A climate of optimism? EU policy-making, political science and the democratization of Central and Eastern Europe (2000– 2015)

Democratic erosion in the EU’s Central and Eastern European (CEE) member states has confounded EU policy-makers. In this paper, we investigate the assumptions behind the climate of optimism about CEE democratization that prevailed in EU decision-making before and after the 5th and 6th enlargements, and the extent to which political science participated in this intellectual climate. Based on a qualitative analysis of EU decision-making in the early twenty-first century and a quantitative analysis of 500 randomly sampled papers published between 2000 and 2015, we find that both policy makers and the most influential research in political science shared a bias towards optimism structured by common assumptions: A procedural understanding of democracy, a rational institutionalist belief in the EU’s capacity to bring these procedures about with the use of incentives and the related assumption that sociocultural dimensions of democracy would eventually follow institutions. We argue that these common assumptions help to explain both the EU’s failure to pre-empt and respond proportionately to democratic erosion, and the failure of our discipline to check that optimism.


Introduction
Someone was trampling on the democratic rights, the ingrained rights of the citizens, and I didn't have the tools (…) No one could imagine that once a country would solve all the problems before becoming an EU member, become a member of the family and keep improving, that they would then dare to take a step back.It wasn't in peoples' minds.
During this time, there was only dismay and bewilderment that this was possible in a democratic constitutional state (…) Everyone had to assume that an integral part of the contract for Hungary were the principles that the country itself had ratified as a member of the EU, which included the Charter of Human Right of the European Union.Everyone had to assume that a member of this community would align with these principles.
In 2011, a two-thirds Fidesz majority ratified Hungary's new 'partisan' constitution (Dani 2013), building the foundation of what would become an avowedly 'illiberal state' at the heart of the EU (Orbán, 2014;Pech and Scheppele 2017).Ten years later, Martin Schultz and Viviane Reding reminisce about the fractures that then appeared in the enlarged EU's constitutional identity.There is a striking candour in their depiction of the EU's unanimous confidence in Central and Eastern Europe (CEE) democratization prior to these events.No one could imagine radical democratic backsliding within the European Union.Everyone assumed that the countries from the 2004 enlargement would uphold their democratic 'contract'.Judging by these accounts alone, a general optimism about the region's future political trajectory appears to have informed EU decision-making either side of the 5th and 6th enlargements.
By the time of writing, the positive expectations that characterized the recent past are much thinner on the ground.The dominant research paradigm concerns democratic backsliding (Müller 2015;Kelemen 2017;Pech and Scheppele 2017), defined by Bermeo as the "state-led debilitation or elimination of any of the institutions that sustain an existing democracy" (Bermeo, 2016, p. 5).As we discuss in this paper, while democratic malaise has been most striking in Hungary and Poland, it is not confined to these countries.We also note, along with Schultz and Reding, that the EU was ill-prepared for democratic backsliding and, until very recently, was largely lacking the legal tools to address this situation (Blauberger and Kelemen 2017;Müller 2015).As many have recognized, this has had dire consequences for the organization's normative integrity and external credibility (Bellamy et al. 2022;Chiru and Wunsch 2020;Kelemen 2017).Considering how much is at stake, it is appropriate to ask what exactly went wrong.
In this paper, we adopt a cultural institutionalist lens to explore this question, focusing specifically on the ideational context within which the EU's fateful decisions were made.More specifically we ask: What key assumptions about democratization in CEE prevailed at the time of the Eastern enlargement, and how might these have shaped policy-making decisions at the time?We acknowledge, and find important, the role of strategic considerations at the EU level, as well as the short-term political interests of individual Member States.Yet we argue that focusing exclusively on the language of interests and incentives fails to fully explain the lack of democratic safeguards put in place by the EU prior to the Eastern enlargement.In this we follow a cultural institutionalist tradition, according to which rules within institutions do not result solely from rational choices but also from the norms, values and forms of knowledge that inform the decisionmaking of actors within them (Bevir and Rhodes 2010;Hall and Taylor 1996).
Our study looks more specifically at the key assumptions about CEE democratization that prevailed within two distinct institutions between 2000 and 2015, the crucial period before and immediately after the 5th and 6th enlargements.First EU institutions themselves, especially the European Commission, which through its monitoring of CEE produced a considerable amount of knowledge about democratization in the region.In the first sections of the paper, we put forward a qualitative reading of the ideas driving the EU's policy actions during this time.Second, we take interest in academia as another key institution, specifically political science and the subfield of democratization studies.For this purpose, we conduct a quantitative analysis of 500 papers published between 2000 and 2015 in the top, generalist political science and in the top post-communist area studies journals.In looking at assumptions about CEE democratization within these two sites, we hypothesise that, taken together, they should provide an accurate picture of the ideational context within which enlargement took place.
Our findings indicate that a climate of optimism about the trajectory of change in the region in the early twenty-first Century existed in both settings.This optimism rested on three key shared assumptions about CEE democratization: A belief in the EU's capacity to bring regime change about with the use of incentives; a minimal and procedural understanding of democracy; and the idea that democracy's sociocultural dimensions are somewhat secondary and would eventually follow institutions.These assumptions emerge from the decisions of EU policy-makers at the time, but also from our quantitative analysis of the literature.While academic work as a whole was diverse, the papers that were most cited shared the EU's inclination towards democratic optimism.In turn, both the EU and the dominant strands of scholarship on political change in the region underestimated the key roles of mass politics, political discourse and civil society in determining not only the quality of democracy but its ultimate survival.We conclude by considering the mechanisms that might explain this intellectual convergence and draw some conclusions about what both EU policy-makers and political scientists can learn from our findings.

Democratic optimism in the EU policy sphere: A qualitative assessment
The EU's insufficient attention to democratic fragility (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) Seen in retrospect, the period of political change in CEE from 2000 to 2015 may be divided into two halves: The period until 2007 when EU monitoring and treaty refinement relating to the 5th and 6th enlargements were taking place, and the period from 2007 when all 10 CEE states were already members, subject to weakened post-accession leverage.We show below that in both periods EU policymakers minimized the threat of democratic erosion: They did not adequately prepare for it, nor did they respond in proportion once it took place.In step with existing accounts, we then acknowledge that various strategic and self-interested factors lay behind these policy failures (Gallagher 2009;Phinnemore 2010;Kelemen 2017;Herman et al. 2021).However, we also stress that a key ideational factor, namely a prevailing climate of optimism about the trajectory of political change in the region, tends to be overlooked in the existing literature.This ideational context made it possible for democratic fragility to be neglected for so long.
Prior to this analysis, it is key to stress one key premise of this paper: That today, democratic malaise is widespread in the region, and not confined to Hungary and Poland.This premise requires justification, as some scholars argue that these two cases constitute influential data points that mask a range of trajectories among other CEE states (Bakke and Sitter 2020;Schimmelfennig and Sedelmeier 2020).This last reading can be questioned on several fronts.First, the earlier positioning of Poland and Hungary as star pupils of EU-leveraged political change cautions against considering them as idiosyncratic cases: Their trajectory is reason enough to interrogate the basis of previous, optimistic assessments (Herman 2016).Second, much empirical evidence suggests that democratic gains throughout the region have at a minimum been halted, and at worst been lost since the pre-accession period.Relying on the liberal and deliberation V-Dem indexes, Gora and de Wilde find 'a deterioration of both rule of law and deliberation in all new EU member states between 2010 and 2019' (Gora and de Wilde 2022).Freedom House Nations in Transit, so hesitant to downgrade states that it took until 2015 for Hungary to slip from the 'consolidated democracy' category, has recorded 2017-2021 falls in every one of the 5th and 6th enlargement EU entrants (Freedom House 2021).This includes deteriorations in countries such as Slovenia, Lithuania, Latvia or the Czech Republic which, much like Poland and Hungary, were previously considered as democratization frontrunners.Finally, the European Parliament has since 2011 debated lack of respect for Article 2 principles ('fundamental values') in the cases of Slovakia, Romania, Bulgaria and Slovenia (and Malta) (Herman et al. 2021).
To what extent then, did the EU anticipate these developments?When Eastern enlargement was set in motion in the 1990s, the EU and its Member States demonstrated some awareness that democracy might be fragile in candidate states.If the overall project was inspired by the liberal optimism that followed the fall of the Berlin Wall (Schimmelfennig 2003), the EU's approach during those years was also tempered by caution arising from the ethnic nationalist violence that engulfed the former Yugoslavia.These events largely defined the context within which the European Council decided, in Copenhagen in 1993, to condition for the first time the opening of accession negotiations on a number of political criteria-namely respect for democracy, the rule of law, human rights, and listed separately to emphasize its importance, minority rights (Rechel 2008, pp. 173-174).The very idea of a formal EU accession process, with its attendant conditionalities and monitoring was an innovation that had not been applied to earlier rounds of entrants.Furthermore, fears that these criteria could not be upheld after accession led to the Article 7 mechanism, introduced with the 1997 Amsterdam Treaty, which allows for suspending the voting rights of a member violating the fundamental values laid out in Article 2.
The EU's monitoring process of CEE states, however, did not match this caution.By 2000, all 2004 and 2007 entrants were judged to have fulfilled the necessary political criteria to start accession negotiations.This included not only frontrunners like Poland, Hungary and Slovenia but also 'illiberal pattern states' such as Bulgaria, Romania and Slovakia (Vachudova 2005) that had been declared 'on course' to fulfil the political criteria, sometimes a matter of months after pro-EU governments had been elected into office.Annual EC Progress Reports repeated this judgement even as promises of implementation went unfulfilled, highlighting a seemingly pre-determined approach to political monitoring that raised eyebrows at the time (Hughes and Sasse 2003).
The permissive approach to political monitoring in the accession states put more emphasis on EU-wide treaties to ensure the safeguarding of democracy in the widened EU.In the years that followed, however, the EU missed several opportunities to strengthen these democratic safeguards (Blauberger and Kelemen 2017;Müller 2015).Having proved unusable in the context of the far-right's invitation to Austria's coalition government in 1999, Article 7 was amended two years later with a 'warning' mechanism prior to full suspension introduced in the Nice Treaty (Freeman 2002).Despite this, the unanimity threshold to enact suspension and the lack of incremental economic sanctions were left untouched.Discussions around the status of the Charter of Fundamental Rights of the European Union in the mid-2000s were another missed opportunity: Prior to the 2007 Lisbon Treaty, the newly formed EU25 decided against holding domestic laws accountable to its principles (Puchalska 2014).
When put to the test, such safeguards proved vastly insufficient to prevent backsliding in the region, or to sanction authoritarian measures once backsliding had taken root.Today, there is a widespread scholarly consensus around the fact that the EU does not have 'the tools', in Reding's words, to deal with de-democratization within its remit (Blauberger and Kelemen 2017;Müller 2015).This is most obviously the case for the most severe cases of backsliding, namely Hungary and Poland.
While Hungary and Poland have been drifting further away from democracy every year since 2010 and 2015, respectively, a warning has yet to be delivered to both countries under Article 7.1. 1The three biggest stumbling blocks to enact sanctions include the unanimity threshold to suspend the voting rights of backsliding Member States under Article 7, the ECJ's weak jurisdiction over Member States' respect for fundamental values, and until recently, the absence of a mechanism for incremental economic sanctions.The new Rule of Law Conditionality regulation, in force since January 2021, partly addresses these insufficiencies but has yet to prove its efficacy in the case of Hungary and Poland.Meanwhile, democratic deterioration beyond these two extreme cases has broadly remained unaddressed. 2o focus only on the inadequacy of Treaty safeguards, however, would be to miss a significant part of this story.Post-accession democratic backsliding in CEE had been consistently abetted by many pro-EU leaders from the EC and EU-15 states.Early attempts to force democratic backsliding onto the EU institutional agenda were mainly initiated by West European centre-left policymakers whose aims were stymied as much by centre-right colleagues as by a lack of tools.Article 7 procedures against FIDESZ were consistently blocked by the European People's Party (EPP) under the leadership of Manfred Weber (Herman et al. 2021;Kelemen 2017;Meijers and Van der Veer 2019).From EC President Jean-Claude Junker's warm personal relationship with Orbán (Wachs 2014) to the continued ideological 'sponsorship' of parties like Bulgaria's GERB by Germany's CDU (Medarov 2020;Szabolcs 2020) backsliding regimes in CEE could count on the political support of many avowedly pro-EU leaders well into the 2010s.
The stakes are certainly high for the EU.The presence of authoritarian states within its remit creates political risks and raises normative challenges (Bellamy and Kröger 2021).Democratic conditionality has value for the EU from an instrumental perspective, as authoritarian states show weak records of mutually beneficial diplomatic and economic relationships with democratic regimes (Ibid).The Fidesz government has provided ample evidence of this since 2010, as it repeatedly vetoed common policy in matters of immigration, environment and institutional reform, systematically misappropriated EU funds (Magyar and Vásárhelyi 2017) and cultivated close ties with semi-hostile states such as China and Russia (Zalan 2019).But democratic backsliding also challenges the EU's intrinsic normative worth, raising legitimacy issues for the union via-a-vis its own citizens (Lacey 2017).The incomplete application of EU law in backsliding states threatens the association's legal integrity, while questioning the fundamental principle of the equality of European citizens in front of EU law (Kelemen 2019).Some have gone so far as to argue that by funding authoritarian states and including them within its decision-making processes, the EU makes itself complicit in deepening democratic deficits at its very core (Kelemen 2017;Müller 2015;Theuns 2020).Given what is at stake, how is it that comparatively little was done either to mitigate these risks or to respond to them when they first manifested?

Strategic and ideational factors in EU policy-making
In explaining the EU's lack of attention to democratic fragility in the region, it is key to stress the role of strategic considerations at the EU level, as well as the short-term political interests of individual Member States.Pressure to speed up the enlargement process, resulting in light-touch political criteria monitoring, stemmed from a variety of sources, often having little to do with preparedness for membership.These ranged from geopolitical concerns as states entered NATO and some participated in the US-British invasion of Iraq (Phinnemore 2010, p. 304), to economic ones with the EU15's impatience to expand their markets eastwards.In addition, as the Treaties were refined, some EU-15 states were reluctant to bring about democratic defence mechanisms that could undermine their own sovereignty in the future (Puchalska 2014).As for the willingness of so many EU decision-makers to provide political and institutional cover for backsliding regimes after accession, strategic motivations are also central.The role of the EPP in preventing censure of Fidesz has been explained as a function of simple political partisanship (Kelemen 2017), while the profit-making interests of German corporations are often cited in attempts to explain the country's curious embrace of illiberal leaders in CEE (Medarov 2020;Szabolcs 2020).
These rational, self-interested factors matter.Yet we argue that focusing exclusively on the language of interests and incentives fails to capture the ideational context within which it was politically possible for the EC to take a permissive attitude to pre-accession monitoring (Hughes and Sasse 2003), or for the EPP to protect FIDESZ into the 2010s (Herman et al. 2021).As highlighted by interpretive and cultural institutionalist traditions, rules within institutions do not result solely from rational choices-there exists an independent, causal role for the norms, values and forms of knowledge that make up the institutional culture and therefore inform the decision-making of actors within it (Bevir and Rhodes 2010;Hall and Taylor 1996).The forms of relationship we posit between ideas about CEE democratization and policy-making at the time are of an 'immanent' nature, whereby material factors, events, discourse and practices are embedded, and affect one another in a looplike fashion (Gofas and Hay 2010).Events trigger ideas and discourse about them.These ideas and discourse are shaped by pre-existing social, economic and political structures.In turn, ideas and discourse give meaning to the ways in which individuals react to triggering events.Finally, these actions give rise to new events, which gradually modify pre-existing structures over time.In line with this, we argue that the lack of attention to democratic fragility can be traced to an institutional culture that conditions how political actors within it understand democratization, as well as the EU's role within this process (Herman 2016).Focusing on the underlying logic of EU conditionality in the run-up to the 2004 and 2007 enlargements, we identify several key assumptions that structure the EU's approach to political change in the region.
First, a key idea present in the mechanisms of accession conditionality is that the EU has the power to bring about democratic political change through the use of both 'sticks' and 'carrots', thereby also assuming the rationality of local political actors.The EU established a number of incentive structures that reveals this logic, including 'gate-keeping' relating to membership itself, the conditional granting of aid and technical expertise and the indirect benefit of gaining market credibility as a result of satisfying the Commission's criteria (Bronk 2002, pp. 17-18).Interim credibility gains in the form of increased foreign investment, lower costs of capital and external aid were assumed to have a direct positive impact on transition, both by improving economic performance and by facilitating access to the financial resources necessary for the enactment of political reforms (Bronk 2002, p. 19 and 31).In short, EU membership was understood as a strong incentive towards political change, and as a way of avoiding the scenario of a 'freezing of reforms', which struck most of the former Soviet Union in the 1990s (Grabbe 2006, p. 101).This logic echoes rational choice premises prevalent in democratization studies in the 1990s and 2000s, according to which national elites in post-authoritarian settings implement democratic procedures when and if they have an interest in doing so (O'Donnell et al. 1986).It is the self-interested nature of elites, then, that would have provided the EU with 'transformative power': By alternating the carrot of incentives with the stick of sanctions the aim was to alter the utility calculations of CEE politicians to bring about convergence upon a common institutional settlement (Börzel and Risse 2012;Pop-Eleches 2007;Schimmelfennig andSedelmeier 2004, 2005;Vachudova 2005).
Second, the criteria for EU membership itself pre-suppose understanding democracy as an institutional arrangement guaranteeing that leaders are chosen through free and fair elections (Schumpeter 1956(Schumpeter [1942]]), rather than as a "regime" (Castoriadis 1997) that pre-supposes deeper ethical commitments from citizens to a democratic way of life.The Commission's major focus has been on the acquis' translation into national law, as well as on formal compliance with human rights legislation in candidate countries; surveying the effective implementation of the acquis or its understanding by decision-makers was far less central (Falkner et al. 2008;Haughton 2007;Hughes and Sasse 2003;Kochenov 2008).As argued by Sasse, this was particularly the case in fields crucial to democratization such as minority rights (Sasse 2008, p. 842).According to her, while Estonia and Latvia had publicly implemented minority rights legislation that satisfied the EU that progress was being made, 'socialisation effects [pointed] in the opposite direction' (Ibid).
Third, and relatedly, the more intangible, sociocultural dimensions of democracy were afforded less emphasis than its procedural dimensions (King 2000;Roberts 2010;Vachudova 2008, pp. 227-228).The weakness of civil and political society in the region was well-known and documented at the time, including by the European Commission itself.Multiple studies in the early 2000s demonstrated the lack of a vibrant civil society, mass distrust in the state and very low levels of political participation in CEE (Howard 2003;Kostadinova 2003;Marková, 2006).These well-known issues were not an obstacle to the 5th and 6th enlargements, which suggests they were not seen as essential to the survival of democracy as an institutional arrangement-it was instead widely assumed that such issues would be resolved through long-term mass and elite socialization into these institutions (Schimmelfennig 2007, pp. 39-40;Vachudova 2008, p. 864).Some authors go further to argue that the rigidity of EU conditionality contributed to deepening these socio-cultural issues by negatively affecting "the 'organic' development of accountable domestic politics, and of public debate over the substance of potential alternatives to the EU's regulatory model" (Grzymalała-Busse and Innes 2003, p. 66; see also Vachudova 2005, pp. 227-228).A degree of optimism was required to assume that, given these inauspicious conditions, key institutions would nevertheless be locked in, and over time, bring about more democratic political cultures.

Democratic optimism in the academic sphere: a quantitative assessment
In the second part of this paper, we pursue our empirical exploration of this ideational context by turning to the key assumptions about CEE democratization held by our discipline, political science, in the period 2000-2015.Our premise here is that taken together, EU institutions and academia should provide a faithful picture of the ideational context within which EU enlargement took place.If similar assumptions prevail within both policy and academic spheres, this will add further weight to the proposition that there existed a broader climate of optimism at the time that conditioned policy-making decisions well into the 2010s.We therefore aim to establish whether political science in the aggregate shared the EU's key assumptions about democracy, democratization and the role of the EU in bringing it about.There are good reasons to work with this hypothesis, as a number of papers have recognized and documented the turn of the century dominance of rational-institutionalism in the discipline generally (Lustick 1994;Marsh and Savigny 2004;Oren 2006;Shapiro and Green 1994), and in the field of Third Wave democratization studies specifically (Alexander 2002;Castoriadis 1997;Herman 2016;Khachaturian 2015).To our knowledge, however, there have been no systematic attempts to verify these claims with a quantitative study of CEE-related scholarship.

Data and case selection
To explore this question empirically, we conducted a quantitative analysis of 500 papers on CEE politics published between 2000 and 2015.Literature for the analysis were randomly selected from the top 20 international political science journals and top 5 East European area studies journals. 3Journals were selected based on Google Scholar rankings derived from the H-5 index, a metric based on citation counts.Second, we ran a search in all 25 journals using the keywords 'Central Europe', 'Eastern Europe', 'post-communist', 'EU enlargement', and each of the individual CEE country names, and their derivative forms.We restricted our search to papers published between 2000 and 2015, which, following the 1998-2000 opening of accession negotiations for all 2004-2007 entrants, tended to focus on democratic consolidation, rather than the earlier debates concerning the transition to democracy from state socialism.Our sample closed a year after Viktor Orbán declared his intention in July 2014 to build an 'illiberal nation-state' (Orbán', 2014), a milestone after which democratic malaise in CEE becomes harder to ignore.
We randomly selected 500 articles with a ratio representative of the entire publication record (over 1200 papers) in each journal type: 306 papers from area journals and 194 from general journals.The selection process included removing papers focused on countries uninvolved in the 5th and 6th enlargement waves; outside our period of interest (analysing events prior to 1989); of an unsuitable format (e.g.book reviews); or outside of our disciplinary remit (e.g., commentary on artistic forms).After removal, we replaced the excluded papers using the same selection method.This process was continued throughout to catch previously unidentified irrelevant papers.In total, 23 papers were replaced in the first batch, 14 papers were replaced in the second batch and 7 papers were replaced throughout the coding process.

Questionnaire and coding
Our questionnaire identified four aspects of academic papers we believe are relevant to our research question: Main topic, theoretical framework, conclusion, and citation count.Specifically for the first three aspects, we treated the resulting categorization into sub-aspects as individual variables and coded each as a binary variable (item = 1, others = 0).We maintained citation count as a single continuous variable (see below).In total, we generated 14 variables which, in line with our exploratory approach, were used alternately as both dependent and independent variables.For a more detailed coding scheme of each variable, see Online Appendix B. Five coders were involved in the coding process, each taking 100 articles in the sample.Two rounds of intercoder reliability tests were performed prior to the start of the process and once more when coders reached halfway into the number of articles for which they were responsible.Both Kappa tests yielded high levels of agreement (see Online Appendix C).
To identify the main topics in our sample, we employed a mix of deductive and inductive coding, beginning with a list of topics that was then refined throughout the coding process.By the end of the process, we had identified six main topics: Civil society (inclusive of minority rights and social movements); government (inclusive of parliaments and courts); EU accession, enlargement, and conditionality; political parties and party systems; political economy (inclusive of economic governance, and corruption); and public opinion/voting.
We coded the theoretical framework into four categories, following Hall and Taylor's (1996) understanding of new institutionalisms in political science.We coded as rational institutionalist (RI) papers that describe institutions as incentive structures that coordinate collective action.Such papers were often identified by the language of cost-benefit analysis or references to individual or institutional interests.Papers were coded as historical institutionalist (HI) when institutions were conceptualized as historically anchored patterns of constraints, with critical junctures creating periods of uncertainty where new outcomes might be locked-in, and eventually, create path dependency.Finally, cultural-ideal frameworks (CI) roughly approximate to sociological institutionalist assumptions, in which institutions are embedded within cultural norms.However, all papers sharing an epistemological focus on meaningmaking activities (discourse and practices) were coded as CI, also encompassing a range of constructivist and context-sensitive approaches that are not typically framed as 'institutionalist' at all.
We chose these categories given the influence of the new institutionalism paradigm in the early 2000s, specifically within the subfield of democratization-and after testing the applicability of these codes to a sample of our data.We nevertheless recognise that this classification simplifies our universe of cases, that not all scholars of politics would use these specific labels to describe their own work, and that each category stands-in for both looser and broader traditions in the discipline.In part for these reasons, we also included a fourth category, Evaluative/Descriptive, to capture scholarship that presented (usually) qualitative descriptions of events or reported findings without applying theoretical assumptions.An example of an evaluative/descriptive paper would be country or election reports.
One of the more difficult and consequential items were the conclusions of a paper, to be coded as 'positive', 'neutral/ mixed' or 'negative'.We assessed conclusions using two questions: 1) Do the authors associate a positive or negative value to the phenomenon under study?; and 2) According to the authors' findings, what is the trajectory or state of this phenomenon?For example, if the authors associate a positive value to European integration and found that Hungary was making progress towards membership, the paper was coded as positive.
Finally, we included citation count as a variable, which we collected from Google Scholar in March 2021.The period for data collection for this variable specifically is restricted to prevent variation and ensure consistency.The mean of citation count in the sample is 54.93 (SD = 69.72;N = 499) with a range between 0 and 488. 4  It is important to note that we do not cast judgements on the merits of individual papers.Not every paper coded as 'positive' about CEE's political trajectory has been discredited by subsequent events.However, we contend that the aggregate level trends that we aim to uncover in this study are revealing of the ideational context at that time and can help us understand the role of ideas in the 5th and 6th enlargements.Regardless of the merits of individual papers, we are interested in whether shared thematic, methodological or theoretical practices are systematically associated with the conclusions of papers, and which type of conclusion is most visible within the discipline as a whole.

Analysis and findings
Overall, we find that, while the literature on Central and East European politics was not overwhelmingly positive over the period 2000-2015, there are several factors which served to make the optimistic narrative a great deal more visible.below shows that the sample barely tilts positive (39.92%) over negative (34.13%), while a quarter (25.95%) remains neutral/ mixed.Four statistically significant patterns confirm this main finding.First, papers examining EU accession, enlargement and conditionality were more likely to be positive, while papers analysing civil society (inclusive of minority rights and social movements) tended to be more negative.Second, EU-accession-centred papers were cited more compared to civil society papers.Third, papers applying the RI framework were less likely to negative.By contrast, those employing CI frameworks were more negative compared to all other frameworks.Finally, the publication of RI papers significantly increased in the run up to the EU accession 'Big Bang' period between 2004 and 2007.
In short, while the literature overall is diverse, through inequalities in the visibility of various forms of research, political science still tended to promote positive accounts that confirmed the EU's monitoring processes while rendering more critical and guarded accounts marginal.

Papers on EU accession tended to be more positive while papers on civil society tended to be more negative
Table 2 below shows how the optimism of conclusions differs systematically depending on the main topic, despite addressing the same countries over the same period.The greatest divergence concerned papers addressing EU accession itself on the one hand and civil society on the other.Papers addressing the EU accession process are 19 percentage points more likely to be positive and correspondingly 13 percentage points less likely to be negative.Conversely, papers addressing civil society-inclusive of minority rights and social movementsare 17 percentage points more likely to be negative (and 15 percentage points less likely to have neutral or mixed conclusions).
For the countries of this region, one key element of the EU's democratic criteria-civil society-was overwhelmingly judged to be on a problematic trajectory.Despite this available evidence, the project of political transformation was still judged to be on track or even fulfilled by most papers, which focused on EU accession process as their main topic.Several papers in our sample illustrate this divergence.At one end of the scale, we find Schimmelfennig and Sedelmeier (2004) who addressed the EU's rule transfer to Central and Eastern Europe as a whole, finding that the EU was able to exert an 'unprecendented influence on the restructuring of domestic institutions and the entire range of public policies in these countries ' (2004: 661).At the other end of the scale, we find that Rechel was wondering what had 'limited the EU's impact on minority rights in accession countries', concluding via a case study of Bulgaria that these reforms had failed to translate to positive on-the-ground outcomes due to a range of factors: 'A lack of internal minority rights standards, an emphasis on the acquis communautaire, missing expertise on minority issues, the superficial monitoring of candidate states, a lack of concern for human rights and a failure in addressing public attitudes towards minorities' (Rechel 2008, p. 171).These two divergent accounts illustrate wider, statistically significant patterns within the field as a whole.EU Enlargement scholars appear to have seen a democratic forest while civil society scholars told us that that the trees making up that forest were not particularly democratic ones.Focusing on the respective Google Scholar citation counts of the two above-mentioned papers, we find that Rechel's paper has a tally of 98 citations while Schimmelfennig and Sedelmeier's paper was cited 1824 times since publication.This disparity is reflective of a statistically-significant relationship, as shown in Table 3.
Relative to other papers in our sample, papers with EU accession as main topic receive an average of 25.25 citations more, while civil society papers have an average of 18.23 citations less.When we directly compare papers from the two topics using a two-sample t test, the mean difference of 35.46 citations-with EU accession papers being cited more-remains statistically significant (t = -3.29;p = 0.001).While a systematic comparison of journal types is beyond the scope of this study, we hypothesize that the higher citation-count of EU accession papers is connected to the higher-ranking political science and EU studies journals they tend to be published in, while civil society papers are often published in lesser-ranked East European area studies journals. 5EU accession papers, which tended positive, enjoyed more prominence and visibility within political science as a discipline than did civil society papers, which tended negative.Rational institutionalist papers less likely to be negative, cultural/ ideal frameworks more likely to be negative In our sample, Rational Institutionalism (38.72%) is the most commonly used theoretical framework followed by CI (24.15%).At first glance, it seems reasonable to state that the emphasis on 'interests' and 'incentives' in RI accounts mirrors the EU's own internal logic of 'sticks' and 'carrots'.RI-inspired authors sometimes used the EU's own metrics-successful closure of accession chapters, 'compliance' with EU directives (Levitz and Pop-Eleches 2010; Sedelmeier 2012)-as evidence of success.
Our analysis provides cautious evidence that the use of RI frameworks is positively correlated with positive conclusions.Table 4 below shows that RI papers are 7 percentage points more likely to reach positive conclusions and 10 points less likely to reach negative conclusions.At the other end of the scale, scholars employing CI frameworks are 12 percentage points more likely to produce negative conclusions.In other words, the likelihood of findings that echoed the EU's optimism increased when scholars relied on frameworks closer to the EU's own technocratic proceduralism.

The predominance of rationalist institutionalism built to a peak at the time of the 5th and 6th accessions
As shown in Fig. 2, RI frameworks are the most popular in almost all years of our study, with a particular dominance during the first half of the period and leading up to the successful accession of 10 CEE countries between 2004-2007.At its peak in 2006, it represents 61.11% of our entire sample.Figure 3 shows that the rising boom in RI during this period corresponds with, and is thus likely to have driven, the rise in positive scholarship in tandem with the successful conclusion of the EU accession process.RI scholarship thus tends to confirm the optimistic bias of the EU accession process but also achieves its greatest prominence in tandem with the successful culmination of that process.
It is certainly true that academic optimism faded after the 'Big Bang' had happened.The average percentage of negative papers increased by 6.76 percentage points between pre-and post-2008 (t = −6.70;p < 0.000).Conversely, the average percentage of positive papers sees a significant decline by 7.24 percentage points (t = 9.76; p < 0.000) from 44.46% prior to 2008 to 37.23% after 2008.

Conclusion
Our analysis has shown that, in the period 2000 and 2015, both academia and EU institutions shared key assumptions about CEE democratization and the role of the EU in this process.Taken together, these shared assumptions suggest that the 5th and 6th enlargement took place in a wider ideational context of democratic optimism.Both sites had in common a view of democratization as driven by rational, selfinterested elites that the EU can incentivize into compliance; a minimal definition of democracy focused on its most procedural aspects; and an understanding of prodemocratic cultural change as following on from procedural transformation.
These assumptions permeate EU decision-making before and after the 5th and 6th enlargements: The European Commission used a logic of 'sticks and carrots' to induce political change, EU conditionality was overwhelmingly focused on legal transcription rather than effective implementation, and clearly observable weaknesses in CEE civil and political societies were not considered obstacles to accession.Our quantitative analyses of the political science literature similarly shows that democratic optimism correlates with a focus on formal procedures over informal norms, elite institutions over civil and political societies, and aggregate data over nose-to-the ground research.While the literature on the political and economic transformation of CEE as a whole was diverse, the articles that were closer in terms of rationalist analytical approach and thematic focus to the EU's own monitoring were overwhelmingly more likely to be rewarded with attention in the form of citations.Advocates of the 'external incentives model', built on rational institutionalist foundations stressing EU leverage, lead with an affirmation of the EU's sticksand-carrots approach to political progress in the region, with problems generally relegated to caveats and disclaimers.Critical studies of democratic progress in the region, in turn, paid greater attention to worrisome dynamics within civil society and political culture in the region, yet these accounts lacked visibility within the discipline.
Democratic backsliding in CEE today suggests that the more substantive dimensions of democracy-both deliberative and participatory-are important conditions for the long-term endurance of constitutional settlements.For the EU to assume culture would eventually follow skilfully designed institutions was a risky bet.It underpinned a series of decisions that EU decision-makers such as Reding and Schultz, as we recorded at the outset of this paper, themselves perceive as policy failures.Had they placed greater value on contextual and qualitative knowledge forms, the EU might today have more robust legal safeguards in place to tackle democratic backsliding.Treaty changes to strengthen such safeguards today are unlikely given the growing number of Member States that have an interest in maintaining weak checks on their undemocratic practices.But EU expansion to the Western Balkans already bears some marks of reflexivity on the outcome of the 2004-2007 accessions: It is now understood that political conditionality should be exercised over a longer period of time, and that more detailed criteria on the rule of law and human rights are needed to monitor democratization (Vachudova 2014, pp. 132-134).These efforts notwithstanding, the logic of the accession process itself could and should change.We hope that some of the energy currently devoted to checking compliance with the acquis could be diverted to a more sociologically aware effort at supporting bottomup democratization.
As for political science, our results demonstrate the potential benefits of greater methodological and theoretical pluralism in the discipline-a more balanced pattern of citations and publications might have contributed, in this specific case, to better outcomes in terms of knowledge building.Our results might also point to the need for greater separation between academia, and the spheres of political power they study.Today, scholars increasingly seek funding from policy actors and are assessed on the extent to which they can demonstrate 'impact' on decision-makers (Wilsdon et al. 2015)-factors that may make them more likely to adopt the language and concerns of the institutions they study.This risk may be heightened in the case of the EU, which is one of the largest funders of social science in the world.
Future empirical studies would need to investigate the mechanisms that explain the parallels we evidence and study these links between the world of EU decisionmaking and the sphere of EU scholarship before and after the 2004-2007 enlargement waves.Shared socialization is likely a key factor, as both established political scientists and policy-makers in the early 2000s would have been educated in a similar context, one in which behaviourism and public management theory dominated the social science curriculum (Hay 2007, pp. 90-123;Ishiyama et al. 2006), and the zeitgeist was shaped by imaginings of the 'end of history' (Fukuyama 1992).But we can also hypothesize both sites influenced one another in the early 2000s, with a two-way relationship between dominant assumptions within scholarship, and dominant assumptions within policy-making (Stoker and Evans 2016, pp. 4-5).Given that the EU, by its own account, relies on expertise as a basis for its actions (Offe 2009;Tsakatika 2007), it would be key to study the extent to which academic work was consulted at the time by European Commission employees and European Parliament research assistants.As highlighted above, there is also room to explore whether the contemporary impetus to produce 'impactful' research and obtain external funding encourages academics to mirror the logic of the institutions they study (Wilsdon et al. 2015).
Figure 1 below shows our finalized sample distribution across 2000-2015.

Fig. 1
Fig. 1 Sampled papers about CEE politics per year

Fig. 3
Fig. 3 Distribution of conclusions over time

Table 2
Impact of main topic on conclusion

Table 3
Impact of main topic on citation

Table 4
Impact of theoretical framework on conclusions Proportion tests show that more civil society papers are published in area journals at a rate of nearly four times more than in general journals (t = −5.64;p < 0.000), while almost double the number of EU accession papers are published in general journals compared to in area journals (t = 3.07; p = 0.002).