2. Party incongruence, satisfaction with democracy, and the effects of misperception

A large literature has investigated the influence of various factors on citizens’ satisfaction with the functioning of democracy (Anderson and Guillory, Reference Anderson and Guillory1997; Rohrschneider, Reference Rohrschneider2005; Hobolt, Reference Hobolt2012; Leiter and Clark, Reference Leiter and Clark2015; Mayne and Hakhverdian, Reference Mayne and Hakhverdian2017; Dassonneville and McAllister, Reference Dassonneville and McAllister2020; Loveless and Binelli, Reference Loveless and Binelli2020; Ridge, Reference Ridge2022). Factors contributing to lower satisfaction include disproportionality and government fractionalization (Christmann and Torcal, Reference Christmann and Torcal2018), voting for losing parties (Curini et al., Reference Curini, Jou and Memoli2012; Singh et al., Reference Singh, Karakoç and Blais2012; Blais et al., Reference Blais, Morin-Chassé and Singh2017; Nemčok, Reference Nemčok2020), and the ideological representativeness of government policies or the overall party system (Kim, Reference Kim2009; Ezrow and Xezonakis, Reference Ezrow and Xezonakis2011; Dahlberg and Holmberg, Reference Dahlberg and Holmberg2014; Dahlberg et al., Reference Dahlberg, Linde and Holmberg2015; Stecker and Tausendpfund, Reference Stecker and Tausendpfund2016; Blais et al., Reference Blais, Morin-Chassé and Singh2017; Mayne and Hakhverdian, Reference Mayne and Hakhverdian2017; Ferland, Reference Ferland2021).

While much of the research in this area is concerned with broader government-citizen congruence (e.g., Powell and Bingham, Reference Powell and Bingham2000; Golder and Stramski, Reference Golder and Stramski2010; Soroka and Wlezien, Reference Soroka and Wlezien2010), a growing body of work examines the consequences of the relationship between individual voters and specific parties, including those they choose to support (Bakker et al., Reference Bakker, Jolly and Polk2018, Reference Bakker, Jolly and Polk2020; Van Egmond et al., Reference Van Egmond, Johns and Brandenburg2020; Goldberg et al., Reference Goldberg, van Elsas and de Vreese2020; Hobolt et al., Reference Hobolt, Hoerner and Rodon2021; Marchal and Watson, Reference Marchal and Watson2022b; Wardt and Otjes, Reference Wardt and Otjes2022). The degree of alignment between the ideological positions of political parties and their supporters is central to the effectiveness of party representation (Mattila and Raunio, Reference Mattila and Raunio2012; Boonen et al., Reference Boonen, Pedersen and Hooghe2017; Dalton, Reference Dalton2018; Carroll and Kubo, Reference Carroll and Kubo2018; Werner, Reference Werner2019; Costello et al., Reference Costello, Toshkov, Bos and Krouwel2020; Costello, Reference Costello2021; Wardt and Otjes, Reference Wardt and Otjes2022). Some consequences of incongruence between parties and voters found in this literature have included decreasing support (Bakker et al., Reference Bakker, Jolly and Polk2018; Marchal and Watson, Reference Marchal and Watson2022b), decreasing antipathy toward other parties (Marchal and Watson, Reference Marchal and Watson2022b), and driving voters to support emerging parties (Wardt and Otjes, Reference Wardt and Otjes2022).

Bakker et al. (Reference Bakker, Jolly and Polk2020) specifically explore the relationship between the representation of voters by parties and citizens’ satisfaction, revealing that party incongruence on issues increases citizens’ dissatisfaction with democracy, leading to support for anti-establishment parties. In closely related work, Van Egmond et al. (Reference Van Egmond, Johns and Brandenburg2020) find a correlation between perceived congruence with the closest party and satisfaction with democracy. Hobolt et al. (Reference Hobolt, Hoerner and Rodon2021) also corroborate the importance of party congruence in influencing such attitudes, conditional on party influence.

Party congruence with supporters can be conceptualized through both objective party positions and through supporters’ subjective perceptions of those positions (Louwerse and Andeweg, Reference Louwerse and Andeweg2020). Some work has defined incongruence based on an objective evaluation of the distance between the parties’ and voters’ views, as gauged by expert surveys (McEvoy, Reference McEvoy2012; Polk et al., Reference Polk, Rovny, Bakker, Edwards, Hooghe, Jolly, Koedam, Kostelka, Marks, Schumacher, Steenbergen and Vachudova2017; Bakker et al., Reference Bakker, Jolly and Polk2020; Marchal and Watson, Reference Marchal and Watson2022a). Perceived congruence, meanwhile, refers to the subjective distance between the positions of parties and supporters, typically measured by surveys of respondent and party left–right placements (Adams et al., Reference Adams, Clark, Ezrow and Glasgow2004; Mattila and Raunio, Reference Mattila and Raunio2006; Ezrow and Xezonakis, Reference Ezrow and Xezonakis2011; Ezrow et al., Reference Ezrow, De Vries, Steenbergen and Edwards2011; Mattila and Raunio, Reference Mattila and Raunio2012; Schumacher et al., Reference Schumacher, De Vries and Vis2013; McAllister et al., Reference McAllister, Sheppard and Bean2015; Adams et al., Reference Adams, Ezrow and Wlezien2016; Boonen et al., Reference Boonen, Pedersen and Hooghe2017; Stiers, Reference Stiers2022). Because voters may have inaccurate or biased perceptions of party positions, perceived congruence can differ from actual congruence.

Citizens’ ability to perceive the ideological positions of political parties accurately is known to be influenced by a wide range of factors, such as education levels and political knowledge or a lack of clarity in party labels (e.g., Palfrey and Poole, Reference Palfrey and Poole1987; Luskin, Reference Luskin1990; Delli Carpini and Keeter, Reference Delli Carpini and Keeter1996; Bartels, Reference Bartels1996; Meirick, Reference Meirick2013; Dahlberg, Reference Dahlberg2013; Banducci et al., Reference Banducci, Giebler and Kritzinger2015; Busch, Reference Busch2016; Carroll and Kubo, Reference Carroll and Kubo2017; Nasr, Reference Nasr2020). In addition to such information gaps, other literature has found that partisan identities can shape information processing or result in motivated reasoning influencing voters’ understanding of policy issues (e.g., Bartels, Reference Bartels2002; Evans and Andersen, Reference Evans and Andersen2004; Carsey and Layman, Reference Carsey and Layman2006; Evans and Andersen, Reference Evans and Andersen2006; Bartels, Reference Bartels2008; Evans and Pickup, Reference Evans and Pickup2010; Tilley and Hobolt, Reference Tilley and Hobolt2011; Jerit and Barabas, Reference Jerit and Barabas2012; Grand and Tiemann, Reference Grand and Tiemann2013), which may distort their perception of party policy positions and the gap between perceived and actual party placements.

These misperceptions of where parties fall on the left–right ideological spectrum can, in turn, distort voters’ assessments of their ideological distance from those parties.Footnote ² Importantly, voter misperceptions about party positions can distort assessments of ideological congruence. When voters inaccurately perceive a party as more ideologically distant from their own stance than objective measures indicate, such misperceptions correspond to greater perceived incongruence. Alternatively, when a voter inaccurately perceives a party as closer to their ideology than expert placements suggest, this misperception might increase their subjective sense of ideological alignment with that party (Merrill et al., Reference Merrill, Grofman and Adams2001; Drummond, Reference Drummond2010).

Here, we explore the relationship between subjective perceptions of representation and attitudes toward the political system. While the objective ideological mismatch between voters and the parties they support may naturally contribute to perceptions of incongruence, we consider whether there is a distinct impact on perceived incongruence separate from the effects of actual incongruence. That is, potentially inaccurate perception of positions may influence satisfaction with democracy by distorting voters’ perceived ideological linkage to parties, even accounting for the actual degree of representation.

In the following analysis, we describe and implement empirical tests to evaluate these questions. The analysis proceeds in two parts. First, we examine the correlation between misperceptions of party positions, actual incongruence, and perceived incongruence. Second, in our main analysis, we investigate whether respondents’ level of satisfaction with democracy decreases with greater perceived incongruence between themselves and the party they support, accounting for actual congruence.

3. Data and measures

3.1 Measuring perceived and actual congruence

While cross-sectional designs are often used to study satisfaction with democracy, this approach may not fully account for the effects of individual characteristics. To address this limitation, we use panel data that allow us to measure changes in key variables for the same individuals over multiple surveys, allowing us to gain better insight into these relationships when holding constant individual-level factors.

Specifically, we use data from the BES data (Schmitt et al., Reference Schmitt, Eijk, Green, Evans, Mellon, Prosser, de Geus and Bailey2021). The UK is useful for studying party representation because of the relative weakness of the party system's representativeness (Brandenburg and Johns, Reference Brandenburg and Johns2014). In particular, in a cross-sectional study of British voters, Brandenburg and Johns (Reference Brandenburg and Johns2014) have found that democratic satisfaction correlates with the lack of perceived proximity to the nearest identified party and not the lack of choices between the major parties. Thus, there is some evidence that UK voters’ attitudes toward democracy are sensitive to how well parties accurately represent their views.

The BES provides periodic surveys of political opinions, perceptions, and preferences, which provides a panel structure appropriate for our study. Because of the variation across regional party systems and contexts in the UK, we restrict the sample used in the analysis to England. Because this study focuses on parties and supporters, only respondents who indicate supporting a party are included. All respondents in these panels were asked to respond to self-reported perceptions of the parties’ left–right positions. Five years of surveys are included in the analysis, from wave 4 in 2015 to wave 18 in 2019. Thirteen waves include the necessary questions about self-reported perceptions of parties' left–right positions. Ten of these waves include the information needed to analyze satisfaction with democracy and these are used in the main analysis below.

Our first aim is to measure perceived and actual incongruence. To measure the actual left–right ideological positions of British parties, we use the mean ideological positions for each party obtained from the Chapel Hill Expert Survey (CHES) for 2014, 2017, and 2019 (Jolly et al., Reference Jolly, Bakker, Hooghe, Marks, Polk, Rovny, Steenbergen and Vachudova2022; Bakker et al. Reference Bakker, Jolly and Polk2018, Reference Bakker, Jolly and Polk2020). While expert placements are still ultimately subjective judgments of parties’ positions, they are independent of voters’ own judgments and reflect experts’ efforts to place parties for analytical purposes. These CHES positions are then matched with the responses from the BES for the closest year of the survey wave (see Appendix A, Table A.1 for the exact survey structure).Footnote ³

Figure 1 presents the average voter's perceived ideological position of the major parties in England on a scale from 0 to 10, where scale 0 represents the “left” in ideology and scale 10 represents the “right.” The Conservative, Labour, and Liberal Democrat parties are denoted by capital letters C, L, and D, respectively. The gray placements correspond to the average perceived positions of voters from the BES data, while the blue placements correspond to the average positions measured by Chapel Hill Survey experts.

Figure 1. CHES expert placements and average BES respondent misperceptions (Note: C = Conservatives, L = Labour, D = Liberal Democrats).

While voters’ perceptions of party positions generally align with expert measurements on average, there is significant variation in how accurately individual citizens perceive party ideology. Figure 2 shows the distribution of the difference between an individual voter's perception and the corresponding actual position for wave 7, as well as continuous lines indicating the fitted normal distributions.

Figure 2. The distributions of misperception (BES wave 7).

3.2 Misperception and incongruence

In this section, we examine the relationship between perceived incongruence, actual incongruence, and misperceptions. We first illustrate these concepts and how inaccuracies in party placements can distort voters’ assessments of representation. We illustrate two scenarios to show how voters can misplace party positions and how this would relates to the actual and perceived political incongruence between themselves and the party they support. First, BES respondents may perceive their own political ideology to be closer to their perceived party placement than to the actual position assessed by CHES experts, as shown in Figure 3b. Conversely, BES respondents may place themselves closer to the actual position than the location they perceive for political parties, as shown in Figure 3a. In this scenario, misperception leads to an underestimation of the degree of representation.

Figure 3. Misperception of party locations: two scenarios, (a) perceived party position farther than actual party position, (b) perceived party-placement is closer than actual party position.

Here, Misperception (${\pi }_{i, t}$) is defined as the absolute distance between an individual respondent's perception of their preferred party's position and the corresponding average position from the CHES expert placements.Footnote ⁴ Specifically, it is calculated as

(1)$${\pi}_{i, t}^{\,p} = \vert \alpha_{i, t}^{\,p}-\bar{\alpha}_{t}^{\,p}\vert ,\; $$

here, for respondent i in wave t, $\alpha _{i, t}^{p}$ represents their perception of the party's left–right ideological position and $\bar {\alpha }_{t}^{p}$ is the average position of the same party reported by the expert survey. This produces a distance, ${\pi }_{i, t}$, between the respondent and the experts, which indicates the level of misperception of the respondent i regarding the position of the party p on wave t. Specifically, ${\pi }_{i, t}$ measures the misperception that voter i has about the party they voted for in the previous general election.

Actual incongruence (γ _i,t) is defined as the absolute difference between the individual respondent's self-placement on general left–right positions and the corresponding average expert placement. This is calculated as

(2)$$\gamma_{i, t}^{\,p} = \vert \alpha_{i, t}^{s}-\bar{\alpha}_{t}^{\,p}\vert ,\; $$

$\alpha _{t}^{s}$ denotes voter i's self-placement in wave t. Perceived incongruence ($\hat {\gamma }_{i, t}$) is measured as the absolute gap between a BES respondent's self-placement and the perceived position of the party they support. $\hat {\gamma }_{i, t}$ is calculated as

(3)$$\hat{\gamma}_{i, t}^{\,p} = \vert \alpha_{i, t}^{s}-\alpha_{t}^{\,p}\vert .$$

Finally, we consider the following panel regression model by including both individual-specific fixed effects and dummies for each wave (v_i and m_t):

(4)$$\hat{\gamma}_{i, t} = \beta_{1}\pi_{i, t} + \beta_{2}\gamma_{i, t} + v_{i} + m_{t} + e_{it},\; $$

where $\hat {\gamma }_{i, t}$ denotes respondent i's perceived incongruence of their own affiliated party in wave t and γ _t,t denotes the actual incongruence between respondent i and their party in wave t. The misperception of respondents about the ideological position of the party they support at time t is represented by π _i,t, accounting for the perceived positions of political parties in the context in which they compete (Wagner and Meyer, Reference Wagner and Meyer2023).

The results of the panel analysis exploring the relationship between perceived incongruence and voters’ misperception are presented in Table 1. We first show the bivariate relationships between misperception and both measures of party congruence. In column (1) of the table, we first show the bivariate relationship between voters’ misperception and actual party incongruence, which we establish has a positive association. That is, voters who misperceive their party's ideology tend to have a larger discrepancy between their own preferences and the positions of the party they support. In column (2), we find a positive bivariate correlation between misperception and perceived party incongruence, with individuals who misperceive their party's position perceiving greater incongruence.

Table 1. Party misperception and perceived and actual party incongruence, BES panel

Note: Robust standard errors in parentheses. *p < 0.10, **p < 0.05, ***p < 0.01.

The subsequent model (column 3) shows the specification described in 3, examining the relationship between misperception and perceived incongruence when controlling for the level of actual incongruence. Here, we see that both actual incongruence and the degree of misperception each explain some proportion of variation in the perception of incongruence among voters. That is, even when controlling for actual incongruence, greater misperception is associated with perceiving a larger gap between themselves and the parties they support, on average. This suggests that misperception plays a distinct role in shaping voters’ perception of incongruence, separate from the influence of actual incongruence.Footnote ⁵

We also performed a cross-sectional analysis using a pooled OLS approach, which accounted for demographic characteristics such as age, education level, gender, survey year, party affiliation, and the number of information sources reported by each respondent. The results of this analysis, presented in Table C.6 of Appendix C, are consistent with the panel findings in Table 1.

4. Perceived incongruence, actual incongruence, and satisfaction with democracy

Having shown the correlation between misperceptions and perceived incongruence in the previous analysis, we now turn to our main investigation of how such perceived incongruence relates to satisfaction with democracy, accounting for actual incongruence based on expert surveys. To investigate this relationship, we consider the following panel regression model, which again utilizes individual-specific fixed effects to control for individual heterogeneity

(5)$$\hat{y}_{i, t} = \alpha_{1}\gamma_{i, t} + \alpha_{2}\hat{\gamma}_{i, t} + \epsilon_{i} + w_{t} + u_{it},\; $$

where $\hat {y}_{i, t}$ denotes the semi-standardized measurement of respondent i's democratic satisfaction.Footnote ⁶ γ _t,t denotes the actual incongruence between respondent i and their party in wave t and $\hat {\gamma }_{i, t}$ denotes the perceived incongruence of the respondent i’ of their own affiliated party in wave t. v _i captures the respondent-specific fixed effects, and m _t captures the time (wave) effect.

Columns (1) and (2) in the upper panel of Table 2 report the estimation results using satisfaction with democracy as the dependent variable. Column (1) considers the case where perceived incongruence is not included as a regressor, while column (2) shows the results when both perceived and actual incongruence are included in the model. In column (1), we see that actual incongruence negatively correlates with satisfaction with democracy.

Table 2. Panel regression: effects of perceived incongruence and actual incongruence on satisfaction, BES panel

Standard errors in parentheses.

*p < 0.10, **p < 0.05, ***p < 0.01.

Once perceived incongruence is also included in the model, the association between voters’ actual incongruence and satisfaction with democracy is no longer statistically significant, while perceived incongruence has a negative and statistically significant association with voters’ satisfaction with democracy. The estimated relationships between perceived and actual party incongruence and satisfaction with democracy are plotted in Figure 4. As shown, a larger perceived incongruence is correlated with a decrease in voters’ satisfaction with democracy, while there is no longer a statistically significant correlation with actual incongruence.

Figure 4. BES: predicted effects of perceived and actual party-supporter incongruence on democratic satisfaction, (a) BES: predicted values of democratic satisfaction by perceived incongruence, (b) BES: predicted values of democratic satisfaction by actual incongruence.

Although the correlation with actual incongruence is not statistically significant when accounting for perceived incongruence, it is important to emphasize that greater objective distances between voters and parties still contribute to dissatisfaction. However, the results suggest that this effect may occur mainly via the influence on perceived incongruence. Consistent with the notion that subjective perceptions are important for attitudes toward democracy, the overall pattern of results suggests that the effect of perceived incongruence remains when accounting for actual congruence.

To evaluate the robustness of this result, we also conducted several additional analyses, shown in the Appendix. First, in column (3) of Appendix Table C.4 we show a model that adds a control for misperception itself to assess its influence alongside perceived and actual incongruence on satisfaction with democracy. We find no statistically significant effect for misperception when actual and perceived incongruence are accounted for, and the effects of perceived incongruence remain nearly the same as those presented above. While perceived incongruence is associated with greater misperception, the effects of the former are present separately from the degree of misperception.Footnote ⁷

Second, we also performed an analysis using a pooled OLS approach, which accounted for demographic characteristics such as age, education level, gender, survey year, party affiliation, and the number of information sources reported by each respondent. The results of this analysis, presented in Table C.6 of Appendix C, are consistent with the findings in Table 1.Footnote ⁸

A third set of additional analyses is intended to partially evaluate the potential endogeneity of perceived incongruence, detailed in Appendix C.4. The first of these uses lagged measures of incongruence and democratic satisfaction, which relates the level of satisfaction to perceived incongruence in the prior survey wave. Similar to our main analyses, the lagged measure of perceived incongruence retains a significant association with lowered democratic satisfaction when including current perceived incongruence and lagged satisfaction with democracy. Second, another analysis examining changes in satisfaction with democracy over time as the dependent variable is detailed in Appendix C.5, which also corroborates the main findings. Third, we also explored an instrumental variable approach, described in Appendix C.6, which is also consistent with the main results. While these supplemental analyses do not eliminate the possibility of endogeneity, they provide some additional evidence suggesting that perceived incongruence may at least partially influence satisfaction with democracy.

A fourth set of supplemental analyses considers a series of alternative measures. First, while expert surveys are useful measures of parties’ “actual” positions independently of respondents, these periodic data limit this measure of congruence to be influenced more by changes in self placements. As an alternative to the expert survey data, we employed an approach that uses average voter placements as a proxy for actual party positions. This method allows for variation in party positions across each survey wave. We conduct a supplementary analysis, detailed in Appendix C.9, where we replace the expert left–right party placements with mean positions derived from BES respondents. To calculate these mean positions, we use both the entire set of BES respondents and a subset of respondents likely to be more politically informed—measured as those with postgraduate degrees or higher. When we use these voter-based measures to measure actual incongruence, the results using these alternative measures remain substantively similar to our main findings, regardless of whether we use the overall respondent average or the more sophisticated subset. The consistency across these measurement approaches provides some evidence that the main findings do not depend on the specific nature of the expert survey data.

Further, we also considered alternatives to left–right ideology measures for respondents and experts and examined an alternative approach to estimate a latent measure of ideological positions based on responses to multiple substantive issue scales using Blackbox scaling (Poole, Reference Poole1998; Poole et al., Reference Poole, Lewis, Rosenthal, Lo and Carroll2016). This facilitates measuring expert and respondent locations based on latent policy preferences rather than interpretations of an abstract left–right scale. For this analysis, we used the BES expert ratings, which include party positions on multiple issues but are limited to certain waves and thus restricted to use in a cross-sectional analysis. We applied this method to BES waves with expert and respondent ratings on four available issues: immigration, redistribution, environment, EU integration. The results using the latent ideological measures mirror the main findings for left–right placements. The details are provided in Appendix C.7. Finally, we also explored the robustness of the aggregate pattern to using each of these policy scales separately, shown in Appendix C.8. The issue-specific results exhibit patterns similar to the main findings using aggregate measure and consistent with the results using the left–right scale.

5. Cross-national analysis of European democracies

As our main panel analysis focuses on a single country context, we also examine whether similar relationships between misperceptions, incongruence, and satisfaction may emerge in a broader set of contexts. To explore this, we use a cross-national sample from the CSES across 14 European countries in Module 5 of the CSES from 2015 to 2021. Examining this broader set of political systems helps assess if perception-driven gaps in ideological representation generally correlate with lower democratic satisfaction when accounting for actual policy incongruence. We utilize CSES data on voters’ perceptions of party positions, self-placements, and satisfaction to estimate cross-sectional models otherwise similar to the main analysis.

We estimate the following specification:

(6)$$\hat{y}_{i} = a_{1}\gamma_{i} + a_{2}\hat{\gamma}_{i} + \theta \tilde{C_{i}} + \eta X_{t} + \phi Y_{i} + \epsilon_{i},\; $$

where $\tilde {C_{i}}$ is a set of demographic characteristics of the respondents, including household income (binned), gender, highest level of education, marriage status, employment status, and household size. We also control for survey years and the country of respondent i by including X _t and Y _i, respectively. The rest of the notation remains the same as Equation (5).

Table 3 reports the findings. Column (1) illustrates that actual incongruence has a statistically significant negative association with respondents’ satisfaction with democracy in the model that does not include perceived incongruence. However, when perceived incongruence is included in column (2), the correlation between actual incongruence and satisfaction with democracy is no longer statistically significant, while the association between voters’ satisfaction with democracy and perceived incongruence is statistically significant. The fitted values are plotted in Figure 5. The results are consistent with the main findings presented above that voters’ perception of the mismatch between themselves and the party they support is especially important to the relationship with satisfaction with the political system, accounting for actual incongruence.Footnote ⁹

Table 3. Regression: effects of perceived incongruence and actual incongruence on satisfaction, European democracies (CSES)

Standard errors in parentheses.

*p < 0.10, **p < 0.05, ***p < 0.01.

Figure 5. CSES: predicted effects of perceived and actual party-supporter incongruence on democratic satisfaction, (a) CSES: predicted values of democratic satisfaction by perceived incongruence, (b) CSES: predicted values of democratic satisfaction by actual incongruence.