Introduction

After the strongly fought, polarizing and problematic presidential election of 2006 in Mexico, the political parties represented in the Federal Congress started a comprehensive evaluation of the electoral process. The goal was to identify and eliminate possible sources of conflict that could lead to an outcome similar to that of the 2006 presidential election, where “at the level of rhetoric and ritual at least, the post-electoral conflict had led to a quasi-revolutionary situation” (Schedler 2007, 92).

The negative campaigns were one of the main targets of this electoral reform. In this article, I use the following definition of negative campaigning as “a speech, act or communication that contains one or more attacks that are personally, voluntarily and directly addressed toward political opponents” (Nai 2013, 46), excluding from this definition attacks on the adversary’s platform or “issue-based attacks.” In 2007, the political parties amended the Constitution and banned the use of negative advertisements as a measure to prevent “further harm to electoral participation,” despite arguments that this measure might impair freedom of speech.

The literature on negative campaigns has not reached a consensus about the effects of negative advertisements on turnout, and the attention of researchers has focused so far on elections in the United States. Some scholars have argued that negative advertising reduces turnout (Ansolabehere et al. 1994; Ansolabehere, Iyengar and Simon 1999), while others have not found any evidence to support those claims (Arceneaux and Nickerson 2010; Finkel and Geer 1998; Lau et al. 1999; Lau, Sigelman, and Brown Rovner 2007; Robideaux 2004; Wattenberg and Brians 1999). There have been a few studies of the relationship between negative campaigns and turnout outside of the United States, but their findings are not consistent: they either point to a fall in voters or show mixed results (Desposato 2007), depending on the context (Nai 2013) or the lack of any relationship (Walter and van der Eijk 2015). Therefore, the purpose of this paper is to assess the effect of negative advertising on voter turnout in a different context: the 2006 Mexican presidential election. Did negative advertising lead to a lower turnout? Are the findings about elections in the United States applicable to other contexts, particularly new democracies? So far there have been few comparative studies of the effects of negativity on turnout across different political systems (Desposato 2007; Nai 2013; Walter and van der Eijk 2015) and examining the case of Mexico may throw more light on the subject. I argue that there are two reasons to expect minimal or null effects of negativity on turnout: the fact that Mexico is a multiparty system, and that the system is rather institutionalized.

In the first section I briefly describe the 2006 presidential election in Mexico. Then, I review previous studies of the expected effects of negative campaigns on electoral participation. Next, using a post-electoral survey, I look for evidence of the effect of negative campaigns on turnout in the case of Mexico. Finally, I conclude that negative advertisements do not seem to have had a significant impact on turnout there.

1. The 2006 Presidential Election and the 2007 Banning of Negative Advertisements

If we consider Alvarez et al.'s (1996) definition of political regimes, in which a system is considered authoritarian if incumbents subsequently hold but never lose elections, then Mexico made a transition to democracy in 2000 when the Partido Revolucionario Institucional (PRI) finally lost the presidential election for the first time since the foundation of the party in 1929.¹

However, the second presidential election held since the transition to democracy ended in an atmosphere of social polarization. The 2006 presidential election was won by the Partido Acción Nacional (PAN) by a narrow margin. The runner-up candidate, from the left coalition Alianza por el Bien de Todos,² did not accept the results and called for massive street protests, and the impartiality of the electoral authority was put in doubt (Schedler 2007). In order to understand that situation and the federal electoral reform of 2007 -which banned negative advertisements-, it is necessary to point out that the 2006 election was shaped by a conjunction of several factors never seen before in Mexican politics.

In the first place, to an extent never seen before in Mexico, that election was overwhelmingly dominated by negative advertisements on television. The three main political parties fully employed that strategy (Aguayo 2010). One reason which enabled them to do so is a Mexican law by which all registered political parties receive generous pubic funding for their ordinary activities and also their campaigns. In the 2006 presidential election, eight political parties received a total public funding of four hundred and thirty five million dollars (Table 1).

In election years, political parties receive an additional amount of money to finance their campaigns, and in 2006 a big proportion of it was used to finance ads in the mass media.³ On average, those parties spent two thirds (66%) of their media budget on television ads, as shown in Table 2.

In 2006, the three main political parties (or their coalitions) bought 38,145 spots on TV, distributed in the following way: 11,904 (PAN), 10,425 (PRI-Green Party) and 16,316 (PRD-Left coalition).⁴ However, instead of focusing on the issues or their opponents´ policies and platforms, the three leading presidential candidates, in line with the typical advice of political consultants nowadays, chose a strategy of attack,⁵ which almost literally consisted of the following steps:

“Advertise early if you have the money [...] 2. Go negative early […]. 3. Appeal to the heart and the gut, rather than to the head. 4. Define your opponent to the voters before he or she can define him/herself or you. 5. If attacked, hit back even harder. 6. it’s easier to give voters a negative impression of your opponent than it is to improve their image of you […].” (Kamber 1997. 46-47

As a consequence, there was a flood of legal appeals by the parties who felt attacked by their rivals, in which they called on the Instituto Federal Electoral (IFE) to prohibit such ads. When the PAN´s ads made negative references to the presidential candidate of the left coalition, the parties of the left claimed they were the victims of “black propaganda” or a “dirty war” and asked the IFE to intervene.

Article 38, letter p, of the COFIPE (federal electoral legislation) prohibited political parties from using phrases which imply “diatribe, libel, dishonor, injury, defamation, or denigrate citizens, public institutions, and other political parties and their candidates.” But there were no clear rules or specific sanctions for negative advertisements. Given the intensity and dynamism of the campaign, the Federal Electoral Court created a specialized procedure to deal with negative advertisements, in which the IFE had eight days to make a decision. The IFE made 23 special rulings and ordered the withdrawal of 22 ads bought by the PAN, the PRI and left coalition, respectively.

The complainant parties said that the IFE acted too late and by the time the ads were withdrawn, the images of their candidates were already damaged. It is also thought that the negative ads exacerbated the polarization of political opinion during and after the campaign.

However, they were not the only reason why that election became so controversial. A second factor was the early start of campaigning, at least informally. As early as the second half of 2003, opinion polls on the potential candidates were being conducted, which forced many politicians to focus their energies on the upcoming campaign, long before its formal start.

A third factor responsible for the polarization we have referred to was the unsuccessful attempt to impeach the then mayor of Mexico City in 2005, Mr. López Obrador, which would have barred him from running for the presidency.

Finally, this was the election with the narrowest margin of victory in Mexican history. According to the final count by the Electoral Tribunal, there was a difference of barely 0.56% (243 thousand votes) between the winner and the runner-up, a marked contrast with the previous two elections. In 1994, the winning candidate (of the PRI) beat the candidate of the PAN by a margin of 23.8% (PAN), and in 2000, the candidate of Alianza PAN-PVEM beat the candidate of PRI by a margin of 6,4% (Table 3).

Given this context of polarization and confrontation, the Mexican federal electoral reform of 2007 tried to address the main sources of electoral conflict. Among other controversial points, the political parties focused on negative advertisements. The draft of the constitutional amendment which the Committee on Constitutional Issues presented to a plenary of the Chamber of Deputies (the Lower House of the legislature):

“The electoral campaigns have turned into a marketing competition dominated by rules of the mass media that are extraneous to politics, in which short spots are the fashion, candidates are treated as a product and citizens are reduced to the position of simple consumers. This is a tendency which renders politics futile, harms democracy and hinders electoral participation. Freedom of speech has clear limits, and as stipulated by article number 6 of the Constitution, this freedom is not synonymous with denigration or slander […] But it is necessary to point out that the new limitations which will be laid down by article 41 of the Constitution do not apply to citizens, the press or the mass media, but the political parties. They are the ones which are prohibited to use expressions which denigrate or slander public institutions and the people.” (Comisiones Unidas de Puntos Constitucionales y de Gobernación 2007)

Even granting that negative ads discourage electoral participation, which is subject to empirical confirmation, the tendency among constitutional tribunals in the world is that free speech should be protected. Constitutional courts have ruled that when two rights, such as the right to honor and freedom of speech, are in conflict, the second has primacy over the first, especially in the case of a public figure. For example, when the Supreme Court of the United States ruled on the issue of free debate, in the New York Times Co. v. Sullivan (376 US 254. 1964), Justice Brennan wrote: “we consider this case against the background of a profound national commitment to the principle that debate on public issues should be uninhibited, robust, and wide-open, and that it may well include vehement, caustic, and sometimes unpleasant sharp attacks on government and public officials.”

In a similar vein, in a landmark case that involved the PRI (SUP-RAP-009-2004), the Federal Electoral Court in Mexico laid the foundations for its view of freedom of speech (Barranco 2010), as follows:

“[…] Political parties, according to article 41 of the Constitution, play a primary role in the promotion and preservation of public opinion, political pluralism and the electoral participation of citizens. As was previously stated, when it comes to the particular duties of political parties with regard to the issues under consideration here, they must be mainly aimed at the discussion and analysis of important topics in the public arena. On that basis, it is easily understood that the messages which are consistent with those aims -meaning that they help to strengthen a free public opinion, improve political pluralism and develop democratic culture in society- enjoy a special protection in the legal order and therefore, the potential negative criticisms that those messages might have are legitimized, even those which turn out to be extremely negative, harsh and intense (SUP-RAP-009-2004).” (Tribunal Electoral del Poder Judicial de la Federación, 2004)

Instead of using such arguments in favor of freedom of speech, the politicians in the Congress supported a limitation of it for reasons of their own when the controversy over the 2006 election led to an urgent call for electoral reform. Since many candidates would be at risk if their unethical behavior in the past were exposed, the constitutional prohibition of negative ads would be beneficial for them.

Political scandal is a permanent threat to all politicians who aspire to positions of power and therefore it would be rational for them to eliminate that risk. According to Ríos Vega (2010), it is absurd to assume that the scandal caused by negative advertisements is not motivated, in part, by party-political interests since the credibility of candidates with doubtful pasts would be at stake.

Despite Mexican legislators’ claim that negative advertisements might damage electoral participation, it seems difficult to attribute the low levels of turnout in federal elections to them. Before the widespread use of negative advertisements in the 2006 election, there was a clear negative trend in turnout levels since 1994, in which presidential elections showed slightly higher levels of turnout than the mid-term ones (Figure 1). Also, it should be noted that turnout rose by almost 17% between 2003 and 2006, right after the most intensive use of negative advertisements in Mexican history. This would suggest that the relationship between negative ads and turnout is precisely the opposite of what the committee on constitutional issues claimed.

Source: Compiled by the author from National Electoral Insitute

Figure 1 Turnout in Federal Elections, 1991-2009 

However, we do not know if the low levels of turnout can be attributed to negative advertisements, or other structural factor like the interest of citizens in politics, different levels of education and income, and so forth. Their negative opinion of the political parties might also explain, in part, why citizens are uninterested in voting. Polls show that political parties are among the least popular institutions in Mexico (Table 4). Citizens often perceive them as profit-seeking groups which are not responsive to their needs and demands.

One example of this dissatisfaction was a call for voters to cast a null vote in the 2009 mid-term election (Cisneros 2016), a movement which resurfaced in the 2012 presidential election.

Theoretical Expectations of Negative Campaigns

Can we expect that the constitutional ban on negative advertisements will improve turnout? Did negative campaigns in fact discourage people from voting in 2006?

Despite abundant research, the evidence about the effects of negative ads on turnout in still inconclusive. There are two broad and opposing schools of thought. One group of investigators suggest that negative advertising reduces turnout. The seminal work of Ansolabehere et al. (1994) claims that negative advertising discourages people from voting because it weakens their feeling of political efficacy: “voters who watch negative advertisements become more cynical about the responsiveness of public officials and the electoral process” (Ansolabehere et al. 1994, 829). To test this argument, they ran an experiment which showed that people exposed to positive advertisements had a mean efficacy higher than those exposed to negative ones. However, these results were questioned by Wattenberg and Brians (1999) who pointed out several flaws in their design and concluded “that Ansolabehere et al. exaggerated the dangers of demobilization posed by attack advertising” (Wattenberg and Brians 1999, 891).

Negative campaigns can also be related to voter turnout by looking at party affiliation. Ansolabehere et al. (1994) suggest that the supporters of the candidate that airs the negative attack will still vote for him or her, whereas the supporters of the candidate who is attacked might be discouraged from voting for that candidate.

Other researchers have sought to define the conditions in which negative campaigning might reduce turnout in a more precise way. The type of message might be an important variable. Personal attacks and the raising of irrelevant issues may alienate potential voters (Kahn and Kenney 1999), and the combination of insulting but relevant messages has an especially powerful effect (Fridkin and Kenney 2011).

Similarly, campaigns which defend the status quo are more likely to discourage the participation of voters who are already bored with or cynical about politics (Nai 2013). Finally, in new democracies with poorly institutionalized party systems, negative advertisements discourage citizens with low educational levels from voting (Desposato 2007).

Several studies have failed to find evidence that negative ads reduce turnout (Arcenaux and Nickerson 2010; Robideaux 2004; Walter and van der Eijk 2015; Wattenberg and Brians 1999). What is more, certain studies claim, oppositely, that negative ads makes citizens more interested in participating in politics. Geer (2006) argues that the emotional nature of negative campaigns stimulates an interest in politics among potential voters and should increase levels of turnout. While Kahn and Kenney (1999) generally agree, they argue that it will only happen in certain conditions, for example, if the negative ads raise issues which citizens regard as important, while if they are merely accusatory and ad hominem, the same citizens will be discouraged from voting: in short, potential voters are able to distinguish between different kinds of political ads.

Kahn and Kenney (1999) believe that negative ads may provide potential voters with more information about politics because they are striking and easy remember and also highlight the potential risks of electing some candidates or supporting certain policies. By the same token, citizens who only watch the t.v. news for entertainment and/ or unintentionally see negative ads on t.v. will acquire more knowledge of politics and that might increase turnout (Prior 2011).

Additionally, Geer (2006) argues that negative ads are more informative than positive ones because they need more evidence to back up their claims and make them more credible, so that the ads do not backfire against their sponsor, and thus they improve the quality of the information available to the electorate.

To summarize, studies of the subject have not reached a clear consensus on the effects of negative ads on turnout. Lau et al.´s (1999) review of 52 academic articles about negative advertising, which applied meta-analysis to the aggregate findings, concluded that the evidence for the effect of negative advertising on turnout is inconclusive: “there is little evidence to warrant the fears of those who believe that electoral participation is imperiled by the increasingly widespread use of negative political advertisements” (Lau et al. 1999, 858).

Expectations for the Mexican Case

In this section I present my expectations for the Mexican case, drawing on previous studies of negative campaigns.

Few have directly analyzed the effects of negative advertising on turnout, probably because the only election where this type of campaign was widely used was the 2006 presidential election, and after that it was constitutionally banned.

Pérez-Damazo (2009) finds that negative campaigns did not have any effect in the intention to vote for the PRI, and because they were not thought to be credible, attacks against the candidate of the left slightly improved the probability of voting for him and reduced the probability of voting for the candidate of the right. Beltrán (2009), in a study of the preferences of voters for the two main candidates in the 2006 presidential election, concludes that the relationship between advertisements and preferences is not direct, but mediated by two mechanisms that are frequently studied by analysts of the mass media: agenda-setting and priming.

Chong et al. (2010), in a field study of the effect of information on voter turnout in municipal elections in Jalisco, find that information about municipal spending that does not comply with federal guidelines leads to a drop in turnout. Although these studies employ either voter choice or turnout as dependent variables, none directly analyzes the relationship between negative advertisement and turnout.

One reason why negative campaigns may only have a minimal effect on turnout, and other variables, like the voter´s choice, is that campaigns have “minimal effects” in general. In the United States, for example, party affiliation is so strong that most voters are immune to campaign messages (Campbell et al. 1960; Zaller 1992). That may be attributed to context: “The United States has an institutionalized two-party system, a stable commitment to issues by the respective parties, and mass partisan preferences” (Desposato 2007, 5).

Some researchers have argued that in multiparty systems the effects of negative advertisements might be weaker because instead of abstaining from voting, citizens who are dissatisfied with their first preference are able to vote for another party. “Therefore, we can expect that in a multiparty system the effect of negative campaigning on turnout will be smaller” (Walter and van der Eijk 2015, 5). This may also apply to parliamentary regimes, since in a post-electoral scenario where is there is no clear majority, some parties will have to form coalitions with others (Dolezal, Ennser-Jedenastik, and Müller 2016; Walter and van der Eijk 2015). Since Mexico is a multiparty system, the effects of negative advertisements on turnout would probably be weaker than in the United States.

The degree of institutionalization of the party system is another factor that might weaken the effect of negative ads on electoral participation. “In weak party systems, most voters are at ‘risk’ of being demobilized by negative advertising, as partisanship and voter information levels are on average very low” (Desposato 2007, 5). In institutionalized party systems voters have strong “priors” about particular candidates or parties and therefore are more resistant to partisan messages (Desposato 2007; Green 2011). The Mexican party system can be considered as fairly institutionalized since it tends to report high levels of electoral stability in both presidential and legislative elections (Mainwaring, forthcoming study). Since the country´s transition to democracy, 80% of the votes have gone to the three major parties (the PAN, PRI and PRD). In addition, those parties have fairly strong organizations and meaningful identities (Moreno 2009). This is another reason why we should expect the effects of negativity on turnout to be minimal.

To sum up, I hypothesize that the effects of negative advertisements on turnout will be minimal or null in Mexico.

The Empirical Evidence

To test the hypothesis about the null effect of negative advertisements on turnout, I used the post-electoral survey (Mexico 2006) by the Centro de Investigación y Docencia Económicas (CEDE) as part of the Comparative Study of Electoral Systems (CSES), which had 1,591 respondents and was held one month after the end of the campaigns, between July 23 and August 2nd, 2006.

My dependent variable, Turnout, was measured in terms of the reported vote by the following question: “In the last presidential election many people were unable to vote because they had to work or travel or another reason: did you yourself vote?”. I coded the variable as 1 if the answer was yes, and 0 if the answer was no. Relying on the respondents´ memory of whether they voted or not might have overestimated the real turnout, due to a social desirability bias. To compensate for that, I excluded people who were not eligible to vote (e.g. the person was not a registered voter, or was under the age of 18 or did not have a voter´s ID).

For my key independent variable, exposure to campaign advertisements, I used the following questions as a proxy: “During the electoral campaign, do you remember any advertisement against Felipe Calderon?” and “During the electoral campaign, do you remember any advertisement against Lopez Obrador?” This resulted in two separate dummies which had the value of 1 when the respondent remembered a negative advertisement and 0 otherwise. Table 5 displays the frequency with which the respondents recalled negative campaigning against the leading presidential candidates.

My final measure of recall (Negative) is an additive scale constructed with the previous two variables. This variable takes the values of 0 if the respondent does not remember any negative advertisement, 1 if he or she remembers only negative advertisements against one of the candidates, and 2 if he or she remembers negative ads against both candidates. I expect that people’s recall of negative political advertisements will not have an impact on the reported vote in the 2006 presidential election.

A word of caution is in order here. Experts in the field agree that one of the shortcomings of the studies of the effects of TV advertisements on public opinion is that the recall data do not measure exposure to campaign advertisements. However, they recognize that this is the most widely used data (Ansolabehere et al. 1994; Wattenberg and Brians 1999).

a. Control variables

Political knowledge (Knowledge). Higher levels of political knowledge should increase the probability of voting (Delli Carpini and Keeter 1996; Verba, Schlozman, and Brady 1995). I measured this variable as an additive variable going from 0 to 3, adding one unit if the voter correctly answered three questions about the political system. 1) What are the Chambers of the Mexican Congress? 2) In general, how many years does a Deputy serve in his or her position? And 3) What is the name of the Governor of your State?

Political engagement (Engagement). The act of voting is also an expression of a person’s psychological involvement in politics (Lewis-Beck et al. 2008). Therefore, this variable should have a positive effect on turnout. I measured this variable with a dummy that takes the value of 1 if the respondent participated in at least one of the following activities: 1) Attached a sticker of a political party to his vehicle. 2) Posted a political advertisement on the window or door of his house. 3) Attended a political rally. The variable otherwise takes a value of 0.

Party Identification (Party ID). Party identification, understood as a psychological identification with the party, could foster turnout. A citizen without a preferred candidate is not likely to vote, whereas voters with a strong party preference are more likely to (Lewis-Beck et al. 2008). I measured this variable with the question: “Do you sympathize with one particular political party?” Again, it takes the value of 1 if the respondent sympathizes with one particular political party or 0 if not.

Education level (Education). Higher levels of education should positively affect the probability of voting. For this variable I used a dummy that takes a value of 1 if the respondent had some high-school education or further education, and 0 if the respondent has a lower level of education. I chose high-school education as a cut-off point because it is at this level of education that people begin to acquire the civil skills that are needed to participate in democracy in a knowledgeable way (Verba, Schlozman, and Brady 1995). For the Mexican case, it has been argued that higher levels of education result in a lower turnout since highly educated citizens do not trust politicians so much (Salazar and Temkin 2007).

Income (Log Income). People with lower incomes vote less than those with higher ones (Lewis-Beck et al. 2008; Verba, Schlozman, and Brady 1995). This variable should have a positive effect on turnout. To measure this variable, I took the natural logarithm of the household income reported by respondents.

Responsiveness (Responsiveness). Another potential reason why people do not vote is that they feel that the parties are not responsive to their interests. Therefore, to test whether the perception of that responsiveness affects turnout, I used the question “Would you say that at least one of the political parties reasonably represents your points of view?” I gave a value of 1 to a “no” and a value of 0 to a “yes”.

b. Results

I estimated the relationship between the recall of negative advertisements and voter turnout using a series of logit models for binary outcomes. Table 6 reports row logit coefficients for these models.⁷Model 1 shows that the recall of attack advertisements does not seem to have a negative effect on voter turnout. The variable recall has the expected negative sign and it fails to have a statistical significance. Thus, this model offers empirical evidence which questions the Mexican legislators’ claim that negative advertisements damage electoral participation.

With the exception of the education variable, all the control variables in this model show the expected sign and are statistically significant, at least at the 0.1 level. This is an expected result since specialist studies expect the engagement variables (Political Knowledge, Political Engagement and Party ID) to be the most important predictors (Verba, Schlozman, and Brady 1995). Contrary to what those studies expect, however, Model 1 shows that education (having a high-school diploma or a higher level of education) has a negative sign, although it is not statistically significant. This model does not seem to show multicollinearity problems,⁸ no matter how often robust standard errors were used to correct identified problems of heteroscedasticity.⁹ The same procedure was applied to the other models given that all of them showed the same results for these tests.

In model 2, I added the income and responsiveness variables. In accordance with my theoretical expectation, the key independent variable -recall of negative advertisements- shows the expected negative sign and fails to show statistical significance. The political knowledge and party identification variables, taken from the civic voluntarism model (Verba, Schlozman, and Brady 1995), lost statistical significance when the new variables were added. Only the political engagement variable remains as an important predictor of voter turnout. Among the new variables, income shows the expected sign, and is also statistically significant. The representativeness variable has a negative sign, suggesting that when citizens do not feel that political parties represent their interests, they may be discouraged from voting; however this variable is not significant.

In model 3 I broke down the variable for party identification into four dummies: one for identification with the three main political parties: PAN (right), PRI (center) and PRD (left), and an additional dummy for identification with the other five smaller parties. In this model, I only included the dummies for the three big political parties, and excluded the dummy of identification with the small parties as a comparison group. All the dummies for party identification show a positive sign, but none of them is statistically significant. The null effect of negative advertisements on turnout holds in this model as well.

Conclusion

Based on the evidence from the United States’ and the theory which explains it, it is unlikely that negative advertising hinders turnout levels in Mexico because of the “minimal effects” of campaigns, the fact that Mexico is a multiparty system and the fairly strong institutionalization of the party system.

My analysis of data from a survey of voters in the 2006 presidential election in Mexico did not support that claim. When the controls for income and responsiveness are not included in the model, the variables of the civic engagement model (Political Knowledge, Political Engagement and Party ID) seem to work very well and do a better job of explaining voter turnout. In conclusion, the evidence -under different model specifications- does not support the idea that negative campaigns reduce voter turnout. Perhaps in other party systems which are less institutionalized than the Mexican ones and/or are bipartisan, negative campaigning might be more effective, a question which deserves further study

Even though the Mexican constitution bans negative advertisements, its practitioners still regard it as an effective campaign tool -one example is the “Peña es un mentiroso” ad sponsored by PAN in the presidential campaign of 2012- and it is possible that they may resurface in the future.

If that were to occur, more sophisticated techniques to measure the recall of ads might be invented to compensate for the current shortcomings of that gauge which are discussed in this paper.

Regardless of the potentially undesirable effects of negative advertisements on turnout, Mexican legislators should have considered aspects of them that may be positive before they decided to ban them, insofar as they may be a deterrent of unethical acts by politicians, provide voters with more information so that they can make better-informed decisions and strengthen the political parties´ oversight of each other.

Future studies should focus on the importance of other factors that may affect turnout more than negative campaigns, like the fact that citizens do not feel that political parties represent their interests. Models two and three seem to suggest that when voters do not believe that political parties are truly representing their interests, they are less inclined to engage in politics. If Mexican legislators have a genuine concern about increasing the levels of electoral participation, they should try to design mechanisms that would bring the political parties closer to the electorate.