Who is willing to participate? Dissatisfied democrats, stealth democrats and populists in the United Kingdom


Address for correspondence: Paul Webb, Department of Politics, University of Sussex, Falmer, Brighton BN1 9SN, UK. E-mail: p.webb@sussex.ac.uk


This article demonstrates that two quite distinctive types of political disaffection – ‘dissatisfied democratic’ and ‘stealth democratic’ – exist among British citizens, with the former being more prevalent. While both types manifest low trust in political elites, the dissatisfied democrat is politically interested, efficacious and desires greater political participation, while the contrary is generally true of the stealth democrat. However, stealth democrats are favourably disposed towards direct democracy, which can be attributed to the populist nature of stealth democratic attitudes. Even so, when given the opportunity to take part in a national referendum, neither stealth democrats nor dissatisfied democrats showed much inclination to vote.


The world's established democracies are replete with talk of political alienation and apathy on the part of citizens. The blame for this state of affairs has been laid at the door of various culprits, including politicians, parties and other institutions and processes of representative democracy, the media and occasionally even citizens themselves. Putative solutions to the problem tend to depend on where critics apportion the blame: those who regard the shortcomings of representative democracy as central to the problem tend to argue that the answer lies in institutional innovations that will facilitate new forms of political participation for ordinary citizens; those who prefer to blame the impact of the media, consumerist culture or the failure of the public to understand the nature of politics, see more potential in better media regulation or improved civic education. This article seeks to shed light on the viability of proposals for greater participation by analysing the attitudes of citizens towards mainstream forms of participation, deliberative democracy and referendums in the United Kingdom. It argues that there is an important distinction between two quite different types of citizen disaffection, and that one of these – which has been referred to as ‘stealth democratic’ elsewhere in the literature – is essentially populist in orientation. While reforms designed to enhance participation may meet the aspirations of some citizens, it is not so clear that they will work for those of a populist stealth democratic mind set. This leaves a considerable challenge for researchers and institutional designers.

The intellectual context

There is now a considerable body of evidence attesting to popular dissatisfaction with politics in the world's established liberal democracies. In the British case, a crescendo of complaint about politicians erupted in the light of media revelations about the extravagant expense claims made by some MPs in 2009 (Kelso 2010), though more general evidence of political alienation has been apparent for far longer. As in many established democracies, anti-party sentiment is widespread and public trust in politicians is low.

Various explanations for the apparent spread of political disaffection can be found in the literature. For instance, Dalton (2004) emphasises two developments common to the advanced industrial democracies. The first is rising expectations of government, especially among younger, better educated, more affluent and postmaterialist citizens. While these are the very groups that have most directly benefited from the spread of affluence, their expectations have increased the most, as has their tendency to criticise political elites, institutions and processes. Yet they do not represent a threat to democracy per se; on the contrary, these ‘dissatisfied democrats’ are driven by a passion for the democratic creed that fosters disillusionment with the way current political processes operate.

A second general source of disaffection is the complexity of contemporary political agendas and mobilisation. New debates over environmental quality, social norms, lifestyle choices, multiculturalism, and other social and cultural issues have led to the triumph of interest articulation over interest aggregation. In such fluid, multidimensional policy space it is very difficult for governments to satisfy most of the people most of the time. Moreover, the mobilisation of ‘dissatisfied democrats’ makes aggregation more difficult still and provokes a demand for reform that goes beyond tinkering with the core institutions of representative democracy (parties, elections, parliaments) to an increase in direct public involvement in the political process. This in turn threatens to exacerbate the imbalance between the ever-growing clamour of articulated interests and the need for institutions that can effectively channel divergent demands into coherent and effective policy programmes.

In fact, there is widespread interest in participatory democracy in general, and in deliberative democracy in particular (see Delli Carpini et al. (2004) and Pateman (2012) for useful reviews of the relevant literature). These are often favoured as solutions to the problem of political alienation, and enthusiasm extends beyond political theorists: in Britain, the last Labour government proposed the introduction of citizen juries and participatory budgeting in local government (Ministry of Justice 2007; Department of Communities and Local Government 2008). At European level, too, there is significant official interest in the potential of participation through e-democracy (Council of Europe 2009). However, there is of course a long tradition of democratic theory, going back to Schumpeter and Weber, which is generally sceptical of the supposed benefits of intensive participation. The most striking contribution in recent years has been made by John Hibbing and Elizabeth Theiss-Morse in their research on American voters (Hibbing & Theiss-Morse 2002). They sternly rebuff the participationists' claims, arguing that people:

do not want to make political decisions themselves; they do not want to provide much input to those who are assigned to make these decisions; and they would rather not know the details of the decision-making process. … This does not mean that people think no mechanism for government accountability is necessary; they just do not want the mechanism to come into play except in unusual circumstances.

Hibbing and Theiss-Morse summarise the orientations of American citizens as a preference for a ‘stealth’ arrangement, whereby citizens know that democracy exists, but expect it to be barely visible on a routine basis. They criticise both the naïveté of popular attitudes towards politics, and the insistence of those who see participatory democracy as the solution to current discontents. The alleged benefits of participatory – especially deliberative – democracy are portrayed by Hibbing and Theiss-Morse as ‘wishful thinking’, and they point out that research tends to reveal that it only works under very limited conditions. ‘Deliberation will not work in the real world of politics where people are different and where tough, zero-sum decisions must be made’ (Hibbing & Theiss-Morse 2002: 207). They cite evidence to debunk three of the major claims of the participationists: that deliberative democracy produces better decision making (actually, the most powerful personalities often dominate, whether or not they are the best-informed or most rational); that it enhances the legitimacy of the political system (rather, face-to-face conflict just exacerbates people's anger and resentment [Morrell 1999]); and that it leads to personal development (again, it just exacerbates the sense of powerlessness, inadequacy and marginalisation of the weakest participants). Indeed, Diana Mutz (2006) has gone so far as to argue that high-intensity deliberation around political differences can actually reduce the inclination of many people to participate in politics because of the desire to avoid conflict.

That said, recent research from the United States now suggests that the pessimism of writers like Hibbing, Theiss-Morse and Mutz may be exaggerated. Using a blend of experimental and survey designs, Neblo et al. (2010) have found that willingness to deliberate in the United States is much more widespread than expected, and that it is precisely the demographic groups that are least likely to participate in traditional partisan politics – and whom we would therefore expect to express the stealth democracy perspective – who are actually most interested in deliberative participation. However, these findings depend crucially on the particular form of deliberation between citizens and elected representatives that is implemented.

Similarly, Bengtsson and Mattila (2009) have found that in Finland people with less education, with less political knowledge and those who feel that the political system does not respond to their needs – again, those we might expect to have ‘stealth democratic’ attitudes – are actually most likely to want greater use of direct democracy in their political system. Perhaps this is because referendum democracy is not at all the same thing as deliberative democracy or high-intensity participation. On the contrary, it has often been regarded as compatible with a populist outlook in which charismatic leaders have a direct and even demagogic relationship with the masses, and thereby largely bypass the institutions of representative democracy. This is a point to which we shall return in due course.

This article seeks to test a major hypothesis that can be derived from this literature. Thus, H1 holds that there are two quite different types of attitude among those who are ‘disaffected’ with politics: a ‘dissatisfied democratic’ orientation (associated with citizens who are higher social status, well-educated, politically interested and efficacious devotees of political engagement and debate), and a ‘stealth democratic’ orientation (associated with those of lower social status, who are less well-educated, less politically interested and efficacious individuals who recoil from political debate and have little desire to participate in political activity beyond the minimum necessary to hold untrustworthy elites in check). The ‘dissatisfied democratic’ stereotype – younger, better educated, more affluent, postmaterialist citizens who are ‘passionate for the democratic creed’ – is derived directly from Dalton's (2004) work. The ‘stealth democratic’ stereotype is derived largely from Hibbing and Theiss-Morse, supplemented by a long tradition of research that associates lack of political interest or efficacy with those of lower social status and education.

If this simple hypothesis is correct, then it could carry significant implications for debates about political reform, for high-intensity forms of participation might at best only be effective in respect of those who fit the dissatisfied democratic profile, but could be counter-productive with respect to those of stealth democratic orientation. While the former may chafe at the participatory limitations of traditional forms of representative democracy and might thrive in a more participative environment, the latter could actually be more vulnerable to political marginalisation, for they are less likely to take to active engagement. They have traditionally depended on parties as key interlocutors and tribunes of their social group interests, but their parties (typically of social-democratic or labour hue) have often lost this role through strategic adaptation. Without representative parties that express their social identities and serve as communities of political learning, these citizens retreat into a disaffected and alienated take on politics. These feelings will only be exacerbated by evidence of ‘feather-bedding’ by self-interested politicians and parties (Hibbing & Theiss-Morse 2002: 121–124).

That said, proponents of high-intensity forms of participation such as deliberative democracy might acknowledge the current empirical reality of the unwillingness of many citizens, especially those of lower socioeconomic class and educational experience, to countenance political engagement, while maintaining that this owes much to the mobilisational shortcomings of existing systems of representative democracy. It is only when such citizens, the argument goes, are actually presented with the opportunity to take part in meaningful political deliberation that they develop greater political interest, knowledge and a new sense of political efficacy. The deliberative experience itself generates greater civic commitment and activism through its educative impact. While the evidence in support of such a contention is somewhat mixed (Delli Carpini et al. 2004), there is undoubtedly some that is favourable: ‘ordinary citizens, given some information and time for discussion in groups of diverse opinions, are quite capable of understanding complex, and sometimes technical, issues and reaching pertinent conclusions about significant public matters’ (Pateman 2012: 9; see also Fournier et al. (2011) for an interesting recent illustration of this point). It is not the purpose of this article to engage seriously with such claims, since they require far more detailed attention than it is possible to provide here. However, it is important to have explained them in order to establish the wider context in which the research reported here is situated. This article, then, constitutes an attempt to test H1 in a British context through the analysis of a specially commissioned dataset.

Measuring stealth democracy and dissatisfied democracy

The data source for analysis is an Internet survey of British citizens conducted in the summer of 2011. This produced a representative quota sample of the adult population weighted by the major demographic factors and electoral turnout. Weighting by turnout was done in order to minimise the risk of over-sampling those who might be inclined to participate in political activity. Research suggests that results from national surveys based on Internet quota samples compare well with probability samples based on in-person interviews in Britain. Sanders et al. (2007) found that a comparison of the two methods revealed few statistically significant differences between the coefficients they generated when used to test parallel models of voter turnout and party preference. Assuming that this holds for the data on which this article is based (and the data were gathered by the same polling company that Sanders used), it suggests that the relationships between the dependent and predictor variables in the models reported below can be trusted as well as if they came from a probability sample.1

The presence of stealth and dissatisfied democratic orientations in the British sample is gauged through a number of relevant attitudinal indicators. The first port of call in this regard must be the battery of measures originally devised by Hibbing and Theiss-Morse themselves in order to tap stealth democracy. Respondents were asked to indicate the strength of their agreement or disagreement with the following statements:

  • Politicians would help the country more if they would stop talking and just take action on important problems.
  • What people call ‘compromise’ in politics is really just selling out one's principles.
  • Government would run better if decisions were left up to non-elected independent experts.
  • Government would run better if decisions were left up to successful business people.

The results reported in Table 1a reveal fairly widespread, though not overwhelming, agreement with these views: on average, 36.8 per cent of the sample agreed/tended to agree, while only 21.8 per cent disagreed/tended to disagree, and 41.6 per cent were undecided. Support is particularly pronounced for the claim that politicians would be more helpful if they stopped talking and just took action on important problems (with 60.4 per cent in favour), while respondents are most wary of the idea that decision making would be best left to technocratic experts (only 19.5 per cent agreeing).

Table 1a. Support for stealth democratic attitudes in the United Kingdom: Hibbing/Theiss-Morse indicators
 Politicians would help the country more if they would just take action and stop talking on important issuesWhat people call compromise in politics is really just selling out one's principlesGovernment would run better if left up to independent non-elected expertsGovernment would run better if left up to successful business peopleAverage
  1. Notes: All figures are percentages. N = 1,355.
Strongly agree19.410.
Tend to agree41.028.816.623.927.6
Neither agree/nor disagree29.739.052.744.841.6
Tend to disagree8.817.916.014.314.3
Strongly disagree1.23.911.912.97.5

In order to subject our data to more rigorous multivariate analysis, these indicators need to be combined into a single additive scale. However, while they are undoubtedly important for tapping a popular sense of exasperation with the way in which elected political elites appear to operate in democratic systems, on their own these items produce a rather low reliability score (alpha = 0.574), and in any case, I would argue, do not capture the full nature of the ‘stealth democracy’ idea. In particular, they fail to take account of the widespread human desire to avoid the conflict and noisy disputation that is intrinsic to politics. Stealth democrats, with their low interest in politics and disdain for politicians, can be expected to be generally conflict-avoidant, while dissatisfied democrats, with their more positive take on political engagement and debate, are more likely to embrace the inevitability of conflict. As Hibbing and Theiss-Morse (2002: 7) say in their original study of stealth democracy in the United States:

People believe that Americans all have the same basic goals, and they are consequently turned off by political debate and deal-making that presuppose an absence of consensus. People believe these activities would be unnecessary if decision-makers were in tune with the (consensual) public interest rather than with cacophonous special interests.

Hibbing and Theiss-Morse (2002: 143) directly connect stealth democracy with conflict avoidance in stating that ‘supporters of stealth democracy believe debate is not necessary or helpful … and they are willing to turn decision-making over to entities that are largely, perhaps completely, unaccountable but that promise efficiency and an absence of contention’. I would argue that none of Hibbing and Theiss-Morse's indicators quite capture the conflict-avoidant dimension of stealth democracy; ‘Politicians would help the country more if they would stop talking and just take action’ comes closest, but any sense of conflict-avoidance that might lie at the core of this view is at best highly implicit. Consequently, Table 1b reports a number of survey items designed by Diana Mutz (2006) to measure conflict-avoidance; these will eventually be incorporated into our stealth democracy index.

Table 1b. Support for stealth democratic attitudes in the United Kingdom: Conflict-avoidance indicators
 I would rather not justify my political beliefs to someone who disagrees with meI do not take it personally when someone disagrees with my political viewsWhen I'm in a group, I often go along with what the majority decides is best, even if it is not what I want personallyWhen I'm in a group, I stand my ground even if everyone else disagrees with meWhen people argue about politics, I often feel uncomfortableI have no problem revealing my political beliefs, even to someone who would disagree with meAverage
  1. Note: All figures are percentages. N = 1,355. * Figures reported here represent the percentage of the sample strongly agreeing (or otherwise) with the conflict-avoidant option. For example, while 1.8 per cent of respondents strongly disagree with the statement ‘When I'm in a group, I stand my ground even if everyone else disagrees with me’, they are reported here under ‘Strongly agree’ because they are the cases who strongly agree with a conflict-avoidant stance.
Strongly agree*
Tend to agree17.46.611.46.820.65.311.4
Neither agree/nor disagree42.425.131.933.235.228.832.8
Tend to disagree27.249.637.143.527.241.537.7
Strongly disagree10.117.718.314.714.422.816.3

In fact, on the whole our respondents do not seem especially concerned to avoid conflict: on average, 13.3 per cent choose the conflict-avoidance options, while 54 per cent prefer the contrary positions. This is most obvious when it comes to people declaring that they do not take it personally when someone disagrees with their political views, that they have no problem revealing their beliefs, even to those who might disagree with them, and that they would be willing to stand their ground even if everyone else in a group disagreed with them. On the other hand, a fifth or more admit that political argument makes them uncomfortable, and that they would rather not justify their beliefs to those who disagree with them. When combined with responses to Hibbing and Theiss-Morse's original items, these conflict-avoidance indicators form a single additive scale that is coded so as to have a theoretical range running from 10 (low stealth democratic orientation) to 50 (high stealth democratic orientation), with a scale mean of 27.4, suggesting that our sample is actually not especially stealth democratic in outlook. The reliability of the index is confirmed by a more acceptable alpha score of 0.618. Thus, we have a scale which captures the incomprehension and frustration of some people at a political process that appears to combine elements of self-regarding disputation and conflict, with sub-optimal back-door deal-making and compromise.

Dissatisfied democracy is gauged through a ‘sunshine democracy’ index, derived from the work of Neblo et al. (2010). Originally designed as a methodological counter to the original stealth index in order to negate the risk of acquiescence-bias among survey respondents, Neblo and his colleagues explain that the sunshine democracy items have the additional virtue of tapping how respondents ‘think that a representative democracy should work in principle, whereas the stealth items tap what they would settle for as a first step away from what they perceive as the corrupt status quo’ (Neblo et al. 2010: 573). The sunshine democracy questions are worded as follows:

  • Openness to other people's views and a willingness to compromise are important for politics in a country as diverse as ours.
  • It is important for elected officials to discuss and debate things thoroughly before making major policy changes.
  • In a democracy like ours, there are some important differences between how government should be run and how a business should be managed.
  • It is important for the people and their elected representatives to have the final say in running government, rather than leaving it up to unelected experts.

Table 2 reports the distributions of respondents' attitudes on these items. Plainly, they are overwhelmingly in favour of all of these statements; on average, 70.4 per cent strongly agree/tend to agree with them. Support is particularly strong (79.6 per cent in favour) for the view that elected officials should discuss and debate things thoroughly before making major policy decisions, which is perhaps a little ironic given how many people also feel that ‘politicians would help the country more if they would just take action and stop talking on important issues’. Clearly, people do not always perceive the apparent contradictions in holding such views simultaneously.

Table 2. Support for sunshine democratic attitudes in the United Kingdom
 It is important for elected officials to discuss and debate things thoroughly before making major policy decisionsOpenness to other people's views and a willingness to compromise are important for politics in a country as diverse as oursIt is important for the people and their elected representatives to have the final say in running government, rather than leaving it up to unelected expertsIn a democracy like ours, there are some important differences between how government should be run and how a business should be managedAverage
  1. Notes: All figures are percentages. N = 1,355.
Strongly agree39.224.021.519.926.2
Tend to agree40.448.842.045.544.2
Neither agree/nor disagree19.424.031.629.626.2
Tend to disagree0.
Strongly disagree0.

Responses are coded so that the additive scale runs from 4 (low ‘sunshine’ orientation) to 20 (high ‘sunshine’ orientation), with a sample mean of 15.7 and an alpha reliability score of 0.762. The sunshine democracy index can be regarded as a valid surrogate for a dissatisfied democracy orientation because people who score highly on this are plainly enthusiasts for democratic values; but at the same time they exhibit by low trust in the political class – a condition which is met in the empirical analysis that follows by the simple device of selecting only those respondents for analysis who report low levels of trust for politicians in general. (Indeed, low political trust is a feature shared by both dissatisfied democratic and stealth democratic attitudes.)

The capacity of the stealth and sunshine indices to capture different underlying orientations is attested to by a bivariate correlation of –0.301 (sig. = 0.000). This significant inverse relationship offers basic support for the argument that stealth democratic and dissatisfied democratic orientations are fundamentally different from each other. However, it is clearly not a perfectly inverse association. This is illustrated by a simple cross-tabulation in which each scale is split halfway along into low and high categories; this reveals that 70 per cent of the sample conforms exactly to the expected combinations of either a high stealth/low dissatisfied democratic orientation (13.3 per cent of total) or a high dissatisfied/low stealth democratic orientation (56.8 per cent). However, while very few respondents (just 4.6 per cent in fact) manage to return low scores on both indices, a significant minority (25.3 per cent) do score highly on both. Overall, there are more respondents (61.4 per cent) who can be placed in the high dissatisfied democracy category than in the high stealth democracy category (38.6 per cent), so it can be inferred that the former type is more widespread among British citizens.

In pondering why the negative relationship between stealth and dissatisfied democracy is not higher, it is helpful to return once more to Neblo's claim that the sunshine democracy items tap respondents' feelings about how democracy should work in principle, whereas the stealth indicators reveal their preferences for ‘a first step away’ from what strikes them as the disappointing reality. Thus, in an ideal world of democracy, politicians would be open to the diversity of voices in society, debating all views thoroughly, and ultimately finding the rational and fair compromises that embody the popular will. In the rather more messy and unsatisfying real world of democracy, however, things often do not work out so perfectly; this being so, many respondents, while still aspiring to the ideal notions of sunshine democracy, are prone to opt for the pragmatic alternative – a prescription that disdains political compromise as unprincipled selling-out, and seeks to cut through the endless clamour of apparently self-interested debate by recourse to the disinterested rule of technocrats or the managerial efficiency of business leaders. This would explain how it is possible for some respondents to simultaneously adhere to stealth and sunshine orientations.

What of the main demographic and attitudinal correlates of these indices? Ordinary least squares (OLS) regression analysis reveals this, as reported in Table 3. This model incorporates key social background factors (socioeconomic grade, educational experience, age and gender), along with measures of political interest and efficacy and basic ideological orientation (left–right, social liberalism–authoritarianism). One can immediately see that, where they are significant, each of the independent variables connects with the two dependent variables in inverse ways; age, education, political interest and efficacy all relate negatively to stealth democracy, but positively to dissatisfied democracy. Thus, as expected, those who score highly on stealth democracy tend to have lower educational attainment, and lack interest in politics or a strong sense of political efficacy; they also tend to be young, female, of lower social grade and to hold socially authoritarian views. By contrast, dissatisfied democrats are likely to be older, better educated, politically interested and efficacious, and to hold leftist views. All of this is consistent with the expectations set out in H1.2

Table 3. OLS models of stealth democracy and dissatisfied (‘sunshine’) democracy
Independent variableStandardised coefficients (significance)
Stealth democracyDissatisfied democracy
  1. Note: n.s. = non-significant.
Left–right ideologyn.s.−0.206 (0.000)
Liberty–authority ideology0.156 (0.000)n.s.
Age−0.086 (0.006)0.090 (0.005)
Political interest−0.246 (0.000)0.269 (0.000)
Political efficacy−0.164 (0.000)0.112 (0.003)
Gender0.067 (0.030)n.s.
Social grade−0.060 (0.063)n.s.
Education−0.079 (0.017)0.111 (0.001)
Adjusted R2 for model0.227 (N = 908)0.196 (N = 908)

Stealth democracy, dissatisfied democracy and political participation

How do these orientations vary with attitudes toward political participation? As a general rule, we would expect those scoring highly on dissatisfied democracy, with their ‘passion for the democratic creed’, to be far keener on all forms of political participation than those who register strongly on stealth democracy, although the latter should not be thought of as entirely averse to participation: they might feel driven to participate if they regard it as a means of keeping the political elites that they mistrust ‘honest’. But what do we mean by ‘participation'? In this article, I distinguish between four main variants: deliberation, referendum democracy, participation in the party-electoral arena, and participation in the non-party arena.

It is common for theorists of deliberative democracy to argue that it is not simply synonymous with ‘participatory democracy’; rather, it is a particular form of participation that entails the active engagement of participants in reasoned political discussion. As Carole Pateman (2012: 8) says of deliberative democracy, ‘individuals should always be prepared to defend their moral and political arguments and claims with reasons, and be prepared to deliberate with others about the reasons they provide’. For Jane Mansbridge et al. (2012: 4–5), a deliberative political system ‘encompasses a talk-based approach to political conflict and problem-solving – through arguing, demonstrating, expressing and persuading’. It is important to distinguish deliberative processes – of which there are many variants, including deliberative polling, citizen juries and citizen assemblies – from other forms of democracy that entail popular participation, such as referendums. Direct democracy by referendum is a particular type in its own right, in part because of its significance for populist and demagogic forms of politics. Referendums can be a way of bypassing the normal channels of representative politics without requiring much active involvement by ordinary citizens beyond a simple yes or no vote. It is a favourite device of anti-establishment populist organisations (including UKIP and the BNP in the UK),3 which contend that mainstream parties somehow betray the people they are supposed to represent. By contrast, there are many forms of participation that we might regard as part and parcel of the orthodox repertoire of activity in representative democracy, and in this article, I shall distinguish between those which relate to party and electoral politics (voting, joining parties, campaigning on behalf of election candidates, etc.), and other non-party forms of participation (such as joining or donating to interest groups, taking part in demonstrations, signing petitions, etc.).

In this section, each of these four variants of political participation – deliberative democracy, referendum democracy, party-electoral and non-party participation – is tested as a dependent variable. More precisely, respondents' hypothetical willingness to participate in each of these ways is regressed on explanatory models that incorporate stealth democratic orientation, dissatisfied democratic orientation, political efficacy and interest, and the main demographic variables. The purpose is to complete the testing of H1.

To this point in the argument, I have sought to demonstrate that stealth democratic and dissatisfied democratic orientations do indeed exist in the electorate, but have done so without checking if these basic attitudinal profiles have their corollaries, as one would imagine they should, in feelings about political participation. It is to be expected that those who score highly on the dissatisfied (sunshine) democracy scale will be eager to take part in any form of political participation, whether it is part and parcel of the orthodox processes of representative democracy, direct democracy or deliberative democracy; by contrast, we would clearly expect those registering high scores on the stealth democracy scale to be far less willing to declare a preference for engagement in the orthodox party-electoral, or non-party arenas, or for deliberative participation. However, it is not so certain that they would be opposed to direct democracy, for there is much in the stealth democratic profile which is intrinsically populist. The stealth democrat is not greatly interested in politics, has little understanding of or patience with its inherent messiness, complexity, adversarial conflict and frequent need for apparently sub-optimal compromises, and has low regard for political elites in general: while the stealth democrat might not have much inclination to get involved in political activity, it is quite conceivable that political actors who hold such views would be drawn to the idea that the ordinary and presumably virtuous people should at least occasionally be able to take decision-making power out of the hands of elites by recourse to referendums.

These varieties of political participation are measured as follows. Respondents' willingness to participate in the party-electoral arena is measured by an additive scale constructed from the following attitudinal questions:

Would you be willing to:

  • Vote in a local, national or European election?
  • Become a member of a political party?
  • Hold office in a local or national pressure group or organisation?
  • Hold local or national party office?
  • Contact a local councillor, members of a devolved assembly, MP or MEP about an issue of concern to you?
  • Donate money to a party or other political organisation?
  • Campaign on behalf of a candidate for local, national, devolved or European election?
  • Be a candidate for an elective post at local, devolved, United Kingdom or European levels?

The scale has a theoretical range running from 7 (low willingness to participate) to 14 (high willingness to participate), with a scale mean of 8.7 (confirming the general unwillingness of respondents to participate in party activity). Alpha for the scale = 0.747, indicating good reliability.

Willingness to participate in non-party forms of mainstream political activity is measured by a similar additive scale formed from responses to the following questions:

Would you be willing to:

  • Hold office in a local or national pressure group or organisation?
  • Sign a public petition regarding a national or local political issue?
  • Take part in a public demonstration about an issue of concern to you?
  • Write a letter to a newspaper editor?
  • Take an active part in a political campaign about an issue of concern to you?
  • Boycott or buy certain products for political, ethical or environmental reasons?
  • Take part in industrial action?

Once again, the scale runs from 7 (low willingness to participate) to 14 (high willingness to participate), with a scale mean of 9.1 (alpha = 0.756).

Willingness to deliberate is measured by a scale derived loosely from the work of Neblo et al. (2010). It is constructed from pooled responses to slightly different variants of a question about hypothetical willingness to participate in deliberative exercises that were asked of randomly selected sub-samples of the dataset. These variants were either worded ‘If you were ever to have the chance of participating in a one-day session where citizens discuss important issues with their local MPs or councillors, how interested do you think you would be?’, or ‘If you were ever to have the chance of participating in a one-hour online session where citizens discuss important issues with their local MPs or councillors, how interested do you think you would be?’ Respondents indicating that they were ‘very interested’ to either of these variants were coded 4, while those indicating they were ‘quite interested’ were coded 3, those opting for ‘not very interested’ were coded 2 and those who were ‘not at all interested’ were coded 1. The scale mean was 2.75 (N = 1,209).

Note that this variable only captures one out of myriad possibilities of what the term ‘deliberative democracy’ might entail. In particular, while it presents a scenario consistent with Mansbridge et al.'s ‘talk-based approach’ to political engagement, it does not involve the need for decision making. To this extent, it is not the deliberative variant that demands most of participants. Then again, it has been argued that the primary function of deliberative democratic processes should not be decision making. For Mark Warren (2012: 12), ‘the common term “deliberative decision-making” confuses the two quite different functions of collective will-formation and collective decision-making’ – while deliberation is ‘a mode of influence in which participants hope to sway one another to their positions’, collective decision making inevitably entails a degree of coercion that is ‘intrinsically contradictory’. Thus, the variable that we are deploying here captures deliberation's strength as a mechanism that facilitates political communication and will-formation.

Finally, support for referendum democracy is measured by responses to the statement: ‘Referendums are a good way to decide important political questions’, where 5 = strongly agree, 4 = tend to agree, 3 = neither agree/nor disagree, 2 = tend to disagree and 1 = strongly disagree (scale mean = 3.67, N = 1,223).

These four models of participation need to be tested in different ways because of the different scales of measurement upon which they are based: the party-electoral and non-party participation variables are additive indices that can be regarded as interval-level scales, for which OLS regression is the most appropriate method; by contrast, willingness to deliberate and support for referendum democracy are ordinal measurements for which ordinal logistic regression is more appropriate.

Table 4 reports the results of the OLS models of party-electoral and non-party participation, while Table 5 reports the results of the logistic regression models of deliberative and referendum democracy.4 Taken together, close examination of these tables reveals that the hypothetical expectations set out above are largely borne out. The higher a respondent's stealth orientation, the less they are willing to participate in party-electoral or non-party forms of mainstream activity (Table 4); similarly, stealth democracy co-varies inversely with willingness to deliberate but, as hypothesised, the situation is different with regard to referendum democracy (Table 5). This latter finding makes for a striking contrast with the other forms of participation, which suggests there is indeed something to the argument that stealth democracy carries undertones of populism. By contrast, the higher a respondent's score on the dissatisfied democracy index, the more willing they are to deliberate, rate referendums highly (Table 5) or participate in other forms of political activity (Table 4). As might be expected, political interest and efficacy both co-vary significantly and positively with all forms of participation, as does education in respect of the party and non-party forms; interestingly, however, it varies inversely with support for referendums – the lower a respondent's educational experience, the more they tend to support them (Table 5). Social grade shares a positive relationship with both deliberative and party-electoral forms of participation: the higher a respondent's occupational class, the more willing they are to take part; again, though, the pattern is a little different with respect to direct democracy, for it is lower social grade respondents (specifically the routine non-manual ‘C1s’ compared to the senior non-manual ‘AB’ categories) who profess a greater liking for referendums (Table 5). Age has little significant bearing on these models, although the young are more likely to favour deliberative democracy (Table 5). Finally, women are significantly less likely than men to want to take part in party-electoral activities (Table 4), but otherwise gender has no significant bearing on any form of political participation. None of these findings should be cause for surprise in terms of the literature or theory – with the exception perhaps of the negative relationship between education and referendum democracy. Once again, though, this does not seem so surprising if we regard direct democracy as part and parcel of the populist repertoire, likely to appeal as much to the disengaged and politically uninterested as to those who have high interest and cognitive understanding of public affairs.

Table 4. OLS models of party-electoral and non-party political participation
Independent variableStandardised coefficients (significance statistics)
Party-electoral participationNon-party participation
  1. Note: n.s. = non-significant.
Stealth democracy−0.216 (0.000)−0.223 (0.000)
Dissatisfied democracy0.066 (0.016)0.117 (0.000)
Political efficacy0.291 (0.000)0.220 (0.000)
Political interest0.146 (0.000)0.127 (0.000)
Gender−0.056 (0.033)n.s.
Social grade0.102 (0.000)n.s.
Education0.085 (0.003)0.126 (0.000)
Adjusted R2 for model0.379 (N = 1,010)0.306 (N = 1,010)
Table 5. Ordinal logistic regression models for deliberative and referendum democracy
Independent variableDeliberative democracyReferendum democracy
Parameter estimates (significance)Parameter estimate (significance)
Stealth democracy−0.039 (0.009)0.034 (0.021)
Dissatisfied democracy0.064 (0.026)0.164 (0.000)
Political efficacy0.176 (0.000)0.067 (0.001)
Political interest0.294 (0.000)0.254 (0.000)
Age−0.008 (0.068)n.s.
Gender (reference category: female)n.s.n.s.
Social grade (reference category: AB)  
DE−0.430 (0.020)n.s.
C1n.s.0.290 (0.077)
Educationn.s.−0.130 (0.006)
Cox & Snell R2 for model0.2570.123
Nagelkerke R2 for model0.279 (N = 945)0.132 (N = 976)

Of course, our findings have focused solely on respondents' claims about how they regard the various forms of political participation tested here. It may very reasonably be pointed out, however, that it is one thing to say that you would be willing to participate if given the chance, or that you think referendums are in principle a good way to decide political issues, but it is quite another to actually participate when confronted with the possibility of doing so. It is therefore interesting to compare the results of our models about hypothetical willingness to participate with those in which actual reported levels of participation are the dependent variables. Do they manifest similar outcomes? Table 6 reports the results of OLS models in which the dependent variables are the ‘actual participation’ counterparts to the two ‘hypothetical’ scales that tap mainstream forms of participation in Table 4 – that is, they are constructed in exactly the same way as the party-electoral participation scale and the non-party participation scale except that instead of each scale item asking respondents ‘would you be willing … ?’, it asks of them ‘have you ever … ?’.5

Table 6. Ordinary least squares (OLS) models of reported participation
Independent variableStandardised coefficients (significance statistics)
Reported actual party-electoral participationReported actual non-party participation
  1. Note: n.s. = non-significant.
Stealth democracy−0.173 (0.000)−0.194 (0.000)
Dissatisfied democracy0.076 (0.007)0.059 (0.040)
Political efficacy0.218 (0.000)0.212 (0.000)
Political interest0.145 (0.000)0.123 (0.000)
Age0.247 (0.000)0.247 (0.000)
Gender−0.045 (0.093)n.s.
Social grade0.097 (0.000)0.068 (0.014)
Education0.056 (0.057)0.101 (0.001)
Adjusted R2 for model0.348 (N = 1,010)0.324 (N = 1,010)

Examination of these reveals little difference between the two sets of results: all coefficients are signed precisely the same way in each model of hypothetical participation and its respective counterpart relating to actual reported participation; the only minor differences are that age becomes significant and positively signed in both of the actual participation models (apparently, it does not make any difference how old you are if it is merely a matter of declaring a willingness to participate, but when it comes to actually participating it helps to be on the mature side), and social grade becomes a significant and positive factor for actual non-party forms of participation (the higher one's social grade, the more likely one is to report actually having engaged in non-party forms of political activity).6

There is no variable in the dataset that parallels the reported actual participation scales for deliberative democracy: very few respondents would ever have had an opportunity to take part in a deliberative exercise involving elected politicians, so it was not feasible to pose such a question of them. However, the timing of the survey (in July 2011) did provide an opportunity for an interesting test in respect of direct democracy. Little more than two months prior to the execution of our survey a UK-wide referendum was held on the question of whether the existing system of single-member plurality elections for the House of Commons should be replaced by the Alternative Vote (AV) electoral system. In the event, the proposal for reform was firmly rejected by the electorate, on a low turnout of just 42 per cent. Although we should not be over-confident in drawing inferences about people's likely general behaviour on the basis of a single referendum, it is nevertheless interesting to see how the results of this event compare with the model of respondents' attitudes towards referendum democracy reported in Table 4.

Table 7 reveals the results of a model of participation in the AV referendum of 2011. Given that the dependent variable is dichotomous (1 = voted, 0 = did not vote), this is assessed through binomial logistic regression. This model displays some notable differences from the model of general attitude towards referendum democracy shown in Table 4 – the most obvious of which is that neither stealth nor dissatisfied democratic orientation had a significant impact on whether or not the respondent actually voted in the AV referendum. Efficacy and political interest had a significant positive impact, as did age and social class; older voters and higher occupational grade respondents were more likely to have participated. Men were more likely to have voted than women, but educational background made no significant difference. The main conclusion that one can draw from this admittedly limited comparison is that, while those with high stealth or dissatisfied democracy profiles may be significantly more inclined to declare support for the general principle of direct democracy, such an abstract claim does not readily translate into actual behaviour. This could in some way be due to the nature of the specific issue at stake in the AV referendum, of course, but we should be aware that the apparent demand for greater participation even from disaffected citizens may not always mean that they will take up the opportunities presented – a point that has been emphasised by some previous researchers (e.g., McHugh 2006).

Table 7. Binomial logistic regression model of reported participation in AV referendum, 2011
Independent variableParameter estimates (significance)
  1. Notes: n.s. = non-significant. N = 1,045.
Stealth democracyn.s.
Dissatisfied democracyn.s.
Political efficacy0.086 (0.001)
Political interest0.410 (0.000)
Age0.042 (0.000)
Gender (reference category: female)0.333 (0.037)
Social grade (reference category: AB) 
DE−0.579 (0.012)
C1−0.414 (0.053)
Cox & Snell R2 for model0.202
Nagelkerke R2 for model0.282

Conclusion: Implications for the potential of participatory reforms

The findings reported here constitute broad confirmation of the main hypothesis, H1, which simply states that there are two quite different types of attitude prevalent among citizens who are disaffected with politics: the ‘dissatisfied democratic’ and ‘stealth democratic’ orientations. The presence of the former is greater than that of the latter in the British adult population, and we have seen that the demographic and attitudinal correlates of these two distinctive orientations differ notably, with dissatisfied democrats being enthusiasts for all forms of political participation, but stealth democrats being far less keen. That said, stealth democrats apparently support the principle of direct democracy. To this extent, the British sample resembles the Finnish data reported by Bengtsson and Mattila (2009: 1045), who noted that many people ‘prefer simultaneously both more direct democracy and more stealth democracy’. While they concede that these things are ‘perhaps not logically mutually exclusive’, they feel that direct democracy and stealth democracy represent different democratic ideals insofar as one ‘stresses direct citizen involvement while the other puts emphasis on efficiency and expert decision-making’. Although this is true, I would suggest that it understates the extent to which direct democracy and stealth democracy are logically compatible through their shared connection with the populist worldview. It is not referendums that stealth democrats shy away from so much as other, more active and time-consuming forms of political participation, both mainstream and deliberative. For these people who are disillusioned with the rule of elected politicians, there are various alternatives that seem appealing: rule by technocrats, entrepreneurs or referendums are all ways of bypassing the politicians. Even so, we have seen that when they are given the actual opportunity to take part in a national referendum, neither stealth nor dissatisfied democrats proved especially inclined to vote. This suggests that demand for the principle of greater participation may owe more to a sense of frustration with, and perhaps incomprehension about, the obscure complexities of the political process, with its all its attendant noise, conflict and undoubted venality, than to an authentic desire to become involved in the detail of political decision making.

What implications do these findings carry for future research? For one thing, the finding that stealth democracy is an identifiable but not an overwhelming presence in the United Kingdom somewhat detracts from the comparative impact of Hibbing and Theiss-Morse's argument. The widespread prevalence of sunshine democratic attitudes in combination with low trust in political elites confirms the prominence of Dalton's dissatisfied democrats in the country – citizens with a professed desire for greater political participation. For these citizens, greater opportunities for participation might in principle be part of ‘the answer’. However, this still leaves us confronting the problem of a significant stealth democratic minority – a minority which is more susceptible to electoral abstention or the appeal of populist alternatives,7 and which may also be further marginalised by any attempts to extend participatory opportunities. As one commentator has suggested:

While the call for more participatory democracy has a visceral emotional appeal, in practice it may only succeed in engaging those already over-represented amongst voters and party members – that is, the educated, affluent and middle-aged. Mechanisms designed to provide greater opportunities for citizens to participate more directly in decision making as a means of increasing legitimacy and reducing the perceived democratic deficit may therefore have the opposite effect. It is likely that those already adept at making their voices heard will use the new structures to continue to advance their particular interests. (McHugh 2006: 551)

Such criticisms of high-intensity participation, especially deliberative democracy, are not uncommon. However, they do not confront the educative argument long favoured by devotees of participation. Could it be that the experience of participating – especially in deliberative democratic exercises – is in itself the very thing that is required to enhance the political interest, efficacy and engagement of ordinary citizens, regardless of background or negative predispositions they may hold about politics? Other researchers claim to have found evidence in favour of such claims (see, e.g., Neblo et al. 2010), and indeed it is certainly possible to point to similar evidence from many studies around the world, including Britain.8 Nevertheless, a particular objective of future research in the field might be to target those with stealth democratic profiles: what forms of participation would work best to enhance their political knowledge, efficacy and engagement? And what should be the central functions of any such innovations in the wider political system – to communicate, educate and facilitate the formation of policy options, or to make binding decisions? If the latter, can such processes work even when participants are in zero-sum situations in which there are winners and losers? If the former, what is the mechanism by which the deliberative process may be effectively and legitimately linked to the decision-making process? In sum, the major objective of this field of scholarship must be to identify forms of deliberation and participation that will work and articulate successfully with existing institutions in ‘the real world of democracy’, and to do so in ways that are sensitive to the nuances of popular attitude.


I am grateful for enormously helpful comments on earlier drafts of this article by Tim Bale, Sabina Avdagic, Lee Savage, Hilde Coffe, Emilie van Haute, Jean-Benoit Pilet, various panel participants at the ‘Elections, Public Opinion and Parties’ conference held at Oxford in September 2012, the EJPR editors and three anonymous journal reviewers. Sole responsibility for the article lies with me, of course.


  1. 1

    This survey was conducted by YouGov using a targeted quota sampling method. An iterative process was used in order to ensure the data were in the correct proportions for each of the major demographics. This achieved a nationally representative sample, with data weighted to the profile of all adults aged 18+ taking into account age, gender, social class, region, political party identification, newspaper readership and election turnout in May 2010. Target percentages were derived from three sources: census data; the National Readership survey (a random probability survey comprising 34,000 random face-to-face interviews conducted annually); and (for party identity) YouGov estimates. The latter were derived from an analysis of more than 80,000 responses to YouGov surveys at, or shortly after, the May 2010 general election, when respondents were asked whether they generally thought of themselves as Labour, Conservative, Liberal Democrat, etc. (party identity); and which party they would support, or had supported, in the 2010 general election. Data were weighted to May 2010 party identity. The final weighted sample size was 1,355. I am extremely grateful to the British Academy for awarding me the grant that enabled me to commission this survey.

  2. 2

    Analysis of residual diagnostics confirms that none of the key assumptions of OLS regarding linearity, homoscedasticity, multicollinearity or autocorrelation are violated in any of the models reported in this article (details available from the author upon request). ‘Left–right ideology’ and ‘Liberty–authority ideology’ are standard attitudinal scales first developed for use in the British context by Heath et al (1994). ‘Political interest’ is measured as a scale of responses to the question ‘How interested, if at all, would you say you are in politics?’ (1 = not at all interested, 5 = very interested). ‘Efficacy’ is an additive scale running from 5 (low political efficacy) to 25 (high political efficacy) created from responses to the following questions: ‘When people like me get involved in politics, they really can change the way that the country is run’; ‘I consider myself well-qualified to participate in politics’; ‘I feel that I have a pretty good understanding of the important political issues facing our country’; ‘I feel that I could do as good a job in public office as most other people’; ‘I think that I am as well-informed about politics and government as most people’. Cronbach's alpha for the efficacy scale = 0.804 and the sample mean = 15.9. ‘Social grade’ is an ordinal scale based on the Market Research Society's model that has been widely used in British social research and is coded here as follows: 1 = occupational categories D & E (i.e., semi- and unskilled manual workers, welfare beneficiaries); 2 = category C2 (skilled manual); 3 = category C1 (routine non-manual employees); 4 = categories A & B (senior and intermediate white collar, managers and professionals). ‘Gender’ is coded 1 = male, 2 = female. ‘Education' = terminal age of full-time education.

  3. 3

    The United Kingdom Independence Party (UKIP) is a nationalist organisation that appeals primarily to the electorate on a sovereignist anti-EU programme. It is careful to aver that it is non-racist, but is widely regarded as populist in view of its critique of the mainstream parties. It won 3.1 per cent of the nationwide vote at the 2010 general election, but no seats in Parliament; however, it did succeed in returning 13 MEPs to the European Parliament in 2009. The British National Party (BNP) is a more overtly far-right organisation with a reputation for xenophobic politics that is both anti-EU and anti-immigrant. It won only 1.9 per cent of the national vote in the 2010 general election, but did succeed in gaining two seats in the European Parliament in 2009.

  4. 4

    In Table 5, gender and social grade are entered into the models as categorical variables, while the remaining independent variables are entered as continuous covariates.

  5. 5

    The scales each run from 7 (low reported participation) to 14 (high reported participation). The scale mean for reported party-electoral participation = 8.27 (alpha = 0.676), and for reported non-party participation = 8.44 (alpha = 0.703).

  6. 6

    It is of course possible that the similarity of the results reported in Tables 4 and 6 owes something to the tendency of respondents to exaggerate their levels of participation, regardless of whether it is supposed to be hypothetical or reported behaviour. Indeed, this is a widely recognised feature of survey-based studies of political participation (see, e.g., Sanders et al. 2007: 163–164). Even if this is true, however, surveys remain valuable for uncovering the significance and direction of relationships between participation and its various predictors, as in this article. Moreover, it is still useful to check for differences between hypothetical and reported actual participation because these differences can be all the more interesting where they do occur. An example is provided in Table 7, where neither stealth nor dissatisfied democracy orientations significantly predict reported turnout in the United Kingdom's Alternative Vote referendum of 2011. This contrasts sharply with the results of respondents' general declared support for referendum democracy in Table 5.

  7. 7

    Note that those who voted for populist party candidates (BNP or UKIP) or who abstained in 2010 had slightly but significantly higher scores on the stealth democracy scale than those who voted for mainstream parties: populist party voters = 27.8; other party voters = 26.8; abstainers = 28.3 (between group significance = 0.000), N = 1,336.

  8. 8

    For instance, the deliberative polling undertaken by James Fishkin for The Power Report 2010 showed that experience of deliberation impacted on the political knowledge and attitudes of some participants (http://cdd.stanford.edu/polls/uk/).