Education-based thinking and acting? Towards an identity perspective for studying education differentials in public opinion and political participation Open Access

Ever since scholars started studying citizens’ opinions and political acting, they have found substantial educational differentials (e.g. Centers, 1949/1961, p. 218; see also Converse, 1972, p. 324; Farnen & Meloen, 2000; Nie, Junn, & Stehlik-Barry, 1996; Putnam, 2000, p. 186). Comparative and longitudinal research has also revealed that these differences remain stable (e.g. Hakhverdian, Van der Brug, & De Vries, 2012) or even grow (e.g. Hakhverdian, Van Elsas, Van der Brug, & Kuhn, 2013; Kalmijn & Kraaykamp, 2007; Stubager, 2010; Van der Waal, Achterberg, & Houtman, 2007). Finally, longitudinal analyses of party manifestos shows that the salience of issues on which higher and less-educated people occupy different positions has increased over time (e.g. Achterberg, 2006).

Not surprisingly, then, in recent years a number of scholars have started to explore whether education is becoming the object of group-based acting and thinking (Bovens & Wille, 2010, 2012; Dekker & Van der Meer, 2009; Stubager, 2009; Tannock, 2008; Van Bohemen, Kemmers, & De Koster, 2011). Despite often large mutual differences, these accounts, we hold, pave the road for a ‘new’ approach to interpreting observed educational differentials in thinking, acting and feeling. This is centred around the idea that the large and longstanding educational differences in (political) thought and action render education a source of social identity. The constituent pieces of such an approach, however, lie scattered over a broad and fragmented literature, rendering it difficult to assess the empirical value of this body of thought. We aim to bring the pieces together and provide this emerging approach with a more robust theoretical background.

In this paper, we draw attention to the particularities of education-based status and subsequently explain how studying educational differentials in political acting and public opinion from an identity approach contributes to a better understanding of these differentials. We also explain how this approach complements existing interpretations of educational differentials, propose a research agenda and present survey data that illustrate the empirical starting-point for such approach.

The framework we propose in this paper seeks to bring together two different bodies of research and thought. The first and more recent strand explores to what extent educational differentials grow into a political consciousness which is conceptualised in a manner fairly similar to traditional class consciousness, suggesting that modern societies are heading toward education-based conflict and cleavage. The second and older one draws attention to the institutional position of education in modern societies, together with processes of the legitimation of social inequality, and suggests that strongly politicised educational identities are unlikely to develop. A brief discussion of both yields the building-blocks for our own approach, which we present in the third section.

The observations that educational differentials in political opinions and acting are (1) substantial, (2) found in most Western countries and (3) relatively stable or even growing over time, have made scholars start to wonder about the ultimate consequences of these differences (e.g. Bovens & Wille, 2010, 2012) and to call for an inclusion of psychological measures that assess education-based group identity (Dekker & Van der Meer, 2009). The latter task has been taken up by others (Spruyt & Kuppens, 2014; Stubager, 2009, 2013). We briefly discuss the work of Bovens and Wille (2010, 2012) and Stubager (2009, 2013) here as an example of this body of research and thought.

Starting from the observation that the less educated are underrepresented on almost every rung of the political participation ladder, Bovens and Wille (2010, p. 418) draw attention to the potential dangers for democracy whereby

Stubager (2009) investigated whether the so-called libertarian-authoritarian value dimension can be considered an educational cleavage. This refers to an alignment of attitudes and issues on which (1) the higher and less educated not only occupy very different and opposing positions, but whereby (2) these groups also become aware of these differences and (3) this awareness and the associated group identification in turn contributes to the widening of the gap between these groups on the same alignment (see also Van Bohemen et al., 2011).

In these accounts, the awareness of educational differences in themselves become the object of group acting and contributes to the reproduction of these differences. Indeed, both Bovens and Wille's and Stubager's approach should be seen as an elaboration of the more general notion that ‘[ … ] everybody knows and everybody knows that everybody else knows that education rules in modern society [our emphasis]’ (Kingston, Hubbard, Lapp, Schroeder, & Wilson, 2003, p. 55). For Bovens and Wille (2010) this lies in their interpretation of the relation between the steep increase of the average educational level of political representatives over time and the low level of political participation of the less educated. According to Bovens and Wille, the latter can be interpreted as a reaction to the former, which thus implies an awareness of educational differences. In Stubager (2009, 2013) the awareness element takes the form of the classical notion of political self-consciousness, that is, (1) the growth of an awareness of the social differences between educational groups, (2) identification with one group whereby (3) this identification and the associated feelings of belonging affect both behaviour and opinions. Despite their differences, both accounts analytically follow a similar reasoning about the awareness of educational group differences. This reasoning strongly relies on the classical approach to the study of the development of a politicized class consciousness (e.g. Centers, 1949/1961) and cleavage approach (cf. Bartolini & Mair, 1990).

Critics of this approach draw attention to the fact that group consciousness – that is, a politicised understanding of one's ‘objective’ location – does not spontaneously develop bottom-up but is instead created by political actors such as political parties or media. With respect to the educational groups, they point to the absence of political parties that mobilise on the educational issue, and the importance of individualism in current societies as a counteracting force to group identification (Hakhverdian et al., 2012). Both elements fit within a vast body of literature on the reproduction and legitimation of relationships of social inequality.

Scholars who study the effects of education from a more general conflict-sociological perspective stress that one of the pathways through which educational attainment influences thinking and acting is by orienting people to social positions and specific socioeconomic contexts in an authoritative way (see Meyer, 1977). Educational credentials do not only perform a gatekeeping role, but do this in such a way that unequal access to certain positions is perceived as ‘just’ and ‘deserved’. These authors, in short, draw attention to the effects of education as ‘an institution’: effects of continuously and authoritatively classifying and ranking people based on their educational attainment (see also Bourdieu, 1984; Solga, 2002; Tannock, 2008). According to Meyer, who did not study this using empirical data, this process fosters what he calls ‘chartering’: a specific way of acting or thinking because one knows that that is the dominant way of acting or thinking among people who have attained a similar position in the educational hierarchy. Indeed, in this account too the awareness of educational differentials plays a key role. What distinguishes the reasoning of Meyer and those who hold similar views from that central in the previous approach is that this awareness itself is thought to be an element in the reproduction of educational differentials, rather than the starting-point of a revolt against them. We now provide two examples – one with respect to the higher educated, another related to the less educated – that will further clarify this argument.

The idea that the more highly educated are aware of educational differentials and that this awareness influences their acting and thought, is clearly present in Jackman and Muha's (1984) reasoning about ideological refinement (see also Kuppens & Spears, 2014; Smith & Seelback, 1987; Wodtke, 2012). Jackman and Muha posit that the more ‘tolerant’ attitudes that are often found among the higher educated result from ideological refinement rather than moral enlightenment. The higher educated know that in contemporary societies being tolerant serves an important indicator for being recognized as educated, and develop corresponding attitudes. This mechanism explains both why the allegedly more tolerant highly educated often act in an intolerant way (e.g. the ‘white flight’ in education) and why educational differentials are not found when more subtle measures for ethnic prejudice are used (Kuppens & Spears, 2014). Clearly, in this example it is assumed that the more highly educated know education's relevance, know what the differences between the higher and the less educated are, and know how they are expected to act. This emphasis on the knowledge component, however, should not mislead us. It merely concerns what Bourdieu refers to as ‘practical’ knowledge, that is, perception schemes linked to labels and classifications that do not necessarily reach the level of coherent political consciousness and discourse. In this regard it is interesting to note that the tolerant attitudes of the more highly educated do not merely result from self-presentation in response to social desirability concerns (Heerwig & McCabe, 2009; Ostapczuk, Musch, & Moshagen, 2009). This means that the education effect is not due to an attempt by the more highly educated to deliberately misrepresent their true attitudes, but results from the ‘practical recognition’ of social hierarchies (Swartz, 2013, p. 81).

There are also examples of the awareness of educational differentials among the less educated. Bourdieu (1984, chapter 8) has already found that political participation – starting with having and articulating an opinion about political issues – is substantially lower among the less educated. Bourdieu explains these differences in terms of what he calls ‘le droit statutaire’. It concerns a feeling of ‘entitlement’, the feeling of having a right (or even a moral obligation) to express opinions in public (see also Myles, 2008). Bourdieu holds that a lack of feelings of entitlement results in self-exclusion (compare Hooghe, 2001). Feelings of entitlement should not be simply equated with the sense of political powerlessness often referenced in political research, and also differ from a realistic assessment of one's own ‘political competences’:

Feelings of lack of efficacy or political competence all refer to individual properties which can be individually expressed on an absolute standard. In contrast, feelings of entitlement are by definition the product of a comparison with others: ‘Technical competence is to social competence what the capacity to speak is to the right to speak, simultaneously a precondition and an effect’ (Bourdieu, 1984, p. 409). What matters then in this view is (1) the (more or less explicit) awareness of one's position in (2) the educational hierarchy. It is the perceived rather than the absolute educational position that is in question, and this perception is also more clearly defined in terms of education-based labels (e.g. ‘highly’ and ‘less’ educated). Bourdieu (1998, p. 50) underscores the importance of these labels when he terms educational credentials ‘objectified symbolic capital’: ‘[ … ] a reputation for competence and image of respectability and honourability’ (Bourdieu, 1984, p. 291). Symbolic capital provides dominant groups with social status; it converts power into prestige by depicting realizations as an outcome of ‘natural gifts’ (Bourdieu, 1990).

Symbolic capital, as seen by Bourdieu (1984), is not an attribute of high-status groups themselves, but relies on recognition by others. As the product of acts of mutual recognition, symbolic capital is by definition unstable. The distinguishing feature of educational credentials when compared to other forms of symbolic capital, then, is that they are ‘objectified’ (Bourdieu, 1998, pp. 50–51):

Consequently, allocation based on educational credentials is expected to be considered legitimate by both the more highly and the less educated. In this way Bourdieu locates education-based labels at the centre of acts of recognition and this of course requires knowledge about these groups, their hierarchy, and one's own group membership and position in that hierarchy.

Meyer, Jackman and Muha, and Bourdieu share with the recent work of Bovens and Wille and Stubager that they all call for a theory about group relations. However, Meyer, Jackman and Muha, and Bourdieu pay attention to the role of power, differences in power positions and the role public actors (such as the media, political parties and education itself) play in the (re)production of these inter-group relations. These elements are almost completely absent in the recent work about the ‘diploma democracy’ or ‘the educational cleavage’. We argue here that they are nevertheless crucial to advance this issue.

This second strand of thought has been criticised on two grounds. First, while Meyer and Bourdieu provide compelling arguments, their theories are not based on empirical data or else they involve post-hoc interpretations. What is lacking are persuasive empirical data which show how these processes themselves (rather than their outcomes) work in practice. Secondly, Bourdieu's work especially has been criticized for leaving too little theoretical space for acts of resistance (e.g. Bourdieu & Eagleton, 1992). Power differentials are thought to elicit more resistance than Bourdieu was willing to admit (for an overview see Swartz, 2013). People are not passive victims of prejudice, but motivated agents attempting to protect their identities from threat (Davies, 1995). The key question, then, becomes how both strands of thought can be reconciled.

The preceding arguments can be combined in three propositions whose combination paves the road for a new type of explanation that is clearly distinguished from, but complementary to, existing explanations for educational differentials in acting and thinking.

The starting-point is, first and foremost, education's centrality in contemporary societies and its associated capacity to authoritatively allocate people to social positions that differ in status. Part of education's importance in current societies derives also from the fact that differences according to educational background are represented as legitimate. This legitimacy of educational differentials is not a static fact, but a practice that is continually reproduced by (1) the application of institutionalized allocation principles on the labour market, (2) in the educational practice itself and (3) in public discourse.

With respect to educational practice, Bourdieu points to the repeated and detailed testing of educational outcomes throughout a person's educational career, the increasing use of intelligence tests which attempt to measure children's scholastic potential and in doing so ‘naturalize’ educational outcomes, the content of education itself, which celebrates creativity and is characterized by an individual-centred approach, and ‘limited’ social mobility which conceals the (often strong) intergenerational reproduction in educational outcomes (e.g. Bourdieu, 1984, p. 415, 1990, pp. 21–22). These characteristics combine in ‘a culture of the gift’ (Reed-Danahay, 2005, p. 488): the representation of scholastic success as the result of superior and natural qualities (Swartz, 1997).

Education-based status gains additional importance in public discourse, in particular through the way in which politicians constantly represent education as a universal problem-solver for all kinds of individual and societal problems (Smeyers & Depaepe, 2011; Spruyt, 2012), a fact epitomised in Tony Blair's statement ‘Ask me my three priorities for government and I will tell you education, education and education’. Wood and Skeggs (2008) also point to the growing importance of reality-shows characterized by an overrepresentation of less-educated participants who are typically put into a trajectory of training and education. The latter, like educational practice itself, in a more subtle way underscores the importance of education in contemporary societies. One should see this as a socializing effect which, however, transcends the educational sphere and becomes part and parcel of the collective representation of societies. Therefore, education-based status cannot simply be equated with general social status.

The way in which educational differentials become connected to legitimised status differentials in practice can also be derived from a content analysis of 824 articles published in 2 Flemish newspapers (Spruyt, 2012). This analysis revealed five different strategies through which the differences between the more highly and less educated are constructed and represented. Common to all these strategies was that educational differentials are presented as large – that is, as a gap – but also as essentially non-conflictual. Three elements contributed to the non-conflictual aspect of the representation: (1) the idea that one's educational position is not definitive (via the existence of lifelong-learning opportunities), (2) the link between education and competence and (3) the idea that current societies operate in a global competition for knowledge and development which renders the educational competition inevitable and endless. These elements explain why interpreting the world through the lens of educational differences rather than through class distinctions becomes appealing for dominant groups: it offers them the opportunity of distinguishing themselves from the dominated in an essentially non-conflictual way. For the very same reasons, education-based categories have little to offer the less educated. One hypothesis that can be directly derived from this reasoning is that the more highly and the less-educated resort to different social categories to interpret societal problems. Whether this is actually the case is of course an empirical question. But the example illustrates nicely how considering the effects of education from an identity perspective, as we propose, involves taking into account the social production of categories and relationships of power.

Secondly, it is assumed that people are aware of their position in the educational hierarchy, for the simple reason that (1) educational credentials often act as gatekeeper for different positions and (2) the importance of education is regularly stressed in media and political discourse. In this way diplomas increasingly correspond to labels (e.g. ‘highly’ or ‘less’ educated) that are used to classify people (for an application to labour market processes see Solga, 2002). This centrality undermines the probability that education-based categories will simply be ignored or downplayed by the less educated. Indeed, as explained by Kuppens, Easterbrook, Spears, and Manstead (2015, p. 1262), what renders being less educated different from many other stigmatised identities is that it is defined in terms of the absence of something (education). This makes it even more difficult to connect it with positive characteristics and further increases the likelihood of dis-identification with this category. Such dis-identification should not be interpreted as the absence of an educational awareness among the less educated, but rather as its consequence.

What matters in all this are not specific educational credentials but the position people occupy in the hierarchy of more highly and less educated. And although the boundaries between these labels are not necessarily fixed or undisputed, an individual's opportunities to increase the likelihood of being recognized as more highly educated are limited, precisely because of the institutionalised (or what Bourdieu terms the ‘objectified’) nature of the classifying practices.

Thirdly, education-based labels are connected with opinions and behaviour considered characteristic of the more highly or less educated (Jackman & Muha, 1984). People do not only know whether they are seen as highly or less educated. They also know, albeit possibly on a practical, non-discursive level, where these labels stand for. This enables them to classify persons (and to be classified) in the educational hierarchy by merely observing their opinions and conduct. It is this awareness that underpins what Meyer calls ‘chartering’. It is also this connection between attitudes/behaviour and educational categories that renders the difference between attitudes – as a cognitive matter – and political behaviour somewhat superficial. Attitudes seen in this way are not so much held by actors but rather posed in social contexts (see also Billig, 1996). A more fruitful distinction in our view pertains to whether the context in which attitudes or behaviour are posed is competitive or not. Taking part in a discussion or standing for election are competitive acts in which people strive for scarce goods (recognition, a position and so on). In such situations it is likely that social categories are more easily activated.

These three propositions fuel the expectation that education-based labels are employed in processes of social distinction. It is precisely because people are familiar with these labels and their social meaning, and know who will and who will not be recognized as highly educated that they, in particular the less educated, become sensitive to references to their educational attainment. From Figure 1, for example, it can be inferred that people will try to avoid the social stigma associated with being classified as less educated. The lower their attained educational level, the more likely people were to classify the level ‘higher secondary education’ as ‘highly educated’. Furthermore, a comparison between people's actual educational attainment and their classification of that level indicated that 49.8% of the respondents considered themselves ‘highly educated’. Such tendencies are well-known in social psychology and referred to as ‘social mobility’. People try to avoid the social stigma of being less educated by identifying themselves with the highly educated (Huddy, 2001, p. 135). It is in this way that education becomes the object of group behaviour without implying a revolt of the less educated. It is also in this way that education-based social identities may independently influence opinions and acting. Revealing in this context are data from an online dating platform which show that similarity in education significantly increased the rate of both sending and replying to initial contacts (Skopek, Schulz, & Blossfeld, 2011). These findings underscore that education-based labels are used in daily interaction processes. Moreover, and specifically related to the ‘dating business’, in regions like Flanders or the Netherlands, some agencies are specifically designed for education-based groups, more specifically the highly educated. So whereas there are indeed no political parties that rally support in terms of education-based labels (see above), these labels are used both in daily interactions and by public actors. We are only about to begin to explore the consequences of this.

The social categorisation approach in social psychology might help us understand how the relation between identities, attitudes and behaviour plays out in practice. In sociological and political research, the conceptualisation of social identities has commonly been restricted to cases in which group members identify very strongly with their group. Such identities are considered (1) permanently salient, (2) acquired early in life and (3) relatively stable over the life course.² Centers (1949/1961, p. 27), for example, describes class as ‘a psychological phenomenon in the fullest sense of the term. That is, a man's class is a part of his ego, a feeling on his part of belongingness to something; an identification with something larger than himself’, and others compare party identification with religious identity (Campbell, Gurin, & Miller, 1971). Such conceptualisation of (political) identities is also central in the cleavage perspective as developed by Bartolini and Mair (1990). Thus far, discussion about the existence of education-based identities has been too strongly framed in this perspective. Scholars who argue against the existence of education-based identities are right when they claim that education is no substitute for traditional class consciousness as defined in the cleavage perspective. Our argument rests on a different conceptualisation of social identities, one we borrow from social identity and categorisation theory.

The early writings of social identity theory also focused to some extent on how group identification becomes a means for collective action for groups who feel unfairly disadvantaged (Tajfel & Turner, 1986).³ Later, however, social identity theory and research developed a more generally applicable framework for social identity processes, also called self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Self-categorization theory aims to predict when exactly people will see themselves as individuals or as members of a particular group. The salience of a particular social identity depends on an interaction between individual and contextual factors. An important individual factor is the individual's identification with a social group. For example, the more one identifies with the highly educated, the more easily this identity will become salient across situations. Taking into account education's central position and the associated social stigma of being less educated, it is reasonable to expect that the highly educated will be more inclined to identify with education-based groups. Contextual factors constitute the fit between the social identity and the current reality. For a group of friends with a variety of educational backgrounds, taking part in a public political debate or organizing a formal meeting of a voluntary association might make those educational differences more salient than if they were only having a drink (because it increases the fit between the education identity and the current situation). For less-educated citizens, experiencing the fact that all politicians clearly bear the attitudinal, behavioural, and linguistic markers of the highly educated (and are therefore likely to be different from other people they know) might make their own educational background more salient than is the case for highly educated citizens. Finally, it is self-explanatory that simply mentioning these group labels in explicit or implicit ways is one of the most efficient ways to increase social identity salience.

The importance of self-categorization as a group member is that it triggers a process of depersonalization whereby everything the person perceives, thinks, and does, s/he does as a member of a group that is compared to other groups (see Voci, 2006). If social identities are not necessarily permanently salient, but rather activated under particular conditions, the key question becomes how to identify these conditions. A priori, any context in which education is mentioned or in which educational differences are apparent has the capacity to make education-based identities salient and trigger the process of depersonalization. It is clear from our earlier discussion of the centrality and awareness of education that such instances are likely to be abundant indeed. But research also shows that especially in politicised contexts, defined here as contexts in which people talk about politics and societal questions, group-based labels are very often used and social distinctions are easily accessible (e.g. Devine, 1992). Therefore, for this reason too politics seems to be a good starting-point for exploring the societal implications of education-based group identity. What categorization theory adds to the study of educational differentials is that it considers these differentials not as static fact, but as something which is realized in concrete situations and as such depends on an individual's characteristics and those of the situation.

How then might all this enhance our understanding of educational differentials? Experimental research into so-called ‘stereotype threat’ has convincingly shown that merely activating the awareness that one is counted by others as belonging to a group for which negative social stereotypes exist, produces threat effects among the people involved (e.g. Steele, 2010). Rather than ‘an internalization of failure’ – as theories about resentment among the less educated and about the ‘losers in the modernization process’ often assume – it here concerns a purely cognitive matter. Importantly, the persons involved do not have to agree with the existing stereotypes or to personally identify with a particular group. Merely knowing that others see them as group members produces stereotype-threat effects. So far, such stereotype-threat models have only been used for the study of differences in educational performances, but they carry the potential to be applied to other areas as well, for example in the context of differences in political participation. Research into stereotypes shows that ‘the educated’ are associated with competence-related traits (Cuddy et al., 2009). The activation of these competence-related stereotypes in concrete situations may explain why the less educated refrain or withdraw from political participation, independently of their personal political interest, knowledge or competence.

The preceding arguments suggest the existence of a vacuum between on the one hand approaches that consider existing educational differentials to have the potential to grow into educational conflict, strong versions of education-based consciousness and political action, and on the other hand those who consider education a form of symbolic capital that elicits little resistance among the dominated (here, the less educated) and even a tacit and unconscious acceptance of one's fate. We believe that exploring the space between both extremes will lead us to a new view on educational differences (in political conduct and thought), which complements existing approaches. In the remainder of this paper we discuss four research directions which are mutually related but clearly distinguishable. Each of them focuses on one of the aspects we outlined as crucial in an interpretation of educational differentials that relies on an identity approach.

Differentials in political thought and behaviour are attributed to different socialization experiences. Although discussion remains about the ‘causal’ effects of education (for an overview Persson, 2015), we argue that that discussion should be broadened. Our approach complements the existing focus in socialization research in two ways. First, it clearly urges for expanding the concept of (political) socialization. If ‘the most powerful socializing property of a school is its external institutional authority, derived from the rules of educational allocation, rather than its network of internal socializing experiences’ (Meyer, 1977, p. 61), then we have to study how and when education instils a feeling of entitlement among young people. In this way we have to go beyond the traditional concepts such as political knowledge, sophistication and feelings of efficacy, which all focus on relatively stable individualistic properties. It seems plausible that these elements contribute to such feelings of entitlement, but how efficient are they in a society that continuously emphasizes the importance of ‘being educated’? Feelings of entitlement have not so much to do with the questions ‘Can I? Am I able to?’ They rather pertain to questions such as ‘Should I? Is it up to me to participate, to articulate an opinion in public, and so on?’ So far, to the best of our knowledge no attempts have been made to measure ‘feelings of entitlement’ directly. One way to do so would rely on measuring the types of emotions people experience – such as the unease or anxiety a less-educated person experiences when (s)he enters a situation where most people are highly educated. In our view feelings of ‘entitlement’, secondly, will be highly context-dependent, that is, they will depend on the salience of educational categories in particular contexts. Traditional surveys or even panel studies are not very useful for tapping into such processes. The increasingly popular experiments in political science, on the contrary, offer interesting opportunities in this regard (Druckman, Green, Kuklinkski, & Lupia, 2011). If people are tacitly aware of social hierarchies and their position within the status structure, subtle manipulations of the salience of ‘education-based labels’ or the awareness of the ‘educational composition of the context’ should result in identifiable adaptations of thought and acting. The outcome of a lack of feeling of entitlement is always the same: self-exclusion. Therefore, this research could ultimately offer new insights into how political participation among the less educated might be increased.

An example will render things concrete. Low turnout rates, high levels of political dissatisfaction and the success of populist parties generate concern about the current state of democracy. One attempt to bring politics back to the people consists of initiatives whereby a group of citizens, so called mini-publics, discuss and search for solutions for societal issues (for an overview see Della-Porta, 2013, chapter 4). The distinguishing feature of deliberative democracy compared to other political innovations is the stress on preference transformation during these discussions. Deliberative democracy entails active interaction among equals so that only the strength of the argument determines the outcome of discussions. No matter how appealing this idea is in theory, scholars have always had reservations about the possibility of creating the right conditions for it in practice (e.g. Holdo, 2015; Olson, 2011). People differ in what Sanders (1997, p. 349) calls ‘epistemological authority’, that is, the capacity to evoke acknowledgement for one's arguments. In addition, we would add that ‘epistemological authority’ will itself be at stake and contested in these discussions. Because education is linked to competence in public discourse, it seems very likely that educational attainment and education-based labels will be used in such discussions. Whether this is actually the case can and should be empirically investigated. Such research would look for answers on questions such as: Do people rely on education-based labels in such discussions, and if so, how? What happens in such groups when education is put forward as a means to solve societal problems? But more basic questions arise too. Who is willing to participate in initiatives such as citizen discussions and under what conditions? Do people prefer more socially homogeneous groups? One can approach this from a static point of view by asking such questions in a simple survey (e.g. Neblo, Esterling, Kennedy, Lazer, & Sokhey, 2010), but our approach suggests relying on a more dynamic design whereby people are given information about who, in terms of educational background, is likely to participate. By manipulating this information and assessing willingness to participate, we may see self-exclusion at work. Such research is likely to enhance our understanding of the persistence of the social reproduction of educational differentials. This is an initial way in which an identity approach may advance our understanding of educational differentials: education yields hierarchical distinctions which correspond to legitimized power differentials so that, when this hierarchy is activated, the highly educated are likely to benefit from it via the compliance of the less educated.

A second implication and research opportunity concerns the need to devote more attention to the ‘constructed nature’ of public opinion when interpreting survey results. Hakhverdian et al. (2012) correctly emphasize that today there are no political parties that muster support on the basis of the educational issue and that this reason alone makes an open educational conflict unlikely. With this argument they align themselves with the idea that the representations and categories that people use to interpret their society are, to some extent, provided to them by public actors such as political parties, but also formal education or media (for very different variants of this argument see Boltanski & Thévenot, 1983; Evans & Tilley, 2012). Authors such as Jackman or Bourdieu radically expand this argument by claiming that these representations and classifications support the interests of dominant groups and thus contribute to the reproduction of existing inequalities. Research inspired by this insight would assess how education and competence are connected in public discourse and how the relationship between the more and less highly educated is portrayed in public discourse (see Spruyt, 2012). One excellent example of this type of analysis can be found in Tannock (2011). He analyses how education-based discrimination intermingles with discussions about immigration. Whereas Tannock himself primarily focuses on the implicit construction of superior and inferior immigrants, his analysis provides a good example about the way education is intertwined with a wide range of public discussions and in this way occupies a central place in public discourse.

Besides research that documents mechanisms resulting in such a general representation of education as a general indicator of competence, we also need research which links representation of this kind to strategic positions in the field. Do political parties for which we know that their electorate differs strongly in terms of educational attainment (e.g. the Green and Populist Right parties) differ in how they talk about education and ‘being educated’? One hypothesis that can be directly derived from our arguments is that parties which attract on average highly educated voters (such as the Green parties and some Social Democrat parties) will be more inclined to use education to indicate social positions and to solve societal problems, whereas right-wing parties will rely on class talk.

Attention to the capacity of societal institutions such as education or the media to instil representations in citizens may also potentially advance an important discussion in social science. Authors who call for more attention to educational differences in public opinion and political behaviour encounter considerable opposition, in particular from authors studying the Anglo-Saxon countries. Characteristic is Evans’ (2010) reaction to Stubager's (2010) attempt, with regard to Denmark, to demonstrate that the association between educational attainment and the position people take on the libertarian-authoritarian value dimension has increased over the last three decades. Evans criticizes Stubager for using too crude an indicator of class (Erikson–Goldthorpe–Portocarero) and urges him to employ a more detailed occupational classification. Interestingly, this is precisely what Houtman (2003) has argued against. Houtman does not deny that the nature of employment changes over time and that it is therefore self-evident that occupational classifications evolve as well, but he asserts that one first has to demonstrate that occupational experiences actually affect the opinions and behaviour of citizens. If not, the effects of education might simply be hidden by a detailed measure of occupation, which of course depends heavily on educational attainment. The discussion about the importance of education versus class has now reached the status of a matter of faith, whereby believers and non-believers simply ignore each other. That is exactly why it seems worthwhile to reconsider this issue from a different angle. The way in which we would like to reopen or resolve this debate is by asking whether it is not plausible that both strands of thought originate in real differences in the way education and class are connected in the different regions from which these opposing researchers come. The fact that these regions differ substantially is indisputable. Higher education in the Anglo-Saxon countries, for example, is characterized by high tuition fees that are almost absent in Flanders, Denmark or the Netherlands. Secondary education in Flanders is even free of charge during the period of compulsory education. The electoral landscape also varies substantially between these regions. It is probably no coincidence that questions about an educational consciousness primarily emerged in regions where low or even absent electoral thresholds yielded a highly fragmented political-party landscape with parties that almost exclusively rally support on issues related to the ‘new’ or ‘cultural’ alignment (comprised of themes such as immigration, the gap between citizens and politics, criminality, feelings of unsafety and so on). The presence of only two large parties in the UK meant that these ‘new’ themes were incorporated into the existing party structure. As a consequence, these issues and the associated educational differentials attained less salience. At a more general level, the history of class is much more prominently present in the Anglo-Saxon countries when compared to the pillarized history of the Low Countries. We could make a longer list of differences, but the bottom line is that all these elements render it plausible that the educational issue and education-based categories are more strongly intertwined with traditional class politics in the Anglo-Saxon countries as compared to regions such as Flanders, the Netherlands or Denmark. Whether this is effectively the case is an empirical question which can and should be the object of empirical research. It is precisely when one acknowledges the constructed nature of public opinion that it becomes problematic to assume that the mechanisms underpinning public opinion formation work identically in all countries. Assessing the relevance of the elements mentioned here could make it possible to advance this discussion.

The fourth and final example of a new research opportunity concerns citizens’ opinions. The idea that opinions are to some extent socially constructed by public actors such as political parties does not imply that public actors create attitudes ex nihilo. In order to be successful, political parties have to address and positively resonate with attitudes and sentiments held by a significant part of the population. Therefore it is worthwhile to assess citizens’ views on the questions raised by the discussion about ‘the diploma democracy’ or ‘the educational cleavage’, that is, ‘Do people construct social identities around their educational level?’ and ‘Do people perceive educational conflict?’ So far, data on this issue have been gathered in Denmark (Stubager, 2009), the Netherlands (den Ridder & Dekker, 2011) and Flanders (Spruyt, 2014). They show that the highly educated perceive less educational conflict than the less educated do, and that the evaluation of the socio-political position of the highly educated by the less educated is far more negative than the evaluation the highly educated make of the position of the less educated. This finding dovetails with the classic conceptualisation of class consciousness. Self-consciousness is always directed at change and dominant groups have everything to gain from the status quo; ‘inertia is on their side’ (Jackman, 1994, p. 379). Since open conflict drastically reduces the possibility of exerting influence, it becomes appealing for dominant groups to represent the dominated as ‘different’ and subsequently express a general appreciation of differences. Individualism, in this view, is seen to be a crucial aspect of relations of dominance. Dominant groups, according to Jackman, develop attitudes that can best be described as ‘paternalism’, a conditional appreciation in which the dominant group retains a positive picture of the dominated group for as long as it does not fundamentally question the dominant group's socio-political position. Paternalism among dominant groups is not seen as the outcome of a deliberately undertaken project of symbolic imposition but as the product of interaction within a symbolic environment of institutionalized inter-group relations. Further research should assess the extent to which education-based identities explain differences in acting and opinions. The social identity approach we referred to earlier predicts in this context that the boundary work that accompanies the existence and activation of education-based identification will further deepen (rather than simply reproducing existing educational differentials, as in the research line associated with a different focus within socialization theory). Such identity-effects, which essentially boil down to conformism with the group one identifies with, are not accounted for in the traditional interpretations of educational differentials and can be considered a second contribution of the identity approach we propose.

It seems likely that research into the existence of traces of education-based identification will show that the often-used ‘level of education’ will prove too crude an indicator to capture education's full relevance. Differences according to the field of study are likely to be relevant as well. More specifically, and following Bourdieu's (1984) two-dimensional conceptualization of social space, we expect that among the highly educated people who have followed education that rewards cultural rather than economic capital (social-scientific education, literature, philosophy and so on) will be much more sensitive to education-based distinctions when compared to their counterparts who followed studies conferring economic rather than cultural capital (business or law, for example).

The latter element also touches upon another important issue. One of the limitations of existing research on the development of an educational consciousness is that it studies education-based identity in isolation. In order to fully take account of the potential asymmetry of group relationships (as explained earlier), we should also investigate how traces of an education-based group consciousness relate to other forms of consciousness. In the survey we used earlier in this paper, people were asked to indicate the amount of perceived conflict between seven pairs of groups. These pairs related to economic, social, gender and regional differences and provided some idea of the relative salience of perceived educational conflict. Table 1 presents the frequency distribution for the seven different groups. We also calculated the differences between three educational groups (lower secondary or less, higher secondary, higher education) in the amount of perceived conflict, and expressed the size of the differences between these three groups in an eta-parameter. Table 1 warrants two conclusions. First, of all pairs of groups considered, educational groups are seen as the most conflictual. More than 46% perceived (very) much conflict between the highly and the less educated. Secondly, differences by respondents’ education in terms of perceived conflict are largest for the educational groups, with the more highly educated seeing very little conflict between educational groups and the less educated perceiving considerable conflict between these groups. Instead of two groups who both perceive a large degree of conflict, an asymmetry appears whereby one group perceives a lot of conflict and the other very little (see Spruyt, 2014 for an elaboration of this point). This pattern (1) is typical for institutionalised relationships of dominance (Jackman, 1994), whereby the dominant group denies conflict in order to preserve status quo, and (2) at the same time illustrates that such dominance does not adopt hegemonic properties. The key question, then, becomes what concrete societal consequences all this has.

When studying this matter we should not forget a lesson Centers (1949/1961, p. 218) taught us a long time ago with respect to class politics:

In this paper we took stock of existing questions and research concerning the growth of an education-based group identity. We presented an approach that combines insights from two strands of literature, formulated several concrete proposals for further research, and provided data that underscored the potential for this research. Studying the four research lines as presented will bring research on public opinion and behaviour more in line with insights derived from theory on power and relations of dominance. It is in this way that we can enhance our understanding of the persistence and reproduction of educational inequalities in actions and opinions.

From a more general point of view, our approach seeks to combine a sociological approach with insights from social-psychological research. Whereas sociologists have provided quite convincing analyses of the socio-structural factors that underpin and reproduce social inequalities, they often assume rather than study the psychological processes at work. Social psychologists, on the other hand, often study psychological process in highly decontextualized contexts so that it becomes difficult to assess the external validity of their findings. Against this background, our approach entails a plea and provides several suggestions for population-based experiments, where these experiments are inspired by or even conducted in real-world situations. The results and interpretations of such studies might have the virtues of combining internal and external validity.

No potential conflict of interest was reported by the authors.