Stakeholder organizations in the European higher education area: exploring transnational policy dynamic

Abstract The study focuses on multi-level and multi-actor characteristics of European governance, concentrating on six European stakeholder organizations in the area of higher education (HE) and research. It analyses how policy positions change over time and how such changes can be accounted for, juxtaposing (1) environmental influence linked to structuration of the European HE policy arena as an organizational field and (2) internal organizational dynamic. Policy positions are seen to comprise three elements – policy issues, policy preferences and normative basis. Theoretically, the study is grounded in policy analysis, interest groups and neo-institutionalism. The empirical material consists of policy documents, subjected to both quantitative and qualitative content analysis. The study identifies changes in policy positions of stakeholder organizations, resulting in both differences and similarities, though the latter do not emerge over time in a way that suggests straightforward convergence. Thus, further research into specificities of policy-making processes within these organizations is necessary.


Introduction
European policy-making is a multi-level and multi-actor governance arrangement in which a multitude of EU, national and regional-level authorities, as well as stakeholder organizations are involved in policy formation, implementation and evaluation (Piattoni, 2010).
Specifically, European higher education (HE) policy coordination comprises two pillarsthe pan-European Bologna Process and the EU-led process that started with the launch of the Lisbon Strategy in 2000, leading to combination of supranational and intergovernmental dynamics (Maassen & Musselin, 2009). There are also stakeholder organizations operating on the European level (such as European University Association [EUA] and the European Students' Union [ESU]) which 'add a transnational flavour' (Elken & Vukasovic, 2014). This complexity with regards to multiple levels and actors is not an idiosyncrasy of HE, but is rather a hallmark of European integration in general (Börzel, 2010 On the one hand, stakeholder organizations are important for policy-makers because they can potentially provide (or withhold) valuable policy resources, such as information, expertise and political support (Braun, 2012). Stakeholder organizations, on the other hand, are given the mandate by their constituents to represent them towards policy-makers at the various levels of governance, e.g. ESU is mandated by its members to represent them towards EU institutions, Council of Europe (CoE) and other international organizations.
Despite having only consultative status in various fora (e.g. the main decision-making structure for the Bologna Process -the Bologna Follow Up Group [BFUG]), European stakeholder organizations have a significant influence over agenda setting and policy decisions (Klemenčič, 2012;Nokkala & Bacevic, 2014;Yagci, 2014).
Yet, despite of their presence and influence, not much is known about their policy positions, or how and why they may be changing over time to affect changes in European policies and initiatives. This mirrors a more general research gap in European studies about stakeholder organizations (also referred to as 'interest groups'). The main focus is on organizations' status and chosen strategies vis-à-vis policy-makers as well as the tension between the logic of influence towards policy-makers and the logic of membership (Kohler-Koch, 2010;Schmitter & Streeck, 1999). Comparatively less focus is put on their internal policy development (but see e.g. Halpin, 2014).
With this in mind, the present study is guided by the following research questions: • What are the policy positions of stakeholder organizations involved in European policy-making in HE and research and how have these policy positions changed over time? • How can these changes be accounted for theoretically?
The theoretical framework of the study combines organizational sociology insights related to institutional isomorphism with political science research on policy analysis and interest groups, providing a conceptual contribution in the form of a robust and nuanced toolbox for analysing multi-actor policy dynamics. The study complements the usual qualitative approaches to analysing policy documents with quantitative content analysis (QCA) in measuring policy positions of stakeholder organizations, thus providing also a methodological contribution. Finally, and in particular in relation to research on European policy-making in the area of HE and research, the study provides an empirical contribution by focusing on so far understudied yet important actors. Specifically, the study is focusing on policy positions of six interest group organizations which have consultative status in the BFUG, namely: the EUA, the European Association of Institutions in Higher Education (EURASHE), the ESU, the European Association for Quality Assurance in Higher Education (ENQA), the Education International (EI) and BusinessEurope and traces them back to 1998 (i.e. before the Bologna Process).

European policy arena as an organizational field
An organizational field is comprised of 'organizations that, in aggregate, constitute a recognized area of institutional life: key suppliers, resource and product consumers, regulatory agencies, and other organizations that produce similar services or products' (DiMaggio & Powell, 1983, p. 148). The 'borders' of an organizational field are to be determined empirically, focusing on the top-down dynamic that stresses rules and regulations as indicative of organizational field boundaries, as well as on the bottom-up dynamic that focuses on organizations themselves and initiatives they take to distinguish themselves as 'a recognized area of institutional life' (Frølich, Huisman, Slipersaeter, Stensaker, & Bótas, 2013). Organizational fields emerge through a process of structuration characterized by increasing interaction between organizations, increasing prominence of inter-organizational coalitions and power relationships, increasing information load and increasing 'mutual awareness among participants in a set of organizations that they are involved in a common enterprise' (DiMaggio & Powell, 1983, p. 148).
To what extent does the European HE policy arena exhibit these characteristics? The six stakeholder organizations (a) are operating on the European level, (b) have been given the mandate by their members to represent their interests on the European level (as expressed in their statutes, founding documents etc.), and (c) were recognized as representative by national governments participating in the Bologna Process, or EU institutions (including the European Commission [EC]). They have been increasingly interacting, not only through official events such as ministerial summits and the so-called 'Bologna seminars' , but also through joint projects (often funded by the EC), regular bilateral or multilateral meetings etc. They represent different constituencies: EUA universities, EURASHE non-university institutions, ESU represents students, EI academic staff, ENQA quality assurance agencies and BusinessEurope employers. Nevertheless, they have in the past advocated jointly for (or against) specific positions, e.g. in the early 2000s EUA and ESU advocated against including HE in the WTO's General Agreement on Trade in Services (GATS), while throughout the Bologna Process ESU, EI and CoE have been pushing for a stronger focus on social dimension of HE in the Bologna Process (Elken & Vukasovic, 2014;Yagci, 2014). Furthermore, there has been also some indication of stabilizing power relationships, not only through formal recognition of status within the BFUG, but also by less formal ways, e.g. EUA and ESU are more prominently featured in programmes of Bologna ministerial summits than other organizations. Increasing number of events and projects, as well as pressure to provide and use information, participate in consultation processes and be aware of policy positions of other organizations leads to an increase in information load. All these organizations indicate very prominently in public (through websites, publications etc.) their own involvement in the building and consolidation of EHEA, as well as involvement of other organizations, thus signalling that they are indeed aware of being 'involved in a common enterprise' , and thus part of a maturing organizational field concerning European HE and research policy-making.
Concerning central organizations in the field, the six stakeholder organizations are essentially independent from each other, but are dependent on the full members of the BFUG (i.e. national governments and the EC) for status recognition, and are also dependent on the EC for recognition within the EU policy consultations and for funding of part of their operational costs and project activities.
In structured organizational fields there is 'an inexorable push towards homogenization' which can be categorized as coercive, mimetic or normative isomorphism (DiMaggio & Powell, 1983, p. 148). Isomorphism concerns 'homogeneity in structure, culture, and output' (DiMaggio & Powell, 1983, p. 147;emphasis added). Culture of an organization can be conceived as 'values shared by the members of an organization … beliefs about organizational purposes and how they should be achieved' (Ashworth, Boyne, & Delbridge, 2009, pp. 172, 173). The point of departure of this study is that, given the basic features of stakeholder organizations elaborated in the next section, their policy positions can be considered as their main organizational outputs and their core aspects, as opposed to their structure (peripheral aspect, see Ashworth et al., 2009).
Coercive isomorphic processes indicate that not all organizations in a field are equal, i.e. some organizations can be considered as more central and other organizations may depend on them for important resources. Mimetic isomorphism appears under conditions of uncertainty and organizations might consider imitating other organizations they deem successful. Normative isomorphism relies on professionalization, in particular in terms of requirement of specific academic credentials and existence of professional networks across organizations which facilitates cooperation between organizations and selection of staff within the organizational field (DiMaggio & Powell, 1983). As will be discussed below, the three different types of isomorphism may be at work in an organizational field at the same time, albeit differently for different organizations and, therefore, with varied outcomes in terms of whose policy positions are becoming more similar over time. This requires looking into what kind of organizations are they and how can we analyse their policy positions.

Stakeholder organizations and their policy positions
Stakeholder organizations are effectively interest groups -organizations which have a mandate to influence policy-makers but do not seek to hold public office themselves (Beyers, Eising, & Maloney, 2008). Their input in this process is based on their own policy positions.
Policy positions can be conceived as public statements identifying which problems are deemed important, which outcomes are deemed desirable and/or what instruments will be employed to solve the problems and achieve the outcomes. 1 In some cases, ideological foundations for specific policy positions may be explicitly formulated as well (what Gornitzka, 1999 refers to as 'normative basis'). While government policies are expected to also include a description of policy instruments (regulation, funding etc.), given that stakeholder organizations do not seek to hold public office themselves and thus do not have to elaborate policy instruments, policy positions of stakeholder organizations may include only a statement on pressing problems and/or on the preferred solutions, with a more or less explicit normative (ideological) basis. Thus, a policy position of a stakeholder organization comprises: (1) policy issues -which aspects of a given sector are referred to in policies, (2) preferences concerning policy issues -what is the desired situation concerning these issues, and (3) the normative basis used to legitimize these preferences. The latter is particularly relevant given that stakeholder organizations justify their policy positions using norms and values of the constituents they are tasked with representing. In essence, policy positions of stakeholder organizations serve at least two purposes: (1) an internal one -defining the normative foundation of the group and guiding (and constraining) its members and representatives in their interaction with those outside the group, and (2) an external one -providing an explicit and public statement to other actors in the policy arena about the organizations' principles, values, norms and policy preferences. Different stakeholder organizations often represent constituencies which may consider different issues as important, have different preferences concerning these issues and have different ideological basis for such preferences (Bunea & Ibenskas, 2015). Policy positions are thus parts of the interest group's policy agenda which the group actively lobbies for (Halpin, 2015).
Given this, it is not likely that there will be significant similarity of policy positions of stakeholder organizations representing different constituencies far from the homogenizing influence of an organizational field. In other words, in light of the structuration of the European policy arena as an organizational field, policy positions of stakeholder organizations are likely to be exposed to the aforementioned 'inexorable push towards homogenization' , i.e. the policy positions may be becoming similar over time. On the one hand, increasing similarity of policy positions -policy convergence 2 -of a stakeholder organization with positions of a central organization (in this case the EC) -the so-called 'delta convergence' -relies on primarily coercive mechanisms of isomorphism. For example, when the EC opened up consultations with the stakeholder organizations 3 on what eventually became 'New Skills Agenda for Europe' , it compelled stakeholder organizations into positioning themselves in relation to the issues put forward in the initial proposal. Similarly, by setting up priority topics to be funded through Key Action 3 of its Erasmus + programme, the Commission also incentivizes the stakeholder organizations to address these topics in their own work if they want to receive the funding and take part in such cooperation (Batory & Lindstrom, 2011). On the other hand, increasing similarity between policies of stakeholder organizations -'sigma' convergence -is primarily a result of normative mechanisms related to creeping professionalization of interest groups and individuals involved with them (Ashworth et al., 2009;Heinze & Knill, 2008). For example, previous research has demonstrated that there are a limited number of individuals who, even after completing their work in one stakeholder organization, remain in the policy arena by working for another stakeholder organization (Elken & Vukasovic, 2014). However, in cases where some organizations are recognized by the main actors in the policy arena as key representatives of a particular stakeholder type later than others, a mimetic mechanism can also be at play given that 'late-comer' organizations might consider organizations that have been a recognized actor in the policy arena longer as examples to emulate in order to succeed. As will be demonstrated below, one of the stakeholder organizations under study has been part of the policy arena since the very beginning, while other five organizations have been admitted gradually over the course of six years.
However, assuming that this is the only dynamic at play would be misleading. Such a view disregards the internal processes within organizations and how the process of developing policy positions may depend on the internal structure of the stakeholder organizations, their main decision-making structures and the mandate their representatives in the policy arena have to act on their own (Beyers, 2008;Bunea & Ibenskas, 2015). This is particularly problematic for European stakeholder organizations which are meta-organizations (organizations of other organizations) constructed primarily for strategic reasons related to pooling resources and exerting influence (Ahrne & Brunsson, 2008). In such organizations, there is a continuous tension between (1) the focus on the internal processes and needs and demands of the members and (2) the external positioning and influence of the policy-makers (see Schmitter & Streeck, 1999 on logic of membership vs. logic of influence). With this in mind, what can also happen is that policy positions reflect primarily the internal dynamic and thus, given the different interests of constituencies of stakeholder organizations, one can expect that the policy positions do not converge over time.
In theoretical terms, convergence of policies of the European stakeholder organizations implies that it is the organizational environment (the European policy arena) that primarily contributes to policy change, while lack of said convergence implies that the policy change is primarily affected by the internal dynamic of these organizations. Given the tension between logic of membership and logic of influence referred to earlier, in an actual empirical situation, the environmental and internal influences overlap and interact, which is why the conceptualization of policy positions as comprising three elements -issues, preferences concerning said issues and normative basis to argue for said preferences -is particularly useful. The distinction between (1) issues, (2) preferences and (3) normative basis means a step-wise operationalization of policy change. From the first to the third step, the extent to which stakeholder organizations are changing their policies increases; the shift in normative basis, i.e. change of ideology which may lead to convergence of normative basis of policies comprises the highest level of policy change. Situations in which there is no policy convergence or it is limited to convergence of policy issues (first order convergence) can be seen as indications that the internal dynamic remains the more important contributor to processes of policy change of stakeholder organizations. Conversely, convergence of preferences (second order convergence) and, even more so, of normative basis (third order convergence) can be interpreted as indications that the environmental influences are overpowering the internal dynamic.

Methodological considerations
Given the interest change of interest groups' policy positions and whether or not such changes reveal a converging pattern, this study is a comparative and a longitudinal one. The policy positions were identified and compared (a) across organizations within the same time period to uncover how similar are policy positions of different organizations at a specific period of time (comparative aspect) and (b) between these time periods to uncover whether similarity is increasing over time or not (longitudinal aspect).
The data-set consists of 310 documents, including: documents explicitly designated as 'policies' or 'positions' by the organizations themselves, declarations specifically addressing the participants of the Bologna ministerial summits and statements specifically addressing the EC (or other EU institutions), all available online (either on organizational websites or websites dedicated to the Bologna Process) published by mid-2015 (the most recent Bologna Ministerial Summit in Yerevan). Such explicit labelling by the organizations themselves indicates that these documents represent the key output of the policy-making processes within these organizations. The 'EU-targeting' documents are usually based on the general policy or position papers, and often include explicit references to them, but they also constitute a data subset necessary to explore the extent of 'delta' policy convergence. For EI and BusinessEurope, which have policy documents concerning other levels of education or issues beyond education, a selection has been made to focus on policy documents which focus (at least indirectly) on HE and research. The data-set includes also official communication concerning HE related issues from the EU institutions (Council and Commission).
The identification of policy positions took into account the specific dynamics of the European policy-making in HE and the differences with regards to timing, frequency, length and comprehensiveness of policy documents of stakeholder organizations. In order to enable comparison across organizations in a specific time period, the policy documents of each organization between two ministerial summits in the Bologna Process were combined into one large document which was used to identify policy positions (issues, preferences, normative basis) of said organization for that specific period. The same was done for policy positions of EU institutions. These constructed documents were used to identify all three elements of policy positions: issues, preferences and normative basis. QCA was employed to identify issues, while identification of preferences and normative basis was done qualitatively.
Identification of policy issues relied on a baseline dictionary developed by manually coding most recent policy positions of all stakeholder organizations (see Appendix 1 for 25 policy issues and their related words and word-stems). Using Yoshikoder and Jfreq, open source software developed to facilitate QCA, 4 prominence of a specific policy issue in policy positions of a specific organization and the EC in a specific time period was measured. This was done by, first of all, excluding the so-called 'stop-words' 5 from the text and by measuring the frequencies of relevant (stemmed) words for a given policy issue. The sum of frequencies of all relevant stems for an issue, normalized in relation to the overall length of text (without the stop-words), was used as an indicator of prominence of a specific policy issue for a specific organization in the specific time period. For this step of the analysis, it is not important whether an organization refers to an issue in a positive or negative sense and how are references to one issue linked to references to another issue. The main aim is to provide an objective assessment of how important a specific issue is for the organization. The underlying assumption is that the more an organization (or the Commission) uses the terms from the baseline dictionary which refer to a specific policy issue, the more important this policy issue is for them. This assumption relies on the idea that variance in use of English language (in particular semantics) in policy documents is minimized by the fact that these documents need to be written in correct English in order to be understandable by a variety of audiences. The comparison of prominences of policy issues between the EU institutions and the six organizations was used to assess 'delta' convergence, while comparison of prominence of policy issues among the organizations was used to assess 'sigma' convergence when it comes to policy issues (first order convergence).
A quantitative approach to measuring first-order convergence was chosen to ensure reliability of measurement in the foundational part of the analysis and because it allows for an assessment of what are the most important policy issues for each of the organizations. However, when analysing the second and third order of policy convergence (convergence of preferences and normative basis), the situation is more complex. Describing the desired situation concerning a particular policy issue requires more than a few words whose relative frequencies can be counted. While there are more elaborate approaches to QCA (the so-called 'bag of words' approaches utilizing Wordfishes or Wordscores software) that have been initially developed to analyse positions of political parties on one-dimension (leftright), they rely on the assumption that opposing positions on an issue can be identified (Klüver, 2009). However, these are considered as unsuitable for analysing policy preferences of stakeholder organizations given that, amongst other, their positions cannot be organized along one dimension and it is not a given that clearly opposing preferences exist (see in

Results and discussion
Before diving into the analysis of policy positions, some basic information on the six stakeholder groups is warranted ( The analysis first focuses on policy issues, which is then used as the basis for analysing policy preferences and normative basis.

Policy issues
First, the analysis focused on the proportion of meaning-carrying words related to the 'baseline' policy issues organized into 25 policy issues in all policy texts. The five most prominent issues for each stakeholder organization for the entire period of time are presented in Table 2.
European coordination, comprising references to the Bologna Process, Lisbon Strategy, EHEA, ERA, EU, Europe, is either the first or the second most prominent issue for all six organizations. Such high prominence of the European coordination as a policy issue is further evidence that stakeholder organizations' are aware of a common enterprise, and that the European policy arena can be considered as a rather structured organizational field. Given this and in order to avoid biasing the analysis in favour of more policy convergence, the policy issue 'European coordination' was excluded from further analysis. Further similarities concern the prominence of research and social dimension policy issues -all organizations apart from ENQA have these two issues amongst their top prominent ones. This similarity will be explored further (see next section). Apart from the similarities, each organization also has its own set of favourite issues. Some of these are arguably expected, given the nature of these organizations, e.g. ENQA focusing the most on quality, ESU focusing the most on the social dimension, EUA focusing on research while its non-university counterpart (EURASHE) focuses significantly on LLL. Strong focus of EI on research is due to its focus on doctoral training and the fact that it has been rather active in responding to Commission's initiatives (European Research Council [ERC], EIT, framework programmes), while BE's focus on research is in relation to how it contributes to building a knowledge economy. Note also that for BE's a highly prominent policy issue is employability, as can be expected from an association of employers' organizations.
While this provides a static indication of similarity between stakeholder organizations, it is not sufficient to analyse the extent of convergence of policy issues that these organizations focus on. In order to assess whether and how much first order convergence there is, prominence of 24 'baseline' policy issues (apart from 'European coordination') was added for each organization for a specific period between two Bologna ministerial summits. The results are presented in Table 3. The numbers indicate the proportion of words related to 'baseline' policy issues in all policy texts of a given organization in a given period, in relation to the total number of words which carry meaning in these texts, expressed in percentages. For example, in 2003-2005 2.07% of meaning-carrying words in EUA policy documents were words corresponding to 24 policy issues identified in the 'baseline' set (see Appendix 1).
Before interpreting the results, a methodological clarification is in order. Percentages reported in Table 3 may seem low at the first glance. What needs to be stressed, however, is that the result mentioned above for EUA in 2003-2005 means that the 95 word stems corresponding to 24 policy issues included in the 'baseline' dictionary comprise 2.07% of  all meaning carrying words utilized in the policy documents in this period (approx. 11,000 words).
As the data show, the prominence of baseline policy issues varies over time for all organizations and not in a way that suggests clearly a growing similarity over time, i.e. steady convergence. Given this, a closer analysis of prominence of the 24 issues over time for each organization was conducted. It revealed that there are differences in how stable the attention to the issues is over time. Figures 1 and 2 illustrate the changes to the two most prominent issues overall -research and social dimension.
More generally concerning the changes in the prominence of specific policy issues 6 : • EUA's attention to research varies significantly; the drop of prominence of all policy issues for EUA in [2005][2006][2007] (Table 3) is primarily a drop in the attention given to 6 detailed data for all policy issues across groups and time is available upon request.  50% pre 1999 1999-2001 2001-2003 2003-2005 2005-2007 60% pre 1999 1999-2001 2001-2003 2003-2005 2005-2007  research. EUA's attention to research also drops towards the end of the analysed period (post 2010) but this coincides with increasing attention to social dimension; • ESU focuses on social dimension rather consistently over the period under analysis, while its attention to research and other issues show significant variance over time. The increase in the attention to social dimension from 2010 to 2012 is potentially indicative of ESU's response to more general developments, i.e. the fact that in many countries in which ESU has members the governments' responses to the 2008 economic crisis were to decrease public funding for HE, including student support systems (Jongbloed & Vossensteyn, 2016). 7 Specifically, early in 2012 (i.e. before the Bucharest Ministerial Conference), ESU adopted an overarching policy paper focusing only on this issue, and also published several reactions to EU initiatives which highlighted 'social dimension' . The possible reasons for the drop in the prominence of social dimension after 2012 will be discussed below; • EURASHE's attention to its favoured topics is not stable; e.g. there is significant variance in how prominent LLL is in its policies with least attention Thus, it seems that there is no clear convergence towards the baseline set of policy issues. Although there are similarities concerning which issues are given most attention overall in organizations' policy positions (Table 2), this sometimes varies significantly over time. The identified 'rise and fall' patterns in prominence of specific policy issues may be, at the first glance, interpreted as indicative of actually changing interest in these issues within these organizations. Specifically, cases in which the prominence of an issue is decreasing over the period under analysis may suggest a waning policy interest by the constituents of the stakeholder organizations. However, closer examination of organizations' policy documents reveals only one instance in the whole data-set in which a later policy position was explicitly adopted to replace an earlier one. Thus, another possibility for decreasing prominence of specific policy issues is that policy positions developed earlier in the process remain valid and used by the groups' representatives throughout the period under study. In other words, what might be taking place is that issues do not really drop off the agenda, but rather that 7 See also the Public Funding observatory established by eUa, http://eua.be/activities-services/projects/eua-online-tools/ public-funding-observatory-tool.aspx (page accessed 20 January 2017).
organizations do not reiterate their existing positions over and over again, unless a change of stance happens, which is actually more relevant for the analysis of policy preferences and normative basis. In light of this, what may be most indicative concerning convergence of policy issues is that at some point in the period under analysis all actors exhibited an increase in how similar their policy issues are with the baseline. In that view, one could argue that there may be indeed evidence of 'sigma' convergence concerning policy issues, but with a more complex temporal dynamic than initially thought.
Concerning 'delta' convergence of policy issues, increasing similarity of stakeholder organizations' issues with the issues raised by the EU institutions, the situation is even more complex (see Table 3). At the first glance, the EU institutions do not seem to dictate the policy focus of the stakeholder organization to an extent that might suggest EU's central position in the policy arena. However, here too interesting 'rise and fall' patterns emerge. In the 2001-2003 period, there was a similarity between the EU and other organizations -most pronounced in the case of BE, ENQA and ESU, least pronounced for EUA. In 2003-2005 BE and EUA are rather close to EU, in 2007-2009 EI, BE and ESU, while towards the end of the period the closest to the EU is BE (with EI, ESU and EUA similarly far). The 2001-2003 similarity is not necessarily surprising, given that it coincides with the Commission's 'The role of the universities in the Europe of Knowledge' communication (European Commission, 2003), in essence the first policy position to elaborate the Lisbon Strategy in the area of HE and to which the stakeholder organizations explicitly responded. Other instances of similarity between EU and some of the stakeholder organizations concerning addressed policy issues can also be traced back to EU issuing positions related to its Modernising Agenda. Thus, (1) in light of the previous discussion about the fact that the 'rise and fall' patterns may be related to internal dynamics of stakeholder organizations, in particular the lack of explicit re-iteration of existing policy positions, and (2) given that at some point each of the organizations was rather close to the EU's position, one could make a similar point about 'delta' convergence of policy issues.

Policy preferences and normative basis
Given the analysis of prominence of specific policy issues presented above, as well as the fact that these were also two issues the EU focuses most extensively on, the following issues were selected for the qualitative analysis of how preferences and normative basis changes over time: research and social dimension. Five out of six organizations (as well as the EU) have these two topics amongst their most prominent ones (bar European coordination). This is not the case for ENQA, but that does not mean that ENQA does not focus on these issues at all (see above).
Concerning the topic of research, EUA highlights rather consistently the need for research and education to be intertwined in universities, highlighting how research is integral to HE and how all learning (including LLL activities) should be research based. For example: Institutions offering research based higher education should ensure that a research component is included and developed in all cycles thus allowing students to acquire research experience and encouraging an interest in research as a possible career. (Lisbon Declaration, 2007) From 2003 to 2005 onwards, EUA focuses also on the importance of quality research training, reflecting the increased focus in the Bologna Process on the third cycle, as well as on allowing for stable conditions for research funding and research careers. In general, EUA's preferences concerning research are rather generic overall, though more specific preferences are formulated in response to EU initiatives. For example, although being in general supportive of the launch of the ERC, EUA also expressed concerns about possible negative consequences for some of its universities (brain drain, 'Matthew effect'). Similar to EUA, ESU also highlights the importance of a research-based HE, from 2005 onwards focusing also on PhD studies, also advocating for PhD training in the non-university sector. ESU also focused from 2007 onwards on research careers, but primarily in relation to gender balance. EURASHE, as can be expected, highlights the importance and value of applied research, also advocates for doctoral education and indicates that the link between education and research is important. EI's focus on research is primarily expressed in relation to academic careers and academic freedom to do research, e.g.: … There is thus great concern over visible signs of increased bureaucracy and control, political control of the use of research resources and the reduction of researchers' free right of publication … current trends which result in a weakening of tenure rights have a devastating effect on academic freedom. (Statement on academic freedom, 2006) The importance of a research-based education and PhD education in general is expressed more clearly from 2003 onwards, while from 2005 EI also picks up the topic of PhD education in the non-university sector. EI also commented specifically on EU's initiatives (such as the European Charter for Researchers and the Code of Conduct for recruitment of researchers). Similar to the EUA, EI supports these initiatives in general, but raises some concerns over the freedom the HEIs will have in implementing these documents and the possible negative consequences this may have for fair recruitment of academic staff. ENQA, as indicated above, deals with research only sporadically, but when it does, the main message is that education and research should be integrated. Contrary to all other organizations, Business Europe does not highlight so much the link research and education, focusing on the role of research in economic development primarily and towards the end of the period under analysis advocating for an 'industrial doctorate' . The European Union (primarily the Commission), presents research as the key element for developing the knowledge-based economy, highlighting the need for universities to modernize so they would also provide a more relevant and more significant research output, which is also the backdrop for launching specific research initiatives (ERC, EIT, Horizon 2020).
Overall, all of the stakeholder organizations (apart from Business Europe) highlight the preferences for research and education to be closely intertwined, suggesting some 'sigma' convergence of preferences. At the same time, stakeholder organizations also maintain the focus on the linkages between the issue of research and issues that are close to their constituencies, e.g. EURASHE highlighting applied research, EI highlighting the issue of academic careers and academic freedom, EUA focusing on the consequences EU's research initiatives may have on the universities, BE advocating for an industrial doctorate etc. Concerning 'delta' convergence of preferences, there are indications this may be present to a limited extent, given the general support of some of the organizations groups (most notably EUA and EI) to EU's initiatives, though in both cases there are also cautionary remarks (i.e. EUA and EI do not share all of the EU's preferences concerning these initiatives).
Concerning the topic of social dimension, EUA focuses on the issue already in 2001-2003 period, but primarily in relation to equity in mobility programmes. Later on, it highlights the importance of equal access as a comparative advantage of EHEA over HE opportunities in other world regions, while social equity in terms of HE completion is added from 2007 onwards, specifically in relation to lifelong learning and the necessity to provide suitable guidance and counselling during studies, in particular for disadvantaged students (European Universities' Charter on Lifelong Learning, 2008). ESU's as the champion of the topic highlights the importance of equal opportunities in HE from very early on, e.g. advocating against using the 2nd cycle as an additional opportunity for student selection, highlighting the need to reduce financial obstacles to mobility and the necessity to organize adequate support for students (financial and otherwise) to ensure equal access, progress and completion. For example, prior to the Prague ministerial summit in 2001, ESU stated: Students therefore are not consumers of a tradable education service, and as a consequence it is the governments' responsibility to guarantee that all citizens have equal access to higher education, regardless of their social background. This means providing students with adequate funding in the form of study grants and the higher education institutions with enough funding to exercise their public tasks. (Student Goteborg Declaration, 2001) EURASHE picks up the issue of social dimension more prominently relatively late and, apart from advocating for more attention to it, it also links the topic to the role HE is expected to play in society (e.g. reducing social inequalities). Similar to ESU, EI focuses on the social dimension from early on, stressing the importance of free and fair access, as well as the social dimension in relation to young staff, also very explicitly warning about detrimental effects of commodification of HE to social dimension. Similar to how it deals with research, ENQA addresses the social dimension of HE in a very limited way, and when it does so, this is primarily in terms of the role QA plays in ensuring that HE responds to the social demands and expectations (reducing inequality, ensuring opportunity). BusinessEurope also discusses the issue rarely and in very specific way: in 2001-2003 period when discussing the importance of new technologies in HE -highlighting the opportunities e-learning may have for widening access, and towards the end of the period under analysis stressing the importance of wider access for ensuring economic prosperity. EU's focus on social dimension is evident in its attention to achieving social cohesion, but primarily highlighting it in more general terms (battling poverty, reducing youth unemployment) and coupling it together with competitiveness and excellence, e.g. as in the 'Youth on the Move' initiative launched in the EU2020 strategy.
Overall, while all organizations do deal with social dimension in one way or the other, they deal with it highlighting the needs of their own constituencies, and there are significant differences in how central the topic is for them. Thus, one can identify rather limited convergence (both for 'sigma' and 'delta') of preferences concerning this policy issue across all of the organizations, with the caveat for ESU and EI, organizations that have already been identified as advocacy coalitions concerning this issue (see earlier reference to Yagci, 2014), and for which 'sigma' convergence of preferences concerning policy issues can be identified.
Concerning normative basis, the qualitative analysis of policy documents reveals that there is significant variety of how it is expressed. Overall, for contentious preferences there are more explicit references to the organizations' norms and values. One such contentious issue is tuition fees in HE. When expressing a preference against the tuition fees, ESU justifies its position by bringing forward the ideas of equity and social justice, while for less contentious preferences (e.g. student-centred learning) the normative basis in ESU's positions is very ambiguous. It is also notable that most of the other stakeholder organizations refrain from discussing the issue of tuition fees explicitly, possibly in reflection of ESU's strong stance against them. One exception is EI that supports ESU's stance for more public funding in HE.
The normative basis for arguing for a stronger focus on the social dimension is somewhat more explicit, though not in all of the stakeholder organizations. Amongst those who do utilize more explicit argumentation in favour of more social dimension in HE, ESU and EI connect it to access to HE or stable employment (respectively) and claim that both are human rights. BE and EU discuss the issue under the term 'social cohesion' but highlight more the economic rationale -social cohesion is important for maximizing the competitive potential, not on its own.
Preferences that all stakeholder organizations seem to share, e.g. that education and research should be intertwined in HE -are presented almost as a given. While they may be accompanied by the claim that such an approach would ensure quality and relevance of education (which are considered desirable in themselves), there is no ideological rationale for combining education and research.
Thus, analysis of policy changes with regards to the normative basis is rather difficult because the policy positions of stakeholder organizations are often rather ambiguous on this matter. If any, 'sigma' convergence concerning normative basis convergence is limited to smaller subsets of stakeholder organizations (ESU and EI), while 'delta' convergence has been identified only for BE (towards EU).

Conclusion
Policy positions of stakeholder organizations exhibit a complex mix of similarities and differences and understanding this mix and how they evolve over time is essential for contemporary European HE policy-making.
A rather comprehensive set of policy issues, all stakeholder organizations focus on has been identified. Given that stakeholder organizations represent different constituencies and thus are likely also to focus on different issues, this result is suggestive of first-order policy convergence. At the same time, there are differences in how significant a specific issue is in the overall policy agenda of the organization. In other words, apart from having positions on all policy issues identified in the baseline dictionary, each stakeholder organization retains primary focus on its own 'pet' issues that reflect the interests of its members. Moreover, the extent to which organizations focus on their most prominent issues varies over time which may indicate that the internal dynamic prevails in the process of shaping policy positions. In order to resolve this ambiguity, a further analysis was conducted which revealed the specificities of the policy-making process -stated policy positions are amended only if a change of preferences or normative basis emerges, otherwise once adopted policy positions remain valid. In light of this, it can be concluded that there is evidence of first-order convergence (convergence of policy issues).
Results concerning change of preferences and normative basis are more mixed -there is both convergence and diversity, suggesting that neither the internal dynamic nor environmental influence can fully account for the identified empirical patterns when it comes to policy preferences and normative basis. Despite indications, that the EHEA is a rather structured organizational field. Overall, the relative importance of environmental over internal influences seems to depend on both the issue at hand and the stakeholder organization in question: stronger environmental influence concerning research, stronger internal dynamic concerning 'pet' and contentious issues.
Conceptualization of policy positions that distinguished between issues, preferences and normative basis enabled a more nuanced analysis of policy change. However, the ambiguity with which stakeholder organization expresses their normative basis makes the analysis of the third element somewhat difficult, at least on the basis of document analysis alone. Concerning the theoretical mechanisms for policy change, while distinction between environmental influences and internal dynamic proved to be important, it is necessary to further unpack the latter. Further research should thus look into the internal dynamic of such organizations, in particular how often and under what conditions are policy positions updated or amended etc.
Apart from delivering the opening insight into organizations that are an integral part of European HE policy-making process, this study also offered the conceptual and methodological tools for analysis of policy positions of stakeholder organizations in other policy domains, in Europe as well as in other contexts marked with both multi-level and multi-actor governance arrangements.