Authoritarian or Simply Disillusioned? Explaining Democratic Skepticism in Central and Eastern Europe

Survey research has revealed that post-communist citizens are skeptical towards democracy. Despite a substantial body of literature that has researched the origins and determinants of these attitudes, consensus has not yet emerged. A major challenge has been to distinguish between individual support for democracy as an ideal political regime and satisfaction with the way democracy is practiced in one’s country. Using structural equation modeling with latent variables, we improve measurement validity and account for feedback effects to better understand the relationship between these attitudes. Consistent with our performance-based theory, we find that positive assessments of political performance drive normative support for democracy. The impact of satisfaction with democracy on democratic support suggests that we should not rush to view post-communist citizens' mindset as anomalous and inherently anti-democratic. Rather, post-communist skepticism of democracy might be generalized to contexts characterized by flawed implementation and unmet expectations of this form of government.


Introduction
Survey research consistently shows that post-communist citizens are highly skeptical of democracy. Central and Eastern Europeans support democracy to a lesser extent than their counterparts from Western Europe, North America, and, more surprisingly, other second and third-wave democracies in Asia, Africa, and Latin America (Pop-Eleches & Tucker, 2017). They seldom engage in politics (Kostelka, 2014) and exhibit alarming levels of distrust of major political institutions (Ceka, 2013;Klingemann et al., 2006). This lack of commitment to democracy in attitudes and behaviors contrasts with the euphoria many felt at the beginning of the post-communist transition (Gibson, 1996;Reisinger, 1994). A rich literature has focused on understanding why this shift occurred.
Scholarship on democratic attitudes falls into two broad streams. One emphasizes deep-rooted legacies, including political culture and communist socialization, which bias citizens against democratic values. The other highlights deficiencies in the performance of political and economic institutions, which turn subjects away from democratic principles and structures. A major challenge for both currents has been to distinguish between an individual's normative support for democracy as a political regime and her evaluations of the way democracy is practiced on the ground. Measuring these attitudes has been elusive, and the relationship between them is not always well specified (Claassen & Magalhaes, 2021).
Our work seeks to address this problem. We argue that satisfaction with the performance of democratic institutions drives support for democracy as a normative ideal. In other words, we attribute post-communist citizens' apparently low commitment to democracy to a sense of disillusionment with the practice of democracy after the collapse of communism. This dissatisfaction can be linked to the remarkable volatility and pervasive flaws in governance that the region experienced in the 1990s and the 2000s. We maintain that Eastern Europeans did not remain oblivious to these flaws. Rather, they allowed these deficiencies to shape their perceptions of the new form of government they faced. This general disenchantment with the reality experienced on the ground thus molded post-communist citizens' democratic attitudes.
We test this argument in two steps. Leveraging survey data from the 1990s and the early 2000s, we first propose a measurement strategy that distinguishes satisfaction from support. We draw on confirmatory factor analysis (CFA) to create latent constructs that capture the multidimensionality of democratic attitudes. In line with (Gunther & Montero, 2006), these constructs are clearly distinct and successfully predict other related political views and behaviors. Using structural equation modeling (SEM), we then show that satisfaction is a meaningful predictor of democratic support in post-communust Europe. How one evaluates her own political system shapes how strongly she normatively embraces democracy as a political regime. Higher principled support, by contrast, is associated with lower satisfaction. Intriguingly, we do not find evidence that life under communism drives democratic support.
Our work makes two important contributions. Theoretically, it builds on performance-based explanations of democratic support. Instead of fixating on Central and Eastern Europeans' supposedly authoritarian mindset and socialization into totalitarian values, we highlight how the imperfect political environment that citizens experienced following the collapse of communism molded their political attitudes. Conceptually, we address a frequent confusion in existing scholarship. The literature has struggled to converge on a common way to operationalize satisfaction and support. In fact, it often views them as different dimensions of the same concept. Drawing on complex measurement and structural equation models, we show that satisfaction and support are distinct but strongly related multifaceted attitudes. We shed light on the directionality of this relationship and discuss the implications of our findings for future research.
This paper is structured as follows. We begin by providing an overview of the literature on democratic support. We proceed to develop our argument linking normative support for democracy to democratic satisfaction. We outline our methodological approach and introduce our latent constructs. We then show that satisfaction and support are distinct concepts that can be reliably compared across countries with similar historical legacies. Once we have validated our latent measures, we test our theory in a structural equation modeling framework. We conclude by discussing avenues for future research.

Theories of Democratic Support: Intrinsic or Instrumental?
Research on Central and Eastern Europe has emphasized the deficit in democratic attitudes, values, and behavioral patterns that characterizes post-communist societies (Pop-Eleches & Tucker, 2013. Despite the initial euphoria surrounding the introduction of democracy after the fall of communism (Gibson, 1996;Reisinger, 1994), normative commitment to democracy has declined over the course of the transition. Existing work indicates that many in the region favor a return to communist rule (Ekman & Linde, 2011;Klingemann et al., 2006;Pop-Eleches & Tucker, 2017). The recent rise of populist, extremist, and protest parties with dubiously democratic credentials and agendas seems to attest to "the rejection of consensual politics" (Krastev, 2007;Vachudova, 2020) and to confirm this questionable democratic orientation. Furthermore, scholarship on the post-communist transition notes that Eastern Europeans lack some of the behavioral characteristics deemed crucial for the normal functioning of democracy, such as trust in political parties and participation in the public sphere (Catterberg & Moreno, 2006;Ceka, 2013;Howard, 2003;Klingemann et al., 2006;Pop-Eleches & Tucker, 2014;Rose, 2009;Sapsford & Abbott, 2006).
A rich research agenda seeks to understand democratic attitudes. Existing work falls into two broad families that highlight either intrinsic or instrumental motivations. Scholars within the first camp attribute individual political attitudes to political culture and societal modernization (Inglehart & Welzel, 2005;Verba & Almond, 1963). In this view, "the distinctive features of a [country]'s history give its citizens a unique pattern of political values" (Reisinger, 1994), leading entire populations to adopt common ways of thinking about political processes (Fitzpatrick et al., 1978). Economic development, in particular, is expected to promote a more pluralist and 1 3 liberal worldview (Inglehart & Welzel, 2010;Lipset, 1959). Although these theories have received empirical support, they rarely specify the duration of the hypothesized effect, account for the heterogeneity of individual attitudes, or explain why countries with similar historical trajectories do not necessarily form identical mindsets.
An alternative argument within this camp instead focuses on socialization. Political attitudes and beliefs are seen as a function of citizens' experiences, learned in early childhood or over their lifetime (Campbell, 1980;Greenstein, 1965;Jennings & Markus, 1984;Mishler & Rose, 1997;Pop-Eleches & Tucker, 2013). According to this perspective, life under communism-and especially under its more rigid Stalinist version-predisposed Eastern Europeans to be less supportive of democracy. Exposure to the teachings, norms, and realities of the totalitarian regime made people accustomed to and accepting of authoritarianism, paternalism, conformity, and dependence (Bialer, 1990). Democratic values, on the other hand, remained foreign. While credible, this argument seldom explains how exactly one's individual experiences affect democratic support. It also fails to effectively account for updating beliefs under democracy or for differences among people who share similar life paths (e.g., Fuchs-Schündeln & Schündeln, 2015;Mattes & Bratton, 2007).
The instrumental strand, on the other hand, links democratic attitudes to specific outputs that citizens associate with democracy. The regime is appreciated for its capacity to offer particular benefits. Approaches in this vein argue that democratic support depends on the performance of political, economic, and social institutions (Dahlberg & Linde, 2018;Dahlberg et al., 2015;Evans & Whitefield, 1995;Kitschelt, 1992;Mishler & Rose, 1997;Przeworski, 1991). People's ties to the system become stronger if political structures deliver positive outcomes that improve wellbeing (Magalhães, 2014). In contrast, if performance is mediocre, the status quo loses legitimacy. As Easton (1975) remarks with respect to specific support, "[w]hat politicians do and how they do it" (p. 437) has implications for their broader feelings toward democracy; "persistent inability to [...] produce satisfactory outputs [...] may well lead to demands for changing [...] the regime" (p. 397). 1 In the context of postcommunist Europe, democratic commitment is therefore viewed as contingent on how the transition unfolded in the 1990s and the 2000s.
One current within the instrumental tradition underlines economic performance, centering on the expectation that many Eastern Europeans held after 1989 that the free-market system would raise living standards (Kitschelt, 1992;Mishler & Rose, 1993, 1997. The recession that hit the region in the early 1990s is argued to have resulted in democratic disillusionment, as hopes for prosperity dwindled (Clarke et al., 1993;Przeworski, 1991). In line with this claim, Duch (1993), Rose & Mishler (1998), Claassen & Magalhães (2021) show that economic factors do indeed influence political attitudes. 2 Other studies disagree, concluding that Eastern Europeans ) 44:1959-1983 do not evaluate democracy simply based on its capacity to improve their economic status (Gibson, 1996;Kunioka & Woller, 2014).
An alternative view connects democratic support to political performance, maintaining that it is the effectiveness of the institutions and electoral processes that emerged after 1989 that determines how people feel about democracy. Evans & Whitefield (1995) find that individual evaluations of democratic systems in the 1990s strongly shaped support for the regime in East-Central Europe. Similarly, Ekman & Linde (2011) show that dissatisfaction with political outcomes drives communist nostalgia in the region. This approach also finds support outside postcommunist societies. For example, Bratton & Mattes (2001) posit that governments' ability to secure basic political rights molds democratic approval in the African context.
This project builds on the instrumental strand, highlighting the importance of perceptions about political performance. In line with Evans & Whitefield (1995), we argue that post-communist citizens' apparent skepticism toward democracy as a political regime is rooted in their general dissatisfaction with the way democracy works in their countries. Research on Central and Eastern Europe shows that the region struggled with the consolidation of democracy (Berman & Snegovaya, 2019). Transition societies confronted substantial electoral volatility (Haughton & Deegan-Krause, 2015;Tavits, 2008), ineffective political parties with vague programmatic appeals (Sikk, 2012), and frequent attacks on democratic institutions (Vachudova, 2005). To this day, many post-communist countries grapple with endemic corruption, low government effectiveness, weak law enforcement, and inefficient institutions (Kornai & Rose-Ackerman, 2004;Petrova, 2021). Our argument views this empirical reality as consequential for political attitudes once it becomes obvious to individuals.
We further argue that personal perceptions and experiences matter for political attitudes. Disenchantment with political structures induces people to withdraw support from these institutions. Evans & Whitefield (1995) maintain that negative experiences with democratic institutions discredit the status quo. Seligson (2002) posits that exposure to corruption erodes belief in the political system. Looking into social policies, Skocpol (1992) shows that the patronage and the inefficiency of the nineteenth-century U.S. administration delegitimized national authorities and prompted Americans to turn to private markets for social services. In a similar vein, Rothstein et al. (2012) contend that bureaucratic malfeasance and incompetence decrease mobilization in support of economic redistribution and lead to a rejection of a more extensive role for the state in socio-economic matters. In all of these cases, irregularities and malpractices paint institutions as dysfunctional, undermine citizens' opinion of the state, and produce political disengagement (Norris, 2011).
We expect democratic support to be similarly influenced by individual experiences and considerations. In particular, we argue that perceptions about the quality of democratic institutions shape commitment to democracy. When citizens believe that their interests are represented, that elections are transparent, and that politicians are held accountable for their actions, they are also likely to feel satisfied with the political status quo. This satisfaction fuels enthusiasm for the political regime. Positive sentiments and evaluations of how democracy functions on the ground thus 1 3 reinforce attachment to broader democratic principles. By contrast, if people think the political system ignores their demands, fails to protect their rights, does not offer effective representation, and does not promote accountability, they are disappointed with the practice of democracy in their countries. The democratic institutions with which they interact fall short of accomplishing their purposes and do not meet citizen expectations. The resulting clash between theory and practice, between expectations and reality, is likely to raise questions about the broader regime, shaking commitment to democracy as a form of government.
This general argument is consistent with patterns of public opinion in post-communist societies. Central and Eastern Europeans generally report low trust in state institutions, high skepticism towards public officials, and low levels of political efficacy. These sentiments strongly contrast with the euphoria immediately after the collapse of communism, when citizens might have had unusually high expectations about their countries' future (e.g., Rose, 2009, Ch. 17). Having spent decades under authoritarianism, East Europeans plausibly expected greater political accountability, strengthened representation, and greater equality before the law under democracy. The political struggles, instability, and scandals that characterized the 1990s instead fueled disillusionment with political institutions (Ceka, 2013), exacerbating dissatisfaction with democracy.
We argue that this dissatisfaction has weakened support for democracy as a political regime across the region. The primary goal of our paper is thus to build on the literature linking individual assessments of the democratic system on the ground to general commitment to democracy as a form of government. Although recent research has alluded to the presence of such a link (e.g., Claassen & Magalhães, 2021), it has rarely examined these attitudes jointly, allowing them to influence the other.
Furthermore, we hold that one's satisfaction with the way democracy works is not necessarily based on the consumption of objective information about political or economic outputs. Instead, satisfaction may reflect complex thought processes and interpretations that draw on highly subjective, personal experiences. Contrary to some previous work that has used country-level contextual variables such as corruption, bureaucratic performance, economic conditions, the extent of civil and political rights, and the quality of governance (e.g., Anderson & Tverdova, 2003;Boräng et al., 2017;Claassen, 2020;Magalhães, 2014), we maintain that aggregate economic and political indicators might miss the true dynamics that underlie postcommunist citizens' deeply rooted skepticism of democracy. In fact, existing scholarship has found that individual political and economic perceptions do not correlate with aggregate-level data in Eastern European countries (e.g., Ceka, 2013;Evans & Whitefield, 1995). Therefore, we use individuals' own evaluations of the political institutions they interact with on a daily basis. 3

Measuring Democratic Support and Satisfaction
We employ a two-stage empirical strategy to test our argument about the impact of satisfaction on support. Our first step is to carefully measure these two attitudes. Public opinion research does not fully agree on how to conceptualize normative support for democracy as an ideal regime and satisfaction with the way democracy is practiced (Gunther & Montero, 2006;Torcal & Montero, 2006). Scholars have different understandings of the specific viewpoints that pro-democratic citizens hold and the concrete outputs that individuals use to evaluate democratic performance. Some studies look into liberal values or psychological traits such as moderation, tolerance of diversity, trust in fellow citizens and political representatives, or belief in the legitimacy of democratic regimes (Dahl, 1989;Doherty & Schraeder, 2018;Verba & Almond, 1963;Waldron-Moore, 1999). Others highlight the embrace of specific institutions such as free and fair elections, political party competition, an impartial judiciary or a representative parliament (Evans & Whitefield, 1995;Ferrín & Kriesi, 2016;Inglehart & Welzel, 2003;Kunioka & Woller, 2014;Seligson, 2002). When it comes to satisfaction, existing work asks what people think about specific procedural and liberal attributes (Dahlberg & Linde, 2018;Linde, 2012;Quaranta, 2018) or whether they are content with the way democracy works in their country.
These diverse ways of thinking about satisfaction and support are compounded by an even greater measurement challenge. While a number of influential works have shown that satisfaction and support are distinct concepts (Canache et al., 2001;Evans & Whitefield, 1995;Gunther & Montero, 2006;Linde & Ekman, 2003), some continue to conflate them or see them as dimensions of the same overarching concept. 4 For example, many studies that in theory analyze support for democracy look into citizens' beliefs about the ability of democratic systems to effectively solve problems, advance order and stability, attain economic goals, or prevent indecisiveness or "quibbling" (Doherty & Schraeder, 2018;Hoffman & Jamal, 2014;Ketchley & El-Rayyes, 2021;Pop-Eleches & Tucker, 2017). We strive to avoid this conflation and, in the vein of Gunther & Montero (2006) and Claassen & Magalhaes (2021), measure satisfaction and support as two distinct attitudes.
To capture the complexity of these attitudes, we use confirmatory factor analysis (CFA) with latent variables. This approach highlights the multidimensionality of democratic views, addresses measurement error, and allows us to combine survey items measured on different scales. We begin by proposing measurement models for support and satisfaction with democracy. We proceed to evaluate the reliability of the two constructs by running separate measurement models for all ten Central and Eastern European democracies in our sample.
We create two unobserved latent variables-support for and satisfaction with democracy-by combining multiple survey questions that reflect different aspects of the two attitudes. CFA tests whether the hypothesized latent construct adequately explains the covariances between its constituent indicators (Bollen, 1989). The measurement models consist of a system of linear equations in which each observed factor (survey question) is assumed to be a function of the latent construct and measurement error. Both latent variables are measured through multiple survey items ( i = 3 or 4). Thus, x i = i + i where x i is an observed factor, is the underlying latent variable, and i is the measurement error term for x i (uncorrelated with and for all i). i is the coefficient for the expected unit change in the observed indicator for a unit change in the latent variable. To scale the latter, we set 1 to 1, assuming that the unobserved construct and the observed indicator share the same scale. To account for the ordinal nature of our data, we use robust weighted least squares, which analyze polychoric correlations rather than covariances.
We draw on the second wave of the Post-Communist Publics Study (PCP). 5 PCP includes numerous questions on political culture, attitudes toward democracy, and individual political and economic experiences. Focusing on the period between 1997 and 2001, when the survey was fielded, ensures that respondents have had time to adjust to and gain experience with the changing political landscape after the fall of communism in 1989. We include ten electoral democracies-Bulgaria, the Czech Republic, Estonia, Latvia, Lithuania, Poland, Romania, Russia, Slovakia, and Slovenia-yielding a total sample of over 11,000 individual observations. 6 The varying speed and depth of the reforms pursued in these countries in the 1990s enables us to explore whether similar attitudes underlie support and satisfaction in different institutional contexts. Survey questions are ordinal, categorical and coded so that higher values equal greater support or satisfaction. Support for democracy captures citizens' normative commitment to democracy. Like Gunther & Montero (2006), we conceptualize it in terms of political legitimacy, or "beliefs that democratic politics and representative democratic institutions are the...only acceptable framework for government (p. 71)." PCP includes indicators that reflect general views about democracy as a political regime as well as opinions about specific institutions traditionally associated with democratic systems. The first survey item, used to scale the latent variable, asks respondents whether they believe that democracy is the best form of government. The second inquires whether elections are the best way to choose a government. The remaining two questions elicit thoughts on whether a parliament and political parties are needed. Figure 1 presents the measurement models. As a latent variable, support is enclosed in a circle. The four observed factor indicators, in rectangular boxes, are attached to separate error terms. Whereas the first indicator is measured on a three-point scale, the other three are dichotomous, giving survey participants the option to agree or disagree with each statement. Higher values indicate higher support for democracy.
Satisfaction with democracy reflects citizens' evaluations of how well democracy performs in their country. Because we recognize that people's perceptions about a form of government and a particular cabinet might be conflated, we factor in views about the broader political context as well as about the current government. We use three PCP questions that ask how satisfied or content respondents are with the quality of democracy in their country and how satisfied they are with the present government. While the first and the third items are measured on a ten-point scale (completely dissatisfied to completely satisfied), the second question only has three categories. Higher values indicate greater satisfaction. Finally, the curved arrow indicates that the two latent variables are correlated. The full survey questions are included in Appendix A.
As a robustness check, we re-run our analysis using a different survey, the 2004 wave of the New Europe Barometer (NEB). The NEB was conducted several years after the PCP, which allows us to explore the attitudes of citizens who have had additional time to adapt to political changes and gain more experience with democracy. The different questions that the NEB asks help us to illustrate how latent constructs can be reliably created from different combinations of available indicators. The NEB results are presented in Online Appendix B.

Measurement Model Results
Table 1 presents the overall fit statistics for the measurement model. We allow the MPlus software to multiply impute missing values to avoid bias. 7 The results suggest that the hypothesized latent variables measure the underlying concepts consistently well across countries. The Root Mean Squared Error of Approximation (RMSEA) never exceeds the accepted threshold of 0.10 (Browne & Cudeck, 1993). The Comparative Fit Index (CFI) and the Tucker Lewis Index (TLI) are above the cut-off value of 0.95 and approach the ideal value of 1 (Hu & Bentler, 1999) for all   (Cheung & Rensvold, 2002). Taken together, the fit statistics indicate good overall fit across post-communist countries. All but two coefficients and R 2 values come out as positively signed and statistically significant at the 0.001 level. These estimates seem relatively consistent across countries, with the factor loadings of Elections, Parties, Parliament, Content, and Satisfied w Gov being roughly comparable in size. As evidence of convergent validity, respondents who agree that democracy is the best form of government tend to see elections, political parties, and national assemblies as necessary for the political process. Similarly, citizens who report higher satisfaction with the way democracy works in their country feel happier with the performance of the current government. As suggested by the R 2 values, variation in the two latent variables explains a large fraction of the variation in the seven indicators. These values do not vary substantially across survey items. This finding implies that, apart from respondents' general feelings toward democracy as an ideal regime and satisfaction with the way democracy performs in their countries, attachment to specific democratic institutions and assessments of particular cabinets are similarly meaningful measures of support and satisfaction.
It is noteworthy that the absolute and relative magnitudes of the factor loadings for both latent variables are similar across all countries in our analysis. Together, these results suggest that our measures are reasonably invariant in our sample of post-communist societies. 9 In other words, individual indicators are structured similarly in relation to the latent constructs and capture the same underlying concept. This helps to at least partially address the concern that citizens of different states might think of democracy in different ways. Further, Ferrín & Kriesi (2016) have shown that Western and Eastern Europeans largely agree on a common normative model of what liberal democracy entails, universally endorsing the same cardinal principles. Our work indicates that, when we zoom in on Central and Eastern Europe, several views on institutions traditionally associated with democratic political systems similarly reflect fundamental attitudes which, given the instruments we employ, we call democratic support and democratic satisfaction. Post-communist citizens' appreciation of elections, parliaments, and parties, for example, is linked to their broader commitment to democracy. The largely consistent results from our country models imply that respondents in the region share a similar conceptualization of the principal features of democratic systems.
While the analysis so far indicates that our measurement models perform well, a lingering question concerns how distinct satisfaction and support actually are. ) 44:1959-1983  According to Table 1, correlations between the two constructs range from 0.16 and 0.38. To further explore this issue, we run exploratory factor analysis comparing two models: one that forces all seven observed factors to load on a single latent variable and another which allows them to load on two different latent constructs. Table 2 shows that the results are consistent with the two-factor model: the 2 , RMSEA, and BIC values associated with the latter are lower while the CFI and TLI indices are closer to 1. All indicators return greater loadings on the latent variable they are hypothesized to capture. This inspires confidence in our claim that satisfaction and support are theoretically and empirically different attitudes. How well do these attitudes relate to other variables that existing research sees as connected to them? In a third step, we test the predictive validity of our latent constructs by examining their association with voter turnout and satisfaction with the market economy. Citizens who support democracy as a political regime are more likely to participate in the political process and see voting as a civic duty (Blais & Galais, 2016). Similarly, given the contemporaneous nature of the political and economic transitions unfolding in the region, Central and Eastern Europeans who are more satisfied with democracy are more likely to report satisfaction with the market economy (Pop-Eleches & Tucker, 2017). To test for convergent validity, we regress turnout and satisfaction with the market economy on our latent measures.
Tables 3 and 4 below show that support and satisfaction are meaningful predictors of these attitudes across the ten post-communist democracies in our sample. The coefficients returned by the latent variables are positively signed and statistically significant in all but a single model. 10 This implies that satisfaction and support successfully predict other political views and patterns of behavior that previous studies have linked to these attitudes. The general fit statistics for these models are excellent, with insignificant 2 p-values for a majority of countries, low RMSEA values, CFI and TFI statistics close or at the ideal value of 1, and low or negative BIC values. 11

Evaluating the Relationship Between Support and Satisfaction with SEM
Our results so far provide evidence that our proposed measures of satisfaction and support capture democratic attitudes fairly consistently across the post-communist countries in our sample. How can we evaluate our performance-based argument that satisfaction affects support? ) 44:1959-1983 Table 3 Satisfaction with democracy as a determinant of satisfaction with the market economy (PCP survey) Dependent variable is satisfaction with the free market in one's country. ***p≤  Political Behavior ( ) 44:1959Behavior ( -1983 To answer this question, we run structural equation models. Structural equation modeling (SEM) relaxes many of the statistical assumptions inherent in linear regression. Namely, it allows bidirectional relationships, differentiates between direct and indirect effects, and accounts for sequential effects within a causal chain. We run a non-recursive model in which satisfaction is a determinant of support and vice versa. This enables us to check whether a feedback loop exists between the two attitudes. While our argument is that satisfaction drives support, it is possible that lack of normative commitment to democracy instead influences citizens' evaluations of the performance of democracy on the ground. Testing for this possibility requires a modeling approach, such as SEM, that permits such complexity. The models in Table 5 include the same measurement models presented earlier, adding a regression component to examine the drivers of the two latent constructs and the connection between them. Model 1 regresses support on satisfaction to test our theory. In a second step, model 2 adds support among the determinants of satisfaction.
Consistent with the literature, we control for a number of individual-level covariates that affect both attitudes. We see support as a function of communist socialization (the number of years a respondent has lived under communism), minority status, religiosity, vote choice (whether the respondent supported the winner of the most recent parliamentary elections), age, gender, education, monthly income, newspaper readership, and community size. Satisfaction, on the other hand, is a function of age, gender, income, education, vote choice, newspaper readership (to gauge political interest), community size, and retrospective and prospective personal and sociotropic economic evaluations. 12 Existing work has shown that electoral defeats and exposure to non-democratic political systems weaken support for democracy (Pop-Eleches & Tucker, 2017). Urban, wealthier, more educated, and more politically informed citizens tend to be more committed to and satisfied with democracy (Anderson & Tverdova, 2003;Magalhães, 2017;Magalhaes & Mattes, 2018;Peffley & Rohrschneider, 2014;Waldron-Moore, 1999;Whitefield & Loveless, 2013). Similarly, more positive economic assessments translate into greater satisfaction with regime performance (Dahlberg & Linde, 2018;Magalhães, 2014). Taking cues from Claassen & Magalhães (2021) and Gunther & Montero (2006), who show that economic performance does not drive democratic support, we exclude those covariates among the determinants of support. Stronger religious attachments, on the other hand, are often associated with conservative-traditional values (Saroglou et al., 2004) or better civic skills and higher political efficacy (Putnam, 2000), which could either dampen or enhance commitment to democracy.
Although many of the individual-level variables that we work with predict both attitudes, minority status, religiosity, and communist exposure uniquely identify support, while the four economic assessments only feed into satisfaction. This allows us to identify the non-recursive model, where support exerts a contemporaneous effect on satisfaction and satisfaction exerts a contemporaneous effect on support. Unfortunately, the Post-Communist Publics survey does not allow us to exploit any temporal variation in attitudes because the two waves of the study rely on different questionnaires. Other multi-wave public opinion surveys do not include enough indicators to enable us to construct our latent variables of interest. We are therefore unable to use time-series data. Our model specification, however, yields an empirical strategy that resembles Claassen & Magalhaes (2021), who use identical equations to model these two attitudes and draw on lagged dependent variables to uniquely identify their non-recursive model. 13 In line with our expectation, Table 5 shows that satisfaction is a positively signed and statistically significant predictor of democratic support. Satisfaction with the way democracy works in practice is associated with greater commitment to democratic norms. This result remains robust when we include additional controls and when we account for the possibility that satisfaction itself can be shaped by support. In fact, Model 2 confirms the existence of a negative feedback loop between the two attitudes: how committed one is to democracy affects one's evaluations of the political regime. Greater support is associated with lower satisfaction. Although it might seem counter-intuitive, this finding is consistent with a scenario in which *p < 0.05, **p < 0.01, ***p < 0.001 R 2 (Support) 0.24*** 0.24*** R 2 (Satisfaction) 0.47*** 0.28*** democratically-minded citizens grow disappointed with the reality on the ground and evaluate democratic practice as failing to meet their high criteria. Our findings thus suggest that respondents' everyday political experience shapes their political beliefs. The latter, in turn, drive assessments of their country's political system. Disillusionment with democratic practice weakens support, making people less invested in non-authoritarian regimes. Simultaneously, high democratic commitment makes individuals more likely to see the political reality they interact with as lacking.
Several other findings stand out. The effect of years lived under communism is miniscule and not robust: the coefficient estimate in the non-recursive model is − 0.002, but in Model 1 the value is 0.001 and fails to reach statistical significance. We therefore cannot conclude that greater exposure to communism undermines democratic support. Consistent with existing work, urban dwellers, highly educated respondents, and citizens who voted for governing parties in the previous elections exhibit higher support for democracy. In contrast, richer, more informed, and more religious individuals are less committed to democratic governance. This could be because the economic elites that emerged during the post-communist transition had strong ties to the totalitarian regime and because exposure to news reveals negative details about the political reality on the ground (e.g., Ceka, 2013).
With respect to satisfaction, wealthier and incumbent-voting citizens tend to be more satisfied with how democracy works. Positive assessments of one's own and one's country's current and future economic situation also translate into higher satisfaction. This is in line with existing work, which has documented a close association between economic and political evaluations (Ekman & Linde, 2011;Gunther & Montero, 2006). As before, more informed respondents, who presumably have access to more detailed information about the functioning of the regime, report lower satisfaction. City dwellers also appear unhappy with the performance of their country's democracy. This could be because of the opportunities they have to observe the political reality. Our results are thus largely consistent with research on democratic attitudes in post-communist Europe.
In Appendix B, we replicate our analysis with data from the 2004/2005 wave of the New Europe Barometer (NEB). Using a different survey does not change our conclusions. Relying on different, but similar questions for the construction of our latent variables, we establish that satisfaction and support are conceptually distinct. The measurement models that we propose capture these concepts well across all the post-communist democracies included in our sample. In a second step, we corroborate our intuition that evaluations of the performance of democratic institutions drive commitment to democracy as a political regime. Indeed, the results from our structural equation models are consistent with the ones presented in the main analysis. As with the PCP survey, the relationship between satisfaction and support in the NEB runs in both directions. How satisfied one is with the reality on the ground affects how committed one is to the political regime at home at the same time as how one feels about democracy shapes how content one is with the performance of democracy in one's country. While moderate to high satisfaction has the potential to promote commitment to democratic norms, persistent disappointment might undermine support. In contrast, more democratically minded citizens often report lower contentment with how democracy works in practice. This finding suggests that having high expectations might have led to disenchantment once Central and Eastern European states initiated political and economic reforms. Future research should examine the evolving link between satisfaction and support over time. Broadly speaking, however, our argument receives empirical support from two different data sources at two different points in time during the post-communist transition.

Conclusion
This article has attempted to further our understanding of why post-communist citizens display highly skeptical attitudes toward democracy. Our argument builds on the literature that links individual-level perceptions of political performance on the ground to general commitment to democracy as a political regime. Specifically, we posit that dissatisfaction with the way democracy has been practiced after the fall of communism drives support for this form of government.
Our empirical approach allows us to distinguish satisfaction from support, to better capture the abstract nature of these complex political attitudes, and to carefully explore the relationship between them across Eastern European societies. Our results confirm that satisfaction and support are distinct, albeit meaningfully related concepts. Furthermore, the structural equation models that we run indicate that satisfaction has a large, direct and positive effect on support. Citizens who feel disillusioned with the way democracy works in their country withdraw support from this form of government. In contrast, individuals who are happy with the functioning of democracy feel more attached to this political regime. These results are robust to different modeling techniques, model specifications, and datasets.
Our findings suggest that we should not rush to view post-communist citizens' mindset as anomalous and inherently anti-democratic. Rather, post-communist skepticism of democracy might be generalized to contexts characterized by flawed implementation of and unmet expectations from this form of government. How citizens feel about democracy has important implications for their political behavior and democratic consolidation. Consequently, understanding satisfaction with and support for democracy sheds light on voter turnout, political mobilization and engagement, the decline of mainstream parties, the rise of populism, and the emergence of new political actors. If dissatisfaction with democratic performance on the ground drives disillusionment with democracy as an ideal regime, the democratic backsliding currently underway in several Eastern European countries might not spur popular reactions against the gradual erosion of democratic principles. Understanding citizens' commitment to and satisfaction with democracy can thus become crucial for the durability of democratic regimes.