Cumulative advantages and disadvantages in attainment of higher education: Set-analytic comparison of asymmetric inequalities in six European countries

Cumulative advantages and disadvantages

The cumulative (dis)advantage hypothesis was first proposed by Merton (1968) to describe a mechanism for allocating social rewards that increases the recognition of established scientists. Dannefer (2003) defines cumulative (dis)advantage as “the systemic tendency for inter-individual divergence in a given characteristic (e.g. money, health, status) with the passage of time” (p. 327). According to this perspective, individuals in competition for scarce positions who have prior resources are better equipped to succeed. Dannefer (2003) marks the affinity between cumulative (dis)advantage and social reproduction, which links social origin to socialization and allocation mechanisms in diverse institutions (e.g. in education). In this view, the initial resources in a parental family have an impact on early sorting and selection, which accumulate over time to reproduce the class structure.

DiPrete and Eirich (2006) describe two main forms of cumulative advantage. The “strict” form is externalized by a process, in which the rate of growth in an outcome is a function of current values of that outcome. Small differences among individuals will grow larger over time. The second form, named by DiPrete and Eirich as a status–resource interaction process, is the direct and indirect effects of initial resource variables on outcomes at different stages of the life course. (Dis)advantages tend to accumulate not only over time but also across different domains.

In empirical research, these hypotheses inspired studies of social inequality over the life course, which show how initial inequalities in social characteristics produce ever greater inequalities over time in the same or other characteristics (e.g. Giudici and Gauthier, 2009; Potter and Roksa, 2013; Ross and Wu, 1996). Another stream in empirical research concentrates on cumulative socioeconomic (dis)advantages and their impact on different outcomes (e.g. Giudici and Pallas, 2014; Kallio et al., 2016; Schoon and Melis, 2019). For example, Kallio et al. (2016) analyze how disadvantages associated with parental background, as measured using multiple indicators, are related to the probability of a child completing secondary school.

Attainment of education is often viewed as highly dependent on advantages and disadvantages passed on from earlier to later generations in terms of unequal resources and socialization patterns (Bukodi and Goldthorpe, 2013; Jæger and Karlson, 2018; Tramonte and Willms, 2010).

Parental resources and educational attainment

Research on the effect of social origin on educational attainment has a long history in Western countries. One of the central research questions these studies are expected to answer is what characteristics of the family signify intergenerational transmission of educational (dis)advantage. At the conceptual level, social origin is usually approached as a multidimensional construct and parental resources, both economic and non-monetary, are considered to be important in intergenerational transmission of educational advantage (e.g. Buis, 2013; De Graaf et al., 2000; Erola et al., 2016; Huang, 2013; Jæger, 2007). Research suggests that the impact of each kind of resource will capture various mechanisms and processes through which educational inequalities are produced (Bukodi et al., 2018; Bukodi and Goldthorpe, 2013).

Sociological theories emphasize the importance of parental non-monetary resources used to secure children’s educational success. The cultural resource hypothesis argues that the effect of social background on educational attainment is also due to the higher level of cultural resources of privileged parents (Bourdieu, 1986; Bourdieu and Passeron, 1990). Parents’ attitudes, values, goals, preferences, and cultural tastes are seen as important factors for children’s educational opportunities because the home environment has an impact on the development of children’s educational preferences and cognitive skills (see Jæger and Breen, 2016; Jæger, 2009; Tramonte and Willms, 2010). Cultural resources are also characterized as one aspect of social status besides social resources (Blossfeld, 2019; Bukodi et al., 2018; Bukodi and Goldthorpe, 2013). However, the cultural capital of resourceful parents is deemed desirable in educational systems because those systems are by design unequitable and pose barriers to children of less educated parents. Researchers often distinguish parental education as a separate resource (Barone and Ruggera, 2018; Blossfeld, 2019; Buis, 2013; Bukodi and Goldthorpe, 2013). Parental education is considered as a critical indicator for parents’ capacity to further their children’s educational career either by creating a supportive learning environment at home or by providing informed guidance concerning navigation in educational system. Highly educated parents will be more familiar with the educational system. They will stimulate their children to do well in school and will more likely secure a high level of educational attainment for their children (for an overview of contributing processes, see Kraaykamp and van Eijck, 2010).

Despite the long tradition of a multidimensional conceptualization of social origin, research generally uses just one measure of parental resources and/or concentrates on net effects of them. Recently, researchers have started to devote more attention to the ways in which these resources interact to support or weaken the educational attainment of offspring (e.g. Erola and Kilpi-Jakonen, 2017; Erola et al., 2016; Huang, 2013). Just as different types of resources might refer to the different ways they impact on educational reproduction, so are combinations of types of parental resources seen as producing a range of effects (Erola and Kilpi-Jakonen, 2017). Previous empirical evidence suggests that parental resources may not simply be substitutes for one another, but rather have a cumulative effect (Conley, 2001; Korupp et al., 2002). Parental education seems to have the greatest, and increasing, relative importance (Bukodi and Goldthorpe, 2013; Meraviglia and Buis, 2015).

Analyses have indicated that it is insufficient to equate social background solely with a father’s position (Albright, 2008; Beller, 2009; Meraviglia and Buis, 2015). Fathers and mothers are shown to contribute separately into educational attainment of children, for example, the mothers’ influence is more often exerted through factors associated with their educational attainment (Erola et al., 2016; Korupp et al., 2002). It is especially true in the situation where the woman’s level of education surpasses that of the man. We decided to include the characteristics of both parents to the analyses to be able to capture the potential accumulation of parental educational resources.

The institutional context of educational inequality

Educational system could mitigate unequal educational opportunities of children with different social background, but previous studies indicate that their role has been quite limited. Shavit and Blossfeld (1993) analyzing social inequalities in 13 countries indicate that despite dramatic educational expansion during the 20th century most studied countries exhibit stability of these inequalities. After an initial equalization largely driven by educational expansion, further equalization has substantially declined (Barone and Ruggera, 2018). However, parents’ transmission of advantages and disadvantages to their children varies across countries and time, depending on a wider context, particularly by institutions (e.g. Borgna, 2017; Breen and Jonsson, 2005; Breen et al., 2009; Esping-Andersen and Wagner, 2012).

We focus on institutional arrangements of educational systems at the levels of secondary and higher education. Stratification, standardization, and decommodification are dimensions of educational systems, which have received the most theoretical and empirical attention. ² These characteristics of education systems have an impact on the costs and benefits associated with educational choices and they are important mediating variables in accounting for the relationship between parental resources and educational attainment (Peter et al., 2010; West and Nikolai, 2013). Therefore, we do not question their relevance in analyzing the institutional context of educational inequality but aim to develop an analytical approach that enables us to comparatively reveal the importance of interplay of these three in different combinations across countries and cohorts. Our case selection is primarily guided by the assumption that our case countries belong to disparate welfare and education regimes (Ferragina and Seeleib-Kaiser, 2011; West and Nikolai, 2013; Willemse and De Beer, 2012) and we aim to analyze how the combinations of stratification, standardization, and decommodification link with the effect of parental resources on children’s educational attainment in these various contexts. Given that we have data from three different cohorts of each country, we are also aiming to explore cohort comparison to capture the potential effect of educational expansion. Thus, we are following most different system design (Przeworski and Teune, 1970; Seawright and Gerring, 2008) which argues that despite the differences across countries, we assume there to be similarities in key variables that explain the outcome. However, in discovering these similarities, we expect to see more than one route to the outcome (equifinality) as different contexts’ tendency to facilitate different institutional mixes is well captured by fsQCA logic. Furthermore, in addition to the “usual suspects” of stereotypical examples of differing regimes, we add two countries from Eastern Europe, a group that is often neglected in comparative welfare and educational research but has important differences in terms of both legacy and timing of educational expansion.

Stratification of an educational system is understood as differentiation into tracks with varying degrees of permeability between them (Allmendinger, 1989). More often, this describes the numerous levels of school curricula in secondary education, for which DE’s system is the most explicit example, as it comprises subdimensions of tracking and selectivity. However, stratification also refers to the way education systems provide various levels of access to higher education (Kerckhoff, 1995). Stratification at the higher education level is also important, alongside stratification in secondary education while the latter often carries over to tertiary level (Borgna, 2017). Thus, the earlier the tracking takes place in the education system, the more problematic in terms of educational inequality and socioeconomic background in explaining it (Van de Werfhorst and Mijs, 2010). Willemse and De Beer (2012) distinguish between two types of stratification at the higher education level. The first is differentiation, which refers to the number of tracks in higher education institutions. Tracks in this context, indicate various educational paths across and within higher education institutions that are associated with educational and occupational life chances (Willemse and De Beer, 2012). The second is vocational specificity, that is, the degree to which a system focuses on general or specific knowledge and skill attainment to prepare for a particular vocation. However, in general the more stratified the education system, the higher is the effects of parental background on educational attainment (see also Van de Werfhorst and Mijs, 2010).

Previous studies indicate that degrees of standardization, that is, the degree to which an educational system follows common, nation-wide standards and are controlled by central government, affects the power of parental resources. Standardization relates to variation in quality of educational institutions (Willemse and De Beer, 2012). Institutions may differ in variables such as budgets, curricula, and examination standards and in the degree to which they meet national standards (Kerckhoff, 1995). It is possible to identify three subdimensions: budget-making, examinations, and school curricula. While tests on the overall effect of standardization on educational attainment yield mixed results (Bukodi et al., 2018; Pfeffer, 2008), it is expected that stronger standardization may attenuate the effects of parental resources on their offspring educational attainment (Van de Werfhorst and Mijs, 2010).

According to Esping-Andersen (1990), decommodification occurs when “a (social) service is rendered as a matter of right, and when a person can maintain a livelihood without reliance on the market” (p. 22). The decommodification of educational systems characterizes the extent, to which education is provided by the state in the form of a public good, rather than being purchased as a private good in the market (Bukodi et al., 2018). There is also the hypothesis that decommodification should play an important role in moderating the effect of parental economic resources (Bukodi et al., 2018). The PIAAC data set does not include any measure of these resources. However, it is possible that decommodification might also influence the impact of parental educational resources on education attainment. The share of private education measures the extent, to which parents have the option to opt out of public education and pay for a private alternative. In general, higher levels of decommodification equate to lower intergenerational transmission of parental resources because access to higher education is open to a wider spectrum of social groups. However, the distributive dynamic of education is different compared with other welfare policies as the extent to which public dominance in educational provisions is associated with wider access to higher education is dependent also on educational inequality, restricted admission in particular (Busemeyer, 2012).

We assume the combination of high shares of decommodification and standardization and a low share of stratification to be the most favorable institutional configuration of a country’s educational system to mitigate the effect of parental resources in guaranteeing the higher educational attainment of their offspring. We will test this assumption in the macrolevel analysis section, being particularly interested to investigate the robustness of that combination or some subset of it in different contexts and across cohorts.

Institutional variation of educational systems in studied countries over time

We have six case countries, each having distinctive welfare and education regimes. In order to characterize educational systems in Britain, DE, IT, and SE we mainly rely on Bukodi et al. (2018), and for EE and CZ on a number of country-specific sources (see Appendix 3).

The British (UK) system is characterized by the gradual elimination of selectivity and tracking at the secondary level. In the 1950s, education was characterized by early tracking and a high degree of selectivity, and comprehensive education was introduced in 1965 (Bukodi et al., 2018). At the higher education level, the UK, among liberal countries has the lowest number of tracks. The pattern of decommodification is characterized by increasing public investments in education in the 1960s and 1970s, and by declining spending levels and the introduction of university tuition fees in the late 1990s. The share of privately educated students at secondary level experienced little change since the 1960s and remained moderate. Until the 1990s, the British education system was less standardized, characterized by mixed local–central budget making, decentralized examinations, and the absence of a national curriculum. The introduction of a national curriculum in 1987 and the gradual standardization of examination in the 1980s changed the level of standardization of the British education system. However, control of budget making remained at both the local and central levels.

In SE, a highly selective, early tracked system changed into a highly permeable system. The gradual introduction of comprehensive schools in the 1960s meant that access to upper secondary education became nearly universal in the 1970s and SE now has a common comprehensive track until the upper secondary level. However, a high degree of selectivity and tracking are characteristic of the upper secondary level, as access to specific tracks is granted based on average grade points (Rudolphi, 2013). Private education was not widespread in SE until the 1990s, but since then it there was a decisive increase in private provision at compulsory school level (Lundahl, 2011). The 1962 educational reform, which introduced a national curriculum and standardized examination, changed the level of standardization. Budget making remained a shared competence of local and central governments.

DE has a highly stratified education system at the lower and upper secondary levels. Its stratified nature has not changed since the 1950s (Neugebauer et al., 2013). The decommodification pattern is similar to the UK, with increases in the 1960s and the 1970s and a gradual decline since the 1980s. University tuition fees were abolished in the 1970s and reintroduced in some federal states in the 2000s. Standardization has not changed since the 1950s. Budget making is centralized at the Bundesland level and curricula are standardized. ³ Examinations depend on the Bundesländer.

According to Bukodi et al. (2018) and Willemse and De Beer (2012), IT occupies a middle position with regards to the three indicators. In the 1950s, the education system in IT was highly elitist. In the early 1960s, educational reform eliminated institutional entrance barriers to upper secondary education (Barone, 2009). This reform also reduced the level of tracking in secondary education by reorganizing the previous four tracks at the lower secondary level into a comprehensive middle school. At the upper secondary level, tracking still exists. Spending levels in education gradually increased, and high levels of private education decreased. There are no tuition fees in universities. Education in IT was historically highly standardized. The central government controlled examinations, curricula, and budgets. In the 1980s, standardization decreased somewhat. Regional authorities are involved in budget-making.

The educational system of EE during the Soviet period combined a model that did not have tracking at the lower secondary level with a strong German model with clear tracks at the upper secondary level (Helemäe et al., 2000). Stratification decreased slightly in the 1990s, but the internal differentiation of general secondary education increased in terms of both regional differences between schools, as well as differentiation between ordinary schools and “elite” schools that select pupils according to individual criteria (Põder and Lauri, 2014). In EE, the degree of standardization was high (standardized since the 1960s) until the 1990s. Overall, while the high standardization of the socialist period was reduced in the early 1990s, the second half of the 1990s witnessed an increase in standardization, most notably in the form of standardized graduation examinations at the end of the secondary school, called “national examinations.” While in recent decade the standardization slightly decreased, overall, the system is still standardized. In Soviet Estonia, the level of decommodification was very high due to egalitarian educational and social policies (Saar and Aimre, 2013), but following the collapse of the Soviet system, higher education policies could be characterized as recommodification. Higher education, since independence in 1991, saw a rapid growth in the numbers of tuition-fee paying students, both in absolute terms as well as in terms of proportion of all students. The tuition-fee free model was restored in 2013 but this shift has not had major impact on existing disparities (Põder and Lauri, 2021).

Until the 1990s, CZ, like EE, featured a high degree of stratification and standardization in its education system. The Czech education system has traditionally been characterized by its stronger emphasis on vocational rather than general education. After 1989 the stratification of the education system became even more pronounced. Upper secondary education retained its structure. However, in lower secondary education, long academic programs appeared (Starkova and Simonova, 2013). Currently, the Czech education system should be characterized as being highly stratified. The degree of standardization of the Czech education system was traditionally very high. Standardization was not ensured at the level of outcomes (there were no standardized examinations at any level), but at the level of processes. Instruction was based on a detailed syllabus, specified textbooks, and extensive guidelines for teachers (Starkova and Simonova, 2013). After 1989 standardization somewhat decreased. Uniform curricula were replaced by less detailed standards. Education at all levels was provided for free during the socialist period. There has been very little private education in the CZ. Public expenditure on education gradually rose after 1989 but remained at quite a low level (Walterova and Greger, 2007). Public universities provide education free of charge.

Figure 1 reveals a picture of institutional variations in six analyzed countries.

OPEN IN VIEWER

Figure 1.

Institutional variation of educational systems in studied countries over time, raw data.

Data and variables

Our data source in the microlevel analysis is the Programme for the International Assessment of Adult Competencies (PIAAC). Developed by the OECD and collected between 2011 and 2012, PIAAC provides internationally comparable data on skills in adult populations in 24 countries, 6 of which are the focus of our analysis: CZ, DE, EE, IT, SE, and UK. The PIAAC background questionnaire also includes a range of information regarding the factors, which influence the development and maintenance of skills such as education, social background, engagement with literacy and numeracy and ICTs, and so on. The survey was implemented by interviewing adults aged 16–65 years in their homes—5000 individuals in each participating country. We distinguish between three cohorts: Cohort 1 (born 1948–1967), Cohort 2 (born 1968–1977), and Cohort 3 (born 1979–1987). The choice of the cohorts aligns with the institutional changes in educational systems. These three cohorts enable us to capture changes in countries’ educational landscapes as outlined in the previous section, and the potential impact of these on higher education attainment. Details on the numbers of respondents in each country and cohort and the descriptions of the main dimensions included in the analyses are presented in Appendix 1.

Our outcome is attainment of higher education, ⁴ which is operationalized and based on the PIAAC question asking for the highest level of formal education obtained by the respondent. We dichotomised our outcome to distinguish between respondents with university degrees versus those without. ⁵ Appendix 1 maps the patterns of the expansion of higher education, that is, we can detect higher shares of higher educational attainment among later cohorts (the difference between the first and the third cohort is approximately 10 percentage points—21 versus 32 percent of respondents on average, respectively have higher education). SE (40%) and UK (39%) stand out as the countries where educational expansion is the highest, while IT (24%) is lagging behind. At the same time, changes in the share of higher educational attainment between older and younger cohorts are quicker in CZ and IT (from 18% to 29% in CZ, and from 14% to 24% in IT).

We have three independent variables characterizing parental family resources: the mother’s educational resources (medu); the father’s educational resources (fedu) ⁶ ; and the family’s cultural resources (books). Parental education is operationalized and based on the PIAAC question about the highest level of education for the respondent’s mother/father. Appendix 1 indicates the following patterns: First, the variability of parental educational indicators is higher than respondents’ educational attainment; second, while in most countries the proportion of higher-educated mothers approaches that of fathers in Cohort 2, this is not the case in DE; and third, CZ, IT, and UK are distinguishable in terms of having a relatively low share of higher-educated parents compared with DE, EE, and SE.

In addition to parental education, we have included the dimension of parental objectified cultural capital, operationalized by the number of books in the parental home when the respondent was about 16 years old. Previous research shows that parents` books are correlated with other aspects of parental cultural resources (e.g. reading behavior and cultural participation) (De Graaf et al., 2000; Sullivan, 2001). This measure is commonly used in education research as a proxy for cultural capital. ⁷ However, a recent critic indicated that self-reported books in the home are subject to sizable and systematic errors of observation (Engzell, 2016). Students from homes with many books perform better but better students accumulate more books and are better informed about their home libraries. There is also country variation: in countries where many books are the norm, the scope of underreporting is larger (Engzell, 2016: 17). Also, we controlled for the potential selective non-response and measurement error stemming from that under-reporting. While the non-response rate is low across our sample (less than 1%), this is slightly biased toward the low-educated. In general, families in IT and UK tend to have fewer books at home than those in CZ, EE, SE, and DE. Unfortunately, the PIAAC dataset does not include any indicators for parental embodied cultural capital. Interpreting our results, we should bear in mind that our parental capital indicators might also incorporate the impact of parental embodied cultural capital.

In the country-level analyses, where we aim to analyze the link between revealed accumulation of parental resources and the institutional setup of education we distinguish between the outcome and three conditions. We have two outcomes (which are the results of microlevel analysis): first, the link (subset consistency) between exceptional parental advantages and the presence of higher educational attainment of offspring (ADV); and second, the link between acute parental disadvantages and the absence of higher educational attainment of offspring (DISADV). We have three macrolevel conditions to distinguish between educational systems: stratification (STRAT), standardization (STAND) and decommodification (DECOMM). Macrolevel indicators for countries and cohorts are calculated using the approach presented in Bukodi et al. (2018) (see Appendices 3 and 5 for the description of the procedure to measure the values of macrolevel indicators and Appendices 4 and 5 for raw and calibrated macrolevel values of each country-cohort and calibration principles).

Analytical strategy

In both steps of our analysis, we rely on the set-analytic approach. In regression approach, the aim is usually to analyze the net effects of independent variables, while controlling for others. In this article, we explore the ways in which parental resources in combination predict educational attainment. The main difference between our approach and classical statistical analysis is that it operates with (co-)presences and absences instead of covariations. This means, for instance, that in our approach we are not relying on correlations between higher educated mothers and higher educated fathers but the overlap between those children who have both parents higher educated and the ones who have one. The higher the overlap, the higher is the degree to which resources cumulate. Overlap is preferred instead of correlations as is better equipped to reveal resources cumulation (Borgna, 2013). Besides, set analysis is better equipped to reveal potential asymmetry in associations under considerations and allow multiple explanatory routes, that is, equifinality, phenomenon relevant for our second step, macro-comparative analysis. However, the article remains explorative.

In the first step, we analyze inequality in parental resources in terms of overlapping and reinforcing advantages versus disadvantages, with a special focus on the different ways advantages and disadvantages are configured by country and cohort. We start with coincidence analysis, which is the assessment of the degree of overlap of multiple advantages and multiple disadvantages. Here the central focus is on set coincidence that is defined as the degree to which two or more groups of respondents (i.e. sets) overlap (i.e. have overlapping memberships in these groups). ⁸ We distinguish between advantageous and disadvantageous sets. ⁹ The former consists of respondents with exceptional advantages—meaning respondents who have two parents (mother’s education indicated by the acronym “medu” and father’s “fedu”) who both have higher education (ISCED5A, 6) and more than 200 books at home. The latter are respondents with acute disadvantages—consisting of respondents with two parents who have either secondary or lower education (ISCED 1, 2, and 3C short) and fewer than 100 books at home. We analyze 3-way advantages and disadvantages. Thus, the groups under investigation are (see also Appendix 2 for the sizes of these groups):

After set coincidence analysis we move on to examine the degree, to which these coinciding advantages and disadvantages are linked to attainment and non-attainment of higher education. Thus, we focus on two types of associations at individual level: first, how consistently the indicators of cumulative parental advantages are associated with the educational attainment of the respondent; and second, how consistently the indicators of cumulative parental disadvantages are associated with the educational non-attainment of respondent. The parameters of fit of these associations are the measure of subset consistency and the measure of outcome coverage. Subset consistency indicates how consistently the individuals who combine exceptionally advantageous/disadvantageous backgrounds attend/do not attend higher education. ¹⁰ Outcome coverage, at the same time, shows the prevalence of advantageous/disadvantageous among all those who have attained higher education/have not attained higher education. ¹¹

Country-level analyses rely on fsQCA ¹² to examine under what combination of the characteristics of educational system is the link between parental resources and higher educational attainment the weakest. In other words, we aim to reveal configurations of the institutional setup of educational systems under which the intergenerational transmission of parental resources is the lowest. We assume, first, that this link is best captured configurationally, as it is more the combination of the institutional characteristics of the education system that brings about the effect, second, there might be different combinations of characteristics to this effect (equifinality); and third, combinations linked to lowering link between advantages and educational attainment might differ from the ones that lower link between disadvantages and non-attainment, that is, we assume asymmetry in associations under study. To employ fsQCA, each dimension of the analysis—the outcome (the link between parental resources and higher educational attainment) and three conditions (stratification, standardization and decommodification) must be calibrated. This means that we transform the available “raw” data into fuzzy-set membership scores. We do that to be able to make two types of distinction for each dimension (set)—the difference in kind and the difference in degree. While raw data contribute for the latter, calibration attaches qualitative meaning for the set. For instance, after calibration we can distinguish between highly, and weakly, decommodified countries. ¹³ For calibration, in the instance of continuous data, we need three qualitative thresholds—fully in, crossover point, fully out. For the calibration procedure, an indirect method based on log odds ¹⁴ was used. We use packages QCA (Dusa, 2019) and SetMethods (Oana and Schneider, 2018) for calibration and analysis in R. For visualizations, we use Stata16.

Following common QCA standards, we proceed from necessity to sufficiency analysis (Schneider and Wagemann, 2012). The study accounts for the asymmetrical nature of the educational attainment by running separate analyses both for the presence of the outcome (attainment of higher education) and the absence of the outcome (non-attainment of higher education). However, while the absence of outcome is usually calculated by the negation of the presence of outcome, we use different subset consistency measures for advantageous and disadvantageous calculated in the first step of our analysis. In analyzing sufficiency, we have 8 (2³) logical combinations of countries’ institutional configurations of educational system because we have three conditions. The analysis of sufficiency proceeds in two steps: first, the evaluation of the consistency (of the evidence that a particular combination is linked to the outcome), and second, the “minimization” of the truth table (based on the Quine–McCluskey algorithm), excluding those rows which are below the threshold for consistency.

We are aware of the potential path dependency of our country-cohorts and its impact on our “ideal-types” and results, indication of which is the aspect that many countries belong to the same configurations across all cohorts (yet, as Table 3 indicates, most countries do “change” their configuration across cohorts).

Coinciding advantages and disadvantages

First, we examine the variability across birth cohorts and countries in the combinations of parental resources. How were parental resources distributed in cohorts that faced different institutional contexts of attainment of higher education? The measures of set coincidences between advantaged sets are given in Figure 2. What we can see from the left-hand side panel of Figure 2 is that the overlap of advantages increases in all countries except IT indicating the increasing accumulation of advantageous parental resources such as higher educated parents and lots of books at home in five of our case countries. The degree of coinciding advantages is highest for younger cohorts, being as high as 0.26 in SE, 0.23 in EE, and 0.21 in DE. This means, that in those countries the overlap between respondents with exceptional advantages (i.e. respondents who have all three categories of beneficial parental resources) and those respondents with only one of them (see footnote 7 on details of the calculation of set coincidence) is more than 20 percent. This indicator is somewhat lower in CZ and UK, and the lowest in IT, which is the only country where we do not see the growing accumulation of advantages. However, this is important to note here that the relative value of educational qualifications has declined over the last century, hence having one or two highly educated parents does not represent the same kind of advantage for older and newer cohorts.

OPEN IN VIEWER

Figure 2.

Coinciding advantages and disadvantages across six studied countries and three age cohorts.

The right-hand side panel of Figure 2 shows that acute disadvantages tend to coincide to a much greater degree than exceptional advantages, being as high as 0.80 in IT in the case of Cohort 1. In other words, the group of respondents who misses all characteristics of parental resources analyzed almost entirely overlaps with the group who misses one of these resources. While this degree of cumulative disadvantages in IT decreases for younger cohorts, it is still the highest—0.58 for Cohort 3. The accumulation of disadvantages of the older cohorts is also quite high in SE, UK, and EE, but this accumulation is sharply decreasing in these countries. Compared with other countries, CZ and DE have moderately low degrees of accumulative disadvantages, especially for the two older cohorts.

Parental resources and attainment of higher education

We proceed with the analysis of subset consistencies to explore the link between parental background and attainment of higher education. Table 1 presents the subset consistencies and outcome coverages of all our case countries. Whereas the former indicates the strength of the link between a particular combination of parental resources (or the lack of it) and educational attainment, the latter shows the prevalence of this combination among the outcomes. We may detect a well-known trade-off between these two parameters of fit: the higher the consistency score, the lower the coverage tends to be. In IT, for instance, the link between exceptional advantages and educational attainment is very high (over 0.9 in the case of Cohort 1) indicating that the advantageous background almost guarantees an attainment in higher education in IT. At the same time the degree to which all respondents who have higher education is “covered” by that exceptionally advantageous background, is low (6%). Thus, while accumulated parental resources are a guaranteed route to higher education (high consistency), there are several other routes to the higher educational attainment at place (low coverage).

OPEN IN VIEWER

Table 1.

Subset consistencies and outcome coverage of advantageous and disadvantageous groups.

	Subset consistency			Outcome coverage
	Cohort 1	Cohort 2	Cohort 3	Cohort 1	Cohort 2	Cohort 3
Advantageous group; has attained higher education
Czech Rep.	0.69	0.91	0.81	0.06	0.11	0.17
Germany	0.63	0.58	0.58	0.11	0.18	0.24
Estonia	0.68	0.63	0.55	0.12	0.27	0.28
Italy	0.94	0.80	0.73	0.06	0.06	0.05
Sweden	0.58	0.64	0.69	0.14	0.26	0.31
UK	0.64	0.76	0.74	0.05	0.11	0.12
Disadvantageous group; has not attained higher education
Czech Rep.	0.93	0.95	1.00	0.15	0.05	0.04
Germany	0.92	0.94	0.92	0.13	0.11	0.07
Estonia	0.90	0.94	0.99	0.43	0.20	0.07
Italy	0.93	0.91	0.90	0.83	0.71	0.59
Sweden	0.89	0.85	0.82	0.47	0.28	0.14
UK	0.87	0.83	0.86	0.53	0.36	0.24

Source: Authors’ calculations based on PIAAC 2011.

One important message from Table 1 is that countries are becoming more diverse in terms of the degree to which advantageous background is linked to the attainment of higher education. In SE, UK, and CZ the link between exceptional advantages and attainment of higher education has increased, in DE it has remained quite stable but in EE and IT has decreased. Still, in EE and DE advantageous social origin has a somewhat lower impact on attainment of higher education, especially for the youngest cohort. For SE, UK, and IT, we see much higher subset consistencies (0.69–0.74) indicating that the link between advantageous background and educational attainment is quite strong for the youngest cohort. Accounting for the recommended level of quasi-sufficiency (0.7–0.75, see endnote 9), we may conclude that in CZ and UK for younger cohorts and in IT for all cohorts, the link between advantageous background and higher educational attainment is systematic. While we might be interpreted it as a sign of expanding educational opportunities: those with more resources are the first ones to exploit such opportunities (Raftery and Hout, 1993), for our design it is important that there are cross-country and -cohort differences independent from educational expansion and motivating us to look for system-level explanations.

We proceed with the subset consistency and outcome coverage analysis of disadvantageous groups and explore the connection between acute disadvantages and (higher) educational non-attainment. Table 1 indicates that the degrees of these connections are much higher compared with advantages. This means that the link between an acutely disadvantageous background and the absence of higher education is much stronger and more prevalent. However, the share of those with an acute disadvantageous background among all those who are not higher educated is decreasing in all analyzed countries. At the same time the patterns of subset consistencies, that is, the degree to which acute disadvantages are linked to the absence of higher educational attainment, are more diverse. SE and IT are countries where the situation is improving for the disadvantageous as the link between the accumulation of disadvantages and the absence of higher education is decreasing across cohorts, in DE the situation is stable in that regard and in CZ, EE, and UK this link is increasing. For younger cohorts in CZ and EE, this means, that being a disadvantageous person, opportunities to get higher education is very limited or even absent. Again, as in the case of advantages, some of these dynamics are present because the condition of having low-educated parents is becoming less and less common, not because the causal link between parental resources and educational attainment is weakened. Still, given the overall high values of subset consistencies, we may claim that for a person not to attain higher education it is sufficient to have acute disadvantages. ¹⁵

To conclude, this section first revealed that with some exceptions, coinciding advantages are an increasing trend and coinciding disadvantages a decreasing one, an expected outcome in the light of educational expansion. Second, the absence of parental resources hinders higher educational attainment of offspring to a much larger extent than the presence of parental resources enables it, the former meeting consistency thresholds across all cohorts only in CZ (also in the older cohorts in IT and UK), and the latter in all country-cohorts. Third, the direction of these two types of linkages is divergent across cohorts, SE being the only exception with diminishing intergenerational transmission of parental resources in both directions. In other words, SE is the only country where the intergenerational transmission of parental background is diminishing as both associations, exceptional advantages in facilitating and acute disadvantages in hindering higher educational attainment are decreasing.

Macrolevel analysis

In this section, we bring together our micro- and macrolevel analyses and examine under what combination of institutional setup of a country’s educational system is the intergenerational transmission of resources, that is, the link between the accumulated parental resources (or the absence of it) and children educational attainment/non-attainment, the weakest. Based on our literature review we assume the configuration of high decommodification, high standardization and low stratification to be the most promising combination from that regard. However, we do not know the viability of this assumption in diverse contexts and in either direction. We employ fsQCA to detect these institutional routes and have two outcomes. First, the link between exceptional advantages and higher educational attainment which is captured by the subset consistencies of advantages (Table 1, upper panel). Second, the link between acute disadvantages and no higher educational attainment, which is captured by the subset consistencies of disadvantages (Table 1, lower panel). In using fsQCA, all dimensions included in the analysis are calibrated (see Appendix 4 for calibration thresholds and raw and calibrated data) to distinguish qualitative differences between the values, for example, to distinguish between country-cohorts with strong and weak intergenerational transmission of resources.

Figure 3 shows the ranges of our two outcome sets and chosen calibration thresholds. As we lack strong conceptual arguments for concrete thresholds, our choices for outcome calibration are empirically driven. While choice of calibration thresholds with strong conceptual grounding would be a preferred choice, the empirically driven alternative is also accepted by QCA scholars (Thomann and Maggetti, 2020), especially when followed by sensitivity analyses as we do (Appendix 6). We know from the previous section that in the case of accumulated advantages country-cohorts rarely meet the sufficiency criteria (0.75) of the strong link between advantageous background and higher educational attainment while in the case of acute disadvantages all countries meet the 0.75 threshold. Thus, we cannot rely on this in transforming our raw data to fuzzy data as it would miss the requisite division between “positive and negative outcome cases.” Therefore, we rely on empirical means and gaps close to these to decide on the relevant variation in the outcome and offer some robustness check in testing the sensitivity of chosen thresholds in the light of our results (Appendix 6). Based on our decisions on thresholds (0.7; 0.905) we have 11 country-cohorts, which have a positive outcome regarding advantages, that is, where the link between exceptional advantages and higher educational attainment is weak (Figure 4). We also have 8 country-cohorts, which have a (relatively) positive outcome regarding disadvantages, that is, where the link between acute disadvantages and no higher educational attainment is relatively weak. We emphasize “relativity,” as we know from the previous section that the link between disadvantageous background and educational non-attainment is strong in all countries. Note also, that we reverse the calibration thresholds in the case of outcomes, that is, the higher the indicator of raw data, the stronger is the link between background and educational attainment meaning a lower score in calibrated outcome.

OPEN IN VIEWER

Figure 3.

Outcome(s) and calibration threshold(s): subset consistencies between parental background and educational outcome across countries and cohorts.

OPEN IN VIEWER

Figure 4.

Outcome(s) and calibration threshold(s): subset consistencies between parental background and educational outcome across countries and cohorts.

For macro-level variables the choice of calibration thresholds is based on the arguments in the related literature (Bukodi et al., 2018 particularly) and the data at hand. In the case of stratification, we have chosen a crossover point at 0.45, leaving us with nine country-cohorts, which have stratified (the prevalence of various tracks at the secondary level and selective entrance to the upper-secondary) educational systems—all cohorts in DE; all cohorts in CZ and the oldest cohorts in IT, UK, and EE (see Appendix 4 for raw and calibrated data). In the case of standardization, the cross-over is set at 0.7, leaving us with 14 highly standardized country-cohorts. We acknowledge this set is skewed and offer a robustness check with an alternative calibration of stratification (Appendix 6). Yet, we admit the limited explanatory power of the dimension of standardization, even after the robustness check, as many countries have similar value on standardization. Thus, country-cohorts that do not have standardized educational systems, that is, do not have centralized budgeting or standardized examinations and curriculum, according to our operationalization and calibration are UK (in the cases of Cohorts 1 and 2) and CZ (in the case of Cohorts 2 and 3). In the case of decommodification we set the threshold at 0.69 leaving us with 7 highly de-commodified country-cohorts—SE across all time periods analyzed, and earlier periods in CZ, DE, EE, and UK.

The necessity analysis (Table 2) reveals that none of the single conditions meet the criteria for necessity (usually recommended to be as high as 0.9, Schneider and Wagemann, 2012). However, there are combinations in both, analyses of advantages and disadvantages, that turn out to be necessary. However, the literature (Oana et al., 2021) emphasizes that in analyzing necessary conditions in addition to consistencies, the empirical relevancy must be considered (measures of coverage and relevance in Table 2) in interpretation and research practice indicates that a RoN close to 0.5 could be reason for concern (p. 74). From that perspective only one necessary combination from the analysis of disadvantage (strat + stand) meet our expectations of parameters of fit. This means that countries to have a low intergenerational transmission of parental resources in case of disadvantages, it is necessary to either have low stratification or low standardization.

OPEN IN VIEWER

Table 2.

The analysis of necessity of institutional characteristics in mitigating intergenerational transmission of parental resources.

Advantages	Parameters of fit
Conditions and configurations	Consistency	Coverage	Relevance
STRAT	0.56	0.69	0.77
STAND	0.79	0.67	0.55
DECOMM	0.49	0.87	0.94
strat	0.52	0.60	0.70
stand	0.31	0.61	0.86
decomm	0.57	0.52	0.51
STAND + decomm	0.91	0.60	0.22
STRAT + STAND + DECOMM	0.93	0.64	0.31
Disadvantages	Parameters of fit
Conditions tested	Consistency	Coverage	Relevance
STRAT	0.36	0.33	0.61
STAND	0.75	0.47	0.44
DECOMM	0.50	0.67	0.86
strat	0.82	0.72	0.77
stand	0.44	0.67	0.88
decomm	0.63	0.43	0.47
strat + stand	0.94	0.65	0.61
start + STAND	0.92	0.51	0.36
strat + decomm	0.93	0.61	0.54

Source: Authors’ calculations based on PIAAC 2011.

To continue with the sufficiency analysis and to examine the link between outcome and configuration of the institutional setup of a country-cohort’s educational system, we employ the truth table algorithm that transforms all data based on calibrated membership scores to configurations. The first column in Table 3 indicates the number identifying each configuration (assigned automatically by the software). The next three columns indicate the presence (1) or absence (0) of each condition. The truth table is based on three conditions representing different institutional characteristics of educational systems and the 18 country-cohorts in this analysis are represented by 6 configurations out of 8 logically possible combinations. ¹⁶ For each, we report the consistency and PRI coefficients. The latter measures the strength of the argument that a certain combination of conditions could also produce the negative outcome. Table 3 indicates, first, that many of our countries changed their policy-mix, except SE that had the policy mix of no stratification, high standardization and high decommodification (strat*STAND*DECOMM) across all three cohorts. Second, we have many configurations that do not meet our consistency criteria—0.75. Third, in the case of advantages, the presence of decommodification is a necessary component in all three highest consistency score configurations (so called INUS conditions; see Oana et al., 2021).

OPEN IN VIEWER

Table 3.

Truth table: the sufficiency of institutional configurations in mitigating intergenerational transmission of parental resources.

Advantages
No.	STRAT	STAND	DECOMM	OUT	Cases	Consistency	PRI	Cases
4	0	1	1	1	3	0.94	0.91	SE_C1, SE_C2, SE_C3
8	1	1	1	1	3	0.84	0.78	CZ_C1, DE_C1, EE_C1
7	1	1	0	0	3	0.72	0.67	DE_C2, DE_C3, IT_C1
1	0	0	0	0	1	0.67	0.48	UK_C2
5	1	0	0	0	3	0.61	0.54	CZ_C2, CZ_C3, UK_C1
3	0	1	0	0	5	0.53	0.38	EE_C2, EE_C3, IT_C2, IT_C3, UK_C3
Disadvantages
No.	STRAT	STAND	DECOMM	OUT	Cases	Consistency	PRI	Cases
4	0	1	1	1	3	0.94	0.91	SE_C1, SE_C2, SE_C3
1	0	0	0	1	1	0.86	0.77	UK_C2
3	0	1	0	0	5	0.57	0.36	EE_C2, EE_C3, IT_C2, IT_C3, UK_C3
8	1	1	1	0	3	0.53	0.15	CZ_C1, DE_C1, EE_C1
5	1	0	0	0	3	0.51	0.28	CZ_C2, CZ_C3, UK_C1
7	1	1	0	0	3	0.35	0.05	DE_C2, DE_C3, IT_C1

Source: Authors’ calculations based on PIAAC 2011.

STRAT (Stratification), STAND (Standardization) DECOMM (Decommodification), Czech Republic (CZ), Estonia (EE), Germany (DE), Italy (IT), Sweden (SE), and the United Kingdom (UK).

Number underlined with country abbreviation denotes cohort; that is, SE_C3 denotes the third (the youngest) cohort in Sweden and so on.

Consistencies and PRIs in bold font denote the ones that meet out threshold criteria for sufficiency (0.75). Country-cohort abbreviations in bold font denote the country-cohorts that are positive outcome cases (set membership score in outcome above 0.5 based on our calibration; see Figure 2 for the visualization).

We proceed by minimizing our truth table, ¹⁷ and as a result we have one sufficient route in the case of advantages and two sufficient routes in the case of disadvantages (Table 4). Starting with advantages, the only institutional combination that is (quasi-)sufficient to mitigate the effect of exceptional advantages in intergenerational transmission of parental resources is the “Swedish way” of combining high levels of standardization and decommodification (Route 1), a route that is followed also by the older cohorts in CZ, DE, and EE. The remaining two parameters of fit, coverage and PRI are also good, thus we may conclude that institutional setup of educational system that combines high standardization and high decommodification is sufficiently linked with the mitigation of the impact of advantageous parental resources on educational attainment. In inspecting truth table (Table 3) we see that in addition to “Swedish way,” configuration 7 that combines high stratification with high standardization and low decommodification (“German way”), has also relatively good parameters of fit but consistency remains slightly below suggested 0.75.

OPEN IN VIEWER

Table 4.

Sufficient institutional configurations that mitigate intergenerational transmission of parental resources.

Cases	Advantages		Disadvantages
		Route 1		Route 2	Route 3
	Membership in outcome	STAND * DECOMM	Membership in outcome	strat * STAND * DECOMM	strat * stand * decomm
CZ_C1	0.64	0.86
DE_C1	0.98	0.53
EE_C1	0.76	0.86
UK_C2			0.98		0.76
SE_C1	1	0.8	1	0.8
SE_C2	0.97	0.8	0.97	0.8
SE_C3	0.64	0.8	0.64	0.8
Route consistency		0.86		0.94	0.87
Route PRI		0.82		0.91	0.79
Route raw coverage		0.44		0.39	0.27
Route unique coverage		0.44		0.32	0.21
Solution consistency	0.86			0.92
Solution PRI	0.82			0.87
Solution coverage	0.44			0.60

Source: Authors’ calculations based on PIAAC 2011.

STRAT (Stratification), STAND (Standardization) DECOMM (Decommodification), Czech Republic (CZ), Estonia (EE), Germany (DE), Italy (IT), Sweden (SE), and the United Kingdom (UK). Number underlined with country abbreviation denotes cohort; that is, SE_C3 denotes the third (youngest) cohort in Sweden and so on. *denotes multiplication (logical AND)

Consistencies and PRIs in bold font denote the ones that meet out threshold criteria for sufficiency (0.75). Country-cohort abbreviations in bold font denote the country-cohorts that are positive outcome cases (set membership score in outcome above 0.5 based on our calibration; see Figure 2 for the visualization).

In the context of disadvantages, we reveal two sufficient routes linked to the mitigation of the effect of acute disadvantages in hindering the intergenerational transmission of higher education. Yet, Route 3 has only one country (i.e. low coverage), and thus we can conclude: for a country-cohort to avoid the intergenerational transmission of acute disadvantages it is (quasi-)sufficient to combine low stratification with high levels of standardization and decommodification. This route is followed by SE across all cohorts and in comparing it with the analysis of advantages, is the subset of Route 1. ¹⁸ Hence, the combination of high standardization and high decommodification works either way, weakening our argument for asymmetries in the associations between the institutional mix that mitigates the link between advantageous background and educational attainment versus the link between disadvantageous background and no education attainment.

We also employ robustness checks to offer some tests on the sensitivity of our results (see Appendix 6).