The politics of early school leaving: how do the European Union and the Spanish educational authorities ‘frame’ the policy and formulate a ‘theory of change’

Abstract The article analyses the interaction between the European Commission and a sample of educational authorities in Spain with regard to the policy against early school leaving. Although this member state scores the highest proportion of early school leavers, apparently it is not adopting some key recommendations issued by the Commission. In fact, while educational policy studies regret this ‘resistance’, studies on EU policies suggest that the EU and the states normally negotiate the ambition and the evaluation of policies in complex ways. In this vein, the article draws on a method of discourse analysis to observe to what extent these educational authorities ‘frame’ the policy in the same terms and share a similar rationale or ‘theory of change’. In brief, the findings point out that the EU, the Government of Spain and two significant regional governments retrieve a similar ‘frame’ but do not agree regarding the ‘theory of change’.


Introduction
Between 2001 and 2010 the Lisbon Agenda put competitiveness and social cohesion high in the agenda of the European Union. Since 2010, the Europe 2020 Strategy has adopted similar goals so as to foster smart, inclusive and sustainable growth. In this vein, the European Council and the European Commission have designed a wide-ranging policy framework addressed to effectively reduce early school leaving to 10% of each cohort by the end of the decade. However, an official report has already noticed that, despite a positive trend, the rhythm of actual reduction is too slow to achieve the target by 2020 (EACEA, European Commission and EURYDICE, 2013, 21, 22).
The analysis of this policy benefits from insights suggested by previous research on comparative education and European policy studies. Experts in education policy have repeatedly argued for an encompassing approach that tackles all the factors that induce students to drop out of either academic or vocational secondary education (NESSE, 2009). As a consequence, the official guidelines recommend to prevent the problem in advance, to intervene on the conditions of the students exposed to this risk, and to compensate actual early school leavers so that they improve their education and skills. However, a recent appraisal of the current trends has observed a widespread 'resistance to EU policies' among the governments of the member states (Timmerman and Willems 2015, 17).
This article draws on studies of both EU policies and education policy to spell out the clues of this 'resistance to EU policies' . The first two sections introduce the theoretical rationale and the method of research. Then, two latter sections present the results of a discourse analysis of the 'frame' and the 'theories of change' used by decision-makers.

The European Union and education policy
Since the Lisbon Agenda was launched in 2001, the Commission and the Council have actively attempted to persuade member states to align their education, labour market and youth policies with a core set of premises. The rationale of this policy is that these three areas have to match with one another in order to both underpin social cohesion and strengthen skills development (Nicaise 2012;Ruiter 2010). This endeavour has eventually contributed to shape a European social field of education policy (Dale and Robertson 2009;Stéger 2014), where the Commission has actively looked for an alliance with the OECD (Grek 2010(Grek , 2013Lingard, Martino, and Rezai-Rashti 2013).
Nevertheless, although the mainstream guidelines remind decision-makers of the varied connections between education and other policy areas, a growing number of studies on education policy find out that member states seldom adopt these recommendations. First, in 2010 the bulk of the Lisbon Agenda goals was included in the Europe 2020 Strategy, but the significant underachievement of the previous phase was neither clearly recognised nor systematically evaluated (Pépin 2011). Second, images of 'fit' between the EU and the member states policies are explicitly constructed with varied intentions. For instance, some small member states and some regions who wanted to distinguish their education policy from the larger state used the Lisbon Agenda to launch their own strategic plans (Rambla 2013). Alexiadou, Fink-Hafner, and Lange (2010) and Alexiadou and Lange (2013) also notice that some governments actively give a low profile to the relation between their education policy and that of the EU. Third, despite the common guidelines set by the Lisbon Agenda and the Europe 2020 Strategy, the understanding of many concepts diverges across member states. For instance, comparative findings have detected disparate understandings of quality assurance in England, Scotland and Sweden (Croxford, Grek, and Shaik 2009;Ozga, Dahler-Larsen, and Segerholm 2011a). Disparities are also remarkable regarding centralisation and decentralisation in England, Finland, France, Germany, Scotland and Sweden (Dupuy 2012;Ozga, Segerholm, and Simola 2011b). Forth, the conclusions of a recent study on governance of educational transitionsconducted across a varied sample of member states, already noticed that different political strategies are currently launched at the EU, member state and local levels. The decisions made at each level are only partially aligned with the decisions made at the other ones. Thus, sometimes educational policy-makers ignore the other levels (Parreira do Amaral, Dale, and Loncle 2015). The outcomes depict a large landscape of empirical variations.
The literature on policy-making in the European Union posits a complementary perspective that contributes to make sense of this apparent paradox. In fact, both the EU and the governments of the member states draw on sophisticate discursive strategies to pursue their interests in the multi-layered space of European politics (Jessop 2007;Keating 2009;Radaelli, Dunlop, and Fritsch 2013). Eventually, these actors shape a social field where they negotiate and influence one another.
The Commission often looks for recognition by calling for impact assessments of its own policies, and engages in reinterpretation by using the corresponding publications to actively craft the public image of its own activities (Adler-Nissen and Kropp 2015;Radaelli, Dunlop, and Fritsch 2013). At the same time, European policies normally respond to the interests of power coalitions that produce and deploy discourses to legitimise their interests (Carstensen and Schmidt 2016). Similarly, the education policy of the EU fashions a complex social field (Dale and Robertson 2009;Parreira do Amaral, Dale, and Loncle 2015) where authorities experiment with varied modes of policy learning. In Spain, educational authorities seem to have adopted EU policies to reduce early school leaving in a 'competitive' way . Instead of reviewing their whole approach (i.e. engaging in 'mutual learning'), policy-makers proclaim their endeavour to meet the official benchmark but deflect the recommendations of the European Council and the European Commission.
This article draws on discursive institutionalism in order to spell out some clues of the currently mainstream policies against early school leaving. In contrast with post-modernist and post-structuralist hermeneutics, critical discourse analysis does not assume that discourses constitute reality. On the contrary, the starting point claims that discourse is a social practice whereby social agents deal with meaning and communication. But these agents undertake other social practices such as weaving social networks or following institutional path dependencies that are not intrinsically discursive. Beyond the scope of linguistics, social sciences may be interested in discourse insofar as it is an entry point to observe some particular phenomena (Schmidt 2008).
Critical discourse analysis highlights some rhetorical devices whereby social agents set the limits of a particular stance. Crucial to this function is the usage of the catchwords that align a variety of concepts with a coherent discourse (Fairclough 2003, 133). This discourse 'frames' the strategy of the EU to tackle early school leaving. Framing consists of organising subjective experience so that the audience associates some concepts automatically with other ones (Goffman 1997). In politics, social movements use frames to justify their vindications and denigrate the position of opposing parties (Benford and Snow 2000). Political actors often frame such themes as new public management or Europeanisation (Diez 1999;Vogel 2012). Therefore, in accordance with the results of other studies on the European Union, it is relevant to inquire whether the Commission and the Council share the same catchwords with the national and regional governments of some member states.
Statements of fact are another key rhetorical instrument of political discourse. These statements convey knowledge exchanges between social agents such as policy actors (Fairclough 2003, 119). A growing strand of research increasingly shows that international development programmes (Stein and Valters 2012), public policies (Stame 2004;Rogers, 2008) and even financial strategies (Jackson 2013) rely on sets of statements of fact that connect definitions of problems with alleged solutions. Normally, these studies label these sets of statements as 'theories of change' . Policy and international relations scholars also capture this feature of political discourse when discussing the growing influence of epistemic communities in linking problems with solutions (Croxford, Grek, and Shaik 2009;Davis Cross 2013).
Unsurprisingly, 'theories of change' are a common resource in European politics (Carstensen and Schmidt 2016;Radaelli, Dunlop, and Fritsch 2013) as well as in European education policy-making (Auld and Morris 2014;Grek, Lawn, Lingard, Varjo, et al. 2009, Grek, Lawn, Lingard andVarjo 2009;Lingard, Martino, and Rezai-Rashti 2013;Simola et al. 2011). Therefore, these researches also suggest that the reception of the message of the EU on early school leaving by member state governments can also be scrutinised by exploring to what extent both parties share the same 'theory of change' .

Context, corpus of discourse and coding
The article explores to what extent, in Spain, a number of educational authorities adopted the EU 'frame' and 'theory of change' regarding early school leaving during the Conservative administration that was in office between 2012 and 2016. This section will briefly report on the main methodological decisions taken by the researcher regarding the sample of educational authorities, the corpus of documents and coding.
The sample of educational authorities is composed by the governments of Spain as well as the regions of Andalusia and Catalonia. Spain is the EU member state with the highest rate of early school leavers. Although some improvement has been noticeable since 2008, progress has not been as significant as in Malta and Portugal, which are the other two extreme cases (European Commission 2012b).
In Spain, since the majority of educational competencies were transferred from the central to the regional governments between 1980 and 2000, policies on early school leaving are also coordinated among layers of governance. Basically, the central government issues framework laws, which must be supported by a qualified majority, to define the curriculum, the legal status of public and private schools, the status of teachers' jobs, the model of school organisation and certain specific programmes fostering coordination between regions. Regional governments are in charge of human-resource management and language policy if Spanish is not the only official language in their territory. These authorities are also responsible for particular aspects of the curriculum, school organisation and their own specific programmes.
The main policy approach to early school leaving was formulated by a framework act approved by the last Socialist administration in 2006 (GovSpain-MoE, 2006). This piece of legislation required to tackle early school leaving by offering remedial courses to students at the last year of the compulsory, comprehensive lower-secondary education. It also asked educational authorities to offer initial vocational training programmes to the sixteen-yearsolds who were either repeating grades or had failed to get their secondary education certificate. Initial vocational training mostly had to encourage students to get back to regular schooling and take their secondary education certificate exams. The 2006 act also intended to coordinate central and regional policies against early school leaving. Thus, both levels of government contributed to fund programmes that delivered academic and leisure afterschool activities to the most vulnerable students (Manzanares and Ulla 2012). But these programmes were severely downsized because of the fiscal adjustments starting in 2010.
In 2013, a Conservative administration reformed the framework education act substantially (GovSpain-MoE 2013a). A number of articles were changed in order to introduce external, compulsory examinations for all schools, and establish a new division of the whole lower secondary education into academic and vocational tracks. Thus, the remedial and initial vocational training programmes established in 2006 were scaled up so as to distribute all thirteen-year-olds either in academic or in these newly vocational tracks. Since most regional governments were led by the Conservative party between 2011 and 2015, many of them welcomed this approach.
Both the governments of Andalusia and Catalonia had previously decided to pass their own Education Act on the grounds of the 2006 Organic Act. This is a plausible criterion to include them in the sample. However, looking at these two regions is also relevant for other reasons. First, they are the most populated of the seventeen autonomous regions. Second, while Catalonia is slightly wealthier than other regions in terms of income per capita, the score of Andalusia is significantly lower. Third, both of them were governed by opposition parties between 2012 and 2016. The Socialist party was in office in Andalusia, as it has been since the eighties. The Nationalist Liberal Party was in office in Catalonia between 2011 and 2015. Although it received the support of the Spanish Conservative Party at the beginning, this alliance collapsed when the Catalan Liberal Nationalists endorsed the campaign for independence in late 2012. Therefore, since a sample including these two regions accounts for socio-economic and ideological disparities, any similarity between their official discourses on education policy is particularly telling.
The corpus of discourse is the list of documents taken into consideration for an analysis (Fairclough 2003). In this research, such corpus was built by means of document search and interviews. Eventually, it contains references to some interviews conducted between 2013 and 2015 in Brussels, Madrid and Barcelona. Although similar interviews were conducted in Andalusia, finally they were not used. Table 1 lists the selected documents. These pieces have been issued by a variety of political actors. The Commission and the Council were in conversation with the OECD and several networks of experts and civil society representatives. The selected educational authorities also produced their own views on their policy. Thus, it is plausible to ask whether this network of political actors shared the same frame and theory of change. Coincidence would be especially relevant, since the very diversity of voices at first suggested that some divergence was likely.
The documents were coded in an inductive way (Corbin and Strauss 1990). First, their references to the main concepts of the EU framework were identified. This operation produced lists of references to 'prevention' , 'intervention' , compensation' and 'coordination between governments' . Second, each list was closely analysed in order to find out similarities and differences. The RQDA package was used for this purpose (Huang 2014). The following two sections report on the results of these analyses. The first one observes to what extent this sample of educational authorities framed their policy against early school leaving in the same way as the European Commission. Then, another section compares the views of the different authorities regarding the 'theory of change' whereby early school leaving was to be reduced.

Common frame but different understandings
At the very beginning of the Europe 2020 Strategy the European Council (2011) issued a Recommendation on policies to reduce early school leaving. That text shares a constellation of references to prevention, intervention and compensation with many other statements of the European Commission (2012cCommission ( , 2012dCommission ( , 2013a. Remarkably, other actors coincide exactly in the same terms, namely the OECD (2012a), some networks of experts (NESSE 2009;Oomen and Plang 2014), and some civil society organisations such as COFACE (2010), ETUCE (2011) and EUNEC (2014. In this view, prevention has to do with wide, societal factors that may reduce the risk of early school leaving. For the EU, preventative measures may deal with general themes as the quality of childhood or the design of an encouraging curriculum. Other measures consist of implementing more specific policies addressing segregation, the needs of schools in disadvantaged areas and the value of language diversity. Prevention also requires designing a strong career guidance system. Intervention and compensation are the other two instruments of a policy that deals with early school leaving. Intervention must consist of early warning systems, a focus on individual needs, systematic support, extra-curricula and outof-school activities, and support to teachers and families. Compensation has to draw on second chance courses, recognition of competences and local, proactive outreach action to find early school leavers (European Council 2011: C191/4-5; European Commission 2013a, 12).
Certainly, over time governments have taken some preventative measures in Spain. For instance, early childhood education and care was notably improved in the nineties. In 2006 the coordinated programmes supported schools in disadvantaged areas. In 2013, the minister of education evoked prevention to argue that the new reform strengthened the flexibility of educational pathways insofar as headteachers could decide which tracks were offered in each school and recruit part of the teaching staff accordingly (Wert 2013, 7).  Commission (2012aCommission ( , 2012bCommission ( , 2012cCommission ( , 2012dCommission ( , 2013aCommission ( , 2013bCommission ( , 2013cCommission ( , 2013d • European Council ( Nevertheless, although all the governments included in the sample retrieved the EU strategy for the sake of legitimation, eventually they understood prevention in a very restrictive way. To start with, when the Government of Spain reported on the measures adopted by the regional educational authorities, the official report coincided with the Commission to portray support to disadvantaged zones as a measure of prevention. But remarkably, it assumed that school strategies targeted to students with either low performance or special education needs were also preventative measures. Interestingly, the Commission classifies the same measures as intervention (European Commission 2013a, 12).
As a rule, regional governments implement school strategies and programmes aiming at academic success by supporting and fostering instrumental forms of learning that lead to better educational results. They also cater to the most disadvantaged zones and populations at risk of exclusion, who therefore are exposed to a higher risk of early leaving. Most regions strengthen academic guidance and remedial courses as soon as any shortcoming in the basic competences of a student is detected. Thus, the education process may pursue properly. An array of programmes catering to special education needs are another widespread preventative measure. (GovEsp-MoE 2013b, 70) The Government of Andalusia designed a particular policy against early school leaving. It followed the central government as far as initial vocational courses for low-performers and coordinated support for vulnerable students were concerned (GovAndalusia-MoE 2007). But it also launched an experiment with pedagogic innovation, and implemented a social benefit targeted to low-income students. This benefit delivers 600€ per year if these students enrol in either academic or vocational post-compulsory programmes (GovAndalusia-MoE 2011, 2012). At first sight, this measure has to do with intervention through support to families and students. But the official description claims that it is a preventative measure.
Scholarship 6000 [supports] low-income students so that they complete their compulsory education and enrol in either academic or vocational post-compulsory courses. In the 2009-2010 school year the Ministry of Education of Andalusia implemented Scholarship 6000 in order to achieve two goals. First, this programme is expected to prevent early school leaving and posit incentives for further education. Second, it has to compensate families and individuals for the income they do not earn when the youth remain at school after sixteen. (GovAndalusia-MoE 2016) In 2012, the Government of Catalonia launched a strategic plan to foster school success. The first line of action somehow retrieved the idea of prevention at the societal level by improving teachers' professionalism. But all the other ones explicitly focused on intervention on intraschool processes. In this view, early school leaving was conflated with absenteeism and had to be mostly addressed by means of compensation.
The following guidlines pursue these goals. The guidelines eventually pattern the action of the Ministry of Education of Catalonia in both compulsory and post-compulsory school programmes.
(2) Providing personalised support to tackle learning problems in early childhood and primary education. (3) Focusing on literacy and reading. (4) Pedagogic innovation in the classroom. (5) Improving communication, intercultural education and multilingualism so as to improve students' language skills. (6) School autonomy. (7) Professionalising the role of principals. (8) Involving families in students' academic and personal development. (9) Relationships with the educational community. (10) Absenteeism and early school leaving: initiatives and training strategies favouring the return to the educational system of person who previously dropped out. (GovCatalonia-MoE 2012, 20-26) Significantly, the Conservative central administration shared this restrictive understanding with the Socialist government of Andalusia and the Nationalist Liberal government of Catalonia. The fact that educational authorities overlooked the definition of prevention circulated by the Commission is not only surprising because Spain scored the highest rate of early school leavers in the European Union. The point is that the prevailing approach in Spain also disregarded the problems of disadvantaged students. Remarkably, in Spain many students repeat grades from the very primary education, and their average academic performance is also quite low compared to repeating students in other OECD countries (OECD 2014: table IV 2.2). Moreover, the percentage of disadvantaged students who attend disadvantaged schools is higher in Spain than in the OECD average (OECD 2011, 93, figure II 5.6;OECD 2014). The mainstream reading of prevention in the country did not take curriculum and teaching practices into account either. However, according to the OECD (2014: table 6.1), in Spain only 33% of secondary education students work in small groups to come up with a joint solution to a problem or task. The figure contrasts with 79% in Denmark,36,8% in France,and 58,4% in England. Similarly,compared to 45% in Denmark,75% in France,and 17% in England, in Spain 85% teachers never observe other teachers' classes and provide feedback (OECD 2014: figure 6.10).
Notably, divergence regarding the scope of prevention did not trigger conflict between the European institutions and the Government of Spain. Thus, the introduction of early tracking within the compulsory lower-secondary programme in 2013, at first sight, departed from the guidelines of the European Commission (2013a). But the role of member states in EURYDICE did not seem to allow for highlighting this disparity so much so that in 2014 the report conceded that the new education act might eventually underpin the flexibility of pathways (European Commission, EACEA, EURYDICE, and CEDEFOP 2014).

Overlapping but disparate theories of change
Critical discourse analysis is helpful to spell out the semantic relations of synonymy and antonymy that the European Commission established between early school leaving, on the one hand, and prevention, intervention and compensation, on the other hand (Fairclough 2003, 133). Firstly, due to the alleged damage for the potential for smart, inclusive and sustainable growth, early school leaving was featured as synonymous with a problem and an economic cost. Secondly, the chain of prevention, intervention and compensation became synonymous with both the solution and a possible economic saving. These semantic relations constructed a 'theory of change' whereby the problem was antonymous with the solution as well as the cost with the saving (European Commission 2013a; European Commission, EACEA, and EURYDICE 2013;European Commission, EACEA, EURYDICE and CEDEFOP 2014).
In 2013 interviewees from the Commission staff admitted that only a few member states were fully implementing the common strategy, but they suggested that this strategy could become a significant criterion to evaluate social funds in the middle term. At that time, the official position stated that multidimensional action was the appropriate response to a multi-faceted problem (European Commission 2013a, 3). The EU claimed that properly aligned measures were likely to reduce early school leaving. Generally speaking, this opinion was widely shared by the OECD (2007), the European teachers union (ETUCE 2011(ETUCE , 2013 and networks gathering school councils (EUNEC 2014) as well as experts on educational studies (NESSE 2009), economics of education (Brunello and De Paola 2013) and career guidance (Oomen and Plang 2014).
The work of the TWG on ESL [early school leaving] has confirmed that good results in reducing ESL can only be obtained if a comprehensive strategy is in place, based on strong and continuous political commitment of all stakeholders in its implementation, at all levels. The TWG has highlighted the importance of striking the right balance between central and local measures, and between preventative, intervention and compensation measures. It has confirmed that the strong involvement and commitment of parents, students, local communities, youth, social and employment services and businesses is essential in developing and implementing policies against ESL. The key policy messages presented on page 3 reinforce the need for a comprehensive approach to ESL. Schools cannot and should not work in isolation. (European Commission 2013a, 25) Somehow, the official European theory of change overlapped with the prevailing one in Spain insofar as careful semantic nuances helped all policy-makers to find some common ground. Thus, in accordance with the insistence on an alleged compliance of central and regional authorities with the European strategy (Government of Andalusia 2007Andalusia , 2011Rigau 2012;Wert 2013), EURYDICE monitoring reports simultaneously reminded of the need of comprehensive action and conceded that Spanish authorities were active in reducing early school leaving (European Commission, EACEA, EURYDICE and CEDEFOP 2014).
Nonetheless, this careful wording could not conceal that in Spain most policy-makers thought that, simply, internal school processes were effective on their own. Thus, the Government of Spain expected that vulnerable students stayed at school if an adapted curriculum stressing literacy, numeracy, organisation skills and academic attitudes catered to their specific needs. In this view, the system was too rigid for those students whose expectations did not comply with the norm. The corollary was that if they were allowed to choose between different tracks, they were more likely to feel comfortable and would decide to continue within the system, thus availing of greater possibilities to improve their personal and professional future development (Government of Spain 2013b, 97864). Compared to the Commission, no reference was made to a comprehensive strategy, community involvement was neglected, and systematic support was not connected to this policy. These discrepancies with the Thematic Working Group (European Commission 2013a, 25) indicated that the educational authorities of the Union and the member state did not adopt the same theory of change. Similarly, in 2013 I interviewed officers of the Ministry who insisted on the potential of diverse tracks to convince teenagers to finish compulsory education and enrol in further education. How does the system deal with children? Why do they leave at a given point? In our view, the problem is that the pathways are neither defined nor adapted to the talents of each child. Thus, at a given point equal opportunities are conveyed by such an uniform approach that they cannot choose the pathway that suits their capabilities best (…) Therefore, the Organic Act on the Quality of Education advances the age when they have to choose a pathway. The European experience shows that this scheme works in countries when the first (but reversible) decision is made at fourteen. Our proposal includes an array of pathways so that a child can go back at any moment if s/he chose (let's say) applied mathematics but afterwards regretted this choice.
When he presented the new Education Act to the Parliament, the Minister underpinned this thesis with another argument. In his terms, an evidence-based review of education policies in the OECD and EU countries eventually showed that this type of flexibility and legal reforms that changed school practices were sound strategies to increase performance.
Definitely, lords and ladies, the countries which were successful in improving their educational performance did so by means of the appropriate legal reforms. These reforms take care of the students' diversity of abilities, deliver an individualised education and a wider flexibility of pathways, put in place more demanding academic standards, and value effort on the grounds of evaluation. Teachers are well trained and schools are allowed to be autonomous enough to specialise and provide students with a larger educational supply. These measures propose changes in the curriculum, the methodology, the pedagogy and the academic standards, but all of them are aligned with the aforementioned three general goals: reducing early school leaving, improving the quality of teaching and increasing employability. They are evidence-based measures. Neither the measures established by the new education act concerning the architecture of the system nor the changes proposed by this act respond to caprice or improvisation. On the contrary, this line of action responds to the educational reforms which have been more successful in a variety of countries. (Wert 2013, 7-10) As mentioned before, the Government of Andalusia implemented a scholarship to tackle early school leaving. Additionally, in 2012 it passed a decree opening a call for schools to be labelled as learning communities, and thus become eligible for extra-support. Clearly, the rationale of this decree emphasised the hypothesis that transforming the internal processes of schools would render the main progress.
Article 2. Definition of 'learning communities' projects 1. A "learning community" is a project of social and cultural transformation of a school and its community. The main goals are improving the academic performance of all and underpinning coexistence and civility. Its distinctive feature is openness to the whole community. Thus, it actively establishes a consensus so that families, associations and volunteers are actively involved in school management and learning activities during the official school timetable. (GovAndalusia-MoE 2012) In Catalonia, the plan of the government aimed at fostering school success by strengthening teachers professionalism, delivering personal support, innovating pedagogic methods, involving families in school life, providing new opportunities to early school leavers and other lines of action that operated within schools (GovCatalonia-MoE 2012). According to the presentation of the minister to the Parliament, this approach was also a conclusion of an evidence-based review of European policies.
How can we achieve these goals? How did we find out the means to achieve them? To start with, we analysed the international literature and adopted the mainstream guidelines. Additionally, we became aware that we had to implement those guidelines simultaneously. They do not form a sequence but a set of simultaneous measures. It is crucial to know which synergies this simultaneous implementation generates. Finally, we took into consideration our expectations for Catalonia as a country, what we want to be in social and economic terms, that is, we adopt a set of simultaneous measures that are grounded on international research and contribute to meet our national expectations. " (Rigau 2012, 8) In the former section a comparison of the official discourses of European and Spanish educational authorities found out that they retrieved a common 'frame' but understood prevention in disparate ways. In this section, it is plausible to conclude that they used different 'theories of change' . While the European institutions highlighted the importance of encompassing preventative policies, Spanish authorities mostly relied on the alleged effect of internal school processes. The former claimed their guidelines drew on the debates hold by an international group gathering experts with policy-makers. The latter claimed their approach was the straightforward result of the available scientific evidence. So, the main finding of this comparative discourse analysis of the two levels of governance notices that both framed the issue in similar terms but disagreed with regard to the best means to achieve the goal.

Discussion and concluding remarks
In a nutshell, a discourse analysis of the guidelines, laws and reports on early school leaving issued by the European institutions and a (small but varied) sample of educational authorities in Spain points at some similarities and differences. Apparently, all these political actors convey their messages within the same 'frame' since they widely agree that early school leaving is a big problem. However, an implicit but striking disagreement emerges with regard to the concept of prevention and the 'theories of change' that underlie their approaches. While the EU recommends an encompassing approach, in Spain most authorities adopt a much narrower perspective. Mostly, EU documents advise to prevent the problem by counter-acting segregation, supporting schools in disadvantaged areas and promoting an encouraging curriculum. In contrast, in Spain authorities think that the main solution lies in diversifying school tracks, experimenting with innovative pedagogies and strengthening the role of principals. These orders of priorities are clearly different. While educational studies highlight the perverse effects of the current misunderstandings with regard to policies against early school leaving, European policy studies highlight how common these misunderstandings are in a variety of policy areas. For sure, it is disappointing that the member state which scores the highest proportion of early school leavers deflects the recommendation to prevent the problem (Timmerman and Willems 2015). However, if 'resistance' indicates that anomalous processes are at stake, the term must be significantly qualified. Whatever the disappointment for objectives with regard to early school leavers, the discrepancy between the EU and educational authorities in Spain is not a sign of tension. On the contrary, this finding unveils the complexity of policy learning. Both experts in EU policy-making (Radaelli, Dunlop, and Fritsch 2013) and EU education policy (Alexiadou and Lange 2013) have convincingly noticed that the involved political actors continuously negotiate the meaning of the policy at stake. This observation also suggests questions on the effective policy learning that results from the transfer of discourses (as shaped by 'frames') and approaches or causal narratives (as indicated by 'theories of change').
By looking at the prevailing policies against early school leaving in Spain, this article finds out two pieces of evidence that contribute to spell out some clues of policy learning. On the one hand, the EU, the Government of Spain and a variety of regional governments share the same 'frame' as portrayed by the same catchwords. This does not seem to posit a particular problem but transmits a sense of agreement. Although their consensus is weak, all the political players eventually state they attempt to follow the same guidelines. On the other hand, albeit subtle, discrepancies with regard to the 'theory of change' are significant. Crucial to my argument is that these discrepancies are likely to affect implementation and outcomes significantly. Alexiadou (2010, 2007)'s typology of modes of policy learning is extremely helpful to make sense of this finding. Between 2012 and 2012 in Spain educational authorities engaged in 'competitive' policy learning of policies against early school leaving. These authorities were really concerned with meeting the official statistical benchmarks although this concern did not entail a full discussion of the underlying 'theory of change' . This is not an exceptional situation but a given state of a social field. It was a stable state for some years, and may be so for longer, but this stability has to be assessed empirically over time.