificates at different vertical levels in the educational systems often refers to the theory of Raymond Boudon (1974) to explain educational background effects. The basic argument goes that pupils from lower socio-economic backgrounds perform less well because they have lower cultural capital transmitted by their parents (Bourdieu 1984) and are less supported by their parents, resulting in fewer competencies (e.g., Stubbe et al. 2016). Hence, owing to their lower educational performance, students from less privileged backgrounds are more likely to drop out at educational transition points (primary effects). Beyond that, even if children from lower socio-economic strata perform equally well in the educational system as their more privileged peer students, they still have a lower probability of continuing school (secondary effects). This is because lower socio-economic strata perceive the costs of additional schooling to be higher and are more likely to finance their studies with student loans. Furthermore, they have on average lower educational aspirations (Authoring Group Educational Reporting 2018). However, after decades of educational expansion, attaining neither an upper secondary nor a university degree is distinct enough at the top of the educational hierarchy. Research by Bachsleitner et al. (2018) on transitioning to still exclusive doctoral programs after finishing tertiary education shows educational background effects. Children from families in which at least one parent has a doctoral degree have a 21.6 percentage points higher probability of transitioning to a doctoral program. This effect is mainly explained by performance-related indicators and factors in the educational biography (e.g., subject choice).
In his theory on cultural reproduction, Bourdieu demonstrates how parents transmit cultural capital to their children, how cultural capital is converted into success by the children in the educational system, and how educational success is finally converted into socioeconomic status, resulting in the intergenerational transmission of privilege. Elaborating on the mechanisms of how status is intergenerationally transmitted, Jaeger and Breen (2016) differentiate between passive socialization and active parental investments: cultural capital is on the one hand transmitted from parents to children by nonreflected components of socialization (e.g., exposure to cultural objects in daily life) and by active investments, on the other hand, where parents "choose whatever amount of investment they think will yield the highest 1 It is an open discussion to which degree the educational expansion in Germany can be considered effective in decreasing social inequalities. Whether educational reforms were a success in this sense depends on the one hand on the understanding of the normative principle of educational systems (see liberal-conservative understanding (Hradil 1999, p. 148), liberal understanding (Dahrendorf 1965), and model of statistical independence (Müller and Mayer 1976, p. 27)). On the other hand, different statistical indicators (e.g., participation ratios, percentage difference, odds ratios) come to different conclusions about the trends concerning social inclusion. return for a given cost" (Jaeger andBreen 2016, p. 1090). Hence, in times of educational expansion, privileged parents have to look for a diverse set of strategies, e.g., educational distinction via horizontal educational decisions. In the following, we elaborate on these strategies and our expectations of how educational background affects the probability of following distinctive educational tracks.

Horizontal Distinction in Upper Secondary Education in Germany
In addition to vertical educational differentiation, socioeconomically privileged parents try to ensure educational advantages for their children through educational choices within the same level of education. Lucas' theory of effectively maintained inequality states that educationally privileged parents secure educational advantages for their families whenever possible; if a particular level of education is no longer exclusive, they ensure their advantage through qualitatively better options within that educational level (Lucas 2001(Lucas , p. 1652. We argue that in addition to different tracks in stratified curricula, within-track differentiation, e.g., via stays abroad and school subject choice, constitutes an important means for distinction. Especially higher educated parents not only have preferences for specific tracks but also the specific school of their children, as these families are more aware of differences in the quality of schools than lower educated parents (Klinge 2016, p. 228 f.). Furthermore, they are not only more aware of quality differences between different schools, but more highly educated parents also have more capital, which allows them to realize their school preferences, either by active interventions at school 2 (Dumont et al. 2014(Dumont et al. , 2019 or by utilizing economic capital to access, e.g., private schools. In countries such as the UK, private schools are a traditional element of the educational system. Attending them is often socially exclusive and graduating from them is associated with better chances in the labor market (see Jackson 2009). In Germany, private schools are on the rise: the number of children attending a private school nearly doubled between 1992 and 2019 from 4.8 to 9.2% (Federal Statistical Office 2019). In 2018, every eighth upper secondary school in Germany was privately funded (Klemm et al. 2018). Although the private schools offer a diverse set of programmatic orientations, e.g., international education or Waldorf education, they have in common that they usually constitute a socially selective learning environment. This is partly the result of the economic hurdle produced by tuition fees, which are rather low in comparison with internationally, and partly because of the programmatic orientation on, for example, transnational education or, for example, Waldorf education, which is especially valued by privileged families with its strong focus on arts and free learning methods (Liebenwein et al. 2012, p. 23 ff.). Because of the resulting social composition effects (Dumont et al. 2013), socially selective schools constitute a learning environment that enhances skill accumulation, thereby offering better educational chances to students who attend private schools. However, Klemm et al. (2018) demonstrate that students at private schools do not differ in their competencies from students at public schools when controlling for confounding variables (see also Hoffmann et al. 2019). As private schools are relatively exclusive and parents might ascribe positive skill development to attendance at private schools (see Thomas and Thomas 1928), we expect to find a positive effect of educational background on attendance at a private school.
Another specific type of upper secondary school in Germany are so-called humanistic Gymnasiums, where pupils must learn Latin from grade five onward and sometimes additionally ancient Greek from grade eight onward. Humanistic Gymnasiums have a long tradition in German educational history. When the Abitur exam was formalized in Germany in 1812, pupils had to learn Latin and ancient Greek. The Gymnasium at that time spent more than 40% of the learning time on these classical languages (Herrlitz et al. 2008). The humanistic Gymnasium was a highly exclusive and socially selective school type and educational reforms leading to educational expansion did not just increase the number of pupils attending a Gymnasium but were associated with a programmatic shift from humanistic to more labor marketoriented education (see Sawert 2018). Today's humanistic Gymnasiums still constitute socially selective learning environments (Baumert et al. 2010) and people who know Latin are ascribed more general and cultural education . Graduating from a humanistic Gymnasium increases the probability of getting into a leading position in the labor market (Sawert 2016). In contrast to private schools, attendance at a humanistic Gymnasium is usually free of charge, as these are most commonly public schools. However, the curriculum with its focus on classical languages might serve as a hurdle for families who are more oriented toward acquiring knowledge in school that is useful in the labor market, e.g., modern foreign languages. Following Bourdieu (1984Bourdieu ( , 1990, attending a humanistic Gymnasium should be a prominent strategy especially among students with highly educated parents and grandparents (see Sawert and Gerhards 2019): higher educational certificates are one subtype of cultural capital in the theory of Bourdieu. Having high cultural capital is associated with a higher probability of developing a habitus that is oriented toward distinction. For successful cultural distinction, cultural practices must fulfill three different characteristics: first, they need to be exclusive; second, the benefit should not be too obvious; third, they need to have a symbolic meaning that has historically been established (Sawert 2018). Learning Latin from grade five onward, most likely by attending a humanistic Gymnasium, fulfills all these characteristics. Hence, we assume that children from highly educated backgrounds are more likely to learn Latin from grade five onward than children from less welleducated backgrounds.
Attending a private school or a humanistic Gymnasium is associated with choosing a specific school. But even after families have opted for a specific school, opportunities for educational differentiation exist. One practice that has been the focus of research is a stay abroad during school time. In a globalizing world, intercultural knowledge, as well as foreign language skills, are an increasingly important resource in the labor market (Tucci and Wagner 2003). Conducting interviews with parents whose children attend a Gymnasium in Germany, Carlson et al. (2017) find that highly educated parents perceive a stay abroad during school time to be a quite common practice, whereas less privileged families do not consider a stay abroad during school time for their children. Weenink (2008), who conducted interviews with parents of children in education, also finds that families not only consider a stay abroad to be a "cosmopolitan vision" (Beck and Sznaider 2006), referring to international values and interconnectedness but they also perceive a stay abroad as an effective strategy for acquiring transnational human capital, e.g., foreign language fluency . Because a stay abroad is associated with high economic costs, a preference for cosmopolitan values, and an orientation toward distinction via the acquisition of transnational human capital, we expect to find a positive association between a privileged educational background and a stay abroad.
Although we derived assumptions about the association between attending a private school, learning Latin from grade five onward, and a stay abroad during school time, our focus does not lie on a single practice. In our analyses, we perceive the three practices as a set for educational distinction in the German upper secondary system, which is in general more likely to be performed by educationally privileged families. Therefore, we formulate the hypothesis that children from an educationally privileged background are more likely to perform at least one of these three practices than children from less privileged backgrounds (H1a). As the habitus associated with distinctive practices is largely transmitted in family socialization (Bourdieu 1991, pp. 64-70), Bourdieu's theory suggests that the socializing conditions of parents, i.e., cultural capital at the level of grandparents, might also influence the likelihood of distinctive practices. Although empirical results on this are still inconclusive and isolating a direct effect of grandparents is not trivial (Breen 2018), the results of recently published studies suggest that there might be an independent effect of grandparents' education on the educational careers of their grandchildren (Sheppard and Monden 2018). Accordingly, we operationalize educational origins in our analyses via parental and grandparental education.
As outlined in the introduction of this paper, Jackson (2009) shows that it is the combination of several privileged practices that is most effective in terms of occupational chances. Hence, we differentiate not only whether children follow none or at least one practice but also analyze whether they follow several distinctive curricular practices, e.g., attending a humanistic Gymnasium and having a stay abroad. We assume that the effect of educational background is even more articulated when they follow several curricular strategies of distinction than when they follow only one such practice (H1b).

Leisure Activities as a Form of Intergenerational Transmission of Privilege
Additionally, we not only consider curricular possibilities of distinction but also include leisure activities in our set of distinctive practices. Lareau (2003) adds to the theory of Bourdieu that unequal distributions of cultural practices, such as visiting the theater or the opera, are not only the result of differences in the habitus but are also produced by differences in active child-rearing practices. Focusing on US middle-and working-class families, Lareau shows that middle-class parents actively organize the leisure of their children, aiming at an increase in cultural capital and a reduction in development risks. She refers to this parenting pattern as concerted cultivation and distinguishes it from a parenting pattern observed in US working-class families, who usually do not intervene in the leisure preferences of their children. Lareau (2003) does not focus directly on active highbrow socialization of the families but her findings can be transferred to this topic in a German context. Reeves and de Vries (2019) demonstrate that cultural consumption, in general, is positively associated with income, even after controlling for factors of the educational and occupational biography. As pointed out by Hartmann (2015) and Jackson (2009), signaling a privileged educational background does particularly pay off in the labor market. One way to signal a privileged socio-economic status is via elite interests (see Jackson 2009). As soon as parents recognize that doing one sport instead of another is perceived more positively in society, it becomes a rational strategy to push the children toward the leisure activity that is perceived to be more prestigious. Focusing on a more passive intergenerational transmission of a highbrow orientation, Bourdieu argues that the socializing environment of the family forms the habitus of the children and that the habitus of the children affects their preferences for distinctive leisure activities. This is outlined by Sawert (2021): Historically academic families, in which not only the parents but also the grandparents have academic degrees, do not have to actively organize the leisure activities of their children, but they produce such a socially selective and homogeneous highbrow environment that their children are socialized in a highbrow cultural environment. This passive socialization mechanism (see Jaeger and Breen 2016) can be assumed to be particularly effective in families in which grandparents also have academic degrees, as this implies that the parents themselves grew up in a household with a high volume of cultural capital on the one hand, and as grandparents might be role models for grandchildren as well.
Two possible leisure activities aimed at distinction are doing specific sports and playing a classical instrument. Whereas it is not distinctive to play football or do bench pressing, sports offered, e.g., by Eton College, such as fencing, tennis, field hockey, or rowing, are usually considered sports that are performed by persons from a privileged socio-economic background. Hwang et al. (2010) demonstrate that golf, tennis, and yachting are socially selective and Engström (1974) shows that additionally for horseback-riding, ballet dancing, and sailing. According to Bourdieu, family is especially important for the intergenerational transmission of a highbrow lifestyle when it comes to actively playing an musical instrument (Bourdieu 1983, p. 187). Learning an instrument is time-and cost-intensive. However, whereas classical instruments such as the violin or the piano are commonly considered highbrow instruments, the same distinctive value is not ascribed to playing the drums or performing pop music. Hence, we formulate the expectation that being from an educationally privileged background increases the probability that the children will do a sport that is considered a highbrow sport and/or have a higher probability of actively playing a classical instrument.
As for the distinctive curricular strategies, we are not interested in the effect of educational background on single distinctive leisure activities but consider them a set of distinctive practices. Hence, we assume that having a privileged educational background increases the probability of following at least one distinctive leisure activity (H2a) and that it, even more, affects the probability of following more than one distinctive leisure activity (H2b).
Our final assumption focuses on the combination of distinctive curricular strategies and distinctive leisure activities. Building on Bourdieu (1984), we expect that a distinctive curricular strategy or a distinctive leisure activity might be followed by less privileged socio-economic classes as well, as they copy the practices performed by the educationally privileged classes. Therefore, we expect the combination of curricular and leisure activities to be particularly selective (H3). With the formulation of this assumption, we bridge the gap to the finding of Jackson (2009) that it is the combination of different signals for a privileged socio-economic background that pays off the most in the labor market.

Data and Operationalization
We use the data of the German Socio-Economic Panel (GSOEP) to test the hypothesis that we derived in the previous section. Since 2000, persons who live in GSOEP households are interviewed during the year they turn 17. The questionnaire contains information about a broad range of topics. One of these topics is educational tracks and leisure activities. We use this information for our study and connect the information of the 17-year-olds with information about their parents. In total, the subsample of the GSOEP contains 7813 individuals living in 5374 households. All respondents are 17 years old at the time they are first interviewed. We restrict our sample to pupils attending the Gymnasium, the highest secondary educational track in Germany, which is usually completed with the Abitur (equivalent to A levels). We focus on distinction at the top of upper secondary education. By doing so, our sample size is reduced to a total of 2905 students attending the Gymnasium. Because of missing values on the dependent variable, the sample size is reduced by 306 cases to a total of 2599 cases. Finally, because of missing values on the main explanatory variable educational background, we lose one case, leaving us with 2598 cases for our main analyses. 3 We focus on the total (causal) effect of educational background on being on a distinctive educational track and on practicing distinctive leisure activities. Hence, our main explanatory variable is the educational level of the parents and the grandparents of the respondents. As cultural distinction via highbrow activities and exclusive educational tracks is expected to primarily take place at the top of the educational hierarchy (see Bourdieu 1984;Bourdieu and Passeron 1977), we only distinguish whether or not parents and grandparents have an academic degree. An academic background at the parents' or/and grandparents' level is ascribed if at least one of the parents or grandparents has an academic degree. If there is missing information about the educational background of one of the parents or several of the grandparents, we only use the information available. Hence, at a minimum we define educational background using information about one parent and one grandparent, thereby underestimating the overall educational level in our sample. We use the information of a maximum of two parents and four grandparents. Hence, we differentiate between a non-academic educational background, in which parents do not have an academic degree, an old academic educational background, if grandparents have an academic degree but the parents do not, a new academic educational background, in which parents do have an academic degree but none of the grandparents and finally, historically academic households. We code a household as a historically academic household if at least one of the parents and at least one of the grandparents has an academic degree. If, for example, the father has an academic degree and the father of the father as well, this household is coded as a historically academic household. In 98.7% of all households, parents' educational level is based on the information of both parents. In 1.0% we only have information about the educational level of the mother and in 0.3% we have only information about the educational level of the father. The data basis for the information on the education level of grandparents is poorer: in 77.0% of all cases, the information on grandparents' education is based on information from grandparents from the mothers' and the fathers' side. In 18.2%, we only have information on the parents of the mother and in 4.8% we only have information on the parents of the father.
We combine five different practices of distinction as a dependent variable. All five variables are measured in the youth questionnaire of the GSOEP (bioage17 dataset).
Attending a private school, being on a humanistic track, or participating in a school stay abroad are curricular distinction practices included in the dataset. We recoded the original variables to dummy variables for all three variables.
For attending a private school, the original variable differentiated between whether a person attends a private school (1), formerly attended a private school (2), or never attended a private school (3). We dichotomized the variable by recoding the actual and former attendance of a private school into one category (Yes: 1), contrasting it with respondents that never attended a private school (No: 0).
Being on a humanistic track was coded based on information of respondents' first and second foreign language. The main characteristic of a humanistic track is that students must learn Latin as their first foreign language at the Gymnasium. That means that they spend more school hours learning Latin than learning English. We dichotomized the answers differentiating only whether a student is on a humanistic track, learning Latin as a first foreign language (Yes: 1) or not (No: 0). Using our data, we cannot differentiate whether students attended a purely humanistic Gymnasium or were on a humanistic track within a normal Gymnasium. Hence, we might underestimate the main effect of education on this practice of distinction, as households oriented toward distinction might prefer to attend a purely humanistic Gymnasium instead of being on a humanistic track at a normal Gymnasium.
Finally, the third variable measuring distinctive curricular practices was a school stay abroad.
We used the original variable in the GSOEP and differentiated between respondents who reported having stayed abroad (Yes: 1) and those who did not stay abroad (No: 0).
Our main interest does not lie in the single strategies but in applying a set of practices of distinction. Hence, in a first step, we analyze whether households apply none of these strategies, at least one of these strategies, or more than one strategy. In a second step, we want to analyze the combination of curricular and leisure strategies of distinction. To do so, we used two variables that measure distinctive leisure practices in the GSOEP subsample.
Our first variable covers whether the respondents are practicing a sport that is perceived as a distinctive sport. In the GSOEP, respondents are asked whether they do sports and if so, which kinds of sports. We used this information and created a dichotomous variable that differentiates between respondents doing distinctive sports (Yes: 1) and respondents doing either no sports or no distinctive sports (No: 0). Based on the works of Hwang et al. (2010), Stokvis (2011) and Engström (1974), and the extracurricular activities offered at elite schools such as Eton college, we defined golf, fencing, rowing, sailing, ice-skating, tennis, ballet, horse-riding, and field-hockey as distinctive sports.
The second distinctive leisure strategy that we analyze is playing a classical instrument. In the dataset, respondents were asked whether they played an instrument and if so, whether they played a classical instrument or not. We combined the two pieces of information and created a dichotomous variable differentiating between playing a classical instrument (Yes: 1) and either not playing an instrument or not playing a classical instrument (No: 0).
As for the curricular strategies of distinction, we created a new variable, combining information on both leisure strategies. This variable differentiates between not participating in either of the two leisure practices, participating in at least one practice, and participating in both distinctive leisure practices.
Finally, for the third part of our analysis, we combined information on leisure and curricular practices of distinction, generating a variable that differentiates four categories: 1. Applying neither curricular nor leisure strategies of distinction 2. Applying only curricular strategies of distinction 3. Applying only leisure strategies of distinction 4. Applying curricular as well as leisure strategies of distinction Besides these variables, all models control for the year the survey was conducted as well as for gender. 4

Results
The following results are structured into three parts. To analyze whether educationally privileged families use and combine distinctive strategies, we first examine educational background differences in distinctive educational strategies (4.1). Second, we focus on the choice of distinctive leisure activities by analyzing the social selectivity of combining distinctive leisure activities (4.2). Finally, we test whether educationally privileged families combine distinctive curricular and leisure activities (4.3). In all analyses, we distinguish between students with a non-academic educational background, an old academic educational background, a new academic educational background, and a historically academic background.
The results are calculated using multinomial logistic regression models and are shown in the tables as average marginal effects (AMEs). They indicate the increase in percentage points of the use of the considered distinctive strategies if the student has an old academic, a new academic, or a historically academic educational background compared with a non-academic educational background (reference category). All models are based on weighted cases with cluster correction for household membership and we control for gender and the survey year in all models. In 4.4 we test whether the results remain robust over different coding strategies. Table 1 shows the effect of educational background on applying distinctive educational strategies. The coefficients in Table 1 are average marginal effects that represent the probability of applying no distinctive educational strategy (outcome 1), one distinctive educational strategy (outcome 2), or more than one distinctive educational strategy (outcome 3) for the different educational backgrounds. Non-academic households are the reference category in all models.

Making Use of Horizontal Differences in the Educational System
Overall, the results in Table 1 indicate strong educational background effects on the probability of applying distinctive educational strategies that are in line with Lucas' (2001) theory of effectively maintained inequality and support H1a and H1b. Children from non-academic households have the highest probability of applying no distinctive educational strategies. The probability of not applying at least one of the three curricular distinction strategies is 7 percentage points lower for old academic households, 15 percentage points lower for new academic households, and 23 percentage points lower for historically academic households. Hence, we observe that the higher the educational background, the less likely it is that children do not apply any distinctive curricular strategy.
Concerning the application of one distinctive educational strategy, the results show that the main line of distinction is between non and old academic households versus new and historically academic households. The results indicate that children from old academic households have a 3 percentage points higher probability of applying one distinctive educational strategy. However, the coefficient is not significant. Children from new academic households have a 12 percentage points higher probability of applying one distinctive educational strategy and for children from historically academic households, this probability is 15 percentage points higher. Hence, for the probability of being either in a private school, or on a humanistic track, or having a stay abroad during school time, the main line of distinction is between having parents with academic degrees versus not having parents with academic degrees. For applying more than one educational strategy of distinction, e.g., attending a private school, and having a stay abroad during school time, we observe that the direction of the effects is in line with hypothesis 1a: children from old, new, and historically academic families have a higher probability of applying more than one strategy. We also observe that children from historical academic households have a higher probability of applying more than one distinctive educational strategy than children from all other educational backgrounds: although the probability of children from old and new academic households is 4 percentage points higher than for children from non-academic backgrounds, the effect for children with a historically academic background is twice as high, with an 8 percentage points higher probability of combining several strategies of curricular distinction. Although these findings are in line with our expectations, it should be noted that the effects of combining more than one strategy are not very substantial.

Leisure Activities as Ways to Cultural Distinction
To test if distinction is also visible in the use of leisure activities, educational background differences in playing a distinctive sport or a classical instrument are presented in Table 2. Again, we focus on the effect of educational background on applying either no distinctive strategy, one distinctive strategy, or a combination of more than one distinctive strategy.
The general tendency of the results supports our hypotheses H2a and H2b. Children from a non-academic background and children from an old academic background do have the same probability of not following distinctive leisure practices. Hence, having academically educated grandparents increases the probability of applying educational strategies of distinction (see Table 1) but does not affect the probability of following distinctive leisure practices. Children from new academic households have a 12 percentage points lower probability of not doing any distinctive leisure activities. For children from historically academic households, this effect is even more articulated, with a 19 percentage points lower probability. Children from a new academic background have a 10 percentage points higher probability of fol- Table 2 Effect of educational background on applying leisure distinction strategies; coefficients are average marginal effects from multinomial logistic regression models, cluster-corrected for households, controlling for gender and survey year; N = 2598 (weighted results)

Educational background
No strategy One strategy More than one strategy K lowing one distinctive leisure practice and a 2 percentage points higher probability of following more than one distinctive leisure practice. In line with our expectations is the finding that children from historically academic households have an even higher probability of practicing one distinctive leisure activity (+15 percentage points, p < 0.05) and also have a higher probability of practicing more than one distinctive leisure activity, that is, playing a distinctive sport and a classical instrument. Hence, although the direction and significance of the results are in line with our expectations; specifically, the size of the effect for applying more than one strategy is rather small.

Combining Strategies of Distinction
The analyses presented in the previous two subsections indicate that in the educational as well as in the leisure domain, applying distinctive strategies is selective and tends to support the argument that the combination of several distinctive practices is even more socially selective than the application of one strategy. In our third and final step of the analysis, we are taking a more holistic view of distinctive strategies; we are interested in the educational background effect of combining distinctive educational and leisure strategies. Table 3 presents the probability for the educational background groups of applying no distinctive educational strategy (outcome 1), only curricular strategies (outcome 2), only leisure strategies (outcome 3), or the combination of curricular and leisure strategies (outcome 4), with non-academic families being the reference category.
The results show that the probability for families of applying no curricular or leisure strategies decreases the higher their educational background. Compared with non-academic families, families with an old academic background have a 6 percentage points lower probability, new academic families an 18 percentage points lower probability, and historically academic families a 34 percentage points lower probability of not applying any strategy. This finding strongly supports the argument that grandparents' education should be considered in future research. Only applying curricular strategies is also affected by educational background; however, children from old and new academic families have the same increase in probability, with 7 and 6 percentage points, in comparison with children from non-academic families, whereas children with historically academic families stand out with a 15 percentage points higher probability. Even more striking, the probability of only doing leisure activities does not differ between non-academic, old academic, and new academic families, but historically academic families have 12 percentage points higher probability. Hence, applying leisure strategies appears to be especially socially selective. Concerning our main interest in the combination of curricular and leisure strategies, the results do not support our hypothesis H3 that the combination of the two strategies is particularly selective. Children from new and historically academic families have the same increase in their probability of combining curricular and leisure strategies compared with children from non-as well as old academic families; thus, the main line of distinction is between families with a non-academic and an academic background.

Robustness Checks
To check whether the findings are robust we re-ran our analysis with (a) income as a mediator (Table 4) and (b) an alternative coding strategy of distinction via leisure strategies (Table 5). As the main argument of this paper is that educationally privileged households reproduce privilege by combining educational and leisure strategies of distinction, we focus only on the robustness of the results presented in Table 3.
First, we tested if the results presented remain robust if income is included in the model as a mediator (Table 4). Income differentials could explain part of the effect of educational background as the educational background is associated with higher earnings and higher earnings increase the opportunity of households to pay, e.g., for tuition fees at private schools and music lessons. Hence, an alternative argument could be that the effects presented in Sect. 4.1 to 4.3 are not the result of differences in orientation toward distinctive strategies, but just a result of different opportunity structures.
Looking at the results when income is included as a control variable, we find that the effects overall become a bit smaller, but the general picture barely changes, with one exception: when combining curricular and leisure strategies, we find that the effect of new academic households becomes 3 percentage points larger, whereas the effect for historically academic households becomes 3 percentage points smaller. Although this is only a moderate change for the individual effects, the overall result is that new academic households are 7 percentage points more likely to combine curricular and leisure strategies than historically academic households. Consequently, in the comparison of households with academic parents, a negative effect is found if the grandparents also have academic degrees, which is a clear finding contrary to our assumptions. Note, however, that the results presented in Table 4 refer to a different analysis population than the results in Table 3 owing to missing values on the income variable. In further analyses, instead of listwise deletion, we imputed the missing values, with multiple imputations with 20 iterations, and the results with multiple imputations yield the same findings as Table 4. In a second robustness check, we examine whether the education effects from 4.1 to 4.3 are robust to coding the leisure variables differently. Specifically, we check whether the results remain the same if, instead of distinctive sports and classical music-making, people do any kind of sports or actively make music in any form. If the results in Table 3 are specific to distinctive sports and classical music-making, then the effect should be substantially smaller with alternative coding.
A first key finding of this alternative approach is that the proportion of those who engage in at least one of the leisure practices considered increases from 22% (see Appendix) to 88%. In our sample of students attending the Gymnasium, playing sports or music is a typical leisure activity. Unfortunately, the SOEP data do not allow us to differentiate between the frequency of playing sports or music.
This change in the distribution has far-reaching consequences for the results, which can be seen in Table 5. If the alternative is to pursue no strategy, the effects are significantly smaller, simply because it is also common among non-academic households to play sports (e.g., soccer) or to make music actively (e.g., pop music). The same picture emerges for the alternative of pursuing only curricular strategies: as almost all households engage in some leisure activity, the differential effects tend to zero here. The previous inequality effects ultimately shift to the option of combining curricular and leisure strategies when coded alternatively. However, this is mainly a statistical artifact, as leisure activities are almost universal, and the effect consequently mainly reflects the effect of additional curricular strategies. Overall, it appears that differentiating sports activity and music-making into distinctive forms and nondistinctive forms is more informative.

Discussion
We focused on the question whether educationally privileged households tend to combine educational and leisure strategies of distinction as a mechanism to inter-generationally transmit a privileged status in times of educational expansion. As research has shown that signaling several "elite" interests is more effective in terms of increased labor market chances than signaling only one "elite" interest, we argued that highly privileged families apply not only one practice of distinction but a comprehensive set of such practices (Jackson 2009). Based on the theory of effectively maintained inequality (Lucas 2001(Lucas , 2009) and Bourdieu's socio-cultural class theory, we derived hypotheses about the effect of having an educationally privileged background on the tendency to attend either distinctive curricular tracks and/or to do/play distinctive sports/music. It is important to note that we focused on these activities based on our theoretical assumptions as well as the given data possibilities and that these practices are just one possible set of practices aimed at differentiation. Several studies have already focused on the association between educational background and specific distinctive practices. However, we add two main contributions to the literature: first, we are not interested in the association between educational background and the likelihood of following the single practices but rather in the association between educational background and the likelihood of combining several distinctive strategies. We make the argument that educationally privileged households are more likely to combine distinctive strategies. Second, in line with recent empirical findings, we focus on the effect of not only the parents' educational background but also the grandparents' educational background.
We are focusing on Germany in our analysis as the German schooling system has the interesting feature that it does not provide specific elite institutions that allow for differentiation and that lead to privileged positions in the labor market. Hence, we perceived the German case as an empirical case in which the need for horizontal differentiation via distinctive practices is particularly high. As data, we used the German Socio-Economic Panel (GSOEP), a large population study, which has existed since 1984, for our analyses. We decided to use the GSOEP, because on the one hand it allows the educational background in the parents and grandparents' generation to be operationalized very well, and on the other hand it allows as many distinctive educational and leisure activities as possible to be considered.
Overall, the results are partly in line with our expectations and partly contradict them: we find that having academically educated parents positively affects the probability of following distinctive curricular and leisure practices. Having not only academically educated parents but also academically educated grandparents increases the probability of following distinctive leisure practices even more but does not increase the probability of combining educational and leisure practices. Hence, the data contradict our assumption that the combination of curricular and leisure activities is particularly socially selective. For the combination of distinctive strategies in both domains, the grandparents' education does not matter. Households in which parents hold academic degrees are more likely to combine strategies from both domains, independent of the grandparents' educational background.
Based on our theoretical assumptions and empirical findings, it is unclear how this finding should be interpreted. One possible interpretation of the overall findings is that it is not the combination of strategies from different domains that is relevant but the combination of different distinctive practices in general. This would be consistent with our findings from 4.1 and 4.2, where at least weak effects of grandparental education on the combination of practices within a domain are found. Moreover, this would be consistent with other work that finds an independent effect for grandparents (see Sect. 2). Nevertheless, it cannot be ruled out that the basic theoretical assumption is simply wrong and that the grandparents' educational background has no independent influence on the pursuit of distinctive practices. Additionally, we cannot rule out the possibility that this observation is an artifact produced by some weaknesses of the database for our research. Although the GSOEP is the best database in existence to research the question at hand, research on distinction at the top of the hierarchy is commonly demanding concerning data quality. In our case, the data do not allow for covering, for example, other curricular and leisure activities, and do not allow for a more comprehensive operationalization of the educational background or other background factors without losing a substantial number of cases.
Our findings fit well with the findings of other research indicating that the combination of different strategies of distinction is one strategy of how academic households intergenerationally transmit their status. However, the results leave the question open whether or not it is relevant to consider grandparents' education in the analysis of social selectivity. The finding that grandparents' educational background is less important for curricular strategies of distinction than it is for distinction via the less institutionalized leisure strategies contributes to the current debate about the relevance of including grandparents' educational level in empirical analyses.
Funding Open Access funding enabled and organized by Projekt DEAL.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4. 0/. Bachsleitner 1985, Dr. rer. pol.,