Democratic Frustration: Concept, Dimensions and Behavioural Consequences

Using insights from the psychology literature, this article introduces and operationalises the concept of ‘democratic frustration’ to shed new light on the pathologies of democratic crises. While political scientists have devoted ample attention to democratic crises and dissatisfaction, this article suggests that citizens’ frequent references to their “frustration” should be taken more literally. Specifically, it suggests that citizens become frustrated when a perceived democratic delivery deficit interacts with a strong democratic expectation or desire. The article tests this model using two original surveys run in the UK during the 2017 General Election and 2019 European Parliament elections. By measuring expectations and delivery deficit separately, the article maps democratic frustration vis-à-vis alternative concepts such as apathy, criticality, and cynicism, and shows that it is more widespread as an expectation–deficit combination than any of them. It suggests that democratic frustration comprises of three dimensions: ideological, institutional and political. Adapting insights from the psychology of frustration that show it usually results in expressions of withdrawal, anger, or aggression, the article then explores how the three dimensions of frustration typically result in different pathologies. Ideological frustration leads to abstention (withdrawal), institutional frustration to peaceful demonstrations or radical vote (anger) and to envisage leaving one’s country, whilst political and institutional frustrations combine and lead citizens to consider taking part in violent demonstrations or even joining a revolution (aggression).


Heterogeneity in Democratic Crises
Democracy is in crisis, or so it is widely thought to be. Abstention worries many nations with turnout merely reaching 41.8% in recent Presidential elections in Slovakia, and decreasing by nearly 14% in the last elections in Slovenia. The rise of extremist and populist parties has been unprecedented in many countries. In Poland and Hungary, populist forces have dominated national politics for much of the recent decade whilst in Germany and Spain, parties such as the AfD (Alternative for Germany) and Vox have emerged in systems where populist parties used to be virtually absent. Conversely, mass protest movements, from Extinction Rebellion or anti-Brexit marches to the Hong Kong uprising against China's increasing influence, the Yellow Vest movement in France, or violent protests in Greece and Chile have rocked many streets, sometimes peacefully, and sometimes violently. In short, contemporary democracies are confronted with a very serious issue: citizens are increasingly disillusioned and disappointed by their democratic institutions.
One concept often characterises this phenomenon in the words of citizens themselves: frustration. However, whilst such frustration is widely acknowledged [1][2][3] this article argues that there would be worth in taking it at face value to the extent that in psychological terms, "frustration" has a rather specific nature, which makes the strength of an existing desire as important as an individual's sense that it is unfulfilled. Frustration also has specific potential consequences, notably in the forms of withdrawal, anger, and aggression, which this article believes can be usefully translated in political behaviour terms to characterise key pathologies of democratic frustration. This article thus theorises the concept of democratic frustration and explains how it can be mapped compared to other frequently used measures of democratic unhappiness such as apathy (or indifference), cynicism and criticality. I suggest that democratic frustration comprises of three important dimensions: ideological, institutional, and political. I then operationalise the concept and its dimensions based on an interaction between democratic expectations and perceived delivery deficit, along the (implicitly interactive) lines of the psychological definition of frustration as an unsatisfied desire. In this article, I then assesses how widespread democratic frustration is compared to some alternative combinations of expectations and perceived delivery deficit, and how robust it is over time using two representative surveys ran during the 2017 General Elections and 2019 European Parliament election in the UK (see Appendix A for details of surveys). Finally, I show how the three dimensions of democratic frustration can lead to different political versions of the expected psychological reactions of withdrawal, anger, and aggression, which I translate into different pathologies of democratic crisis such as abstention (withdrawal), extremism (anger) and violent protest (aggression). Overall, the article claims that democratic frustration is a critical, and indeed dominant, component of democratic crisis in the UK, and that its different dimensions can have different behavioural consequences likely to result in diverging individual challenges to the existing democratic order. This can give us a precious indication of why some unhappy citizens choose to abstain in elections whilst other will prefer to vote for radical parties or even engage in a revolution.

Existing Models
The crisis of democracies has of course been a key focus of attention for the political behaviour literature. Authors have seen it as symptomatic of the distrust [4] and cynicism of citizens towards political systems, institutions, and social elites [5-8,]. A growing sense of dissatisfaction [9][10][11] has accompanied a decline in turnout [12,13] and party and union memberships [14][15][16] in parallel to a resurgence of populist and extremist behaviour [17] and mass protest movements. A sense of powerlessness, inefficacy cynicism and alienation alongside a lack of interest [18,19] have been found as key factors to such crisis behaviour. The labels used and phenomena described may sometimes be confusing referring to dissatisfaction, distrust, or even apathy all of which have different theoretical implications.
The literature also shows that the democratic crisis may sometimes particularly affect some categories of citizens, notably young people, who are often vocal in their criticisms of how democracy works, sometimes opting for non-electoral forms of participation [10]. Young people in France also signalled a form of democratic frustration during the Presidential Election in 2017. The top two ballot choices for young people aged 18-29 were Mélenchon and Macron. Indeed, both of these candidates advocated "new ways" of doing politics with a promise to overhaul the existing power structures. There has also been ample evidence that both economically-deprived and ethnic minority populations have lower turnouts than average and lower trust in democratic institutions [13].
The idea of a democratic deficit or democratic under-delivery often implicitly (and less frequently explicitly) underlines the importance of citizens' expectations in the literature. For instance, Norris points out that the perceived delivery of electoral democracy often 'lags behind' citizens' expectations [10]. However, even when acknowledging the implicit existence of expectations that are unfulfilled, many models empirically focus on perceived democratic delivery or delivery deficit, without explicitly measuring the specific expectations of citizens when it comes to democratic processes, personnel, and outcomes. Implicitly, those expectations are treated as though they were constant or irrelevant. Yet, from a psychological point of view, variations in desire and expectations are at the heart of frustration, which so many citizens refer to when it comes to their democratic experience [20]. What if we took those references to frustration at face value applying what psychology models tell us about frustration and adapting that model to the question of democracy.

The Psychological Concept of Frustration
The psychological concept of frustration is based on a 'failure to satisfy a motive' [21]. A sense of frustration is reported when an individual is prevented from attaining a certain objective or goal. Frustration is thus sourced from a failure to satisfy a conscious or indeed subconscious desire. That centrality of desire is of critical importance because it suggests that an individual will not feel frustrated about something that they do not care about, or to go a little further, that the potential for frustration increases the more one cares (or indeed obsesses) about something. That centrality is emphasised by Lacan who redefined the psychoanalytical concept of frustration and its relationship to desire through three layers: symbolic, imaginary, and real [22]. The corresponding 'level' of satisfaction or frustration is directly related to the strength of the need, which confirms the operationalisation of democratic frustration model.
The satisfaction deficit is thus only one of the two components of frustration alongside importance, which is here referred to as "expectation" (note that in this context, this is expectation meant as "emphasis" or "caring", not as a standard), so that frustration practically works as an interaction between the two as follows:

Democratic Frustration = Expectation * Perceived Delivery Deficit
Beyond psychology, the link between desire or expectation and frustration has also been noted in arts, for instance by Smuts in 'The desire-frustration theory of suspense', which discusses how Hitchcock and Truffaut intuitively went against traditional aesthetic models to create suspense, instead, by 'seeding information' that generates a desire on the part of the spectator which can then be more effectively frustrated [23]. Research in criminology, organisational behaviour, and communication have also found frustration to be influenced by psychological [24][25][26][27], sociological [28,29], and socialisation determinants [30][31][32], in addition to specific stimuli [33,34]. There is an important subconscious element to its expression [35], which, crucially, is often displaced from its direct source and thus risks being misdiagnosed.
Thus, according to the psychology literature, frustration must be treated as a naturally endogenous and largely subconscious variable, with psychological, social, experiential, and contextual sources, and multiple emotional, attitudinal, and behavioural outcomes. Conversely, this model of democratic frustration focuses on those very democratic desires and aspirations that remain unfulfilled as much as on the more traditional question of the perception of delivery deficit itself. In that sense, the paradox of citizens' frustration (as opposed to criticality or disengagement) will stem from necessarily strong democratic desire and expectations which will be unfulfilled as opposed to being compatible with a lack of desire or interest. Indeed, frustration requires a powerful desire, and its characterization lies at the heart of understanding the frustration itself and what solutions can be proposed that would reconcile desire and perceived delivery gap.
As democratic frustration necessarily implies that people care and desire democracy and that there is a mismatch between expectation and perceived reality, it assumes the simultaneous existence and variation of both expectation and perceived delivery rather than solely focusing on the latter whilst implicitly assuming the former to be constant. The democratic frustration model can then inform the conceptualisation, causality, and pathologies of frustration and link it to insights from the political behaviour literature on the crisis of participation and populism, that is, on the dimensions and consequences of democratic frustration.

Democratic Frustration and Other Expectations-Delivery Gap Combinations
As we have just seen, most measures of citizen disengagement rely on the concept of dissatisfaction and tend to focus on the perception of democratic 'delivery'. The concept of democratic frustration is centred instead on the mismatch between expectation and delivery deficit. I therefore propose to separate the measure of the democratic expectation and the measure of its perceived shortfall. The concept is essentially interactive in nature. If an expectation of democracy is not desired, then I assume that it will not lead to frustration. If an expectation is perceived as being fulfilled, 'delivered' by the system, this will not be a source of frustration. Instead, democratic frustration is present where the democratic delivery does not meet the expectation, resulting in a democratic delivery deficit that creates democratic frustration.
Democratic frustration is defined here as the interaction between citizens' expectations of their democratic system and their perceptions of a deficit in the democratic delivery of those expectation. In this way, both expectations and assessments of the democratic system can vary together (positive or negative correlations) or independently. The interactive element means that those with higher expectations, will care more about negatively perceived delivery, which thus becomes more likely to create frustration. Consequently, there can be no frustration where the perceived delivery is positive, but also, critically, as per psychological models of frustration, there cannot be frustration where there is no high expectation in the first instance. By contrast, where perceived delivery is negative, expectations come to 'weight' frustration.
By importing and transposing psychological insights the model thus fully integrates the notion of democratic expectations alongside perceived delivery deficit into the measurement of democratic frustration. In turn, those two dimensions enable me to uniquely differentiate between democratic frustration and other patterns of negative feelings about democracy discussed earlier which do not, by contrast, require either the same starting expectations, or the same level of perceived delivery deficit. The resulting conceptual map is shown in Table 1. It suggests that whilst high expectations and high perceived delivery deficit create the target concept of democratic frustration, a similar perception of deficit combined with low expectations would instead constitute cynicism. When the low expectation is combined with medium perceived delivery deficit, it would instead match the definition of apathy. Note that I suggest that one combination (low expectation with low perceived delivery deficit) in that matrix is unrealistic. This is because existing research in psychology and marketing suggests that there is a conditional and asymmetric relationship between experience and expectations. Poor experience may not lead to lower expectations, but high experience will lead to expectations being re-evaluated [36,37]. As a result, someone experiencing consistently high democratic delivery would not be able to retain modest democratic expectations. Instead, their experience would immediately lead them to be more demanding, a bit like someone who having experienced what they consider to be the employment of their dreams would never able to consider that all works are equally boring thereafter. Notes: *Existing research in psychology and marketing confirms that there is a conditional and asymmetric relationship between experience and expectations. Poor experience may not lead to lower expectations, but high experience will lead to expectations being re-evaluated [36,37].
This conceptual model allows expectations and perceived delivery to vary simultaneously and independently. My model also uses democratic theory, representation and behavioural research to theorise the dimensionality of democratic frustration based on the three dimensions derived from a previous pilot study (ideological, political, and institutional).

Dimensions of Democratic Frustration
In this context, specific categories of individuals may be more susceptible to (and differently affected by) frustration than others, and the taxonomy of frustration relates those variations to emotive elements [25] but also the diverse nature of the objects that frustration relates to. If such a thing as democratic frustration exists, it thus becomes essential to consider what could be its dimensions. To do this, we consider the different ways in which citizens are known to use democracy and elections.
There is an abundant body of democratic theory literature which informs us of the various potential functions of elections [38][39][40][41] and criteria of democratic representation [42][43][44], legitimacy [45], and accountability [44,46]. While this literature uncovers multiple discrete components of democracy and potential bases to evaluate its quality, it is possible to emphasise three important dimensions that come recurrently. The first is ideological congruence, which can give citizens an impression that their substantive preferences are represented by the system. The second pertains to the importance of institutional processes, transparency, and effectiveness. Finally, a third is related to the perceptions of political trustworthiness and integrity [4,47].
Based on those three components, we thus derive three possible dimensions of democratic expectations and frustration: • Ideological, which pertains to the existence of a congruent offer to respond to a citizens' substantive preferences, • Institutional, when it comes to the existence of adequate processes capable of effectively and transparently achieving democratic linkage, and • Political, which relates to agency, political personnel, and the integrity of their behaviour, ethos, motivations, and democratic service. Each of those components is first a source of potential democratic "value" for citizens and thus expectations and "democratic desire" in the sense of frustration theories. It is also secondly a potential basis of evaluation and perceived shortfall of delivery. In other words, there can be variation for each of the two constitutive components of democratic frustration as defined in the previous section, within each of those three dimensions

Consequences of Democratic Frustration
Finally, as described at the beginning of this article, the different symptoms of what is globally termed a crisis of democracy vary significantly, both in terms of the human choices that they represent, and of the challenges they raise for democratic systems. Interestingly, however, I argue that those varying responses also seem to quite precisely echo what psychology research has identified as the three main types of human responses to frustration.
At the heart of this model, Sargent [48] describes a sequence of behaviour that features emotion as the central dynamic factor of three key behavioural consequences of frustration: withdrawal, anger, and aggression. Those three potential responses to frustration are confirmed over and again by the psychology literature. Berkowitz [26,29] and Bandura [49] conclude that frustration is indirectly facilitative to emotional responses such as aggression. Anger is deemed similar to the extra punitive response behaviour [25] and a measurable consequence of intolerance of frustration [50]. By contrast, Dollard and Miller [51] connect the frustration of a desire as the source of aggression, leading to the "frustration-aggression hypothesis" developed by Berkowitz [26,29].
The parallel with symptoms of democratic pathologies is rather striking. De-participation (abstention, membership decline) is close to the psychological concept of withdrawal. By contrast, populist and extremist voting can easily be interpreted as symptoms of what psychologists describe as anger, whilst engaging in violent protests or revolutions meets their criteria for aggression. Thus, thinking of current systemic pathologies as symptoms of democratic frustration can explain violent protest but also radicalisation and disengagement. This is all the more relevant that when the source of the frustration is not clear to the subject, violence (regardless of its forms and expressions), anger or withdrawal are typically displaced on an innocent target, especially if the subject feels ignored or humiliated.
We thus highlight three types of behavioural consequences which represent fundamentally different yet predictable responses to potential frustration from a psychological point of view, namely withdrawal, anger and aggression. Those typical patterns of behaviour that political scientists have associated with a sense of democratic unhappiness can thus be "translated" alongside the psychological definitions of withdrawal, anger and aggression quite easily. A typical form of withdrawal could be abstaining in elections, voting for an extremist party matches the definition of an expression of anger, and accepting the possibility of violence to act out on one's unhappiness that of aggression. I consequently derived 6 possible reactive patterns, two each (one "soft" and one "hard") for the three types of psychological reactions considered. I then asked respondents how likely it would be for them to engage in them in view of their potential frustration with a political situation to assess the potential consequences of each dimension of democratic frustration. In addition, I also use "voting for an opposition party" as a control reaction which I described as cooperation (as opposed to withdrawal, anger, and aggression). This operationalisation is summarised in Table 2. Table 2. Typology of the consequences of democratic frustration.

Reaction Level Withdrawal Anger Aggression Soft
Abstention Peaceful demonstration Violent demonstration Hard Leaving the country Extremist vote Revolution The model explores how the different dimensions of democratic frustration may affect different behavioural patterns, notably assessing the circumstances under which democratic frustration will lead to 'pathologies' such as disengagement, extremism or violent protest, defined here as the potential democratic equivalents to the psychological symptoms of withdrawal, anger, and aggression. As my model suggests that the three dimensions of democratic frustration are hierarchized, with ideological frustration being perceived as more "redeemable", political frustration as the most personalised, and institutional frustration the most fundamental, I expect the ideological dimension to result in more "benign" pathologies resulting in withdrawal, the more personalised frustration resulting in corruption and dishonesty of the elite personnel to be more likely to lead to anger, and the more fundamental institutional frustration to be more likely to lead to anti-system reaction, i.e., be more likely to result in aggression. In short, the three dimensions will lead to increasingly severe frustration reactions. The expectation is also that whilst levels of the three dimensions of democratic frustration will affect the six democratic withdrawal, anger, and aggression symptoms, they will not be predictors of cooperation in the form of voting for a regular opposition party.
The model can therefore be summarised as a series of simple hypotheses, which will be tested in this article using data from two surveys (see Appendix A for details of surveys).

•
Hypothesis 1: when conceptualising democratic evaluations as a product of democratic expectations and perceived delivery gap, a large proportion of citizens will appear as democratically frustrated (high expectations, high perceived delivery gap) compared to other combinations of democratic expectations and perceived delivery gap. • Hypothesis 2A: Democratic frustration is deeply held and therefore, it will be stable at the individual and aggregate levels over time 1 . • Nevertheless, aggregate levels of democratic frustration will increase in periods of political crisis (Hypothesis 2B).

•
Hypothesis 3A: High levels of ideological frustration will result in patterns of withdrawal. • Hypothesis 3B: High levels of institutional frustration will result in patterns of anger. • Hypothesis 3C: High levels of political frustration will result in patterns of aggression.

Methodology, Data, and Operationalization Strategy
This article is based on two surveys that were designed to understand the nature of democratic frustration. Using fully representative samples of British citizens, I use this data to test the operationalisation of the concept and its three dimensions and assess their consequences by adapting a psychological model of withdrawal, anger, and aggression. The surveys were conducted during the 2017 UK General Election (including both open-ended and close-ended data) and the 2019 European Parliament elections in the UK. These two surveys followed from a qualitative pilot centred upon the discourse of democratic expectations and perceptions of democratic delivery deficit, which was conducted in France using election diaries during the 2017 Presidential election. An additional UK qualitative pilot enabled me to gain an insight into how people spontaneously think and speak about democratic frustration and informed the question wording in the two surveys analysed in this article.
The surveys used in this article were conducted online with representative samples of the British population of 2000 respondents each. I designed multiple items to measure the proposed interactive concept of democratic frustration. Each pair of items measured a democratic expectation and the evaluation of its perceived delivery deficit (how bad the system is at delivering). They were measured independently using 0-10 scales. The frustration items were operationalised as an interaction between those two components so that frustration item 1=democratic expectation 1 x and thus varies from 0 to 100. Complete unimportance or perfect delivery equate to a level of frustration of 0, whilst utmost importance combined with worst delivery deficit will take the level of frustration to 100.
I created mean indices for each of the three dimensions based on a theoretical expectation as to which dimension of frustration it was measuring based on the existing literature. For instance, the items that referred to the political dimension focused on the critique of the perceived ethos and integrity of the political elite, whilst the institutional items were derived from an opposition to a perceived inadequate system and process. Finally, items centred on the ideological dimension referred to a perceived lack of ideological congruence. I also used factor analysis to confirm the internal coherence of each variable (please see appendix for full model and factor loadings). I then compared the mean and standard deviations for each dimension of democratic frustration for the population as a whole as well as specific sub-categories.
Running the surveys two years apart (similar sampling but no panel element) enables us to assess the aggregate level stability of democratic frustration over time, of its dimensions, and of its basic social and demographic correlates. This produced remarkably similar results over the period, intuitively suggesting (though this must be assessed at the individual level using panel data) that democratic frustration is generally stable at the societal level, in terms of the hierarchy between its three dimensions, and in its social and demographic characteristics.

Scoping Democratic Frustration
The first claim of the model was that we can map key pathologies of democratic crisis using the two axes of democratic expectations and perceived delivery deficit. The idea is that the two dimensions vary independently, so that some citizens could combine different levels of aspirations and perceived delivery with only high desire and with high perceived delivery deficit combining into democratic frustration. I thus distinguish between citizens marked by high and low democratic expectations and high, medium, and low perceived delivery deficit. The asymmetric number of categories stems from the actual distributions of respondents' answers, with much less nuance when it comes to expectations, which tend to be skewed in their distribution but also rarely "average". I then proceed to identify the distribution of respondents along the typology discussed earlier for each of the three dimensions of interest in this article: ideological, institutional, and political. The results are presented in Figure 1. The analysis of the data confirms that democratic frustration can be conceptualised as the interaction between democratic expectations and the perceived deficit in the democratic delivery. The findings confirm a naturally endogenous multidimensional variable and highlights the interaction between democratic expectations and perceptions of systemic shortfall. The figure highlights some interesting and important results. First, it confirms that as theoretically expected, high perceived delivery (i.e., a perception of low deficit) is incompatible with low expectations. That cell is virtually empty. By contrast, the figure also shows that democratic frustration is by far the most common combination of expectations and deficit assessment amongst British citizens confirming Hypothesis 1. It concentrates between 40.1% (ideological dimension) and 54% (political dimension) of total respondents alone. The second largest category are satisfied citizens (high expectations, and low perceived deficit) followed by critical citizens (high expectation and medium perceived deficit), indifference (low expectation, medium deficit) and cynicism (low expectations, high deficit).
We note that the distribution varies among the three dimensions. Frustration is far more dominant a combination on the political dimension than on the other two and in particular, in institutional terms, citizens' expectations tend to be significantly lower than in political and to a lesser extent, ideological ones.
Altogether, this first phase of the test shows not only that democratic frustration exists, and that it would be a mistake not to consider how much citizens care when assessing the nature and effect of perceived democratic shortfalls, but that it is, in fact, by far the most common democratic crisis pathology that we need to analyse and understand at the individual level.

Stability and Evolution of Democratic Frustration over Time
To test the robustness of the concept, I then look at the structure and distribution of the three dimensions of democratic frustration over time using the results from the 2017 and 2019 surveys. As discussed in the methodology section, the two surveys did not constitute a panel so it is impossible to assess stability at the individual level, but as they used the exact same measurement and samples recruited in the exact same way by the same polling company, it is possible to compare them at the aggregate level. For that section, I use the mean indices of the three dimensions of frustration rather than the factor scores because of the intuitive nature of the 0-100 scales and because factor scores using the two different surveys would not be fully comparable as they would be based on the idiosyncratic data structure within each survey. After confirming similar structure of the data, I looked at distributions and evolution over time. The results are shown in Figure 2.
First, the figure confirms a certain stability in the hierarchy between the three dimensions of democratic frustration at the aggregate level, consistently with Hypothesis 2A (though the data available does not permit to assess the more important question of stability at the individual level). The political dimension registering by far the highest scores compared to institutional and ideological frustration, which is lowest in both cases. The levels of standard deviation are also slightly higher for the political dimension of frustration. The second important finding is that on the whole, average levels of democratic frustration seem to have increased substantially in the UK between 2017 and 2019. Differences in means are all statistically significant using ANOVA tests for the ideological (p = 0.015), institutional (p = 0.007) and political dimensions (p = 0.000). This is thus particularly true of the political dimension which has worsened substantially faster than for the ideological and institutional dimensions. On the whole, political democratic frustration has increased by nearly five points during the period (+ 4.56), which is roughly twice that of institutional (+ 2.38) and ideological (+ 1.97) dimensions of frustration. It is worth remembering that the evolution coincides with a dramatic period in British politics, whereby the anger of many citizens towards politicians in the context of the Brexit crisis seems to have reached unprecedented levels. The specific blaming of politicians and political elites for the continuing Brexit chaos, worsening political and societal divisions, and perceptions of lack of control may also explain why the political dimension of democratic frustration seems to be affected so much more seriously than its ideological and institutional counterparts. This would be consistent with the expectations of Hypothesis 2B, but we have no way of compellingly assessing causality based on only two time points.

Behavioural Consequences of Democratic Frustration
Finally, the conceptual model proposes that democratic frustration can trigger several emotional and beyond political behavioural reactions. As discussed earlier, I operationalised six potential behavioural reactions in reaction to democratic frustration ranging from abstention (withdrawal) to extremist voting (anger) and violent protest (aggression). In political behavioural terms, as pathologies of democratic crisis, those consequences represent increasing levels of radicalism, and my model thus expects the three hierarchized dimensions of frustration to lead to these increasingly severe consequences.
In this way, democratic frustration may lead to a sense of injustice or inadequacy and exclusion. This second causal component of the model considers democratic frustration as an independent variable, and its impact on behavioural consequences. As a reminder, we suggested that a high score on the ideological dimension could result in feelings of representative mismatch and thus be more likely to result in democratic withdrawal. By contrast, higher levels of political frustration are more likely to result in a sense of unfairness and personal betrayal by political elites and beyond in protest attitudes and behaviour, and institutional frustration by a sense of fundamental democratic failure legitimising anti-system reactions. Those behavioural reactions could present real threats to democratic practice and underline the essential need to understand expectations, perceptions of democratic dysfunction and the associated pathologies of frustration. I test the six models of expected consequences of democratic frustration, as well as the control model of cooperative reaction (i.e., voting for mainstream opposition parties) using OLS regressions with the 2019 European Parliament elections data. The three democratic frustration dimensional predictors are the factor scores of ideological, institutional, and political dimensions. Note that because we use factor scores, signs are largely irrelevant due to factor indeterminacy as long as the variables behave coherently. We also note some multicollinearity between the frustration dimensions (obviously, the three dimensions of frustration are naturally correlated) which imposes some caution in the way we analyse the relative predictive strength of the three dimensions when they are jointly relevant. The regressions also include basic social and demographic controls as well as psychological controls for hostility and sympathy towards alternative voters shown by Harrison (2017, 2020) to have some impact on attitudes towards democracy. The results are presented in Figure 3 and regressions in Appendix A. The findings-which as discussed, must be interpreted with caution-are mixed but overall intriguing. They suggest that indeed, there is some level of differentiation on the impact of the three dimensions of frustration on different democratic reactions. First, none of the three dimensions of democratic frustration explains cooperative reactions in the form of support for opposition parties. This is noteworthy, and indeed, the overall model of cooperation has a very low R 2 of 0.03 almost exclusively attributable to control variables in the model.
The ideological dimension of frustration is the most differentiated. It only explains abstention, which we conceptualise as the archetypical withdrawal reaction, which none of the two other dimensions of frustration explain. This matches Hypothesis 3A. Total R 2 for the model is 0.09.
By contrast, institutional frustration is the only one with a clear effect on both soft and hard anger reactions (engaging in peaceful protest and voting for radical or extremist parties). This, in turn, confirms Hypothesis 3B. Political frustration has a limited effect on the latter but only significant at 0.05 level, which is not very convincing with a sample of over 2000, and no effect on peaceful demonstrations. Ideological frustration does not explain either. Those two models have a limited R 2 of 0.07 and 0.08 respectively. On balance, it seems that when citizens' frustration stems from a perception of failure by the political elites as a whole, they include radical parties' politicians and civil society groups in their assessment, which makes a radical vote or peaceful demonstration an unlikely answer to political frustration.
By contrast, when it comes to soft and hard reactions of aggression (which, in effect, include a potential for violence), they are jointly explained by the institutional and political dimensions of frustration, which does not fully uphold Hypothesis 3C (which expected a predominant impact of the political dimension of frustration, whilst the institutional dimension effectively proves as important). This is where the risks of multicollinearity make it hard to ascribe variance with precision to one or the other dimensions, but where both seem to lead to citizens considering somewhat "desperate" measures. When citizens' frustration is based on the desire for both functioning institutions and decent politicians to be failed by major perceived delivery gaps is when they consider all other option but aggression to potentially having been exhausted. As mentioned, however, this finding in particular needs to be analysed with caution and further tests will be needed in other contexts Those two models have larger adjusted R 2 , with 16% of total variance explained by the models.
Finally, modelling reasons why citizens would consider leaving the country would seem to betray our theoretical expectations as per Hypothesis 3A, if it is considered a hard version of withdrawal. Indeed, a willingness to leave the country when feeling frustrated is instead mostly explained by institutional frustration, with limited effects of the political dimension, making it more similar to anger reactions. The total R 2 for that model is 0.14. In retrospect, this is probably explained by the radical and active nature of expatriating as a reaction to democratic frustration. While exiling oneself for political realities is quite literally a form of withdrawal in that it extracts a citizen from the frustrating situation that (s)he is trying to avoid, it does not constitute withdrawal in the psychological sense of the term as that would pertain to an eminently passive reaction. As a result, that particular item probably has validity issues as a form of withdrawal, which explains why it tends to resonate in citizens' minds as a form of radical anger towards their country of citizenship.
Ultimately, and warning caution once again, the model suggests that the three dimensions lead to possibly divergent types of democratic reactions on the part of citizens. These findings confirm the importance of the dimensionality of democratic frustration, showing that feeling democratically frustrated in the ways highlighted can lead to different types of negative reactions, whilst not explaining why citizens avail themselves of the power of accountability by voting for pro-system alternatives. One significant limit is the relative permeability between the political and institutional dimensions in the sense that ultimately, citizens who are consistently frustrated by the behaviour and standards of their political elites will soon feel frustrated by the institutional system which has produced them in the first place. This may impose further ways of differentiating between those dimensions and their specific consequences. With those limits notwithstanding, it looks like different dimensions of frustration have differentiated impacts on some but not others of the democratic embodiments of withdrawal (abstention), anger (peaceful demonstrations, voting for radical parties) or aggression (resorting to violent demonstration or joining a revolution). Our model of considering leaving the country as a form of withdrawal is also explicitly disproved and citizens' reactions suggest that they think of it far more as an expression of democratic anger.
Each of these three modes of reaction represents important challenges for democratic systems, which may even be considered threats to the entire fabric of democratic processes. In that sense, understanding how modes of frustration result in each of them and in different degrees is critical when it comes to the possible political and institutional resolution mechanisms that political systems could use to try and address their citizens' sense of frustration.

Conclusions: Understanding Desire to Understand Disappointment
Democratic frustration may represent a missing link with neighbouring concepts. Most existing models (disengagement, dissatisfaction, distrust, etc.) focus on how voters position themselves visà-vis the system, i.e., a judgemental approach. They often treat citizens' own democratic desires as either constant, or at best as control. Yet, in a frustration model, perceived dissatisfaction is endogenous (and not exogenous) and caused by them in their interaction with a perceived delivery deficit.
Further investigation would need to be completed to test whether the dimensions of democratic frustration may also represent different layers or depths of democratic frustration and may thus be structured as a Mokken scale. Understanding whether the dimensions are covariant or structured by a Mokken scale is critical to avoid error in the measurement of the dependent variable. It is anticipated that the ideological dimension would be the shallowest layer, related to the failure to meet democratically expected levels of congruence, whilst the political element may represent an intermediate level in which political elites fail to meet citizens' expectations. The institutional dimension is expected to be the most fundamental layer as it implies a chronic disapproval of institutional procedures.
A second limitation of this article is that it only deals with one case study, the UK, which is both institutionally specific in its strong majoritarianism, and politically idiosyncratic given the unprecedented ways in which the Brexit divide is redefining lines of political discussion and behaviour in the country. By contrast, institutional designs are likely to influence democratic expectations as well as perceptions. It is of little use to consider that citizens would hold monolithic expectations of what democracy should deliver it terms of representation, accountability or indeed what it symbolically means to them as citizens to participate in a democratic system. Instead, in countries such as the Netherlands or Belgium which use Proportional Representation based on large constituencies, the democratic system may make it easier for citizens to find parties ideologically meeting their expectations, and by contrast, may give citizens less control of Government formation as it is largely negotiated by the parties themselves and less ability to punish incumbents that have disappointed them. Therefore, substantive democratic expectations may differ in proportional and majoritarian democracies, just as they may differ between emerging and consolidated democracies.
This article aims to take the concept of democratic frustration literally by highlighting how psychological models of frustration make it necessary to look beyond perceptions of democratic delivery gaps to capture and integrate the concept of democratic desire which may vary significantly across individuals. This offers new ways to read what democratic unhappiness may mean and how it emerges, but also, how it can be answered by institutional systems and political elites. It lays the ground for a different understanding of the impact of democratic frustration on emotions, attitudes, and behaviour. Crucially, it also highlights how democratic frustration may be a missing link to explain how and why unhappy citizens end up engaging in radically different behavioural phenomena ranging from disengagement and abstention to support for extremist parties or protest, or widespread feelings of exclusion and apathy mirroring the psychological model of withdrawal, anger, and aggression reactions. Those potential reactions are fundamentally different individually and psychologically, and of vastly divergent consequences for democratic systems. The possibility that different dimensions of democratic frustration, despite some potential intercorrelation may finally explain why such different behavioural consequences are reached would be a major improvement to our understanding of democratic pathologies and their consequences.

Survey Questions:
Q8: How important is it that British democracy should make you feel the following ways? Please answer using a scale from 0 to 10 where 0 means not important at all and 10 means extremely important.
[0 = not important at all… 10 = extremely important] Row 1. Represented by people whose ideas are close to mine 2. That politicians are genuinely interested in listening to citizens 3. That politicians are genuinely more interested in what is best for citizens rather than what is best for themselves 4. That I have a genuine choice between a range of political alternatives 5. That politicians are transparent and honest 6. That politicians consider the long-term interest of citizens 7. That I am involved in the democratic process 8. That if citizens are unhappy with the way they are governed, they can get rid of leaders who have not performed well 9. That I feel respected by politicians Q9: And how good is British democracy at making you feel the following ways at the moment? Please use the same scale from 0 to 10 where 0 means not good at all and 10 means extremely good.
[0 = not good at all… 10: extremely good] Row 1. Represented by people whose ideas are close to mine 2. That politicians are genuinely interested in listening to citizens 3. That politicians are genuinely more interested in what is best for citizens rather than what is best for themselves 4. That I have a genuine choice between a range of political alternatives 5. That politicians are transparent and honest 6. That politicians consider the long-term interest of citizens 7. That I am involved in the democratic process 8. That if citizens are unhappy with the way they are governed, they can get rid of leaders who have not performed well 9. That I feel respected by politicians

Factor Loadings
Below are the full factor loadings for the three individual factors generated based on the 2, 3, and 4 items measuring ideological, institutional, and political dimensions for the purposes of the multivariate regressions. 1). Ideological factor: As there are only two items measuring ideology, measurement implicitly functions as a mean index. However, eigenvalue is a useful measure to assess the items covariance between the ideological items compared to the other two dimensions, In this case, eigenvalue is 1.64 explaining 81.99% of total variance and the component matrix score is 0.91 for both items. 2