• 1 Introduction

  The existence of a ‘democratic deficit’ as a global phenomenon has been widely documented in a number of recent studies (Foa et al., 2020; Norris, 2011). Citizens tend to be more dissatisfied with democracy and have lower levels of support for and trust in the government, political elites and representative institutions (Bedock, 2017; Dalton, 2004; Norris, 2011). And yet, a high level of political trust among the public is often considered to be crucial for the long-term stability of democratic systems (Marien & Hooghe, 2011). This is even more true for trust in the parliament, an institution through which citizens can exercise their power over public decisions and which, therefore, plays a crucial role in the process of political representation (Holmberg, Lindberg & Svensson, 2017).
  This article aims to determine whether citizens’ level of trust in the parliament depends on the quality of the representative linkage. Much of the literature has studied political representation as a dependent variable and overlooked the extent to which different levels of representation produce different outcomes, especially in terms of citizens’ political trust. The focus of this research is all the more crucial because of the inequalities in representation highlighted by a growing body of literature. Political representation is not uniform and homogenous, and some citizens, especially the most affluent and educated ones, tend to enjoy better representation (e.g. Bartels, 2008; Gilens, 2005).
  Do citizens respond to different levels of political representation? In this article, building on Belgian survey data collected by scholars from Universiteit Antwerpen (UA) and Université libre de Bruxelles (ULB), we measure the quality of representation as policy opinion congruence, that is, the degree of ideological proximity between citizens and their political representatives (Golder & Stramski, 2010) on a set of specific policy statements. Opinion congruence is a structuring element of political representation, as it helps to translate citizens’ policy preferences into political decisions (e.g. Miller & Stokes, 1963), especially when legislators take their own preferences into account when undertaking legislative activities (Levitt, 1996). In other words, the congruence between the policy preferences of citizens and those of their representatives is considered to be a relevant indicator in measuring the extent to which citizens’ opinions are ‘made present’ in the legislature (Pitkin, 1967).
  The analysis presented in this article is innovative for two main reasons. First, it is worth noting that the vast majority of studies analysing the attitudinal effects of congruence focus on satisfaction with democracy (e.g. Dahlberg & Holmberg, 2014; Ezrow & Xezonakis, 2011; Mayne & Hakhverdian, 2017). These studies found that citizens evaluate democracy more positively when the political system is more congruent with or responsive to their views. In contrast, we focus here on another measure of political support: trust in the parliament. Trust is a subjective evaluation of a relationship (Van der Meer, 2010) and is, in that regard, directly related to the representative relationship between citizens and legislators. Second, whereas most studies measure opinion congruence on the left–right scale (e.g. Ezrow & Xezonakis, 2011; Kirkland & Banda, 2019; Mayne & Hakhverdian, 2017), we present an analysis based on opinion congruence measured on specific policy statements. Our measure of policy opinion congruence takes the multidimensionality of the political space into better account (e.g. Kriesi et al., 2006; Lesschaeve, 2017; Lutz, Kissau & Rosset, 2012).
  We find clear evidence that policy opinion congruence is positively associated with citizens’ political attitudes. More specifically, citizens whose policy preferences are closer to those of their party’s legislators tend to have higher levels of trust in the parliament than citizens whose preferences are completely at odds with those of their representatives. We also find that the effect of opinion congruence on trust is not moderated by education but rather by political interest. Unexpectedly, citizens who consider themselves to be very politically interested trust the parliament independently of their level of policy opinion congruence.

• 2 Theoretical Background

  Decades of studies have shown that advanced industrial democracies are facing an erosion of political support (Dalton, 2004). Citizens tend to be more distant from political parties, more critical toward institutions, and less positive regarding governments (Dalton, 2004; Norris, 2011).
  Academic literature generally assumes that political support is a necessary element for the legitimacy and stability of political systems. This idea goes back to the 1970s, when it was thought that democratic regimes might not survive high levels of political distrust (Crozier, Huntington & Watanuki, 1975). However, more recent studies have shifted the focus by considering that political distrust or dissatisfaction is a reflection and not a cause of democratic malaise (Pharr, Putnam & Dalton, 2000; Van der Meer, 2010; Van der Meer & Zmerli, 2017).
  According to these works, low levels of political support are related to the actual functioning of democracy. Political support is thus considered to be endogenous to the political context and is dependent on the way citizens evaluate the functioning of the political system (Martini & Quaranta, 2019). Although citizens continue to be attached to democratic values and principles (Dalton, 2004), and to the mere existence of representative institutions, they have been increasingly dissatisfied with the political performance of these institutions (Pharr et al., 2000). While a large body of research has investigated the connection between political support and the outputs of political systems (e.g. Dahlberg & Holmberg, 2014; Hobolt, 2012; Wagner, Schneider & Halla, 2009), this article focuses its attention on the way the quality of representative linkage affects political support.
  In representative democracies, the quality of representation can be operationalised as opinion congruence (Martini & Quaranta, 2019, p. 7), that is, the ideological proximity between citizens and representatives (Golder & Stramski, 2010). Theories of democratic representation presume that there should be some level of congruence between the policy preferences of the represented and those of the representatives (Miller & Stokes, 1963). Previous research studying the connection between opinion congruence and political attitudes has found that citizens who are more congruent with political elites tend to have higher levels of political support (Ezrow & Xezonakis, 2011; Ferland, 2021; Kim, 2009; Mayne & Hakhverdian, 2017; Stecker & Tausendpfund, 2016).

  2.1 The Effects of Policy Opinion Congruence on Political Trust

  The extant literature on the attitudinal impacts of opinion congruence has been “limited almost exclusively to satisfaction with democracy” (Martini & Quaranta, 2019, p. 61). Despite being widely used in comparative research (Kriesi, 2013), this indicator has been criticised for being interpreted differently by respondents and for capturing both specific and diffuse support (Kriesi, 2013; Linde & Ekman, 2003). A conceptually more specific indicator of political support is political trust. Indeed, political trust systematically refers to a very specific set of political actors whereas democratic satisfaction is rather broad and is likely to refer to different dimensions of democracy (Kriesi, 2013).
  In this article we suggest that, conceptually, political trust is a better indicator to relate to opinion congruence. Indeed, trust always involves an interpersonal relationship between a truster and a trustee (Hardin, 1999). Whereas it is possible for citizens to have various institutions or democratic processes in mind when sharing their level of satisfaction with democracy, the indicator of political trust asks them to evaluate a very precise relationship with specific political objects (Van der Meer, 2010). Trust is, in that regard, the product of citizens’ subjective assessment of the nature of their relationship with various institutions or political actors, and of how these actors behave (Norris, 2011). Conceptualising trust as a relational concept supports the claim that citizens’ trust in representative institutions is related to the nature of the relationship between them and their political representatives.
  Moreover, previous research has found that political trust is a more volatile indicator than satisfaction with democracy (Martini & Quaranta, 2019; Norris, 2011), suggesting that it is more susceptible to be affected by external factors and variations, such as congruence.
  While political trust can be conceptualised in multiple ways (i.e. trust in the government, parties, parliament, politicians), our focus here is on trust in the parliament (Grönlund & Setälä, 2007; Holmberg et al., 2017; Van der Meer, 2010). Indeed, parliaments are a critical body in the functioning of representative democracies and are central institutions in charge of policymaking. The indicator of trust in the parliament is expected to be associated with the way citizens’ policy preferences are represented in the parliament, an institution whose primary function is to link citizens’ preferences to policy decisions.

  2.2 Conceptualising Policy Opinion Congruence

  This article studies the relationship between opinion congruence and trust in the parliament. Opinion congruence between citizens and political representatives can be operationalised and measured in very different ways (see Golder & Stramski, 2010). Political representatives’ preferences can be operationalised as the preferences of the government (Curini & Jou, 2016; Mayne & Hakhverdian, 2017), those of the median party in the parliament (Kim, 2009) or those of individual legislators (Miller & Stokes, 1963). In this article we adopt a meso-level approach: we measure the proximity between citizens’ opinions and those of legislators belonging to the political party they voted for (see Dahlberg & Holmberg, 2014). This approach better accounts for the crucial role of political parties (Dalton, Farrell & McAllister, 2011), especially in Belgium, a country often described as a partitocracy (Van Haute, Amjahad, Borriello, Close & Sandri, 2013).
  On the citizen side, opinion congruence can be operationalised as the proximity between the position of the median voter and the mean position of elected representatives (e.g. Dassonneville & McAllister, 2020; Martini & Quaranta, 2019, p. 107). Although this median citizen and many-to-one measure has been widely adopted in the literature on political representation (e.g. Dalton et al., 2011), it does not consider the diversity of policy preferences within a constituency. In contrast to this, congruence can be operationalised at the individual level as the ideological distance between every citizen and their political representatives (one-to-one measure). This allows the study of different individual levels of political representation and, therefore, the assessment of their attitudinal impacts.
  Most of the studies on the relationship between political support and congruence rely on a measure of congruence constrained to the general left–right axis (Ezrow & Xezonakis, 2011; Kim, 2010; Mayne & Hakhverdian, 2017). Although the left–right dimension remains dominant and structuring in established democracies, extensive research has shown that it does not encapsulate all the dimensions of the political space (Stecker & Tausendpfund, 2016). Whereas the left–right scale captures the positions of citizens on socio-economic issues rather well, it overlooks issues such as European integration, the environment or migration. As it reduces the political space to a single dimension, a congruence measure focused solely on the left–right axis may be biased. One might miss what Thomassen (2012) named the ‘blind corner of political representation’, consisting of voters with inconsistent preferences (Otjes, 2016), or cross-pressured voters holding conservative positions on cultural issues and liberal ones on socio-economic questions (Lefkofridi et al., 2012; Van der Brug & Van Spanje, 2009).
  Therefore, it is crucial to move beyond the unidimensional focus of the left–right axis and use a measure of policy opinion congruence based on different policy issues (Lesschaeve, 2017). Previous studies on the relationship between democratic satisfaction and congruence have noted that including specific policy issues is highly relevant (e.g. Hall & Evans, 2019; van Egmond, Johns & Brandenburg, 2020). For instance, measuring congruence on five issue dimensions, Stecker and Tausendpfund (2016) found that citizens who are more distant from their government are more dissatisfied with democracy. To the best of our knowledge, there has been no study on how different levels of congruence on specific policy issues affect citizens’ levels of political trust.

  2.3 Explaining the Effect of Policy Opinion Congruence on Trust in the Parliament

  We expect citizens’ trust in the parliament to be affected by their levels of policy opinion congruence with the political party they voted for in the last elections. The policy proximity between voters and political parties is a crucial element of the responsible party model of representation (see Dalton et al., 2011), which holds that parties play a crucial role in articulating and translating citizens’ policy preferences in the policymaking process. We expect citizens to trust the parliament more if their opinions are represented in the legislature (Dunn, 2015; Grönlund & Setälä, 2007; Martini & Quaranta, 2019). More precisely, individuals are expected to have an “intrinsic satisfaction” derived from the feeling that their views are shared by their party’s representatives (van Egmond et al., 2020, p. 2), regardless of whether that party is in charge. While the extant literature reported a significant positive effect of opinion congruence on satisfaction with democracy, we expect congruence to also affect trust in a parliament, which is a more precise indicator capturing citizens’ evaluation of the performance of their political representatives.
  The relation between policy opinion congruence and political trust can be explained by the role played by citizens’ perceptions of their own level of policy proximity. However, doubts can be raised as to the extent to which citizens accurately know their representatives’ positions on specific policy issues. Previous research has indeed noted that the quality of the political information citizens have at their disposal is poor (Lau & Redlawsk, 1997; Miller & Stokes, 1963; Talukder, Uyttendaele, Jennart & Rihoux, 2021) and that the cost for gathering extensive political information is quite high given the fact that “neither the media nor political elites produce or report on measures of ideological congruence” (Mayne & Hakhverdian, 2017, p. 827). Although we do not expect citizens to know their representatives’ position on every policy, we expect them to use cognitive heuristics or available information on a few policies to estimate their overall level of proximity with representatives. On the one hand, cognitive heuristics, and more specifically partisan and ideological cues, allow citizens to infer parties’ positions on the basis of a party label or of a left–right ideological placement (Foos & De Rooij, 2017; Lau & Redlawsk, 2001), and on the other hand, citizens are likely to use the information on salient issues received through social media, the media or discussions with peers to estimate their overall level of proximity with a party. In sum, we expect citizens to use cognitive heuristics and limited information about parties’ positions in order to estimate their overall level of policy opinion congruence, which, in turn, will affect their level of political trust.

  Hypothesis 1 – The policy opinion congruence between citizens and their party’s legislators is positively associated with trust in the parliament.

  Furthermore, we expect the relationship between policy opinion congruence and political trust to be contingent upon some characteristics that are specific to the individuals. Previous research has emphasised the impact of education levels in moderating the effect of the quality of democracy on political trust (Kolczynska & Bürkner, 2021) or satisfaction with democracy (Mayne & Hakhverdian, 2017). For instance, Mayne and Hakhverdian (2017) found that the effect of congruence on democratic satisfaction is larger for citizens with higher levels of education.
  The conditioning effect of education can be theoretically explained by two elements. First, it can be related to the cognitive mobilisation theory (Dalton, 2007). Indeed, a higher level of education is likely to affect the way citizens receive and process information about their representatives’ policy preferences. If citizens do not receive accurate information, or do not process it well, we cannot expect policy opinion congruence to influence political trust. According to this explanation, education impacts the cognitive capacities of individuals to identify levels of policy opinion congruence and, by extension, the relationship between congruence and trust in the parliament. Second, the conditioning effects of education can also be related to the level of support of democratic values attached to the different levels of education (Mayne & Hakhverdian, 2017). Indeed, as citizens with higher levels of educational attainment tend to support democratic principles more (see Hibbing & Theiss-Morse, 2002; Verba, Schlozman & Brady, 1995), they should pay greater attention to their level of policy opinion congruence.

  Hypothesis 2 – The effect of policy opinion congruence on trust in the parliament is larger for citizens with higher levels of education.

  Education can be used as a proxy for political sophistication (e.g. Mayne & Hakhverdian, 2017). Political sophistication is a construct capturing the extent of citizens’ knowledge about politics and is composed of elements such as interest, motivation, awareness or expertise (Carpini & Keeter, 1996; Gallina, Baudewyns & Lefevere, 2020; Luskin, 1990). Although education and political sophistication tend to be associated (Mayne & Hakhverdian, 2017), having a high level of education does not necessarily mean that you are systematically interested, knowledgeable or aware of political issues. Put differently, although education may impact citizens’ cognitive skills and, therefore, their ability to understand politics, it does not mean that all educated citizens are willing to spend time gathering information about politics. Thus, we contend that education and political sophistication must be analysed as two distinct variables.
  In this article we rely on political interest, which is an important component of political sophistication (Gallina et al., 2020). Politically interested citizens are likely to spend more time following politics and, as a consequence, might be better able to perceive their level of policy opinion congruence. Therefore, we expect the level of political interest to moderate the relationship between congruence and trust in the parliament. The more citizens are politically interested, the more their level of trust will depend on their level of policy opinion congruence (Stecker & Tausendpfund, 2016).

  Hypothesis 3 – The positive effect of policy opinion congruence on trust in the parliament is larger for citizens with higher levels of political interest.

• 3 Data and Operationalisation

  These hypotheses were tested by data from Belgium. The Belgian political system is of particular interest when it comes to political representation. Indeed, the consociational system (Van Haute & Wauters, 2019), combined with its proportional representation, gives smaller parties larger opportunities to access the parliament and, therefore, a higher probability for citizens to find a congruent political party (Lijphart, 2012). However, while we acknowledge the specificities of the Belgian case, we do not expect a different effect between policy opinion congruence and political trust in other countries.
  We used a Belgian data set collected in the framework of the POLPOP project,¹x The POLPOP project is a collaboration examining politicians’ perceptions initiated by Stefaan Walgrave from the University of Antwerp and funded by the Flemish national science foundation (FWO: grant number G012517N). In Francophone Belgium, it is led by Jean-Benoit Pilet and Nathalie Brack (ULB). which conducted both a citizen and a political elite survey. The elite survey targeted federal and regional Members of Parliament between March and June 2018. MPs filled the questionnaire during a face-to-face meeting. The response rate was 76.8% among all Dutch-speaking legislators, and 74.8% among all French-speaking ones. The citizen survey was conducted online with the help of a survey company (Survey Sampling International). The survey population consisted of Dutch- and French-speaking citizens over the age of 18 with access to the internet. The sample size was 2,389 observations for Dutch-speaking citizens and 2,371 observations for French-speaking citizens. The survey sample aimed to be representative of the population with regard to age, gender and education.²x There is a small bias in favour of older and higher educated citizens, something that we account for in our analysis. More information about the representativeness of the sample is presented in Appendix 1.
  Our dependent variable was trust in the parliament. We measured the level of trust in the Belgian federal parliament in order to match this variable with our measure of policy opinion congruence between citizens and their federal legislators. The question read as follows:

  Can you indicate on a scale from 0 to 10 how much you personally trust the Belgian federal parliament? 0 means you do not trust the institution at all, and 10 means you have complete trust.

  The first hypothesis was tested using a measure of citizen–legislator policy opinion congruence. In the POLPOP project, both citizens and political representatives were asked to give their opinion on eight policy statements.³x The eight statements were as follows: ‘National armies should be replaced by one European army’; ‘Voting should remain compulsory’; ‘The most polluting cars should be forbidden in cities’; ‘Company cars should be more heavily taxed’; ‘The right to strike should be restricted’; ‘Belgium should never expel someone to a country where human rights are violated’; ‘The full income of all parliamentarians should be published yearly’; ‘The retirement age may not exceed 67 years’. Citizens and political elites had five answer options: totally disagree, rather disagree, rather agree, totally agree or undecided (neutral or no opinion). This strategy provided a unique opportunity to measure opinion congruence on the basis of specific policy issues, instead of the left–right dimension. We measured policy opinion congruence as the policy proximity between each citizen and legislators from the party they voted for in the last elections.
  More specifically, we compared the position of an individual citizen (whether the respondent agreed or disagreed with a statement) with the position of their party. Parties’ positions equal the majority position of their federal legislators.⁴x Except for the party PTB, as only regional MPs responded to the survey. However, we are confident about the reliability of this measure as the average majority positions of federal and regional MPs are identical most of the time in our data set. Regional and federal legislators disagreed in only 5 out of 104 party issue dyads (4.8%). For each citizen*issue dyad, respondents received the value 1 if their position was similar to that of their party’s legislators and 0 otherwise. The final opinion congruence variable was the average score of the eight policy statements and ranges from 0 (complete incongruence) to 1 (complete congruence). More precisely, the citizen–legislator opinion congruence variable was calculated as follows:

  

  where i refers to a respondent, Citizen’s position is the position (1 = agree; 0 = dis­agree) of citizen i on the policy proposal k and Position of the party is the position (1 = agree; 0 = disagree) of citizen I’s political party on the policy proposal k.
  For the second and third hypotheses, we included an interaction effect between our measure of policy opinion congruence and the citizens’ level of education (H2) and political interest (H3). The education binary variable captured whether the citizens’ highest level of education was secondary education at best (0), or whether they had a higher non-university or a university degree (1). The political interest variable was measured as a categorical variable with the following levels: not at all interested, not interested, somewhat interested, interested and very interested.
  Beyond these variables, several control variables expected to affect the citizens’ level of trust in the parliament were included in the models. The first was the electoral success of political parties. Anderson, Blais, Bowler, Donovan and Listhaug (2005) showed that citizens who perceive that their political party won the election tend to express a higher level of political support. To operationalise the electoral winner variable, respondents who voted for a political party that joined the governing coalition (N-VA, CD&V, Open VLD and MR) were coded as 1 and 0 otherwise. The second set of control variables were socio-demographic variables. Although the direction of the effects is not consistent throughout all studies, it is widely recognised that political trust is likely to be affected by gender (Glaeser, Laibson, Scheinkman & Soutter, 2000), age (Brewer, Gross, Aday & Willnat, 2004) and social status (Schoon, Cheng, Gale, Batty & Deary, 2010). Therefore, we included the income of individuals (whether citizens belonged to the bottom third, to the middle category or to the upper third of the income distribution), their gender (women were coded 1 and men 0) and their age.⁵x Including the variables of age as well as education in our models allowed us to control for a potential overrepresentation of these groups in the sample. Moreover, we controlled for respondents’ political extremism. Previous research in social psychology found that extreme ideological left and right positions correspond to lower levels of trust (Krouwel, Kutiyski, Van Prooijen, Martinsson & Markstedt, 2017). The political extremism measure is the citizens’ absolute distance between their own position on the left–right dimension (0-10 scale) and the mean left–right position of our sample (5.11). Lastly, we controlled for the language group of the respondent, a crucial variable in the case of Belgium. Indeed, the country has two distinct media landscapes and party systems, as well as different public opinions, which are often distinguished empirically (Hooghe & Dassonneville, 2018; Uyttendaele, Jennart, Talukder & Rihoux, 2020). We present an overview of all variables in Appendix 1.
  Finally, it is worth noting that some of these control variables might also have had an effect on our main independent variables. Previous research has found that the preferences of elected representatives tend to be closer to those of richer and more educated citizens (Gilens, Phillips & Lax, 2011; Rosset, Giger & Bernauer, 2013), including in Belgium (Lesschaeve, 2017). This created a risk of endogeneity, as the values on our congruence variable were likely to be related to both income and education, along with our measure of political trust. The inclusion in the models of the potentially confounding variables income and education allowed us to take this into account and to predict the independent effect of congruence on political trust.

• 4 Results

  To test the hypotheses developed in the theoretical section of this article, our analysis consists of a set of linear regression models predicting citizens’ level of trust in the parliament. The results are presented in Table 1. Model 1 tests for the direct and independent effects of policy opinion congruence on citizens’ trust in the parliament. Model 2 includes the different control variables. Model 3 tests for the interaction effect between education and policy opinion congruence whereas Model 4 tests for the interaction effect between political interest and policy opinion congruence.

  Table 1 Models Predicting the Level of Political Trust Among Belgian Citizens

  	Trust in the Parliament
  	Model 1	Model 2	Model 3	Model 4
  Congruence	0.779***	1.335***	1.405***	2.080***
  	(0.209)	(0.227)	(0.302)	(0.705)
  Age		−0.003	0.004	−0.003
  		(0.003)	(0.003)	(0.003)
  Female		0.057	−0.097	0.072
  		(0.093)	(0.093)	(0.092)
  Income (middle)		0.163	0.185	0.186*
  		(0.112)	(0.115)	(0.111)
  Income (high)		0.190	0.280**	0.259**
  		(0.121)	(0.124)	(0.119)
  Electoral winner		1.092***	1.133***	1.105***
  		(0.099)	(0.101)	(0.098)
  Extremism		−0.081***	−0.032	−0.085***
  		(0.030)	(0.030)	(0.030)
  Flanders		−0.012	0.034	−0.043
  		(0.094)	(0.096)	(0.094)
  Higher education		0.246***	0.272
  		(0.093)	(0.313)
  Political interest (Not interested)		0.968***		0.921
  		(0.196)		(0.637)
  Political interest (Somewhat interested)		1.662***		1.987***
  		(0.161)		(0.523)
  Political interest (Interested)		1.893***		2.253***
  		(0.172)		(0.554)
  Political interest (Very interested)		1.851***		4.099***
  		(0.214)		(0.685)
  Congruence × Higher education			0.121
  			(0.447)
  Congruence × Political interest (Not interested)				0.092
  				(0.947)
  Congruence × Political interest (Somewhat interested)				−0.488
  				(0.780)
  Congruence × Political interest (Interested)				−0.505
  				(0.815)
  Congruence × Political interest (Very interested)				−3.122***
  				(0.967)
  Constant	3.630***	1.288***	2.272***	0.892*
  	(0.146)	(0.268)	(0.280)	(0.502)
  N	3120	2630	2630	2637
  R-squared	0.004	0.123	0.074	0.126
  Adj. R-squared	0.004	0.118	0.071	0.121
  Residual Std. Error	2.410 (df = 3118)	2.277 (df = 2616)	2.337 (df = 2619)	2.277 (df = 2620)
  F Statistic	13.905*** (df = 1; 3118)	28.144*** (df = 13; 2616)	21.037*** (df = 10; 2619)	23.604*** (df = 16; 2620)

  ***p < 0.01; **p < 0.05; *p < 0.1; Standard errors in parentheses.

  Citizens’ predicted values for trust in the parliament at different levels of policy opinion congruence, depending on their level of education.
  

  With regard to our first hypothesis, we expect political trust to be related to citizens’ level of policy opinion congruence with their party’s legislators. The results in Table 1 confirm this hypothesis. Models 1 and 2 display a positive and statistically significant main effect for policy opinion congruence. In line with our hypothesis, a higher level of policy opinion congruence between citizens and legislators is associated with an increase in trust in the parliament. This finding is consistent throughout all models, indicating that policy opinion congruence is a significant predictor of political trust. The effect of congruence on trust remains statistically significant even after controlling for socio-demographic variables, as well as for the effect of having voted for a winning political party. Moreover, policy opinion congruence continues to have a significant impact on trust even when we control for citizens’ level of political interest.

  However, although we found a significant and positive effect of congruence on trust in the parliament, it is worth mentioning that it was smaller than the effect of political interest or the effect of having voted for a winning political party, two factors often considered to be important predictors of citizens’ political support (e.g. Anderson et al., 2005; Holmberg et al., 2017). Indeed, a shift from one standard deviation in policy opinion congruence is associated with a 0.54-point increase in political trust. By contrast, a shift from one standard deviation in political interest is associated with a 1.09-point increase in trust in the parliament, and a shift from one standard deviation in the electoral winner variable is tied to a 1.04-point increase in political trust.
  Models 3 and 4 explore the moderating effect of education and political interest, respectively. We expected the effect of policy opinion congruence to be larger for highly educated and politically interested citizens. In Model 3, a high-education dummy was interacted with our measure of policy opinion congruence. The coefficient for our measure of opinion congruence remained positive and statistically significant, indicating that, contrary to our expectation, attitudes of respondents with a lower level of education were affected by their level of opinion policy congruence. The non-significant interaction term further rejected our hypothesis. To facilitate the interpretation of this interaction term, Figure 1 displays citizens’ predicted values for trust in the parliament at different levels of policy opinion congruence, depending on their level of education. Our results suggest that the effect of congruence between citizens and their party’s legislators on trust is not larger for citizens with higher levels of education. The implication of these results will be further discussed in the final section.

  Citizens’ predicted values for trust in parliament at different levels of policy opinion congruence, depending on citizens’ levels of political interest.
  

  In Model 4, the political interest variable was interacted with our measure of congruence. We hypothesised that the effect of congruence on trust would be larger among citizens with a high level of political interest. The results of Model 4 rejected this hypothesis. First, the coefficient for our measure of policy opinion congruence was positive and statistically significant. This indicates that among citizens who have no interest in politics (i.e. citizens who declare that they are not at all interested in politics), policy opinion congruence continues to have a significant and large effect on trust in the parliament.
  More importantly, the effect of policy opinion congruence on trust is different depending on the citizens’ level of political interest. We found an important distinction between respondents with a very high level of political interest (those who declared being ‘very interested’ in politics) and all the other respondents. Indeed, the interaction coefficients in Model 4 were not statistically different from one another for all levels of political interest, except for the ‘very interested’ citizens. By contrast, for highly politically interested citizens, the interaction coefficient was significantly different, and negative. This suggests that, for citizens who declared they were very interested in politics, the effect of policy opinion congruence on trust in the parliament is smaller than for all other citizens.
  The nature and magnitude of the moderating effect of political interest are illustrated by Figure 2, which displays the citizens’ predicted values for trust in the parliament at different levels of policy opinion congruence, depending on their levels of political interest. It confirms that, contrary to our expectations, policy opinion congruence has a large effect on trust for all respondents expect for very politically interested ones. Moreover, Figure 2 shows that the effect of congruence on political trust is no longer statistically significant for the very politically interested citizens. In other words, regardless of whether their policy preferences are congruent with those of their party’s legislators, the level of trust for very politically interested citizens remains stable. Therefore, our third hypothesis was rejected by the data. The implications of this finding will be discussed further in the following section.

  Finally, we conducted a series of robustness checks. We first re-estimated the parameters of each model using a new operationalisation for our dependent variable, which is the proximity between a citizen’s position and the majority preference in the parliament. We expected trust to be positively related to that collectivist measure of congruence (Kim, 2009). The results continue to support our previous findings: citizens whose positions match the majority position in the parliament have a higher level of trust, except for very politically interested citizens (Appendix 3). We then replicated our models with a continuous measure of congruence instead of a dichotomous one. For each statement we compared citizens’ positions with the percentage of MPs from their party who agreed with the statement in question.⁶x The score for each statement has been computed as follows: congruence = 1 - | c - p | , where c is the position of the citizen (1 if agrees and 0 otherwise), and p is the percentage of MPs (from the party the respondent voter for) who agree. The final variable is, for each citizen, the average of the citizen’s scores on the eight policy statements. That measurement allowed us to account for political party division. The results are presented in Appendices 4 (citizen–party congruence) and 5 (citizen–parliament congruence) and lead to a similar conclusion.

• 5 Discussion and Conclusion

  Following a recent stream of literature on the attitudinal effects of political representation, our research aimed at investigating the linkage between policy opinion congruence and citizens’ level of trust in parliament. This article, therefore, contributes to the literature that studies parliamentary representation from citizens’ perspective (Mayne & Hakhverdian, 2017; Stecker & Tausendpfund, 2016). However, contrary to much of the work on citizens’ satisfaction with democracy, we focused our attention on trust in the parliament, which can be considered as a more specific measure of support towards one crucial institution of modern representative democracies. Our article further contributes to the literature by measuring opinion congruence on specific policy issues. To that end, we used a data set collected in Belgium in which both citizens and legislators gave their opinion on eight different policy statements.
  In our research we investigated the extent to which a lack of policy opinion congruence for citizens is associated with their level of trust in the parliament. In that regard, we found that policy opinion congruence matters. The more citizens’ policy preferences are congruent with those of their party’s legislators, the more they tend to trust the parliament. In that regard, the growing levels of inequalities in representation documented both in the United States and in Europe (e.g. Bartels, 2008; Rosset et al., 2013) cannot be analysed in a black box. They are likely to have a direct and important effect on how much citizens trust representative institutions and, therefore, on the extent to which they will find policy decisions legitimate (Marien & Hooghe, 2011).
  We further expected the relationship between policy opinion congruence and trust in the parliament to be moderated by certain individual characteristics. Because a higher level of education may increase an individual’s capacity to identify levels of congruence, we expected the effect of policy opinion congruence on trust to be larger for highly educated citizens. This hypothesis was rejected by the data. Indeed, our analyses provided no evidence that education is a moderator of the relationship between congruence and trust in the parliament. This finding contradicts that of Mayne and Hakhverdian (2017, p. 836), who showed that there is “evidence that ideological congruence has a larger effect on the higher educated than on the less educated”.
  We suggest three potential explanations for these divergent findings, all of them related to the operationalisation of our variables. First, we measured opinion congruence on specific policy issues instead of on the more abstract left–right axis. It is possible that citizens with higher levels of education are indeed cognitively more capable of gauging their congruence on an abstract left–right scale (see e.g. Lesschaeve, 2017), which thus affects their level of political support more significantly. However, our findings suggest that all citizens, regardless of their education level, are likely to have some knowledge of their representatives’ position on specific policy issues. Second, we operationalised political support as trust in the parliament and not as satisfaction with democracy. One cannot rule out that the highly educated, who often report higher levels of support for democratic principles (Dalton, 2004), are ‘even more’ satisfied with democracy when they are congruent with their representatives. By contrast, when looking at a more specific indicator of support, we found that trust in the parliament is affected by policy opinion congruence for all citizens, independently of their educational level.
  Finally, while Mayne and Hakhverdian (2017) used education as a proxy for political sophistication, we argue that education may capture cognitive abilities but not necessarily political sophistication. For that reason, we also tested the moderating effect of political interest, a well-established indicator of political sophistication⁷x Descriptive statistics from our sample confirm that, although political interest and education are positively correlated, it is still the case that 42.77% of highly politically interested citizens have a low level of education. (Gallina et al., 2020). We expected the effect of policy opinion congruence on trust in the parliament to be larger for politically interested citizens. However, we found no evidence of this. On the contrary, the individual level of policy opinion congruence had no effect on political trust for citizens who are very interested in politics, while it had a significant effect for all the other respondents.⁸x This finding is robust as it holds when testing with citizens’ declared level of political information, another key component of political sophistication (see Appendix 2).
  Previous research has found that politically interested citizens are more likely to have higher levels of trust in the parliament (Holmberg et al., 2017). Our results led us to distinguish further between citizens who are ‘interested’ and those who are ‘very interested’. Although both groups have quite similar levels of trust in the parliament, only very politically interested citizens have a level of trust that is not affected by their level of congruence. Put differently, citizens with a very high level of interest in politics have a ‘stock’ of political trust that, according to our analysis, continues to be significantly high, regardless of the quality of representation. We propose two main explanations for this finding, but further research is needed to back up these claims. First, it may be related to the fact that highly interested citizens, compared with all other respondents, are more likely to understand the constraints of multidimensional representative politics. Therefore, although they may perceive the incongruence between their views and those of their representatives on a specific set of policy statements, they simultaneously acknowledge that perfect opinion congruence in a multidimensional political environment is extremely rare. Second, it may be that highly interested citizens have a bias ‘in favour of’ representative institutions and, therefore, their level of trust does not necessarily result from an evaluation of the quality of representation.
  Our results are reassuring as well as worrying. On the one hand, trust in the parliament for less politically interested citizens is affected by the extent to which their party’s political representatives share their views. Although our study did not test a direct causal mechanism between low levels of representation and political distrust, it nevertheless shows that trust in the parliament is responsive to the way citizens’ policy preferences are reflected in the legislature. By contrast, very politically interested citizens do not have different attitudes depending upon their level of policy opinion congruence.
  Our findings have implications for further research. First, the unexpected findings regarding the moderating effect of education signal the need for additional studies. In that regard, further research should draw on both measures of political support (satisfaction with democracy and political trust) and use several indicators of congruence (left–right axis and specific issues) in order to better disentangle the different effects of education. Second, the moderating effect of political interest should also be further investigated. In that regard, future studies should focus on distinguishing objective and subjective congruence. Indeed, integrating both perceived congruence and objective congruence in a single research design would help determine whether the very politically interested have a more accurate knowledge of their congruence and, therefore, whether they continue to trust political representatives even though they know their legislators are incongruent. Finally, policy opinion congruence was measured on a limited number of statements, and we acknowledge that our study would benefit from including more of them. This would provide more robust results and allow us to test issue-level characteristics, such as salience.

References

• Anderson, C., Blais, A., Bowler, S., Donovan, T. & Listhaug, O. (2005). Losers’ Consent: Elections and Democratic Legitimacy. Oxford University Press.

• Bartels, L. M. (2008). Unequal Democracy (STU-Student edition). Princeton University Press; JSTOR. doi: 10.2307/j.ctt7t9ks.

• Bedock, C. (2017). Reforming Democracy: Institutional Engineering in Western Europe (New product edition). Oxford University Press.

• Brewer, P. R., Gross, K., Aday, S. & Willnat, L. (2004). International Trust and Public Opinion about World Affairs. American Journal of Political Science, 48(1), 93-109. doi: 10.2307/1519899.

• Carpini, M. X. D. & Keeter, S. (1996). What Americans Know about Politics and Why It Matters. Yale University Press. Retrieved from www.jstor.org/stable/j.ctt1cc2kv1.

• Crozier, M., Huntington, S. P. & Watanuki, J. (1975). The Crisis of Democracy (Vol. 70). New York University Press.

• Curini, L. & Jou, W. (2016). The Conditional Impact of Winner/Loser Status and Ideological Proximity on Citizen Participation. European Journal of Political Research, 55(4), 767-788. doi: 10.1111/1475-6765.12161.

• Dahlberg, S. & Holmberg, S. (2014). Democracy and Bureaucracy: How their Quality Matters for Popular Satisfaction. West European Politics, 37(3), 515-537. doi: 10.1080/01402382.2013.830468.

• Dalton, R. J. (2004). Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies. Oxford University Press.

• Dalton, R. J. (2007). Partisan Mobilization, Cognitive Mobilization and the Changing American Electorate. Electoral Studies, 26(2), 274-286. doi: 10.1016/j.electstud.2006.04.009.

• Dalton, R. J., Farrell, D. M. & McAllister, I. (2011). Political Parties and Democratic Linkage: How Parties Organize Democracy. Oxford University Press.

• Dassonneville, R. & McAllister, I. (2020). The Party Choice Set and Satisfaction with Democracy. West European Politics, 43(1), 49-73. doi: 10.1080/01402382.2019.1609286.

• Dunn, K. (2015). Voice, Representation and Trust in Parliament. Acta Politica, 50(2), 171-192. doi: 10.1057/ap.2014.15.

• Ezrow, L. & Xezonakis, G. (2011). Citizen Satisfaction with Democracy and Parties’ Policy Offerings. Comparative Political Studies, 44(9), 1152-1178. doi: 10.1177/0010414011405461.

• Ferland, B. (2021). Policy Congruence and Its Impact on Satisfaction with Democracy. Electoral Studies, 69, 102204. doi: 10.1016/j.electstud.2020.102204.

• Foa, R.S., Klassen, A., Slade, M., Rand, A. and R. Collins. 2020. The Global Satisfaction with Democracy Report 2020. Cambridge, United Kingdom: Centre for the Future of Democracy.

• Foos, F. & De Rooij, E. A. (2017). The Role of Partisan Cues in Voter Mobilization Campaigns: Evidence from a Randomized Field Experiment. Electoral Studies, 45, 63-74. doi: 10.1016/j.electstud.2016.11.010.

• Gallina, M., Baudewyns, P. & Lefevere, J. (2020). Political Sophistication and Populist Party Support. The Case of PTB-PVDA and VB in the 2019 Belgian Elections. Politics of the Low Countries, 2(3), 265-288. doi: 10.5553/PLC/258999292020002003003.

• Gilens, M. (2005). Inequality and Democratic Responsiveness. Public Opinion Quarterly, 69(5), 778-796. doi: 10.1093/poq/nfi058.

• Gilens, M., Phillips, J. & Lax, J. R. (2011). Representational Inequality in the US States. APSA 2011 Annual Meeting Paper.

• Glaeser, E. L., Laibson, D. I., Scheinkman, J. A. & Soutter, C. L. (2000). Measuring Trust. The Quarterly Journal of Economics, 115(3), 811-846. doi: 10.1162/003355300554926.

• Golder, M. & Stramski, J. (2010). Ideological Congruence and Electoral Institutions. American Journal of Political Science, 54(1), 90-106. doi: 10.1111/j.1540-5907.2009.00420.x.

• Grönlund, K. & Setälä, M. (2007). Political Trust, Satisfaction and Voter Turnout. Comparative European Politics, 5(4), 400-422. doi: 10.1057/palgrave.cep.6110113.

• Hall, P. A. & Evans, G. (2019). Representation Gaps: Changes in Popular Preferences and the Structure of Partisan Competition in the Developed Democracies. Unpublished Manuscript, Minda de Gunzburg Center for European Studies, Harvard University.

• Hardin, R. (1999). Do We Want Trust in Government? In M. Warren (Ed.), Democracy and Trust (pp. 22-41). Cambridge: Cambridge University Press. doi:10.1017/CBO9780511659959.002.

• Hibbing, J. R. & Theiss-Morse, E. (2002). Stealth Democracy: Americans’ Beliefs about How Government Should Work. Cambridge University Press.

• Hobolt, S. B. (2012). Citizen Satisfaction with Democracy in the European Union. JCMS: Journal of Common Market Studies, 50, 88-105. doi: 10.1111/j.1468-5965.2011.02229.x.

• Holmberg, S., Lindberg, S. & Svensson, R. (2017). Trust in Parliament. Journal of Public Affairs, 17(1–2), e1647. doi: 10.1002/pa.1647.

• Hooghe, M. & Dassonneville, R. (2018). A Spiral of Distrust: A Panel Study on the Relation between Political Distrust and Protest Voting in Belgium. Government and Opposition, 53(1), 104-130. doi: 10.1017/gov.2016.18.

• Kim, M. (2009). Cross-National Analyses of Satisfaction with Democracy and Ideological Congruence. Journal of Elections, Public Opinion and Parties, 19(1), 49-72. doi: 10.1080/17457280802568402.

• Kim, S. (2010). Public Trust in Government in Japan and South Korea: Does the Rise of Critical Citizens Matter? Public Administration Review, 70(5), 801-810. doi: 10.1111/j.1540-6210.2010.02207.x.

• Kirkland, J. H. & Banda, K. K. (2019). Perceived Ideological Distance and Trust in Congress. Social Science Quarterly, 100(5), 1810-1827. doi: 10.1111/ssqu.12659.

• Kolczynska, M. & Bürkner, P.-C. (2021). Political Trust as a Cause and Consequence of Democracy: Longitudinal Analysis of European Data. Retrieved from osf.io/chx6a.

• Kriesi, H. (2013). Democratic Legitimacy: Is There a Legitimacy Crisis in Contemporary Politics? Politische Vierteljahresschrift, 54(4), 609-638. doi: 10.5771/0032-3470-2013-4-609.

• Kriesi, H., Grande, E., Lachat, R., Dolezal, M., Bornschier, S. & Frey, T. (2006). Globalization and the Transformation of the National Political Space: Six European Countries Compared. European Journal of Political Research, 45(6), 921-956. doi: 10.1111/j.1475-6765.2006.00644.x.

• Krouwel, A., Kutiyski, Y., Van Prooijen, J.-W., Martinsson, J. & Markstedt, E. (2017). Does Extreme Political Ideology Predict Conspiracy Beliefs, Economic Evaluations and Political Trust? Evidence from Sweden. Journal of Social and Political Psychology, 5(2), 435–462. https://doi.org/10.5964/jspp.v5i2.745.

• Lau, R. R. & Redlawsk, D. P. (1997). Voting Correctly. American Political Science Review, 91(3), 585-598. doi: 10.2307/2952076.

• Lau, R. R. & Redlawsk, D. P. (2001). Advantages and Disadvantages of Cognitive Heuristics in Political Decision Making. American Journal of Political Science, 45(4), 951-971. doi: 10.2307/2669334.

• Lefkofridi, Z., Giger, N. & Kissau, K. (2012). Inequality and Representation in Europe. Representation, 48(1), 1-11, doi: 10.1080/00344893.2012.653227.

• Lesschaeve, C. (2017). The Predictive Power of the Left-right Self-Placement Scale for the Policy Positions of Voters and Parties. West European Politics, 40(2), 357-377. doi: 10.1080/01402382.2016.1229088.

• Levitt, S. D. (1996). How Do Senators Vote? Disentangling the Role of Voter Preferences, Party Affiliation, and Senator Ideology. The American Economic Review, 86(3), 425-441.

• Lijphart, A. (2012). Patterns of Democracy: Government Forms and Performance in Thirty-Six Countries (2nd ed.). Yale University Press.

• Linde, J. & Ekman, J. (2003). Satisfaction with Democracy: A Note on a Frequently Used Indicator in Comparative Politics. European Journal of Political Research, 42(3), 391-408. doi: 10.1111/1475-6765.00089.

• Luskin, R. C. (1990). Explaining Political Sophistication. Political Behavior, 12(4), 331-361. doi: 10.1007/BF00992793.

• Lutz, G., Kissau, K. & Rosset, J. (2012). Policy Congruence Beyond Left-Right: Evidence from the Swiss 2007 Election. 27-29.

• Marien, S. & Hooghe, M. (2011). Does Political Trust Matter? An Empirical Investigation into the Relation between Political Trust and Support for Law Compliance. European Journal of Political Research, 50(2), 267-291. doi: 10.1111/j.1475-6765.2010.01930.x.

• Martini, S. & Quaranta, M. (2019). Citizens and Democracy in Europe: Contexts, Changes and Political Support. Springer.

• Mayne, Q. & Hakhverdian, A. (2017). Ideological Congruence and Citizen Satisfaction: Evidence from 25 Advanced Democracies. Comparative Political Studies, 50(6), 822-849. doi: 10.1177/0010414016639708.

• Miller, W. E. & Stokes, D. E. (1963). Constituency Influence in Congress. The American Political Science Review, 57(1), 45-56. doi: 10.2307/1952717.

• Norris, P. (2011). Democratic Deficit: Critical Citizens Revisited. Cambridge University Press. Retrieved from https://books.google.be/books?id=slFcAQAACAAJ.

• Otjes, S. (2016). What’s Right about the Left–Right Dimension? The Causes and the Consequences of Ideological Inconsistency on Economic Issues in Germany. German Politics, 25(4), 581-603. doi: 10.1080/09644008.2016.1223841.

• Pharr, S., Putnam, R. & Dalton, R. (2000). A Quarter-Century of Declining Confidence. Journal of Democracy, 11, 5-25. doi: 10.1353/jod.2000.0043.

• Pitkin, H. F. (1967). The Concept of Representation. University of California Press. Retrieved from https://books.google.be/books?id=AgUVWLswTNEC.

• Rosset, J., Giger, N. & Bernauer, J. (2013). More Money, Fewer Problems? Cross-Level Effects of Economic Deprivation on Political Representation. West European Politics, 36(4), 817-835. doi: 10.1080/01402382.2013.783353.

• Schoon, I., Cheng, H., Gale, C. R., Batty, G. D. & Deary, I. J. (2010). Social Status, Cognitive Ability, and Educational Attainment as Predictors of Liberal Social Attitudes and Political Trust. Intelligence, 38(1), 144-150. doi: 10.1016/j.intell.2009.09.005.

• Stecker, C. & Tausendpfund, M. (2016). Multidimensional Government-Citizen Congruence and Satisfaction with Democracy. European Journal of Political Research, 55(3), 492-511. doi: 10.1111/1475-6765.12147.

• Talukder, D., Uyttendaele, L., Jennart, I. & Rihoux, B. (2021). The Impact of VAAs on Vote Switching at the 2019 Belgian Legislative Elections’ More Switchers, but Making their Own Choices’. Politics of the Low Countries, 3(1), 73-94. doi: 10.5553/PLC/.000010.

• Thomassen, J. (2012). The Blind Corner of Political Representation. Representation, 48(1), 13-27. doi: 10.1080/00344893.2012.653229.

• Uyttendaele, L., Jennart, I., Talukder, D. & Rihoux, B. (2020). Les utilisateurs du test électoral 2019 en Belgique: Quels profils, quelles motivations et quels effets sur la connaissance politique? In Les Belges haussent leur voix: Une analyse des comportements électoraux du 26 mai 2019. PUL Presses Universitaires De Louvain.

• Van der Brug, W. & Van Spanje, J. (2009). Immigration, Europe and the ‘New’ Cultural Cleavage. European Journal of Political Research, 48, 309-334. doi: 10.1111/j.1475-6765.2009.00841.x.

• Van der Meer, T. (2010). In What We Trust? A Multi-Level Study into Trust in Parliament as an Evaluation of State Characteristics. International Review of Administrative Sciences, 76(3), 517-536. doi: 10.1177/0020852310372450.

• Van der Meer, T. W. & Zmerli, S. (2017). The Deeply Rooted Concern with Political Trust. In Handbook on Political Trust. Edward Elgar Publishing.

• van Egmond, M., Johns, R. & Brandenburg, H. (2020). When Long-Distance Relationships Don’t Work Out: Representational Distance and Satisfaction with Democracy in Europe. Electoral Studies, 66, 102182. doi: 10.1016/j.electstud.2020.102182.

• Van Haute, E., Amjahad, A., Borriello, A., Close, C. & Sandri, G. (2013). Party Members in a Pillarised Partitocracy. An Empirical Overview of Party Membership Figures and Profiles in Belgium. Acta Politica, 48(1), 68-91. doi: 10.1057/ap.2012.25.

• Van Haute, E. & Wauters, B. (2019). Do Characteristics of Consociational Democracies Still Apply to Belgian Parties? Politics of the Low Countries, 1(1), 6-26. doi: 10.5553/PLC/258999292019001001002.

• Verba, S., Schlozman, K. L. & Brady, H. E. (1995). Voice and Equality: Civic Voluntarism in American Politics. Harvard University Press.

• Wagner, A. F., Schneider, F. & Halla, M. (2009). The Quality of Institutions and Satisfaction with Democracy in Western Europe—A Panel Analysis. European Journal of Political Economy, 25(1), 30-41. doi: 10.1016/j.ejpoleco.2008.08.001.

Appendix

Appendix 1 Representativity of the Sample and Descriptive Statistics of the Variables of the Model

Flanders			Wallonia
Gender	Population	Sample	Gender	Population	Sample
Male	0.4946	0.4991	Male	0.4946	0.4612
Female	0.5054	0.5009	Female	0.5054	0.5388
Education	Population	Sample	Education	Population	Sample
None or primary school	0.2317	0.0636	None or primary school	0.3050	0.0853
Secondary school	0.4008	0.5210	Secondary school	0.3455	0.4759
Age category	Population	Sample	Age category	Population	Sample
18-24	0.1317	0.0921	18-24	0.1470	0.0912
25-34	0.1472	0.1286	25-34	0.1687	0.1648
35-44	0.1520	0.1408	35-44	0.1645	0.1580
45-54	0.1704	0.1735	45-54	0.1685	0.1775
55-64	0.1616	0.2460	55-64	0.1514	0.2116
65-74	0.1238	0.1871	65-74	0.1132	0.1706
75-84	0.0797	0.0295	75-84	0.0640	0.0234
85-94	0.0336	0.0023	85-94	0.0226	0.0029
Party	Population	Sample	Party	Population	Sample
Groen	0.0742	0.0795	PS	0.2481	0.2243
Spa	0.1211	0.1197	MR	0.2050	0.2238
CD&V	0.1602	0.1034	Ecolo	0.0701	0.0922
Open VLD	0.1336	0.0833	cdH	0.1060	0.0692
N-VA	0.2798	0.3391	PTB-GO	0.0419	0.0741
Vlaams Belang	0.0502	0.0879	DéFI	0.0334	0.0273
PvdA	0.0245	0.0327	PP	0.0323	0.0190
Other	0.0190	0.0187	Other	0.0585	0.0614
Did not vote	0.0938	0.1356	Did not vote	0.1253	0.2087

Descriptive Statistics: Variables of the Models
Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max
Congruence	3,125	0.665	0.206	0.000	0.500	0.833	1.000
Female	4,745	0.539	0.499	0.000	0.000	1.000	1.000
Age	4,760	48.872	16.279	18	35	62	107
Extremism	4,746	1.725	1.601	0.108	0.108	2.892	5.108
Flanders	4,760	0.502	0.500	0	0	1	1
Electoral winner	4,760	0.335	0.472	0	0	1	1

Descriptive Statistics: Distribution of Political Interest
	Freq	% Valid	% Valid Cum.	% Total	% Total Cum.
Not at all interested	778	16.393	16.393	16.345	16.345
Not interested	656	13.822	30.215	13.782	30.126
Somewhat interested	1,861	39.212	69.427	39.097	69.223
Interested	1,118	23.557	92.984	23.487	92.710
Very interested	333	7.016	100	6.996	99.706
<NA>	14			0.294	100
Total	4,760	100	100	100	100

Flanders			Wallonia
Gender	Population	Sample	Gender	Population	Sample
Male	0.4946	0.4991	Male	0.4946	0.4612
Female	0.5054	0.5009	Female	0.5054	0.5388
Education	Population	Sample	Education	Population	Sample
None or primary school	0.2317	0.0636	None or primary school	0.3050	0.0853
Secondary school	0.4008	0.5210	Secondary school	0.3455	0.4759
Age category	Population	Sample	Age category	Population	Sample
18-24	0.1317	0.0921	18-24	0.1470	0.0912
25-34	0.1472	0.1286	25-34	0.1687	0.1648
35-44	0.1520	0.1408	35-44	0.1645	0.1580
45-54	0.1704	0.1735	45-54	0.1685	0.1775
55-64	0.1616	0.2460	55-64	0.1514	0.2116
65-74	0.1238	0.1871	65-74	0.1132	0.1706
75-84	0.0797	0.0295	75-84	0.0640	0.0234
85-94	0.0336	0.0023	85-94	0.0226	0.0029
Party	Population	Sample	Party	Population	Sample
Groen	0.0742	0.0795	PS	0.2481	0.2243
Spa	0.1211	0.1197	MR	0.2050	0.2238
CD&V	0.1602	0.1034	Ecolo	0.0701	0.0922
Open VLD	0.1336	0.0833	cdH	0.1060	0.0692
N-VA	0.2798	0.3391	PTB-GO	0.0419	0.0741
Vlaams Belang	0.0502	0.0879	DéFI	0.0334	0.0273
PvdA	0.0245	0.0327	PP	0.0323	0.0190
Other	0.0190	0.0187	Other	0.0585	0.0614
Did not vote	0.0938	0.1356	Did not vote	0.1253	0.2087

Descriptive Statistics: Variables of the Models
Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max
Congruence	3,125	0.665	0.206	0.000	0.500	0.833	1.000
Female	4,745	0.539	0.499	0.000	0.000	1.000	1.000
Age	4,760	48.872	16.279	18	35	62	107
Extremism	4,746	1.725	1.601	0.108	0.108	2.892	5.108
Flanders	4,760	0.502	0.500	0	0	1	1
Electoral winner	4,760	0.335	0.472	0	0	1	1

Descriptive Statistics: Distribution of Political Interest
	Freq	% Valid	% Valid Cum.	% Total	% Total Cum.
Not at all interested	778	16.393	16.393	16.345	16.345
Not interested	656	13.822	30.215	13.782	30.126
Somewhat interested	1,861	39.212	69.427	39.097	69.223
Interested	1,118	23.557	92.984	23.487	92.710
Very interested	333	7.016	100	6.996	99.706
<NA>	14			0.294	100
Total	4,760	100	100	100	100

Appendix 2 Models Predicting the Level of Political Trust Among Belgian Citizens – Main Independent Variable: Political Information

The political information item in the Belgian POLPOP reads as follows: “To what extent are you, in general, informed about politics? Where would you place yourself on a scale from 0 to 10, where 0 means that you are not at all aware of politics, and 10 that you are fully aware of politics?” We treat the variable as a continuous variable.

	Trust in Parliament
	Model 1	Model 2	Model 3	Model 4
Congruence	0.779***	1.278***	1.405***	3.371***
	(0.209)	(0.227)	(0.302)	(0.606)
Ageo		−0.003	0.004	−0.004
		(0.003)	(0.003)	(0.003)
Female		0.147	−0.097	0.153*
		(0.093)	(0.093)	(0.092)
Income (middle)		0.133	0.185	0.143
		(0.112)	(0.115)	(0.111)
Income (high)		0.132	0.280**	0.191
		(0.122)	(0.124)	(0.119)
Electoral winner		1.069***	1.133***	1.075***
		(0.099)	(0.101)	(0.098)
Extremism		−0.108***	−0.032	−0.108***
		(0.030)	(0.030)	(0.030)
Flanders		0.099	0.034	0.068
		(0.094)	(0.096)	(0.094)
Congruence × Education (high)			0.121
			(0.447)
Education (high)		0.227**	0.272
		(0.093)	(0.313)
Congruence × Political_information				−0.348***
				(0.098)
Political information		0.270***		0.507***
		(0.023)		(0.068)
Constant	3.630***	1.305***	2.272***	0.003
	(0.146)	(0.257)	(0.280)	(0.444)
N	3,120	2,624	2,630	2,631
R-squared	0.004	0.123	0.074	0.126
Adj. R-squared	0.004	0.119	0.071	0.122
Residual Std. Error	2.410 (df = 3,118)	2.277 (df = 2,613)	2.337 (df = 2,619)	2.276 (df = 2,620)
F Statistic	13.905*** (df = 1; 3,118)	36.523*** (df = 10; 2,613)	21.037*** (df = 10; 2,619)	37.607*** (df = 10; 2,620)

***p < 0.01; **p < 0.05; *p < 0.1

Appendix 3 Models Predicting the Level of Political Trust Among Belgian Citizens – Citizen–Parliament Collective Congruence

	Trust in Parliament
	Model 1	Model 2	Model 3	Model 4
Congruence	1.440***	1.295***	1.775***	1.182***
	(0.186)	(0.201)	(0.266)	(0.445)
Age		−0.010***	−0.004	−0.010***
		(0.003)	(0.003)	(0.003)
Female		−0.067	−0.232***	−0.045
		(0.079)	(0.080)	(0.079)
Income (middle)		0.114	0.157	0.125
		(0.093)	(0.096)	(0.093)
Income (high)		0.160	0.264**	0.222**
		(0.103)	(0.106)	(0.102)
Electoral winner		0.977***	1.115***	0.989***
		(0.087)	(0.089)	(0.087)
Extremism		−0.073***	−0.026	−0.076***
		(0.025)	(0.025)	(0.025)
Flanders		0.092	0.136	0.060
		(0.081)	(0.084)	(0.081)
Higher education		0.240***	0.761***
		(0.081)	(0.289)
Political interest (Not interested)		1.117***		0.833*
		(0.145)		(0.448)
Political interest (Somewhat interested)		1.677***		1.481***
		(0.118)		(0.367)
Political interest (Interested)		1.911***		1.700***
		(0.132)		(0.415)
Political interest (Very interested)		1.843***		3.142***
		(0.178)		(0.589)
Congruence × Education (high)			−0.596
			(0.421)
Congruence × Political interest (Not interested)				0.490
				(0.680)
Congruence × Political interest (Somewhat interested)				0.330
				(0.557)
Congruence × Political interest (Interested)				0.395
				(0.616)
Congruence × Political interest (Very interested)				−1.780**
				(0.845)
Constant	2.984***	1.716***	2.379***	1.849***
	(0.127)	(0.204)	(0.223)	(0.316)
N	4,520	3,631	3,631	3,647
R-squared	0.013	0.143	0.084	0.142
Adj. R-squared	0.013	0.140	0.081	0.138
Residual Std. Error	2.460 (df = 4,518)	2.304 (df = 3,617)	2.382 (df = 3,620)	2.309 (df = 3,630)
F Statistic	59.721*** (df = 1; 4,518)	46.551*** (df = 13; 3,617)	33.112*** (df = 10; 3,620)	37.612*** (df = 16; 3,630)

***p < 0.01; **p < 0.05; *p < 0.1; Standard errors in parentheses.

Appendix 4 Models Predicting the Level of Political Trust Among Belgian Citizens – Citizen–Party Congruence (Scale)

	Trust in the Parliament
	Model 1	Model 2	Model 3	Model 4
Congruence	1.284***	2.177***	2.180***	3.910***
	(0.285)	(0.318)	(0.416)	(0.997)
Age		−0.002	0.004	−0.003
		(0.003)	(0.003)	(0.003)
Female		0.041	−0.114	0.058
		(0.095)	(0.095)	(0.095)
Income (middle)		0.156	0.164	0.178
		(0.115)	(0.118)	(0.114)
Income (high)		0.169	0.247*	0.242**
		(0.124)	(0.127)	(0.122)
Electoral winner		1.118***	1.167***	1.134***
		(0.104)	(0.107)	(0.104)
Extremism		−0.064**	−0.017	−0.064**
		(0.031)	(0.031)	(0.031)
Flanders		−0.144	−0.122	−0.187*
		(0.096)	(0.099)	(0.096)
Education (high)		0.255***	0.020
		(0.095)	(0.400)
Political interest (Not interested)		1.047***		1.383*
		(0.205)		(0.822)
Political interest (Somewhat interested)		1.681***		2.684***
		(0.167)		(0.683)
Political interest (Interested)		1.924***		2.967***
		(0.179)		(0.721)
Political interest (Very interested)		1.855***		4.614***
		(0.223)		(0.895)
Congruence × Higher education			0.525
			(0.611)
Congruence × Political interest (Not interested)				−0.549
				(1.316)
Congruence × Political interest (Somewhat interested)				−1.637
				(1.092)
Congruence × Political interest (Interested)				−1.663
				(1.142)
Congruence × Political interest (Very interested)				−4.184***
				(1.367)
Constant	3.421***	0.820***	1.922***	−0.133
	(0.186)	(0.300)	(0.327)	(0.643)
N	2,934	2,466	2,466	2,472
R-squared	0.007	0.123	0.075	0.124
Adj. R-squared	0.007	0.118	0.071	0.118
Residual Std. Error	2.386 (df = 2,932)	2.258 (df = 2,452)	2.317 (df = 2,455)	2.260 (df = 2,455)
F Statistic	20.289*** (df = 1; 2,932)	26.371*** (df = 13; 2,452)	19.945*** (df = 10; 2,455)	21.721*** (df = 16; 2,455)

***p < 0.01; **p < 0.05; *p < 0.1; Standard errors in parentheses.

Appendix 5 Models Predicting the Level of Political Trust Among Belgian Citizens – Citizen–Parliament Collective Congruence (Scale)

	Trust in the Parliament
	Model 1	Model 2	Model 3	Model 4
Congruence	1.284***	2.177***	2.180***	3.910***
	(0.285)	(0.318)	(0.416)	(0.997)
Age		−0.002	0.004	−0.003
		(0.003)	(0.003)	(0.003)
Female		0.041	−0.114	0.058
		(0.095)	(0.095)	(0.095)
Income (middle)		0.156	0.164	0.178
		(0.115)	(0.118)	(0.114)
Income (high)		0.169	0.247*	0.242**
		(0.124)	(0.127)	(0.122)
Electoral winner		1.118***	1.167***	1.134***
		(0.104)	(0.107)	(0.104)
Extremism		−0.064**	−0.017	−0.064**
		(0.031)	(0.031)	(0.031)
Flanders		−0.144	−0.122	−0.187*
		(0.096)	(0.099)	(0.096)
Education (high)		0.255***	0.020
		(0.095)	(0.400)
Political interest (Not interested)		1.047***		1.383*
		(0.205)		(0.822)
Political interest (Somewhat interested)		1.681***		2.684***
		(0.167)		(0.683)
Political interest (Interested)		1.924***		2.967***
		(0.179)		(0.721)
Political interest (Very interested)		1.855***		4.614***
		(0.223)		(0.895)
Congruence × Higher education			0.525
			(0.611)
Congruence × Political interest (Not interested)				−0.549
				(1.316)
Congruence × Political interest (Somewhat interested)				−1.637
				(1.092)
Congruence × Political interest (Interested)				−1.663
				(1.142)
Congruence × Political interest (Very interested)				−4.184***
				(1.367)
Constant	3.421***	0.820***	1.922***	−0.133
	(0.186)	(0.300)	(0.327)	(0.643)
N	2,934	2,466	2,466	2,472
R-squared	0.007	0.123	0.075	0.124
Adj. R-squared	0.007	0.118	0.071	0.118
Residual Std. Error	2.386 (df = 2,932)	2.258 (df = 2,452)	2.317 (df = 2,455)	2.260 (df = 2,455)
F Statistic	20.289*** (df = 1; 2,932)	26.371*** (df = 13; 2,452)	19.945*** (df = 10; 2,455)	21.721*** (df = 16; 2,455)

***p < 0.01; **p < 0.05; *p < 0.1; Standard errors in parentheses.

Noten

• * Authors names are listed in no particular order.

• 1 The POLPOP project is a collaboration examining politicians’ perceptions initiated by Stefaan Walgrave from the University of Antwerp and funded by the Flemish national science foundation (FWO: grant number G012517N). In Francophone Belgium, it is led by Jean-Benoit Pilet and Nathalie Brack (ULB).

• 2 There is a small bias in favour of older and higher educated citizens, something that we account for in our analysis. More information about the representativeness of the sample is presented in Appendix 1.

• 3 The eight statements were as follows: ‘National armies should be replaced by one European army’; ‘Voting should remain compulsory’; ‘The most polluting cars should be forbidden in cities’; ‘Company cars should be more heavily taxed’; ‘The right to strike should be restricted’; ‘Belgium should never expel someone to a country where human rights are violated’; ‘The full income of all parliamentarians should be published yearly’; ‘The retirement age may not exceed 67 years’.

• 4 Except for the party PTB, as only regional MPs responded to the survey. However, we are confident about the reliability of this measure as the average majority positions of federal and regional MPs are identical most of the time in our data set. Regional and federal legislators disagreed in only 5 out of 104 party issue dyads (4.8%).

• 5 Including the variables of age as well as education in our models allowed us to control for a potential overrepresentation of these groups in the sample.

• 6 The score for each statement has been computed as follows: congruence = 1 - | c - p | , where c is the position of the citizen (1 if agrees and 0 otherwise), and p is the percentage of MPs (from the party the respondent voter for) who agree. The final variable is, for each citizen, the average of the citizen’s scores on the eight policy statements.

• 7 Descriptive statistics from our sample confirm that, although political interest and education are positively correlated, it is still the case that 42.77% of highly politically interested citizens have a low level of education.

• 8 This finding is robust as it holds when testing with citizens’ declared level of political information, another key component of political sophistication (see Appendix 2).