How challenger parties can win big with frozen cleavages: Explaining the landslide victory of the Servant of the People party in the 2019 Ukrainian parliamentary elections

In the Ukrainian parliamentary elections of July 2019, Volodymyr Zelensky’s new party, Servant of the People (SN) won a majority of seats in the context of what had been a ‘frozen cleavage’ dividing party voters along a single geo-cultural dimension: pro-West/anti-Soviet versus anti-West/pro-Russian positions. Analysing a unique set of surveys of public and expert opinion, we find that its unprecedented success stems from the extreme weaknesses of the existing and often discredited parties rather than ideological shifts. Our findings also question whether challenger parties in other contexts, including consolidated democracies, must compete on new issue dimensions in order to succeed electorally.


Introduction
In the Ukrainian parliamentary election of July 2019, Servant of the People (SN), the new party of Ukraine's recently elected outsider president, Volodymyr Zelensky, won a majority of seats and 43% of the popular vote. This result was unprecedented in post-Communist Ukrainian politics, and occurred against a background of what we (Chaisty and Whitefield, 2018) have evidenced as a 'frozen cleavage' in party competition. Frozen cleavages ought to make some strategies of new entry parties more difficult. How then can we explain the success of Servant of the People?
We address this question through the lens of the broader comparative literature on party competition. First, we consider whether SN followed a theoretically anticipated strategy of new challenger parties elsewhere by positioning itself on a new salient dimension on which established parties were less able to compete (Bergman and Flatt, 2019;Meguid, 2005;Meyer and Miller, 2015;Rovny and Edwards, 2012;Zons, 2016). Some recent scholarship pointing to growing multi-dimensionality in Ukrainian politics since 2014, when war and crisis transformed the country's politics, could lead us to expect this possibility. 1 However, we do not find evidence for the emergence of new cleavages dividing party voters. Second, we consider whether, in line with some directional approaches to party competition, SN was able to appeal to voters in large numbers by the clarity of its stance on one side of the main cleavage (Rabinowitz and Macdonald, 1989). Again, we find no evidence that this strategy might explain SN's success. Instead, the party adopted a highly centrist stance on this dimension, which was much more in line with what Rovny (2012) calls a 'blurring strategy'. Third, consistent with demand-side approaches to electoral change (Hobolt and de Vries, 2015;Hobolt and Tilley, 2016) we consider whether the success of SN resulted from a significant compositional shift in the electorate towards the centre of the distribution on the main cleavage. While compositional change did occur in Ukraine, following Russia's annexation of Crimea and the de facto disenfranchisement of a substantial proportion of voters in rebel-held areas in the east of the country, this explanation also lacks support. Finally, in line with catch-all approaches to party strategy that emphasise valence-based strategies (Abney et al., 2013;Green and Hobolt, 2008), we consider whether SN was more or less likely than parties in previous elections to win voters across the spectrum of attitudes on the main dimension. This hypothesis does appear to hold up well for two reasons. First, the party's connection to Zelensky, which gave it a competitive advantage over rival parties that tried to adopt the same strategy. Second, the strength of SN's anti-establishment appeal, and its credible alternative to the existing and often discredited parties (Pop-Eleches, 2010).
The structure of the rest of the paper seeks to develop these theoretical issues with original empirical evidence. First, we briefly overview the Zelensky phenomenon and the 2019 parliamentary election; present the comparative literature on challenger parties, and discuss what we might expect of SN in the Ukrainian context given our theoretical priors about post-Communist party competition. Second, we introduce our data and measures and then the empirical analysis that allows us to address our main theoretical expectations. Finally, we conclude by reflecting on the implications for Ukrainian party development and for challenger parties in a broader comparative context.

Theorising challenger parties in post-Communist circumstances
Even by the standards of Ukraine's high electoral volatility, Volodymyr Zelensky's election to the post of president on 21 April 2019 was remarkable. 2 Zelensky became president less than 5 months after announcing his candidacy on New Year's Eve 2018.
Zelensky's embryonic party, the Servant of the People (SN), had little time to organise for the subsequent parliamentary elections held on 21 July 2019. 3 It faced powerful local networks (Herron and Sjoberg, 2016); competition from another well-resourced outsider party 'Voice', led by charismatic rock star Svyatoslav Vakarchuk (Zimmerman, 2018); and the prospect of low turnout from Zelensky's youthful voters (Kudelia, 2019). Yet, SN won the July election with the highest seat share of any post-Soviet Ukrainian political party. The party gained 43% of the popular vote and 254 out of 450 seats in the Verkhovna Rada, Ukraine's parliament. The scale of SN's victory was unprecedented. Although new parties do emerge on a regular basis in Ukraine (Rybiy, 2013), 4 none had managed to win the outright majority achieved by SN.
In explaining the success of challenger parties like SN, the bulk of the scholarly literature pertains to consolidated democracies. Political divisions between established parties in these circumstances are often long-standing and may even relate to the foundations of democratic competition in the country (Lipset and Rokkan, 1967). Parties have well-known positions on the principal cleavage (Meguid 2005;Meyer and Miller, 2015), and have deep social and organisational ties to the electorate. In the famous phrase of Lipset and Rokkan (1967), divisions within party systems may become frozen. Consequently, new parties wishing to enter the electoral arena face significant obstacles to success since established parties have advantages in respect of the bases over which elections are fought, ownership of the issues, and linkages to voters (Rovny and Whitefield, 2019). How do new parties breakthrough in these conditions? How appropriate are these explanations in the Ukrainian context?

Cleavage-transforming strategy
The literature on the strategies of challenger parties in the context of established democracies focuses largely on how they position themselves on new issue dimensions that are highly salient to an electorate that divides across established party lines. Consistent with directional theories of voting (Rabinowitz and Macdonald, 1989;Rovny, 2012), challenger parties signal the salience of the new issue by adopting more extreme and clear positions. This poses a challenge to established parties. New issues fit poorly on to existing divisions, and can be a source of conflict within established parties and their electorates. In other words, existing lines of cleavage may defrost sufficiently to allow new parties to mobilise on new issues.
Of course, established parties are not without strategies that they can deploy to reduce the challenge that new issues and new parties can pose. In an influential piece of research, Meguid (2005) has argued that established parties may seek to marginalise challengers by dismissive, accommodative or adversarial policy approaches, with varying degrees of success depending on the national context in which established parties pursue these strategies. Other research has shown that established parties may deal with new issues by 'embedding' them alongside the issues on which they are most advantaged (Kriesi et al., 2006;Whitefield and Rohrschneider, 2019). Furthermore, established parties may seek to shift their positions on the new issue in order to win support from voters who are now most motivated by it, (Meijers and Williams, 2020), though both theory and evidence suggest that this may be a suboptimal move, as it may appear as a concession to the challenger party on an issue over which it is likely to have superior ownership (Rohrschneider and Whitefield, 2016).
How plausible and applicable are these understandings of challenger party strategies -and established party responses -in the Ukrainian context and in the 2019 election in particular? In the context of a strengthening of civic identities since 2014 (Kulik, 2018;Pop-Eleches and Robertson, 2018), Zelensky's social and economic libertarianism created the possibly for a new left-right division between SN and more conservative and economically paternalistic parties (Ishchenko, 2019). Issues like SN's pledge to end the moratorium on agricultural land privatisation, for instance, had the potential to open up a new cleavage. However, as our previous research has shown (Chaisty and Whitefield, 2018), there are good theoretical reasons to expect otherwise. Notwithstanding the inchoate nature of the Ukrainian party system since 1991, significant shifts in the supply of parties in Ukraine has proven quite compatible with stable underlying frozen cleavages, a finding supported by other work on post-Communist systems (Rovny and Polk, 2018). The reasons for this cleavage stability in the context of party organisational and electoral flux centres on two interconnected aspects.
First, it is precisely in the context of highly unstable party organisations that both parties and voters are likely to find an ideological focal point in the most clearly established lines of political division (Whitefield and Evans, 2000). To take a stylised example: at the very onset of a new party system in which parties and voters have only a very limited understanding of their mutual issue positions, the most likely convergence will be on issues of most historical and contemporary prominence. These issues may be stabilised electorally over time as party-voter linkages are built and as policy feedbacks come into play; parties will pursue policies that reward their supporters and so constituencies emerge that reinforce the main line of cleavage. The absence of party stability over time, however, need not -and we argue likely will not -undermine the cleavage basis that emerged at the first election. Rather, the problem of a focal point facing voters deciding on new parties -and new and weakly organised parties -remains similar to that found in the early election, but with anything even clearer evidence to parties and voters that the existing division is the best way to appeal.
Second, organisational weaknesses of parties within the post-Communist context with weak voter linkages means that the incentives to challenger parties to appeal on a new issue dimension are much reduced because existing parties have much lower reputational connections (Pop-Eleches, 2010). Because these parties fail to assert issue ownership, it is much easier for challenger parties to stake their ground on existing lines of division rather than needing to shift the terrain of politics to an area in which existing parties have fewer advantages. Indeed, in the post-Communist context in which political problems are endemic and established parties are often held in disrepute, the party system is much more open to challenger parties to lay claims to resolve issues on the main cleavage that existing parties have failed to deal with (Hanley and Sikk, 2016;Haughton et al., 2011).
Therefore, while the literature on challenger parties in established party systems points to good reasons why such parties might pursue new issues -issue ownership, potential to divide existing parties and constituencies -the same factors in post-Communist systems suggest that challenger parties, such as SN, are much less incentivised to pursue a strategy of pursuing new issues. Instead, in these conditions, challenger parties are much more likely to compete with established parties on the existing cleavage structure. We therefore expect to reject H1: H1: The success of SN was the result of its appeal to voters on a new issue cleavage.

Centrist strategy
Why then did SN do so much better than previous challenger parties in the 2019 election? Was it perhaps, in line with directional theory, that it positioned itself much more clearly on the main salient dimension? The intuition here is that in conditions of considerable uncertainty about where a new party stands, voters will be attracted to a party that stands out. We doubt that this will be the case in the Ukrainian context. In our view, there are strong theoretical reasons to expect that a directional strategy will not be effective when parties compete over existing cleavages in a polarised electorate. A strategy of attracting attention by emphasising extremism on this cleavage is as likely to alienate voters who have already made their minds up that they stand at the opposite extreme. After all, the existing cleavage is one on which, almost by definition, voters have already established their position on both sides of the issue dimension. Therefore, assuming that a new issue dimension did not emerge in 2019, we do not expect that, in the absence of a major change in the views of the electorate, SN would have chosen to place itself at one extreme of the existing dimension. Faced with a polarised electorate on the existing cleavage, such a position would be unlikely to produce the large margin of victory that SN obtained. We therefore expect to reject H2a: H2a: The success of SN resulted from taking a stance at a more extreme position on the existing cleavage.
An alternative approach for a new party would be to position in the centre ground of politics. We can think of myriad examples in which political parties respond to changing voter preferences that enable them to improve their electoral performance, notably as parties move right or left with voters' 'thermostatic preference shifts' (Wlezien, 1995) or as parties respond to growing social liberalism and conservatism (Inglehart, 1990;Kitschelt, 1994). This was plausible in the Ukrainian context. Zelensky provided the possibility for a centrist alternative to the nationalism of the incumbent President Petro Poroshenko, by promising to heal the social wounds created by the war in the east of the country. The support of SN for a negotiated solution to the conflict and more inclusive language policies had obvious centrist appeal.
However, a successful centrist appeal also requires a large and electorally dominant constituency of voters that congregates in the political centre, and is unattached from existing parties. And, for such a constituency to produce the scale of SN's success in 2019 would have required a compositional shift in the electorate of unprecedented proportions. We are sceptical about this possibility. Stances on established cleavages are ordinarily sticky, particularly when as in the case of Ukraine voters' views are rooted in identities and social locations, such as region, ethnicity and language (Chaisty and Whitefield, 2018;Rovny and Polk, 2018). Without this compositional shift, a centrist strategy would be unable to win big. As with the directional strategy, a centrist appeal would still risk alienating large swathes of voters clustering on either side of the dominant cleavage. We therefore expect to reject H2b: H2b: The success of SN resulted from a centrist appeal that capitalised on a large compositional shift of the electorate to the political centre.

Catch-all strategy
Our explanation for SN's success builds on insights into catch-all party strategies (Mainwaring, 2019). Catch-all parties aim to attract voters with very different attitudinal stances, appealing to a very large part of the electorate. They do so by positioning themselves in the ideological centre of the main cleavage and by shifting their appeal to a valence capacity -political or economic competence, trust in leadership, etc (Green and Jennings, 2017). Note that there is no reason to expect that voters will agree about how well a party instantiates a valence issue, and indeed parties may seek to compete on differing valence issues, e.g. competence versus positivity (Kosmidis et al., 2019). In the Ukrainian context, these sorts of valence appeals also include opposition to corruption. We do not consider anticorruption appeals as a possible new cleavage, since it is difficult in theory and practice to find pro-corruption parties. Therefore, unlike a purely centrist appeal, this strategy captures a much wider range of voters. The key characteristic of valence issues is that they are not programmatically spatial.
In the context of the widely perceived failure of the old parties to manage a country's challenges, a new party might be most successful in winning the votes of disaffected voters not by pushing new issues or by adopting extreme positions but by deemphasising ideological divisions on the existing cleavage. In Jan Rovny's phrase, challenger parties may compete 'by muddying their position' (Rovny, 2012). However, while Rovny considers muddying to be an effective strategy of challenger parties on the main dimension while they take an extreme position on a new issue, we consider a case in which a challenger party competes on the main dimension and is simultaneously catch-all. In this case, a party would seek to appeal successfully to a broadly shared valence concern that is salient across the electorate as a whole, regardless of where voters stand on the main cleavage dimension. That is, it takes a centrist muddy position while emphasising a non-ideological valence difference with other parties on a set of policies that it can credibly claim to deliver.
This strategy is unusual from the perspective of the comparative literature. Challenger parties would normally lack the credible reputation to deliver competence or the organisational capacities to reach such a diverse set of voters. However, the opening for a challenger party to pursue a catch-all strategy may be most present in the conditions that we have described: when new issues are most difficult to pursue and in any case their pursuit is electorally unnecessary, and when old parties are highly discredited. All that may be required in that context is for voters across the spectrum of the main political cleavage to share that sense of disaffection and for a credible centrist/muddy political party to emerge. Thus, SN had the opportunity to leverage a catch-all appeal around a powerful valence issue: transformation of the existing party system (Hosa and Wilson, 2019). We contend that this appeal explains its huge success. We therefore expect to find support for H3: H3: The success of SN may be attributed to its ability to offer a catch-all appeal on a valence issue.
In what follows, we put these theoretical expectations to the test. We first consider whether the 2019 election reproduced well-established partisan divisions or whether it introduced a new dimension to political competition. We then consider where SN and other parties positioned themselves on the main cleavage, before analysing whether there was a compositional shift in the attitudes of the electorate towards the centre of the distribution. Finally, we investigate where SN positioned itself on the main issues and how it distinguished itself on non-ideological valence issues.

Method and data
We investigate the electoral success of SN by analysing both original mass and expert survey data. The survey research provides evidence for testing the demand-side explanations for hypotheses 1, 2a, 2b and 3, while the expert survey provides evidence for testing the supplyside explanations that underpin hypotheses 2a, 2b and 3.

Mass survey
Our public opinion analysis is based on the results of a mass survey of 2040 Ukrainian adult citizens (18 years and older), which was conducted by the Ukrainian polling agency, the Kyiv International Institute of Sociology (KIIS) in August 2019. The survey was representative of all regions, apart from the rebel-held areas of eastern Ukraine. The sampling details and response rate are included in the Online Appendix.
We replicate the research design of the analysis of previous Ukrainian elections in 1994, 1998, 2007and 2014(Chaisty and Whitefield, 2018 to assess both the cleavage transforming and centrist strategies. Our dependent variable, the party preferences of respondents, is measured by the question 'Which party did you vote for in the elections to the Verkhovna Rada of Ukraine on 21 July 2019?' The distribution of the support for those parties voted for by 20 or more respondents is summarised in Table 1. 5 Our explanatory attitudinal variables focus on aspects of the geo-cultural dimension of voting that predominated in the analysis of parliamentary elections before 2019 (e.g. Birch, 2000;Chaisty and Whitefield, 2018;Clem and Craumer, 2008;Herron, 2014). They cover ideological views of citizens towards the Soviet past and independence, and relations between Ukraine and the West (EU): 6 (1) 'The dissolution of USSR was a good thing' 7 (2) 'Ukraine should integrate as far as possible with the European Union' OR 'Ukraine should remain isolated as far as possible from the European Union' 8 In addition, we include three questions, which address aspects of civic nation building that has arguably strengthened since 2014, and which correlate significantly with party support in other analysis (Chaisty and Whitefield, 2018: 163): (3) 'Tell us, please, what do you think about the idea that a democracy, in which multiple parties compete for power, is the best system for governing Ukraine?' (4) 'And what do you think about the idea that a market economy, in which there is private property and economic freedom for entrepreneurs, is the best system for Ukraine?' (5) 'All minority ethnic groups in this country should have to be taught in Ukrainian' Consistent with earlier work (Chaisty and Whitefield, 2018: 162), we use a combination of discriminant analysis and likelihood ratio analysis derived from multinomial logistic regression to estimate the most significant attitudinal predictors structuring party voting. We also use the same attitudinal questions to analyse whether significant shifts to the centre had occurred.
To assess the catch-all strategy, we analyse mass survey data to assess the strength of SN's valence appeal. Using ordinal logistic regression of a question about the ideological distinctiveness of Ukrainian parties, we estimate to what extent supporters of SN were more likely to reject the existing party system.

Expert survey
In addition to public opinion surveys, we use expert surveys to assess the positioning of SN on issues that comprise the main party cleavage, and to analyse the leadership appeal of Zelensky compared to that of other party leaders. This analysis provides a test of the supply-side evidence for the cleavage-transforming, centrist and catch-all strategies. Our expert data enable over-time comparisons. We analyse four waves of expert surveys in Ukraine: 2002Ukraine: , 2007Ukraine: , 2013Ukraine: , and 2019. Each survey involves 10 experts recruited from a master list of scholars who had published a peer-reviewed article or book on the Ukrainian party system in the past 10 years. They form part of a comparative project of party systems in 27 European states, which has involved over 1000 experts since the early 2000s (Rohrschneider and Whitefield, 2020). 9 Given the significant number of small parties that exist in many party systems, including Ukraine, we use two criteria to determine the inclusion of parties in the expert survey: (1) whether they had representation in a national parliament, and/or (2) whether they received at least 2% of the national vote in the previous election. The 2019 Ukrainian expert survey, which was well underway before the start of the 2019 parliamentary election campaign covered nine parties. This number includes the most successful parties in the election -SN, Opposition Platform-For Life, the Petro Poroshenko Bloc (European Solidarity), 10 and the All-Ukrainian Union Fatherland. It also includes parties that eventually gained only a handful of seats -the Opposition Bloc, the All-Ukrainian Union Freedom, and the Union Self Reliance -as well as parties that gained no seats at all -Civic Position and the Radical Party of Oleh Lyashko. The survey omits Voice, which emerged after the presidential election in May 2019.

Cleavage-transforming strategy
Our analysis of public opinion data finds little evidence to suggest that SN's success was the result of the transformation of the existing cleavage structure. Table 2 summarises the results of both the discriminant analysis and the multinomial logistic regression analysis. The discriminant analysis finds that attitudes determining or discriminating between the party choices of respondents were still overwhelmingly concentrated on one dimension. Variables covering attitudes towards integration with the EU, dissolution of the USSR and the teaching of Ukrainian language in schools, accounted for almost 95% of the variance in party support. The results of the likelihood ratio test, which we derive from multinomial logistic regression, support these findings. The one-dimensional nature of the party cleavage is consistent with the analysis for earlier elections. We find that this geo-cultural dimension explained 90% or more of the variance in support for parties in parliamentary elections that took place in 1995, 1998, 2007and 2014(Chaisty and Whitefield, 2018. The only notable difference with the results for 2014 is the position that voters took on the teaching of Ukrainian in schools, which loaded onto this first dimension in 2019. This may be a consequence of the more prominent position that language issues played in the presidential campaign, with Petro Poroshenko's nationalist appeal to 'Army, Language, Faith'. As in 2014, the position that respondents took on EU integration was the most powerful predictor of their party support. Table 2 indicates the significance of this effect by the relative size of the correlation coefficients and chi-squares generated by the discriminant analysis and likelihood ratio tests.
Therefore, SN achieved its success by mobilising support within the existing cleavage structure. In contrast to the analysis of contender parties in more consolidated democracies, there is little evidence to suggest that the emergence of a new dimension accounts for SN's majority vote. Our discriminant analysis finds that the position that respondents took on the question of private property -the second main dimension in 2019 -accounted for only 3% of the variance in the support for different parties. Hence, we reject H1.

Centrist strategy
Rather than transforming the existing cleavage, both the mass and expert surveys suggest that SN sought to maximise its support by orientating towards the existing cleavage structure through an appeal that avoided concentration on either extreme. Figure 1 plots the mean scores produced by the discriminant analysis for the voters of each party on both the main dimension and the private property dimension. The chart clearly illustrates SN's success in avoiding the polarisation that had restricted the appeal of parties in previous elections (Chaisty and Whitefield, 2018). Unlike the voters of winning parties in earlier years, the plot of SN voters on both dimensions does not cluster at either extreme.  According to our expert survey, the mean position of SN's voters correlated with the centrist stance that the party took on the main issues that comprised the dominant cleavage. This was evident on the issue of European integration, which was the most significant discriminating issue in 2019. Figure 2 compares expert assessments of SN's position on this issue with expert assessments of the positions (on the question of European integration) of all parties that gained the largest share of the vote in parliamentary elections held in 2002, 2007 and 2012. 12 More specifically, the chart shows the difference between the aggregate expert score for each party and the Ukrainian median party score for each expert wave. The perfect correlation with the median party score (indicated by the absence of a bar) clearly indicates SN's centrism in comparison with the largest parties in earlier elections. Hence, we reject H2a.
However, centrist positioning alone cannot explain the scale of SN's success in a system of frozen cleavages. There is little evidence to suggest that compositional changes produced a large shift in attitudes towards the centre. Our survey responses to the most significant attitudinal questions (each with a five-point scale from 'strongly disagree' to 'strongly agree') show no growth in respondents at the median of the distribution. On the contrary, the median positioning of voters in 2019 shows the political centre to be smaller than was the case in 2014. For instance, on the question of European integration, the war in the east of the country significantly moved the median towards strong support for EU integration in 2014: over 55% of the voters that we surveyed strongly agreed that Ukraine should integrate as far as possible with the EU (Chaisty and Whitefield, 2018: 167). In 2019, this median group had shifted to those who would 'rather integrate' than isolate, and it comprised just 20% of the party voters in our sample. 13 Moreover, the dispersal of survey responses on European integration was greater in 2019 than in 2014, with higher median absolute deviation. Hence, we reject H2b.
Instead, analysis of the EU question and other issues that comprise the main party cleavage show that SN won large support by attracting voters who occupied all positions on these issues. Figure 3 illustrates this by comparing the percentage of SN voters across each response category (from 'definitely isolate' to 'definitely integrate') on the question of European integration, with the survey responses of those voters who supported the largest party in previous elections. As can be seen, SN secured high levels of support from voters who occupied the full gamut of positions on European integration. In contrast, the support of the largest parties in earlier elections was skewed by voters concentrated on one side of the 'definitely isolate'/'definitely integrate' divide. By attracting large support on issues like this, SN was able to secure a landslide in 2019 while parties in earlier elections did not. This further supports the rejection of H2a.

Catch-all strategy
To pursue such a successful catch-all approach in the absence of a large political centre or a programme capable of transforming the existing political cleavage, two factors appeared to be of critical importance: the presence of a charismatic leader, and the existence of a valence issue that spanned the entire electorate.
Without Zelensky's victory in the presidential elections, it is unlikely that SN would have achieved a landslide in the parliamentary contest. Zelensky's success in the presidential ballot was pivotal in making SN the focal point for the political centre. The decision by Svyatoslav Vakarchuk not to run in the presidential contest significantly weakened the challenge of his contender Voice party. The expert survey clearly illustrates the importance of this leadership factor. Figure 4 cross tabulates the mean expert scores on two questions: the importance of party leadership in determining policy, and the importance of the leader's charismatic appeal. On each question, experts assessed SN as both the most leader-dominated party and the party with the leader who had the most charismatic appeal. In all the expert surveys for previous elections, parties that secured the highest share of the popular vote did not score so highly on these questions. However, as the fate of Oleh Lyashko's Radical Party illustrates, the presence of a dominant and charismatic leader is a not sufficient condition for electoral success. Lyashko's party failed to secure a seat in 2019. Rather, the catch-all strategy of SN was successful because it was based on the anti-establishment appeal of Zelensky. The valence issue of replacing the existing party system and political establishment dominated the election, and Zelensky and SN were unrivalled as the political force believed by voters to be the most capable of delivering it.
We analyse the strength of attitudes towards the existing party system by comparing the responses of SN voters with those of the supporters of other parties to the question 'All parties in Ukraine offer the same programmes'. Our analysis in Figure 5 shows the extent of the anti-establishment mood among SN voters. Holding at their means the most significant socio-economic and demographic predictors of voting in Ukraine -region, language, age, religiosity, and economic prosperity (Barrington, 2002;Clem and Craumer, 2008;Khmelko and Wilson, 1998) -we find that SN voters were significantly more likely than the supporters of other parties to 'agree' or 'fully agree' that all parties offer the same programmes. We also interpret this as further evidence that SN voters were not ideologically distinct from other parties, since respondents presumably took into

Definitely isolate
Rather isolate Neither/nor Rather integrate Definitely integrate Communist (1995) Communist (1998) Regions (2007) Poroshenko (2014) Servant of the People (2019) Figure 3. Proportion of each category of respondents on the question of European integration (by response categories from definitely isolate to definitely integrate) who voted for the largest party. account SN's position when making judgements about differences among parties. 14 Given the high level of cynicism about political parties in Ukraine, and the political class more generally, these marginal effects are striking. When we conduct the same analysis for elections in previous years, voters for parties that secured the largest share of the popular vote were no more likely than those who voted for other parties to believe that parties were all the same. This highlights the extent of the anti-establishment mood that SN was able to capture across the entire society, and adds further support to our theoretical expectation that its success can be attributed to its ability to offer a catch-all appeal on this valence issue. Hence, we find support for H3.

Discussion and conclusion
The analysis above has addressed the issue we set in the title of the paper. We find that it is indeed possible for an entirely new challenger party to win a convincing parliamentary majority even when an election is based on 'a frozen cleavage' on which parties had previously been highly divided and polarised. Servant of the People did exactly that in the 2019 Ukrainian election. It competed and won an election in which the main line of division in how voters chose parties remained over identity and system-level issues relating to the fundamental orientation of the country. If the literature suggests that challenger parties should compete by politicising new issues, Ukraine provides evidence to the contrary.
We also find, however, that the success of Servant of the People was contingent on other factors and strategic choices. Given that voters remain polarised on the geopolitical issues facing the country and that there was no significant compositional shift to the centre, the success of SN's muddy position on the main cleavage was contingent on other factors at play. This may include a degree of exhaustion with the military conflict and political crisis associated with the literal tug-of-war between the Russian and European poles of the cleavage dimension. But the main conditioning factor was likely the broader collapse in the prior party system under the weight of its inability to resolve the country's main problems. In that context, a centrist position on the main cleavage meant that SN did not alienate voters at either extreme while at the same time it offered them an alternative valence-based appeal across the entire spectrum. That appeal was the perceived competence, freshness and charisma of the party's leader Zelensky. Whether Zelensky's charisma is a necessary condition for the success of a challenger party in this context is worthy of further consideration. As we suggest below, Emmanuel Macron's charisma may also have played a role in a similar challenger party insurgence. But whether necessary or not, SN was able to outcompete other challengers positioning themselves in the same way; Zelensky's stunning victory in the presidential contest was the ideal focal point for that strategy.
What are the implications of these findings, first for Ukraine and then more broadly? In Ukraine, the main cleavage appears as frozen as ever. While some scholars (Fedorenko et al., 2016) Figure 5. The predicted probabilities that supporters of Servants of the People and those of other parties 'fully disagree' to 'fully agree' with the statement that all parties in Ukraine offer the same programmes (ordinal logistic regression).
view, is suggestive of the possibility of a return to a much more polarised politics in Ukraine around identity and geostrategic views of the country. When the conditions that allowed the huge victory of SN -exhaustion, collapse of faith in the existing party system, and a powerful valence candidate and party -are absent, a centrist position to build a catch-all appeal may no longer be optimal. Absent these particular conditions, in fact, we might expect that parties may succeed by reverting to a much more extreme position on the main dimension. Much of course may depend on how SN and Zelensky manage their political opportunity. Competence in government, success in fighting corruption, and perhaps above all delivering on those while pursuing a centrist strategy -between Russia and the EU -might well prevent repolarisation and indeed lead to the sort of compositional shift to the centre of the cleavage for which we found no evidence. We might speculate that if that happened, a new cleavage dimension might be more likely to emerge in Ukraine. But no party has yet managed the geo-strategic and domestic challenges that Ukraine faces. What finally does our analysis contribute to understanding party politics outside Ukraine? Its applicability to other post-Communist countries that also display 'frozen cleavages' (Rovny and Polk, 2018) seems obvious. In many of these societies, parties are also in considerable disrepute with weak voter linkages and with openings to challenger parties to politicise corruption and government failure, if a suitable focal point for catchall appeal can be found. Winning big may depend most on the last of these conditions. However, we must also consider the possibilities of challenger parties in consolidated democracies winning big without radical shifts in the cleavage structure. In these countries, of course, established mainstream parties have much stronger grips on historic divisions and therefore greater ability to outcompete challengers. Established parties for the same reason may have greater difficulties in competing on new issues because these can cut across existing ideologies and constituencies. These two conditions may make it much more likely that challenger parties will choose to try to build support on new rather than old divisions. Nevertheless, these incentives may not be as powerful as they once were for challenger parties in consolidated democracies, where the failures of established parties to deliver on both old and new issues appears to be driving greater volatility in both the supply of parties and voter attachments. Therefore, in consolidated democracies, challenger parties with charismatic leaders against a background of deep voter dismay at party performance -Macron and La Republique En Marche perhaps -may find electoral success not by transforming cleavages but by offering themselves as able to deliver solutions to a political malaise.
it. In discriminant analysis for 2014, which replicates the analysis discussed below, we find that attitudes towards government policy on Russia load on to the first discriminating dimension alongside attitudes towards the West (EU). Attitudes on this question also correlate strongly at the bivariate level with questions about the West (EU) and the Soviet past. This is entirely in line with our expectations given the nature of the issue cleavage. We have no reason to believe, therefore, that the war had subsequently become cross cutting or that attitudes towards the West (EU) or the USSR are not connecting with a 'geo-cultural' dimension. These results are available on request. 7. Questions 1, 3-5 have five response categories for each statement: (1) strongly disagree/oppose, (2) disagree/oppose, (3) neither/nor, (4) agree/support, (5) strongly agree/support. See Online Appendix for descriptive statistics. 8. The response categories for Question 2 are (1) definitely the first opinion, (2) the first opinion rather than the second, (3) neither/nor, (4) the second opinion rather than the first, (5) definitely the second opinion. We reverse the order of coding in the analysis and in the descriptive statistics included in the Online Appendix, so that 'Ukraine should integrate as far as possible with the European Union' is the second opinion. 9. See Online Appendix for further details. 10. The name of this party changed to European Solidarity after the presidential election in May 2019. 11. (*) indicates the largest absolute correlation between each variable and the first discriminant function (i.e. the dominant cleavage) . 12. The survey asks experts to position the stances of parties on the question of integration with Europe on a seven-point scale. We converted the results to a 0-1 point scale for ease of interpretation. 13. On other attitudinal questions -on the market, democracy, language in schools -the size of the median response category for all party voters was smaller than in 2014 and closer to that of pre-2014 surveys in 1995, 1998 and 2007. And, on the question of the dissolution of the Soviet Union, the size of the median response category was identical to 2014. Moreover, our findings are robust to the inclusion of nonvoters. In contrast to 2014, when voter turnout was depressed in the east, turnout fell largely uniformly across the county in 2019. 14. The results of the full ordinal logistic model are available in the Online Appendix.