The Electoral Impact of Direct-Democratic Practices

In recent years there has been a growing interest in the integration of mechanisms of direct citizen participation in the institutional structure of representative democracy, particularly at the local level. This essay examines the electoral impact of mechanisms of direct citizen participation. Although it is often considered that participatory schemes can be a means to achieve electoral success in the hands of politicians seeking re-election, quantitative analyses of 65 Spanish municipalities demonstrate that electoral success is far from being an immediate consequence of direct democratic practices (DDPs). The qualitative analysis of four cases shows that electoral consequences directly attributable to participatory devices depend on their design and on how they fit into the whole political process. Participatory processes that are too rigid and those, especially, that generate expectations that cannot be translated into real policies may end up having a negative effect. On the other hand, DDPs may account for network-building and improved information among citizens that, in turn, may have electoral consequences. DDPs are thus neither a blessing nor a cure per se in their electoral effects. Instead, as with representative democracy, their consequences and success will ultimately depend upon their procedural dimension.


Introduction
In recent years there has been a growing interest in the integration of mechanisms of direct citizen participation into the institutional structure of representative democracy, particularly at the local level.This has increased the attention paid to different aspects of these mechanisms, from their causes to their organizational characteristics and their various consequences.This essay tries to explore an aspect that has not been considered by the previous literature; it examines the potential electoral impact of mechanisms of direct citizen participation.We will proceed as follows.
We will proceed as follows.In the first section, we will contextualize the research question through a discussion of the role that mechanisms of citizen participation play in local governance.The second section will develop the research questions and present our main hypothesis.The third and fourth sections will present the empirical research, including a quantitative analysis in the third section and four case studies in the fourth.Finally, the conclusions will present our main findings and discuss their implications for local democracy.

Participation and local democracy
Although almost all definitions of democracy include references to participation, there is far less agreement regarding how much participation there should be and how this participation should take place.How, and to what extent, citizens should participate in the making of political decisions is a question that has long been debated by political theorists.The debate has traditionally opposed those who assert that participation should be limited to the representative channels of democracy (mainly voting) and those that argue that meaningful participation should be continuous and intense.For this latter group, the local arena has always been ideal for developing most of the more intense participatory undertakings.
Without necessarily embracing the premises of participatory democracy, many authors would agree that today's political systems are in trouble when they try both to provide solutions to the very complex problems they face and at the same time to maintain the legitimacy of institutional structures.We find, on the one hand, heterogeneous societies in constant flux for which no general solutions are available, a globalized economic power increasingly deregulated within states, and a lack of technical and scientific means capable of guaranteeing the effectiveness of policies.On the other, as many studies have concluded (Norris, 1999;Torcal and Montero, 2006), there is increasing evidence of the level of political disaffection felt by citizens, 'the subjective feeling of powerlessness, cynicism and lack of confidence in the political process, politicians and democratic institutions, but without questioning the political regime' (Di Palma, 1970: 30).Indeed, recent studies relate this increase in levels of dissaffection to 'a democracy with a well-established record of adopting exclusive rather than inclusive institutions and deliberative processes' (Torcal, 2003: 18) Defenders of a greater degree of citizen participation in the public sphere have increased in number and have occupied an increasingly central position in the debate.Barber (1984) affirms that a 'strong democracy' is necessary, as opposed to a 'weak democracy' which limits participation to elections and the expression of citizen preferences to choosing between parties.Greater citizen involvement, a fuller exercise of citizenship and greater effectiveness of public policy are the expected benefits of more intense participation.Direct and continuous participation by citizens in the regulation of society and the state is, for some authors, the most effective way of promoting freedom and of assuring that institutions work correctly (Macpherson, 1987).
The debate between the liberal and participatory conceptions of democracy that has long proceeded at a normative level is today empirical.One of the main criticisms raised by those who defend a democracy based exclusively on representative institutions is that it is difficult to put the intense participation defended by some into practice while keeping it extensive.Moreover, they suggest that attention should be paid to the actual (and not only the expected) consequences of participation.Would direct-democratic practices reinforce or undermine the legitimacy of representative institutions?Would more participation reduce political disaffection among citizens and increase the legitimacy of political institutions?Which specific type of participation would bring these benefits?
The defenders of a greater degree of citizen participation are indeed a heterogeneous group with often conflicting opinions, since the question is not merely one of greater participation, but also of how to articulate that participation.Thus, we find authors who highlight the importance of participating in associations and the virtues of a diverse and active social fabric (Putnam, 1993;Cohen and Rogers, 1995); others who favour mechanisms of direct democracy (Budge, 1996); and those who place emphasis on the possibility of deliberation, even though this might mean a restriction of the number of citizens involved (Fishkin, 1991;1995).Even so their intention is not to provide an alternative to the representative system but rather to find ways to improve it.In fact these theorists agree that their proposals are 'complementary and compatible with the main representative institutions of mass society' (Barber, 1984: 342).
But although there is an acknowledgement of the complementary nature of such proposals, there is a lack of studies relating to the degree to which such a relationship is possible -that is, to how the institutions of representative democracy can be combined with more direct, demanding and intensive mechanisms of citizen participation.While there is a considerable amount of work describing the expansion of new mechanisms of citizen participation (Kahn, 1999;Lowndes et al., 2001;Font, 2003) and some on why it emerged (Akkerman, 2003;Font and Blanco, 2005), the effects of citizen participation on the political system have not received a great deal of scholarly attention.Particularly, the political consequences of the direct-democracy processes that have been put into practice particularly at the local level remain quite unexplored.
This essay intends to fill in one of these gaps by asking how these processes might affect one specific aspect of representative democracy.Their possible effects could be quite diverse, ranging from cultural effects on participants to organizational changes in local government or changes in the contents of policies, to mention just a few.However, we will address a much more limited question: how do these direct-democratic procedures affect the electoral results of the parties that put them into practice?This question, as far as we know, has not been previously addressed by the literature on local democracy.We do not claim that it is more important than others, but we suggest that the answer to it may be crucial for the development of local mechanisms of citizen participation.If developing these mechanisms has an electoral cost, very few politicians will agree to implement them.By contrast, large and widespread electoral gain could produce a considerable number of new users, not greatly interested in participation as such, perhaps, but only in one of its consequences.

Direct-democratic practices: concepts and research questions
Although the Spanish Constitution of 1978 does not make any meaningful reference to direct democracy,1 there has recently been a tendency, particularly at the local level, to promote new methods of citizen participation in public affairs (Font, 2001;Subirats et al., 2001;Collet et al., 2005).Direct-democratic practices (DDPs) are mechanisms through which the citizenry is called to participate directly in everyday public decision-making on a voluntary basis.DDPs are meant to be a way of both democratizing politics and legitimizing certain policies or policy outcomes and are usually kept to the local level, which is regarded as the most suitable level for direct democracy to work.However, DDPs are enormously eclectic: their methodology, duration, purpose and political scope vary sharply from one to another.
The scope of this essay is limited to the impact of DDPs on electoral behaviour.More specifically, as local authorities' decisions to implement DDPs depend on cost-benefit calculations with regard to their possible electoral rewards, we will focus on the impact DDPs might have on citizens' voting preferences.Our aim is not to identify their impact on levels of turnout, but to clarify if, to what extent and in what way DDPs can be held accountable for the electoral gains and/or losses political parties might suffer in local elections.
The results of the 2003 local elections in Catalonia and in Spain came as rather a surprise since several governing parties that had explicitly promoted DDPs suffered significant electoral losses.Thus, two questions came inevitably to the fore: (1) to what extent do DDPs influence the citizens' preferences in terms of party choice; and (2) through which mechanisms.The present essay seeks to establish whether or not an empirical pattern can be identified that enables us to state that some form of relationship exists between DDPs and the citizen's vote.
We expect that, in any case, this effect will be limited.First, because the literature on electoral behaviour tells us that the range of factors affecting the vote is quite large and the effect of participation policies on the vote is just one among many.Second, because among all the possible electorally relevant policies, the development of DDPs is unlikely ever to be a crucial one.Previous research has shown the electoral effect of policies that are a central concern for many citizens, like social or economic policies (Fiorina, 1981;Lewis-Beck, 1988;Fraile, 2005).But participation is a crucial issue for very limited sectors of society.2Third, because DDPs are not as visible and well known as other local policies, only citizens who have some information about them will consider their existence as a factor in whom they choose to vote for. 3or this purpose we carried out a statistical analysis of the electoral results of local government parties in 21 municipalities where direct-democratic practices were considered significant and compared them with similar cases where these DDPs did not occur (see the next section).But beyond the question of whether this electoral impact can be appreciated, there is the further question of how DDPs operate so as to influence voters' preferences.In order to disentangle how this effect takes place we analyse in depth four cases where the possibility of DDPs having an impact was particularly significant.If DDPs have any electoral effect, it is no less important to understand through which causal mechanisms that effect operates than to establish its existence.These in-depth studies should allow us to reinforce the conclusions of the empirical results, to better understand the micro-mechanisms relating to DDPs and electoral effects, as well as to formulate hypotheses for further research.

Quantitative study
The analysis of the impact of participatory experiences on electoral results is undoubtedly a difficult task for several reasons.First, we know that multiple factors have an influence over voting behaviour and that, in addition, it is difficult to isolate the relative effect of each of them (Anduiza and Bosch, 2004).Second, in order to reach general conclusions that apply beyond specific cases, it is necessary to choose data that are easily identifiable and comparable.This forced us to obviate context-specific aspects that will be reconsidered in the qualitative analyses presented later.The quantitative research strategy has thus consisted in a quasi-experimental approach.We have identified a number of municipalities where DDPs took place and compared their electoral results with municipalities with similar characteristics but where no such procedures were carried out.In the following section we explain the details of the case selection.Source: Own elaboration based on data given by experts

The selection of cases and variables
The first step was to draw up a list of municipalities where significant DDPs had taken place during the term of office previous to the elections of interest (1999)(2000)(2001)(2002)(2003) both in Catalonia and the rest of Spain.4This list was presented to seven experts5 on citizen participation who were asked to evaluate DDPs on the basis of two criteria: visibility and intensity.By visibility, we understand the level of public awareness attained by the procedures, or, in other words, their importance in the public debate.Intensity refers to aspects related to the operational dimension of procedures, such as the number of participants and their diversity, the number of meetings or phases established, the time required, the importance of the issues concerned, etc.We consider intensity and visibility to be key elements of DDPs and, as such, potential conditioning factors in elections.
For DDPs to have relevant effects on electoral results, they should be visible and intense.Experts were then asked to score each case between one and ten for each criteria.The highest-rated cases (fourteen Catalan and seven from the rest of Spain)6 were selected for analysis.As Table 1 summarizes, the DDP's mean visibility in the 21 selected municipalities is 6.8, while the evaluation of the intensity of the experiences presents a mean of 6.5.The standard deviation is relatively low, which indicates a considerable degree of agreement among the experts' evaluation.As control cases we wanted a group of municipalities that were as similar as possible to the selected cases, but without DDPs.Thus, between two and four municipalities in which no DDPs had been carried out were chosen for each one of the municipalities selected.These controls were selected taking into account their similarities with the municipalities with DDP in terms of three sets of variables: local government context, socio-economic characteristics and territorial proximity.As a result, the controls were as similar as possible to the selected municipalities with DDPs in terms of the number and political orientation of parties in government, population size, disposable income and geographical area.Overall we had a total of 65 units of analysis, 21 cases with DDP and 44 controls. 7he phenomenon that we wish to explain is the unequal electoral developments between the elections of 1999 and 2003 that were, presumably, partially influenced by whether or not DDPs had been put into practice.Thus, at an operational level, the dependent variable of the study is the variation in electoral results between 1999 and 2003.This variation, in turn, can be expressed in terms of two variables.The first (XVOT1) reflects the difference in the percentage of votes won by the main party in government over the 1999-2003 period.The second (XVOTGOV) measures the variation in votes of all governing parties in this period.Obviously, where a single party governs the value is the same for both variables.The distinction is made in order to analyse two different hypotheses: DDPs affect (positively or negatively) only the main governing party; DDPs affect all governing parties.On the other side, the independent variable of interest (PARTICIP) is the existence of direct democratic practices defined in terms of a dichotomy (coded 1 if the municipality had a significant DDP during the term and 0 otherwise).
There are other potentially relevant independent variables that may have influenced electoral outcomes, and that must be taken into account to ensure that differences between cases with DDPs and controls cannot be explained by other elements from the political or economic context.These are precisely the variables used to select the controlling municipalities and are included in the analyses.Variables related to the characteristics of the local government are considered because these may also be associated with electoral results.We want to be sure that an effect potentially due to DDPs is not due instead to the fact that being part of a coalition, for instance, or having an absolute majority in the local chamber penalizes parties electorally.Thus, we explore the effects on election results of the number of parties in government (NPPGOV99, scoring from 1 to 4), the relative weight of the main party in government between 1999 and 2003 (PRESREL99, percentage of councillors from the main party compared with other governing parties), and whether the government had an absolute majority or not (MAJOR99, coded 1 if so and 0 otherwise).8Socio-economic characteristics are included as standard controls.We consider the number of inhabitants of the municipality (HABIT1000, in thousands) and two indicators of economic wealth, the index of disposable household income per capita (RENDAD, scoring from 1 to 10) and the available income (NECON, in 1,000s of euros).9So is there any relationship between putting DDPs into practice and electoral change?Do DDPs reinforce or weaken local governments in terms of their electoral results?To begin with, we might consider the totality of the 65 cases and see whether governing parties that promoted DDPs did, in general, better or worse in the elections than those that did not.Table 2 shows that incumbent parties did worse in most municipalities, whether they had organized DDPs or not.The parties with most votes in 1999 in those municipalities where DDPs were put into practice fared worse in 2003 than their counterparts in municipalities where no such procedures had taken place.
While the former suffered an average decline in vote of 8.3%, the latter recorded a fall of just 4.3%.
At the same time, the decline in the overall vote of governing parties in municipalities where participatory schemes had been carried out was 8.6% compared with just 5.7% in municipalities with no DDPs.Though these differences are not statistically significant at conventional levels, considering the number of cases available for analysis, they are substantially relevant.Looking at these results, DDPs do not seem to provide better electoral outcomes for incumbent local governments.
A closer look at the data (see Appendix 2) allows us to qualify this conclusion.There was one particularly strong case of electoral decline, St. Bartomeu del Grau, where the large electoral shift was related to the emergence of new parties.If we consider this to be an extreme case and take it out of our analysis, we find that the differences in means calculated above are considerably reduced (see Table 3).The difference derived from DDPs is now less than 2% in the case of the largest governing party, and just over 0.5% for governing parties taken together.While there remains a difference in favour of those municipalities with no DDPs, it has been reduced.
Two more analyses can be done to qualify or confirm these results.Firstly, we can use multivariate regression analysis to see if these differences remain the same after controlling for other variables that may account for electoral decline by.Secondly, we can choose to no longer treat the 65 municipalities as a single entity and instead compare each one with its respective controlling cases.
The results of the regression confirm previous evidence, as Table 4 shows.DDPs reduced the vote for parties with most votes and for parties in government even after introducing the controlling variables, as the two full models show.Indeed, the effects of participation increase their negative value when the controlling variables are introduced.Beyond this substantive interpretation, none of the coefficients is statistically significant at conventional levels owing to the small number of cases as well as to the high correlations between some of the variables.The selected models are more parsimonious in order to partially overcome this problem.They confirm that DDPs have a negative impact on the party with most votes, and to a smaller extent on the government coalition, though we are still moving beyond the limits of conventional statistical significance.The fact of having an absolute majority seems to be a relevant variable for the electoral decline of the main party, while political variables do not seem to affect the electoral result of the governing parties in a significant way.The multivariate analysis thus confirms the bivariate results: participation has a weak, negative, statistically nonsignificant effect on electoral results of both the main party and governing parties.
When we analyse the participative municipalities in light of their respective control municipalities, our results must be even further qualified (see Table 5).We now compare the mean values of each DDPmunicipality and their respective controlling municipalities in terms of both leading government party and coalitions as a whole.In the case of leading government parties, of the 21 participating municipalities, 8 obtained an average score above that of their controls, with 13 doing worse.Thus, 60% of DDP municipalities had worse results than their controls.In terms of governing coalition parties, the results are even closer: in 9 municipalities governing coalitions scored better results in 2003 than in 1999, with 10 cases doing worse than their respective controls.*A value of 0 means that the town with participation and its controls have equal differences between the two elections.A negative value means that the party/parties in government suffered a larger electoral loss in towns with participation.A positive value means that the party/parties in government had a better result in towns with participation.(*)The break-up of the coalition Entesa per Sabadell, from which ICV formed its own group, makes it impossible to compare the results of the two elections, even though they are clearly favourable to the parties in power

Source: Own elaboration
On the basis of these data we might think that the relationship shown so far could be due to a few cases of heavy electoral falls conditioning the overall results.If we turn once more to the table of differences in voting between 1999 and 2003 (see Appendix 2), municipalities such as St. Bartomeu del Grau, Arbúcies or Cabezas de San Juan might be the clearest examples of such interference.Summing up, comparing mean values and regression coefficients leads us to a first conclusion: in municipalities that had promoted DDPs, both the leading party and the sum of coalition parties suffered electoral losses of greater proportions than in municipalities where no DDPs had been promoted.The introduction of controlling variables does not reduce such effects, but rather increases them.At the same time, we should also note the lack of significance of the coefficients and the existence of extreme cases.On the other hand, a comparison of the mean values of each municipality with those of the control cases shows an even less homogeneous relationship.This heterogeneous picture, then, confirms the need for a further question: Have participatory schemes really been the cause of the electoral shifts that we have found?This question requires the analysis of case studies, the aim of which is to further explore whether electoral change can indeed be attributed to the promotion of DDPs and, if so, to suggest possible factors that explain how and why.

Case studies
In order to obtain a more precise picture, we undertook in-depth studies of four municipalities in which, a priori, we could not rule out DDPs having a certain impact on the electoral results.The cases chosen were Córdoba (300,000 inhabitants), Esparreguera (20,000 inhabitants), Rubí (70,000 inhabitants) and Sitges (23,000 inhabitants), four quite diverse cities governed by left-led coalitions in the first three cases and a right-wing coalition in the fourth.In the first two municipalities the governing parties increased their percentage of votes and number of councillors, so they could be taken as examples of local governments benefiting electorally from the promotion of DDPs.The other two, Rubí and Sitges, are municipalities where the governing parties lost office, due to a heavy loss of both votes and councillors (see Appendix 4).
The case studies are based on eight interviews in each municipality and on the analysis, where possible, of other sources, such as press and other secondary literature.The interviewees were chosen according to four profiles: independent observers, elected politicians from different parties, local government civil servants and participants in some of the DDPs.The objective was to be able to cover the whole range of relevant actors, from politicians to participating citizens.The inclusion of independent observers should ensure an input from local experts not directly related to the procedures, but with sufficient analytical capacity to evaluate the possible electoral impact of DDPs or other factors such as the broader political context.Asemi-structured interview was carried out, generally of an hour in length, that included questions concerning the DDPs, the position taken by the leading political actors, the media repercussions and a subjective evaluation by the interviewee of the potential electoral impact of DDPs. 10he results of the four case studies reinforce the conclusions of the first part of this essay: the electoral effects of DDPs seem to be rather weak and difficult to distinguish from other factors accounting for the citizen's voting behaviour.Nonetheless, when they operate they seem to act through a complex causal mechanism, which can be summarized as follows: As can be read from Figure 1, the electoral impact of DDPs cannot be considered immediate.DDPs certainly cannot be regarded as directly responsible for electoral outcomes.Instead, they have an indirect influence.Direct-democratic practices can be held responsible for (1) promoting networks and (2) diffusing politically relevant information.

Figure 1
The electoral impact of DDPs (source: own elaboration) As to the network-generating effect, in Sitges and Rubí new and electorally successful political parties (see Appendix 3), Acció per Sitges and Alternativa Ciutadana de Rubí, were created shortly before the 2003 local elections.From our interviews we know that the DDPs were crucial arenas as regards both the creation of these parties and their subsequent electoral success.Meetings, deliberations and exchanges of information could be held responsible for the articulation of new groups (i.e.parties) by individual citizens or other pre-existent groups, in order to improve power resources and political skills.In Córdoba and Esparreguera too, networks were created and/or reinforced thanks to DDPs.Although the broader political context did not provide meaningful incentives to create new parties, all interviewees agreed in stating that networks have improved since DDPs were put into practice.Thus, inter-associational communication, organizational skills and the availability of information to citizens and associations would have improved thanks to the promotion of DDPs.
As to the loud-speaker effect, direct-democratic practices contributed to the diffusion of politically relevant information mainly in three ways.First, the promotion of participatory schemes was in itself a way of putting certain issues at the centre of public attention.DDPs happened to be publicized as major democratic events in which, it was said, the citizenry was to be empowered and representative institutions constrained.Second, information was transmitted at meetings organized as part of the participatory schemes themselves and processed through deliberation.Finally, as DDPs are most often linked to specific issues (urban planning, budgeting, etc.) relevant information was transferred from local authorities to citizens as documentation was handed out at the meetings.As a consequence, through increased information flow, citizens developed new skills in specific policy fields and thus were better prepared to critically address the local government's performance.
However, to what extent the creation of networks and the diffusion of politically relevant information are to be held responsible for positive or negative electoral results is yet another story.What our research has been able to identify are two effects directly attributable to DDPs: networkbuilding and an improvement of the information available to citizens.If newly created networks effectively evolve into parties standing for election, and if these parties happen to be successful and achieve representation in the Town Council, that is something different.Networks and an increased flow of information are electorally neither positive nor negative in themselves, but turn into potentially government-reinforcing or -weakening features, depending, on the one hand, on the presence of pre-existing problematic issues and on the nature of personal and political relationships between the main actors -two things that can be conceived of as being theoretically distinct, but are empirically hard to keep separate-and, on the other hand, on the design and status of DDPs.
Thus we distinguished between a substantive dimension and a procedural dimension in the DDPs.The substantive dimension comprises the topics dealt with in the context of the DDPs and, thus, the broader political context in which DDPs operate.The procedural dimension, by contrast, takes in methodological issues to do with the design of the DDPs and their status vis-à-vis representative democracy.
Pre-existent problematic issues can exert a vast influence over the final perception that participating citizens and the wider public have of the DDPs and the local government's performance during its term of office.The perception citizens have may be positive or negative, depending on how sensitive issues are dealt with by local authorities.In Rubí and Sitges urban planning had evolved as a sensitive political issue around which actors' discourses, interests and positions had been articulated.Contrariwise, as neither urbanplanning nor any other especially salient or sensitive subject existed in Córdoba and Esparreguera, any issue could, a priori, be debated through participatory mechanisms.Furthermore, and intimately related to the first point, depending on the attitude of the main political actors, DDPs will be either accepted as a valid framework of political participation and decision-making commonly agreed upon, or rejected by opposition parties on the grounds that they are a means of government control over civil society and, thus, turn into a powerful political weapon.In Rubí and Sitges political actors had sharply conflicting points of view on many subjects, especially, as previously mentioned, on how to enhance urban planning.Personal and political relationships worsened during the political term and had spillover effects on DDPs.A tense previous political context and deteriorating personal and political relationships were crucial to the way that the citizens of Rubí and Sitges perceived the local government's performance.Yet, a tense political context was not exclusive to those towns.Córdoba and Esparreguera's local governments and oppositions also had conflicting points of view on many issues and tense personal and political relationships.However, Córdoba and Esparreguera managed not to reproduce those conflicts in the participatory schemes.According to our interviewees, DDPs were gratifying experiences in which deliberation was successful and participants' opinions were taken seriously into account.
This brings us to our last consideration: how and why are DDPs able to produce networkbuilding and information-diffusing processes that lead to government weakening or governmentstrengthening outcomes in specific situations?The answer, according to the interviewees, is to be found in the design of the participatory schemes and their status vis-à-vis representative institutions, that is, in the procedural dimension of the DDPs.Design refers to the organizational structure of a DDP, understood as its ability to enable participants to define or redefine the terms of discussion and the aims of the process as well as to freely and extensively deliberate in order to enhance the exchange of valuable information.Rubí is an example of an extremely rigid design.There, the terms of discussion in most DDPs were set by the local authorities and did not accord with the criteria of many of the participants.The interviewees revealed a great deal of dissatisfaction and mistrust arising from a clear perception of manipulation.Córdoba and Esparreguera, by contrast, are examples of how participants were able to define or redefine some of the basic rules as well as to introduce new issues for deliberation.But this type of flexibility presupposes political commitment and courage, scarce attributes in participatory politics so far.
Political commitment and courage, in turn, lead us to the consideration of what status DDPs have or should have vis-à-vis representative institutions.The question is whether DDPs, once implemented, should be able to produce collectively binding decisions or whether they should be restricted to being consultative procedures.Whatever the answer, local authorities committing to participatory politics should be aware of the consequences entailed if it is done 'wrongly': DDPs promoted with the aim of legitimizing previously taken decisions run the risk of failing to do so, when participant citizens realize they are being manipulated.Moreover, local governments that take decisions that contradict those taken by DDPs also run the risk of generating mistrust and adverse reactions.Local Agenda 21 (LA21), set up by the local government of Sitges to discuss a new framework for the city's urban planning from a sustainable point of view, is an example of this.Even before LA21 came to its decisions, the local authority decided not to take its work into account, as it seemed to run counter to its initial expectations and interests.Something very similar happened in Rubí, antagonizing not only participants, but also wider public opinion.

Conclusions
In this essay we have examined the relationship between participatory politics and electoral change both quantitatively and qualitatively.Specifically, we have addressed the possibility that participatory devices induce electoral changes at local elections.
The quantitative data do not show any conclusive results regarding the way and the extent to which DDPs are responsible for variations in the electoral results of parties standing for local elections.The results rather point to a limited and irregular effect.Contrary to what is often considered a commonplace, namely that participatory schemes are a means to achieve electoral success for politicians seeking re-election, the analyses show that electoral success is far from being an immediate consequence of DDPs.They have, at most, a limited and irregular effect, and cannot be considered responsible for significant electoral changes that are usually the result of many context-specific characteristics of the local political setting.This is not an unexpected result.Voters may approve the development of expanded opportunities for citizen participation, but these mechanisms are only one among many factors influencing the local vote.They compete in this role with other things such as the identity and character of the candidates, national political issues, political loyalties, etc.They are also less visible, less well known and for many citizens less important, than other local issues, from traffic to social affairs.
Notwithstanding this first conclusion, DDPs and electoral results may be related under certain circumstances.As we have shown in the qualitative analyses, and despite the secondary role of participatory devices in local politics, DDPs may have indirect electoral effects if certain requirements are met.DDPs may be responsible for network building and improved information diffusion among citizens that, in turn, may work in favour of or against local governments at elections.Thus, as a second major conclusion, electoral consequences directly attributable to participatory devices depend on the design of DDPs and on how they fit into the whole policy process.Participatory processes which are too rigid and especially those that generate expectations that do not translate into real policies, may turn out to have a negative effect.On the other hand, when these processes are coherent with a new style of doing politics, they may become a visible symbol that reinforces the electoral standing of local governments.
Ultimately, the effects will depend on how participatory experiences are dealt with in different political settings and, broadly speaking, how direct democracy fits into the wider notion of representative democracy.Thus, one may conclude that DDPs are per se neither an electoral blessing nor a cure for the shortcomings of representative democracy.Instead, as with representative democracy itself, success will ultimately depend upon the procedural dimensionan interesting question to be taken into account by political theorists and policy-makers alike.
These results have implications for two fields of analysis.First, our conclusions affect the debate about which factors explain the local vote and, thus, up to which point local elections can be used to hold local governments accountable.Previous research had considered the effect of some very central policies and our research helps to broaden the range of policies analysed.From this point of view, our results suggest that policies which are not highly visible are likely to have limited and not very regular electoral effects.If local elections were a good opportunity to hold governments accountable, this exercise would only deal with some of the most central and visible local policies.Second, the cases analysed suggest that if one is interested in the consequences of DDPs, there are many other factors to be considered beyond their effect on electoral results.The development of new mechanisms of citizen participation will not necessarily hurt any local official, but neither will it necessarily give them special help.The debate on whether or not to develop them should then be fought on other issues, like their empowerment effects or their consequences for the quality of local policies.This is possibly good news for local participation since it obviates both the fear of developing them because of their potential negative effects, and the temptation to implement them only in order to win elections.Eva Anduiza, (eva.anduiza@uab.es),Institut de Govern i Polítiques Públiques, Departament de Ciència Política i Dret Públic, Edifici B -Campus de la UAB, Universitat Autònoma de Barcelona, E-08193 Bellaterra (Cerdanyola del Vallès), Spain, Joan Font (jfont@cis.es),Centro de Investigaciones Sociológicas, Montalban, 8, 28014 Madrid, Spain, Pau Mas (pau.mas@uab.es),Institut de Govern i Polítiques Públiques, Departament de Ciència Política i Dret Públic,Edifici B -Campus de la UAB, Universitat Autònoma de Barcelona, E-08193 Bellaterra (Cerdanyola del Vallès), Spain, and Sergi de Maya (sergiode.maya@uab.es),Departament de Relacions Institucionals i Participació,Generalitat de Catalunya, Avda.Diagonal 409, E-08008 Barcelona, Spain.

Table 1
Visibility and intensity of selected DDPs

Table 2
Comparison of the means of the two dependent variables for municipalities with and without DDPs

Difference 99-03 for party with most votes in 1999 (XVOT1)
Source: Own elaboration

Table 3
Differences in means without outlier

99-03 for governing parties since 1999 (XVOTGOV) Total Without participation With participation Difference T test P
Source: Own elaboration income is the income available to families of a given territory for consumption or saving.It represents the difference between resources and expenditure.

Table 4
Multiple regression: effect of participation on the differences between 1999 and 2003 in the results of parties with most votes and governing parties Source: Own elaboration

Table 5
Difference in mean values* between municipalities with DDPs and their respective control cases

Table C
Results 1999 and 2003 local elections in Rubí a Governing parties for the 1999-2003 period b Party that left the local government during the 1999-2003 period Source: Ministry of the Interior