(2018) Money, sex and broken promises : politicians' bad behaviour reduces trust.

This article develops and tests empirically a theory of the effect on political trust of forms of behaviour that violate social, political and legal norms about how politicians ought to behave. These include taking money for favours, over-indulging in private life and making misleading promises to win votes. The evidence comes from a specially designed survey in Britain, France and Spain, countries where popular distrust of politicians appears greater than illegal political behaviour. Bad behaviours, especially abandoning election promises once in ofﬁce, have a much stronger effect on distrust of political parties that do differences in partisanship. Comparing national regressions shows that the impact of bad behaviours is very similar in Britain, France and Spain.


Introduction
For citizens to decide whether politicians are to be trusted or distrusted requires applying standards to evaluate their behaviour. Standards are social constructs; bad behaviour can refer to breaking legal, social or political standards (cf. Tänzler et al. 2012). If the behaviour of politicians is consistent with laws and social norms, this should encourage popular trust. But if popular representatives break these standards, this should encourage popular distrust (cf. Mishler and Rose 1997;Rose 2014).
The positivist approach of contemporary social science research favours defining standards by reference to reliable and verifiable measures. Laws are particularly suitable for this purpose, since they are readily available formal statements of standards. In addition to those applicable to all citizens, officeholders are subject to laws specific to their office, such as anti-bribery measures prohibiting the abuse of public office for private gain. If politicians act within the law, this may be deemed to make them trustworthy. Politicians whose behaviour is subject to public controversy favour this approach because they can defend what they do by emphasizing that their actions are legal. However, this standard reflects what Jennings (1985) has described as 'moral minimalism', because it excludes ways in which politicians can break public standards without breaking any law (cf. Jennings et al. 2016).
Social norms are culturally constructed by combining informal standards held in the mind and legal standards formally inscribed in statute books. Norms establish social psychological expectations about how people ought to behave. In their relations with fellow politicians, officeholders are expected to comply with informal 'rules of the game' if they are to be trusted by colleagues (Fuchs 2007). The norms that ordinary citizens apply to their political representatives can combine standards used in their personal relations and specifically political standards. Violating informal norms can be described as bad behaviour, while violating formal standards is an illegal act (Rose and Peiffer 2018, chapter 1). 1 Insofar as politicians comply with social norms about how they ought to behave, this should encourage political trust. Insofar as they behave badly, it should encourage distrust.
Trust, a word with meanings varying with context (Levi and Stoker 2000), is important for political institutions to implement collective decisions (Easton 1965). If there is a widespread belief that in unforeseen situations politicians will act as beneficially as possible, politicians can expect widespread popular acceptance of their decisions (Newton 2007;Dowding 2018, p. 33f). Trust is particularly important in theories of representative democracy. Citizens give direction to government by casting their vote for the candidate or party that they trust will be most likely to act in accord with their interests and values (cf. Miller and Stokes 1963). Theories of principal-agent relationships assume that voters, as political principals, can trust the representatives they elect to act in accord with the mandate that voters give them. However, a variety of realist theories and empirical studies question whether politicians can be trusted to act honestly and to represent voters' views (cf. Wlezien 2004;Gailmard 2014;Allen et al. 2016).
This article is innovative in developing a theory of how different types of bad behaviour affect political trust and testing it with a specially designed survey questionnaire. It differentiates legal, social and political types of bad behaviour 1 The term corruption was historically used to refer to violating standards of all kinds; this meaning survives in reference to a computer file being corrupt. Because the term corruption has been loaded with many different meanings (see Heidenheimer and Johnston, 2002;Rose 2014;Philp 2015;Heywood 2017;Ardigo and Hough 2018;), in this article the term bad behaviour is used to refer to the violation of informal social norms about actions of politicians. and different types of punishment for each. To guard against contextual effects confounding generalization, the data come from national sample surveys in Britain, France and Spain, three countries where corruption in the legal sense of bribery is not widespread but where there is evidence of limited trust in democratic representatives (Dogan 2005;Norris 2011). In all three countries, statistical analysis shows that it is politicians' violating their electoral mandate that gives the biggest boost to distrust, Moreover, this effect remains strong after controlling for differences in respondents' partisanship and socio-economic status.

A theory of bad behaviour and distrust
The theoretical importance of trust invites the question: What causes individuals to trust representative institutions central to democratic politics? Many theoretical explanations have been advanced (Zmerli and van der Meer 2017). They range from social psychological predispositions to trust face-to-face relations and perceptions of the political and economic performance of political institutions to democratic institutions promoting trust (Warren 1999).
Theories tend to focus on political trust; 2 distrust and scepticism are treated as residual categories (Mishler and Rose 1997). Because of our focus on the effect, if any, of breaking standards, our model focuses on their effect on distrust ( Figure 1). Breaking three different types of standards-laws, moral social norms or a political mandate-can increase distrust. In addition, the more severe the punishment deemed appropriate for breaking a standard, the greater the increase in distrust. The model also controls for the potential effect of an individual's partisanship and socio-economic characteristics. Our first hypothesis calls attention to three different types of standards that, if broken, can increase distrust. By including social and political as well as legal standards, it allows for a broader range of influences than an exclusive focus on legal violations, such as taking a bribe. It also leaves open to empirical investigation whether the effect of each standard is independent of the others, whether they form a single underlying attitude, or whether some standards have a significant effect on trust while others do not (cf. Seyd 2016).
H1: The more people see politicians as breaking legal, social or political standards, the more likely they are to distrust representative institutions.
Laws set out clear and enforceable rules for assessing whether a politician's behaviour breaks a legal standard, and a law court can make a judicial determination about whether or not a politician's behaviour has broken a law. Lawyers are not the only social scientists relying on laws as the major criterion for discriminating between good and bad behaviour. Many social science definitions of corruption instance violations of laws against bribery as a prime example. The theory and practice of public administration focus attention on the adoption of laws and bureaucratic regulations that will promote behaviour by officeholders that complies with laws and regulations (see e.g. Rose-Ackerman and Soreide 2011). Deciding what statutes define as legal and illegal is a responsibility of politicians. Where anti-corruption laws affect their own interests, such as financing the cost of political campaigning, politicians can include loopholes that effectively allow them to accept money in ways that are legal but may be inconsistent with social norms. National practices vary in the extent to which politicians are subject to special jurisdictions, as has been the case in Britain when Parliament acts as the judge of the behaviour of errant MPs (Hine and Peele 2016) or have immunity from prosecution while in office (Wigley 2003).
Social norms set standards of behaviour that are appropriate and those that are not. Instead of being set out in black-and-white legal statutes, they are in the minds of individuals, comprising cultural values and beliefs relevant to social and political roles (cf. Welch 2013). In a democratic political culture, for example, norms emphasize that citizens ought to vote when an election is held and that elected politicians ought to represent the views of their voters. The behaviour of individuals in parliament and government reflects not only formal rules but also informal norms (March and Olsen 1989;North 1990).
By definition, a democratic political system requires politicians to act in keeping with their role as popularly elected representatives. Insofar as these standards are widely shared in the population, they reflect non-partisan rather than partisan values, for example, not making ethnic or racist slurs about fellow citizens. When deciding policies, politicians are expected to respect the mandate that they receive from their voters (cf. Thomson et al. 2017). Politicians are also subject to norms about personal behaviour that is inappropriate because it brings their public office into disrepute, such as appearing drunk on television or using unflattering obscenities to describe other politicians, voters or countries (cf. Allen and Birch 2015).
Informal social and political norms cannot be enforced in a court of law but they can be judged in the court of public opinion. Print, television and social media can hound politicians by publicizing activities that are deemed to violate informal norms, and public opinion polls can act as a quasi-jury rendering a popular verdict about whether the accused has engaged in bad behaviour. If politicians are ashamed by the exposure of behaviour they thought would be kept private, they can resign office voluntarily. If they are hesitant to do so party leaders can offer an embarrassing colleague a choice between resigning or being sacked (Jacquet 2015).
Legal and normative standards of behaviour can be in conflict, since opinions of citizens about how politicians ought to behave are subjective popular judgments, while decisions about what is illegal are made in the courts. For example, participants in the #MeToo movement have publicized politicians behaving in ways that they judge as violating contemporary standards of gender relations. Even though few politicians have faced court charges for such activities, many who are the object of #MeToo complaints have accepted that their behaviour is shameful and apologised or left office. A similar disjunction between legal and informal popular standards was demonstrated when the parliamentary expenses claimed by almost 400 British MPs were leaked. The media headlined as scandalous expense claims, such as that for cleaning the moat around one's stately home. Only a handful of MPs were convicted of making an illegal claim, but most MPs repaid some expenses that met parliamentary standards but that they did not want to defend in public, and some decided not to stand for re-election (cf. Kavanagh and Cowley 2010, p. 311ff, 398ff;VanHeerde-Hudson 2012).
The punishment appropriate for breaking standards varies in accord with the legal maxim that it should fit the crime. The more serious the violation of a standard is, the more severe the punishment should be. Moderate punishments can be assigned to activities that are considered inappropriate for a public officeholder but not damaging to public policy. Some activities may be tolerated and not result in any punishment. For a politician to be photographed drunk in public does no harm to anyone but himself or herself, while a politician who accepts money for fixing a contract for a constituent who builds an unsafe bridge at a grossly inflated cost violates both legal and social standards.
H2: The more people see the violation of standards as serious, the more likely they are to distrust representative institutions.
The punishment of Chris Huhne, a British Cabinet minister, illustrates gradations of seriousness. When his car was caught by a speed camera, his wife pleaded guilty to a speeding offence, an experience common to many Britons and not requiring an apology. When months later Huhne was indicted for perverting the course of justice by pressuring his wife to lie to the court to avoid losing his motoring licence, he was so shamed that he immediately resigned as a Cabinet minister. After pleading guilty to the charge, he also resigned his seat in Parliament and was given an eight-month prison sentence.
Since bad behaviour by politicians may not be the only stimulus to political distrust, our model controls for the effect of partisanship, which is often a significant source of disagreement in the application of standards. A politician accused of breaking standards may counter-charge by saying the attack is motivated by partisan opponents. This response was employed by President Bill Clinton after being impeached by the Republican-controlled House of Representatives and is being used daily by President Donald Trump to defend himself in the court of public opinion. Partisan loyalty can influence citizens to ignore bad behaviour by other members of their party in accord with the maxim 'my party, right or wrong'.
Our model also controls for socioeconomic status, since sociological theories hypothesize that differences in status influence attitudes towards corruption (cf. Heath et al. 2016). A British study has found that higher-status people are more likely to be tolerant of the behaviour of politicians and have more trust in political institutions (Allen and Birch 2015, p. 117). Inglehart (1989) has developed theories about younger and older citizens differing in the social norms that they apply in judging behaviour (cf. Schoon and Cheng 2010). Gender differences in representation may lead women to be less trusting because fewer women hold elected office.

Public perceptions of bad behaviour
Since democratic politicians are meant to represent citizens, a national sample survey is an appropriate means for obtaining evidence about how the public perceives the behaviour of their representatives. To complement the concentration on breaking formal anti-bribery laws found in Transparency International's Global Corruption Barometer (www.transparency.org/), we designed a questionnaire to measure the extent to which people see politicians engaging in bad behaviour and how serious they think these activities are.
To avoid basing conclusions on a single national case, the Politicians Behaving Badly (PBB) survey was conducted in three West European democracies, Britain, France and Spain. They were chosen because, by the conventional measure of illegal corruption, the payment of bribes is low in all three countries. A Eurobarometer Survey (2014) found that only 2% of French respondents, 1% of Spaniards and 0.4% of British respondents reported that they had paid a bribe to a public official in the past year. In the global Corruption Perceptions' Index produced by Transparency International, all three countries are rated in the top quarter, but they also differ from each other. The CPI rating of Britain is 82, France 70 and Spain 57. The French-based survey organisation, Efficience3, conducted telephone interviews with a random stratified sample of 1004 Britons between 4 and 22 January 2016; with 1003 French between 11 and 29 December 2015; and with 1000 Spaniards between 11 and 22 December 2015.
While each country has a distinctive history and political culture relevant to corruption and scandals (cf. Della Porta and Yves 1997), in the years leading up to the survey in all three countries there was substantial media publicity about politicians breaking standards. The British media competed in headlining stories of PBB in their private lives, offering to use their office in exchange for the payment of large fees, and misleading voters by making policy U-turns (VanHeerde-Hudson 2012; Hine and Peele 2016). In France, Le Canard Echainé has regularly published exposes of bad behaviour by leading politicians. The financial affairs of former President Nicolas Sarkozy and economics minister and later IMF head Christine Lagarde were subject to scrutiny and the sexual affairs of President Francois Hollande were well publicized. In Spain a substantial number of cases have been publicized about the illegal payment of money to major politicians in established parties (Orriols and Cordero 2016). At the time of the survey Prime Minister Mariano Rajoy was campaigning for re-election amidst allegations of involvement of corruption and the former economics minister was on trial for corruption and subsequently convicted.
To assess trust in representative institutions, a single question was asked: To what extent do you trust political parties? Replies were coded on an 11-point scale ranging from 0, no trust, to 10, complete trust. The focus on parties rather than individual politicians provides a common reference point across electoral systems that differ in whether people vote for a party list, an individual candidate or have a combination of choices. It also avoids the risk of contamination because of the popularity or unpopularity of a locally elected MP or national party leader at the time of fieldwork. National respondents differed only in the degree to which they withhold trust from parties that represent them. The mean British respondent gave the least negative rating, 4.2. In France the mean score was 3.7 and in Spain 3.0.

Breaking standards
Only a very small percentage of people are likely to have first-hand experience of a politician behaving badly, while almost everyone is exposed to media stories about politicians that can be used to evaluate them. Therefore, the PBB questionnaire asked people about their perceptions of politicians' behaviour. While this is not evidence of how politicians actually behave, it is sufficient to influence the subjective judgments that people make about whether politicians are to be trusted (cf. Figure 1).
The PBB question about politicians taking money-How many Members of Parliament take money from people who want political favours?-was intentionally ambiguous. It leaves open whether the money was an illegal bribe or was given as a material reward to an MP who had legally done a political favour. The former interpretation is implied when the British media videos a sting in which an MP offers to take money to help a fictitious business interest (see e.g. Insight 2018). When this is publicized, an MP usually claims that the payment was justifiable remuneration for representing an interest and no legal charge is made against the MP (Insight 2018) Only one-third think that most or all MPs take money for doing favours; the median respondent thinks that less than half do so, and those saying hardly any do so outnumber the proportion saying all MPs take money for favours (Table 1). Notwithstanding differences in how MPs are elected in Britain, France and Spain, there is little difference between countries in the extent to which MPs are seen as taking money for favours.
In all societies there is a distinction between behaviour that may be accepted in private life, but which, if it becomes public, may be judged as a violation of social norms about how politicians ought to behave (cf. Sarmiento-Mirwaldt et al. 2016). While the media are quick to headline-specific examples of a politician's private life as scandalous, to find out if such stories are generalized to politicians as a class the PBB survey asked: How many politicians in their private life overindulge in drink, sex or drugs? In all three countries, two-thirds or more thought that only some or hardly any politicians over-indulged in their private behaviour, and cross-national differences in perceptions were limited (Table 1).
A basic assumption of representative democracy is that citizens can compare the different policies that politicians offer and vote for the one that comes closest to their view, confident that, if elected, their choice can be trusted to deliver the mandate that their voters give (Schumpeter 1952). To determine whether voters trust politicians to do what they promise, the PBB survey asked: How many politicians promise to do one thing if elected and then do the opposite after being elected?
The question made no allowance for politicians pleading extenuating circumstances when they execute a U-turn and abandon an election promise. Insofar as voters see politicians as unscrupulously seeking their vote, they are less likely to be trusted. Of the three standards asked about, the one most often seen as violated is the political standard that politicians should not make misleading election promises. Among Britons, almost three-quarters see most elected representatives as political hypocrites; the proportion rises to more than five-sixths in France and seven-eighths in Spain (Table 1).
All in all, the public tends to see the behaviour of politicians in shades of grey rather than in black-and-white terms. Less than 2% think all politicians break all three standards of behaviour and less than 1% think hardly any do so. Although respondents were making an evaluation in their national context, their judgments tended to differ more between types of activities than between countries. In Britain there is a difference of up to 43 percentage points between the perception of most MPs misleading voters and most over-indulging in their private lives, In France, the contrast between misleading voters and taking money or over-indulgence rises to 58%. Spaniards likewise discriminate in their perception of bad behaviour: There is a 68 percentage point difference between those seeing politicians misleading voters and those seeing them over-indulging in their private lives.

Seriousness of bad behaviour
The PPB survey included a nuanced measure of the seriousness of corruption. Immediately after asking about the frequency with which a standard was broken, people were asked what should happen to a politician who violates a particular standard. Up to five alternatives were offered. They ranged from the most severe in terms of its consequences-going to jail or leaving their office-to nothing need be done. Intermediate categories included paying a paying a fine or publicly apologising. Since these punishments are not mutually exclusive, respondents could and often did endorse multiple punishments. There is overwhelming agreement that breaking each type of standard is serious; on average <1% say nothing should be done. Even though making a public apology is considered necessary, it is not deemed sufficient to deal with the violation of informal standards about how politicians ought to behave (Table 2). There is little difference between Britons, French or Spaniards in the seriousness with which people treat taking money for favours. Five-sixths think that politicians who take money for favours should lose their post and more than half think they should also go to jail. A majority likewise think that politicians who overindulge in their private lives should forfeit their public office, but few think such behaviour deserves a jail sentence. The cross-national assessment of the minority who over-indulge shows significant differences. In Spain, 71% think those who do so privately should lose their public office, in France 56% take this view, and in Britain 45% endorse severe punishment.
PBB respondents talk tough when asked: If someone you voted for did this (that is, broke a promise), what would you do at the next election? More than two-thirds of Spaniards and about three-fifths of Britons and French say they would vote for someone else. Only one in nine say they would vote the same. However, replies to this hypothetical question do not appear to be matched by the proportion of

Testing the effects on political trust
Even though our three European countries appear similar in attitudes towards PBB, it does not follow that the determinants of political trust are the same in each country. To assess whether this is the case, we conducted separate regression analyses in each country. 3 If the same indicator is significant in all three national contexts, this is especially robust evidence of support or rejection of a hypothesis and, if results are the same in two of the three countries, this shows substantial support. To provide a clear focus on substantive relationships, we also report the results of a regression analysis that pools the three national surveys (Table 3). Given sample sizes of one thousand, we use 0.05 or lower as our standard for statistical significance in the three national analyses and 0.01 in the larger pooled database.
Since the impact of significant variables on trust can vary substantially, we calculate predicted probabilities for the effect on trust of an independent variable changing from its lowest to its highest value. 4 Coefficients with a minus sign show that the variable reduces trust, but the size of their impact varies. Thus, a coefficient of À3.61 for the effect of politicians lying to get votes reduces Spanish trust in political parties by more than three and one-half points on the 11-point trust scale, while a coefficient of À0.93 for the effect of over-indulging reduces trust by just under one point.
Altogether, empirical analysis provides strong support for our model of the effect of corruption on individual trust in the political parties that represent them. In all three national analyses, the OLS regression accounts for a high level of variance: 27% in Britain, 26% in France and 25% in Spain in the pooled three-nation analysis. Moreover, notwithstanding many historical differences between the countries, there is consistency in the national evaluation of influences. The three 3 While a theoretical case can be made that corruption and trust have reciprocal effects on each other, to test this with a Structural Equation Model requires variables that can serve as control instruments and these do not occur in the PPB survey (cf. Wroe et al. 2013m, p. 179ff). types of standards are significant in all three countries. So too is partisanship. In contrast, the seriousness of punishment for violating standards is of limited significance in all three countries, and so too is the effect of educational status. There is extremely strong support for our first hypothesis (Table 3). If people perceive politicians as breaking any of the three informal standards, they are significantly more likely to distrust political parties. Moreover, the effect is found in all three countries. The standard that generates the most distrust is central to the theory of representative government: elected representatives should keep the promises that they made to voters when seeking votes. After controlling for all other influences, the pooled data analysis estimates that trust in politicians will fall by 2.6 points among people who see politicians saying one thing to win votes and doing the opposite once in office. The effect is biggest in Spain; dishonesty in making political promises lowers trust by 3.6 points.
The effect of politicians being seen as taking money for favours is second in impact. In the pooled data analysis it lowers trust in political parties by an estimated 1.48 points. While this is substantial, the effect is two-fifths less than that Note: All independent variables have been recoded to a range of 0-1. Regression coefficients show the maximum effect on the 11-point scale for trust of an independent variable moving from its lowest to its highest value. of politicians breaking promises; moreover, the impact varies substantially between countries. It is twice as high in Britain as in France. If citizens think most of their politicians over-indulge in their private lives, this has a significant effect on trust too; however, this effect is less than breaking other standards. After controlling for taking votes and taking money, in all three countries it lowers trust by under one point (Table 3). Contrary to hypothesis 2, measures of the seriousness of bad behaviour have no consistent effect on political trust. Endorsement of the strictest punishmentsend violators to jail or eject them from office-has no significant effect in the pooled data set or in France and Spain. While significant in Britain, the size of the effect is much less than that for misleading voters or taking money (Table 3). Moreover, in all three national surveys, people who see corruption as an important problem do not differ significantly from their fellow citizens in their level of political trust. The significant effect in the pooled data set is small and may simply be a by-product of having three times as many interviews when calculation significance (Table 3). The contrasting findings for the first and second hypotheses shows that it is not the seriousness of what is done but whether a politician behaves badly that depresses trust.

Partisan effect
The importance of controlling for party loyalties is robustly confirmed. At the time of the survey, the governing party in Britain was the Conservatives, in France the Socialists held the presidency, and in Spain the Popular Party was in office. Notwithstanding major differences between these parties, in all three countries those who voted for the governing party are substantially more likely to trust parties (Table 3). The effect holds after controlling for the negative impact on trust of believing that politicians who hold office renege on the promises they make when campaigning to win office. While the halo effect of supporting governments of different ideologies has a positive impact on trust, boosting it by as much as 1.56 points in the pooled database, it does not offset the negative impact of politicians breaking their promises to voters. Moreover, in Spain, where the governing party has been mired in all kinds of allegations of bad behaviour, the boost to trust among its supporters is less than one-third of the depressing effect on trust of politicians misleading their voters.
In all three countries those who do not vote for the governing party considerably outnumber government supporters. They include supporters of established opposition parties that have been in government and hope to return; new parties that have never been in government; and those without any party preference. Established parties have the most grounds for being distrusted, since they have previously been in government; they include the Labour Party in Britain, the Socialist Party in Spain and the Republicans in France. After controlling for all other effects, in all three countries supporters of the governing party, whether Socialist or conservative, tend to be significantly more trustful and supporters of the established opposition parties significantly less trustful. Insofar as holding office encourages trust, this implies that the alternation of established parties between government and opposition will lead partisans to alternate between sometimes trusting and sometimes distrusting a government that in theory is democratically representative. In other words, the impact of partisanship is not ideologically based; it reflects the temporary electoral success of competing parties.
In preliminary regressions, we tested whether those who do not support established parties rotating between government and opposition tend to be more distrustful of parties. This is consistent with non-established parties such as UKIP in Britain, the National Front in France and Podemos in Spain being described as protest parties. In fact, it makes no difference in trust, even in Spain, where protest parties were strongest at the time of the PBB survey. One possible explanation is that in Spain the overall level of trust in parties was very low. In Britain by the time of the survey at the beginning of 2016 the Conservative government was seeking to regain supporters by implementing UKIP's flagship policy of having a referendum on the UK withdrawing from the European Union. Supporters of protest parties were not significantly more distrustful than the national average and the same was true for non-voters. In other words, those who have stopped voting for the established parties of government are not so much angry as apathetic about the claims of governors to be trustworthy.
Notwithstanding the emphasis in sociological theories on the pervasive effect of socio-economic status on popular attitudes, empirical tests of the effect are inconclusive (cf. Heath et al. 2016;Zmerli and van der Meer 2017). None of the three demographic indicators-high status, low status and being a woman-has a significant effect on trust in the pooled analysis. The consistent failure of socioeconomic status to show a significant effect indicates that people who are more informed about politics are just as likely to distrust political parties as citizens with a secondary school education. The same is true for those with the low socioeconomic status, except in France, where the effect is marginally significant but the impact is slight (Table 3).
Gender has no significant influence in the pooled analysis, Britain or Spain. French women are inclined to be less positively trusting but the size of the effect is small. In a preliminary regression analysis, we tested the effect of age which the PBB survey divided into four groups. In all three countries the youngest group, age 18-24, did not differ significantly in their trust from the oldest, over 55; the same was true of the intermediate age groups. This implies that distrust in government is an all-age phenomenon independent of specific events in early socialization or of any inter-generational changes in standards of private morality.

Implications for representation
Our evidence shows that to confine the effect of corruption on trust to actions breaking laws is to misread the minds of voters. Breaking informal standards is a major cause of distrust too. The most important informal influence on trust is politicians saying one thing to get elected and doing the opposite once in office. This implies that attempts to improve trust in politicians by adopting formal laws will be inadequate, since major informal influences on distrust are within the law.
Although our theory is in principle applicable to all political systems, the evidence comes from only three European countries. Notwithstanding differences between the histories, institutions and party systems of Britain, France and Spain it is striking that separate regression analyses come to the same conclusion. The violation of informal standards about how politicians ought to behave substantially depresses trust in democratically elected politicians. Whatever the standard broken, the minimum punishment tends to be loss of office. Moreover, people who think a less serious punishment is acceptable are just as likely to distrust parties as those who consistently favour severe punishments.
A single survey cannot show whether perceptions of bad behaviour have been increasing. Surveys over the past several decades show that distrust in political parties and politicians is long-standing (Klingemann and Fuchs 1995; Norris 2011; van der Meer and Hakhverdian 2017). Whiteley et al. (2016) argue that short-term changes in trust are simply fluctuations around a long-term equilibrium of limited trust. This suggests that many citizens have long-established predispositions to trust or distrust their representatives, views that are, at most, only temporarily altered by events or the behaviour of a particular party leader (cf. Allen et al. 2016).
Because standards of bad behaviour are informal, they are easily contestable. In the absence of substantial evidence of wrongdoing, a politician accused of acting badly can reject allegations. In the face of evidence, a politician can claim that there is no wrongdoing as long as there is no violation of a formal legal standard. When allegations and evidence of bad behaviour imply that laws have been violated, as in the investigation of Russian meddling in the 2016 US election, the response can be, as in the case of the Administration of Donald Trump, a counter-charge against critics of corrupt or bad behaviour. Counter-charges create a race to the bottom among parties 'normalizing' their behaviour by making their opponents appear to be acting as they do. Insofar as countercharges are convincing, this is likely to fuel popular distrust in all politicians.
Partisanship qualifies the extent to which individuals distrust political parties (cf. Anderson et al. 2005). Among those supporting the government of the day, 64% see all or most politicians as reneging on their promises, and this is the case for 82% favouring establish opposition parties, 87% if without a party affiliation and 89% of supporters of protest parties that have never been in government. While the effect of being in government makes supporters more trusting, the fact remains that a substantial majority in citizens, whatever their partisan status, see politicians as misleading voters.
Distrustful citizens do not accept the classic Schumpeter (1952) model of democratic elections as offering a choice between alternative parties of government. Instead of seeing degrees of corruption that allow for a choice of one party as a lesser evil (Cordero and Blais 2017), they see both alternatives as forming an untrustworthy cartel (cf. Katz and Mair 2009). Confronted with a choice of parties that cannot be trusted to keep their pre-election promises, citizens can decide not to vote. However, while election results show fluctuations in turnout, there has been no consistent downward trend in turnout in West European countries. However, there has concurrently been a big rise in votes for unestablished parties that have never been in government (Mudde 2016;Wagner and Meyer 2017). In France, outsider candidates from the right and left took almost half the vote in the initial round of the 2017 presidential ballot. In Spain, the two established parties of government together won just under half the vote in 2016. In Britain, an outsider party, UKIP, has been so successful in getting the Conservative Party to adopt its major policies that it lost its own electoral support.
Although the bad behaviour of politicians has destabilized party systems, it has not led to a loss of trust in democracy as an ideal (cf. Ferrin and Kriesi 2016; Kumlin and Esaiasson 2012). Distrustful citizens are not voicing an attack upon their democratic political system. They are expressing dissatisfaction with the extent to which politicians fall short of formal and informal standards of democratic behaviour that they would like political elites to change (Vasilopoulou 2018). In doing so, citizens are endorsing Winston Churchill's (1947) defence of democracy as 'the worst form of government except for all those other forms that have been tried'.

Funding
The work of the first-named author was supported by an Economic and Social Research Council grant number ES/I03482X/1 about the global experience of corruption. The PBB survey was funded by the Wissenschaftszentrum Berlin and the Centre for the Study of Public Policy, University of Strathclyde. Earlier versions were presented to the Parliamentary Standards Committee of the UK Parliament