The Conditional Ideological Inducement to Campaign on Character Valence Issues in Multiparty Systems

Abstract

Acknowledging the role played by character valence issues in affecting parties’ fortunes, several recent papers have investigated the possible intentional use of such issues in electoral contests. A corollary of this line of research has focused on identifying conditions under which parties are expected to invest more in valence campaigning. In this article, we focus on the role played by parties’ relative ideological positions in a multiparty setting. We identify the existence of an inverse relationship between the distance of a party from its ideologically adjacent competitors and its incentive to campaign on character valence issues. However, the extent of this relationship can be conditional on institutional and electoral factors. We test these hypotheses by focusing on the emphasis a party places in its electoral manifestos on the specific character valence issue of corruption. Statistical results largely confirm our hypotheses.

Keywords

political parties elections public opinion and voting behavior

Since Stokes’ seminal article (1963), a substantial literature has developed that highlights the importance of valence issues in political competition, and how they are used to link parties with some condition that is positively or negatively valued by the electorate. Valence issues have been analyzed both from a theoretical and an empirical perspective. Regarding the former, a number of contributions have incorporated valence issues into their analysis with the aim of exploring the implications of valence considerations on parties’ strategies (Adams & Merrill, 2008, 2012; Ansolabehere & Snyder, 2000; Aragones & Xefteris, 2012; Enelow & Hinich, 1982; Groseclose, 2001; Schofield & Sened, 2006). With respect to the latter, it has been shown that the valence attributes of parties (and/or candidates) can have notable effects on their electoral support (Abney et al., 2013; Buttice & Stone, 2012; Clark, 2009, 2014; Clarke, Sanders, Stewart, & Whiteley, 2009; Green & Jennings, 2012).

Acknowledging the impact of valence issues on parties’ fortunes, several recent papers investigated the possible intentional use of valence issues by candidates or parties in electoral contests (see, among others, Meirowitz, 2008; Schofield, 2003; Zakharov, 2009). Moreover, a corollary to this line of research attempted to determine the conditions under which parties are expected to invest in valence campaigning to a greater extent (Ashworth & Bueno de Mesquita, 2009; Curini & Martelli, 2010; Damore, 2002; Elmelund-Præstekær, 2010; Skaperdas & Grofman, 1995; Serra, 2010).

This article assumes the same perspective on parties’ intentional adoption of valence issues. In particular, we will focus on how parties’ relative ideological positions and electoral considerations affect their emphasis on valence issues. Valence issues can be both policy-based (for example, issues such as reducing crime or increasing economic growth) and non-policy based (for example, trustworthiness, credibility, integrity, campaigning ability, etc.). In both cases, all voters hold identical positions (preferring more to less or less to more, depending on the issue). In contrast, policy issues involve conflicts of interest among groups of electors (being in favor of or against the welfare state, gay marriage, etc.). The voluminous literature on “issue ownership” focuses on the former, policy-based type of valence issues (Budge & Farlie, 1983). In contrast, the present article will concentrate on the latter type of valence issues, i.e., non-policy-related factors that can provide parties (or candidates) with an electoral advantage (or disadvantage) independent of their policy positions (for a discussion in this regard, see Clark, 2014). Specifically, we follow the terminology employed in Adams, Merrill, Simas, and Stone (2011; see also Stone & Simas, 2010) and investigate the importance of “character valence” attributes such as honesty and competence rather than “strategic valence” attributes such as fundraising skill, name recognition, and campaigning ability. The reason for this choice is that while strategic valence issues are directly instrumental for winning elections, it may seems at first sight that character valence issues are intrinsic qualities that cannot be intentionally adopted as competitive weapons.

With respect to the existing literature, our article presents three main novelties. First, unlike most extant works on valence campaigning that focus on bipartisan systems, we consider the more commonly observed multiparty system. By doing so, we demonstrate that a party’s electoral incentives to campaign on character valence issues increase when the ideological distance from its adjacent competitors decreases. Second, while the role of ideological considerations in valence campaigning has been sporadically considered in other studies (see in particular Ashworth & Bueno de Mesquita, 2009; Serra, 2010), we show how certain intervening factors (institutional and electoral) can potentially mediate this relationship. Finally, we employ data from the Comparative Manifesto Project to control for the above-mentioned relationship between parties’ locations on the ideological spectrum and their incentive to adopt character valence issues in political campaigning. In particular, we analyze the spatial determinants of the emphasis a party places on the specific character valence issue of political corruption in its electoral manifesto.

The (Conditional) Ideological Incentive for Character Valence Campaigning

Let us begin by sketching the decisional environment we model to understand a party’s incentives to campaign on character valence issues. More exactly, we focus on the consequences that a change in the relative ideological locations of parties can have on a party’s one-shot decision to adopt a campaign based on character valence issues in both a two-party and a multiparty scenario. In our discussion, we do not consider situations in which parties can change their location on the left–right political spectrum when facing valence competition (for game theoretical models applied to two-party systems that allow parties to modify their policy platforms in such situations see Adams & Merrill, 2013; Serra, 2010). Despite its simplicity, our stylized decision-theoretic model is nonetheless able to produce a clear hypothesis that can be empirically tested irrespective of the “direction” of a party valence campaign (i.e., positive or negative alike: see below).

Let us suppose that an electorate holds policy preferences summarized as ideologies along a portion of a left–right line, say the segment [0,1], and voters’ ideal points z are evenly distributed along that segment.¹ Furthermore, let us assume that all electors vote sincerely and have single-peaked and quadratically decreasing utility functions that are identically spaced from their ideal points. We first focus on two parties L and R that attempt to maximize votes and are acquainted with electors’ preferences. L and R propose ideologies that are respectively at x and x + d on the segment [0,1]. In the case of pure ideological confrontation, that is, if both parties have equal valence endowment, Party L receives all votes from 0 to the midpoint z’ ≡ x + d / 2, while Party R receives votes from that point to 1. Let us now suppose that the electorate perceives Party L as exhibiting a higher endowment of character valence issues. Following a common notation (see Groseclose, 2001), the utilities of the generic voter with ideal point z relative to L and R are, respectively,

u_{L} (z) = k - a {(x - z)}^{2}, u_{R} (z) = - a {(x + d - z)}^{2},

where d is the distance between L’s and R’s policy proposals, and k and a are parameters reporting, respectively, the valence superiority of Party L and the relative weight of ideology with respect to the valence embedded in voters’ utility.²

Relaxing the restrictive assumption that parties’ valence endowments are independent of their choices, the problem arises as to which ideological conditions alter parties’ incentives to invest in campaigning on socially shared values such as character valence issues. To address this problem, let us assume that Party L can purposely undertake a valence campaign. Irrespective of the direction of such a valence campaign (positive, i.e., aiming to increase its own valence endowment, or negative, i.e., aiming to decrease the valence endowment of the competing party), the following reasoning applies. We begin with the positive valence campaign scenario. Figure 1 illustrates how the ideological distance between parties impinges on Party L’s incentive to exploit this opportunity. Both panels of the figure depict the same valence predominance of Party L (the same value of k). However, in the left panel, a smaller distance between parties’ ideologies can be observed, while on the right, one observes a much larger distance. This implies very different consequences: In the latter case, Party L’s vote advantage ΔV_L is modest, while in the former case, this increment is considerably larger.

Figure 1.

Greater importance of character valence issues on votes (left) when party ideal points are close to each other; lesser importance when parties are far (right).

Turning from graphical intuition and equalizing the two expressions in (1) to determine the spatial location of the indifferent voter when k > 0 (i.e., when Party L displays a valence advantage), we find that Party L receives electoral support from voters whose preferences are between 0 and z* ≡ x + d / 2 + k / 2ad, and Party R receives the remaining votes from that point to 1. If we assume that the parties’ utility coincides with the votes they receive, such that U_L ≡ x + d / 2 + k / 2ad, we can define the variation in Party L’s votes caused by a unitary increment of k (i.e., the derivative ∂U_L / ∂k) as a measure of the party advantage produced by its conquered valence superiority. Therefore, the conjecture that a party can deliberately influence its valence endowment leads to the following formula for Party L’s incentive to compete on valence:

L ’ s incentive \equiv \frac{\partial U_{L}}{\partial k} = \frac{1}{2 a d} .

Figure 2 illustrates Party L’s incentive to campaign on character valence issues as a function of the distance d between L and R.

Figure 2.

Party L’s incentive to campaign on character valence issues grows hyperbolically as the distance between parties shrinks.

As is clear from the hyperbolic relationship displayed in the figure, when L and R are very ideologically proximate to one another, the incentive of a party to campaign on character valence issues is high and extremely responsive to distance variations. However, when the two parties lie beyond a certain distance from one another, such an incentive will be much lower and nearly flat. This conclusion is coherent with that underlined by Ashworth and Bueno de Mesquita (2009, p. 194) in a game theoretical setting applied to a two-party system. They show that the marginal return to valence accumulation inversely depends on the degree of ideological polarization.

In our framework, we are not modeling the costs of valence campaigning in terms of the financial and human resources required to detect and highlight the presence or absence of shared character values (contrary, for example, to the already quoted Ashworth & Bueno de Mesquita, 2009; Serra, 2010). Nevertheless, assuming such a costly investment in character valence issues does not alter the point noted above. Indeed, a party with a fixed amount of resources to be spent in the forthcoming electoral campaign will only undertake a valence campaign when its expected benefits are substantial. As long as parties’ ideologies are sufficiently distant, the amount of resources to be invested in a campaign based on character valence issues would have to be quite large to generate significant benefits in votes for a party. In such a situation, a party could find it more appealing to invest its (limited) resources in other activities such as highlighting specific aspects of its political program in an attempt to affect the campaign agenda (see Sides, 2006; Sigelman & Buell, 2004). Assuming increasing rather than constant marginal costs of valence campaigning (as in Serra, 2010) strengthens this consideration.

We now extend this result to a multiparty system. To observe this, we consider a system with three parties L, C, and R that possess ideological positions x, x + d₁, x + d₁ + d₂, respectively. This means that electors’ utilities from voting for them are

u_{L} (z) = - a {(x - z)}^{2}, u_{C} (z) = - a {(x + d_{1} - z)}^{2}, u_{R} (z) = - a {(x + d_{1} + d_{2} - z)}^{2} .

If we focus on the extreme parties in this party system (i.e., L and R), relaxing the two-party hypothesis does not change the functional form highlighted in (2) regarding the incentives a party has for investing in a positive valence campaign. Once again, this motivation will be affected by the ideological distance separating an extreme party from its adjacent competitor (Party C in the example above for both parties L and R). The position of the non-adjacent party (Party R if we focus on Party L and Party L if we focus on Party R) plays no substantial role.

However, moving to a multiparty setting implies new insights if we focus on inner parties that may compete with two adjacent parties, that is, Party C in our example. Indeed, Party C’s incentive depends on the total gain obtained from L and R. Before its positive valence campaign, Party C receives votes between the point at which u_L(z) = u_C(z) and the point at which u_C(z) = u_R(z), that is, the interval (d₁ + d₂) / 2 (see Figure 3, left panel).³ After campaigning successfully on its own values, C receives all votes between the point at which u_L(z) = u_Ck(z) and the point at which u_Ck(z) = u_R(z) with k typically being the extent of its success in valence campaigning and u_Ck(z) the electors’ utility from voting for the apparently more morally reliable Party C (Figure 3, right panel).

Figure 3.

Votes obtained by the centrist Party C before (left panel) and after (right panel) a successful positive valence campaigning.

It is therefore easy to observe that Party C’s support (and utility) is $U_{C} \equiv d_{1} + d_{2} / 2 + k / 2 a (1 / d_{1} + 1 / d_{2})$ , and hence that its incentive $\partial U_{C} / \partial k$ to compete on valence issues is given by the formula: $1 / 2 a (1 / d_{1} + 1 / d_{2})$ (4) represented in Figure 4.

Figure 4.

Party C’s incentive to campaign on character valence issues is a function of two variables d₁ and d₂ (where d₁ + d₂ = constant).

The incentive for an inner party to compete on valence therefore appears to be very strong when both its adjacent parties are located near it (i.e., when an inner party is “squeezed” between two closely adjacent parties), moderately strong if at least one of the two adjacent parties is close, but declines when both adjacent parties are further away.

Analogous consequences can be derived in the case of successful negative campaigning in a multiparty setting based on character valence issues.⁴ For example, if L’s negative campaigning succeeds in reducing party R’s valence endowment of quantity h, the benefit accrues entirely to C, and hence Party L has no incentive in attacking non-adjacent Party R on values. However, if the same negative campaign succeeds against C, Party C will lose votes to Party L, while Party R can also benefit from this. Thus, in the case of negative campaigning, the incentive of an extreme party to compete on valence issues is also inversely related to the distance from the adjacent party and unrelated to the distance from non-adjacent parties. By the same reason, an inner party may profitably attack one or the other of its adjacent parties (or both), while similarly failing to derive any benefit from negative campaigning against non-adjacent parties.

Let us call nearest neighbors two parties having no other party located between their ideological positions. Then, from (2) and (4), we can derive the following hypothesis:

Hypothesis 1 (H1): The incentive of a party to compete on valence issues grows proportionally with the reciprocal of the distance between it and its nearest neighbor (for an extreme party) and with the sum of the two reciprocal of the distances between it and its nearest neighbors (for an inner party).

The model discussed thus far does not investigate whether the impact of ideological considerations on valence campaigning is mediated by any intervening variables. A possible refinement of the model would require a more direct accounting for institutional factors (Debus, 2008; De Winter, 2002). In this respect, an important (institutional) difference is apparent between incumbent and non-incumbent parties with respect to valence campaigning in multiparty systems. While all parties may prefer to maintain their ideological distinctiveness from their allies in government or opposition, it seems reasonable to assume that a governing party cannot attack the moral standing of a fellow coalition member or attempt to distinguish itself from its partners as being for example more honest (see below) without serious consequences for the stability of the coalition. For analogous reasons, it is realistic to assume that a non-governing party would primarily cast blame on the moral shortcomings of one or more parties in the government coalition. Consequently, it can be argued that the incentive of Party i to engage in campaigning on character valence issues may be conditional on the status characterizing the neighbors of that party. In this context, the term “status” refers to the type of relationship each party has with the incumbent government (i.e., a cabinet or a non-cabinet party). That is,

Hypothesis 2 (H2): Ideological considerations are expected to play a stronger role in a party’s incentives to invest in a valence campaign when the status of its nearest neighbor(s) differs from its own.

Second, we have thus far assumed a uniform distribution of voters. However, we could reasonably hypothesize that the impact of ideological considerations would be greater every time the share of votes at stake for a given party involved in valence campaigning (i.e., the number of voters ideologically located close to a party) increases. Therefore, we can advance a second refinement of H1:

Hypothesis 3 (H3): Ideological considerations play a stronger role in a party’s incentives to invest in a valence campaign when there are more votes at stake.

Testing the Empirical Content of the Ideological−Valence Relationship

The relationship envisaged in the preceding section between ideological positions and character valence campaigning concerns a number of items that are undefined in principle. However, political corruption stands out as one of the most prominent examples in this respect (see Stokes, 1963, 1992).⁵ Focusing on corruption as a character valence issue presents another important advantage: Doing so allows us to take advantage of the widely used Comparative Manifesto Project (CMP) data set (Budge, Klingemann, Volkens, Bara, & Tanenbaum, 2001) to empirically examine our hypothesis.

It is well known that CMP coding procedures involve sorting all politically meaningful expressions in each party’s manifesto into a group of categories (welfare, defense, law and order, etc.), then taking the percentages in each category as a measure of the party’s priorities (Budge et al., 2001). Among these categories, Political Corruption (per304 following the CMP denominations) explicitly includes all references to the “need to eliminate corruption and associated abuse in political and public life.” As a party’s leadership carefully considers which specific subjects and contents to incorporate into its manifesto, the (amount of) emphasis placed on the Political Corruption category reveals the importance of the corruption issue to the party in the forthcoming campaign. Thus, the choice of CMP data set suits our task of attempting to examine the spatial incentive that motivates a party to highlight valence issues in their campaigns.

Note that with respect to the Political Corruption category, one cannot distinguish between positive and negative statements on corruption in the CMP data. However, this does not prevent us from testing the hypotheses highlighted in the previous section. In fact, according to the above discussion, the relationship envisaged between the incentive to emphasize character valence issues and ideological positions applies regardless of whether valence campaigning is positive and negative.

Still, it is true that the focus on positive and negative campaigning on character valence issues (political corruption, in the present case) can potentially lead to further hypotheses. For example, an incumbent party will not have an incentive to blame a coalition partner for corruption (i.e., the “negative side” of valence campaigning), but it may have an incentive to emphasize its own record for integrity in office (i.e., the “positive side” of valence campaigning). The opposite is the case for a minor or extreme party without a coalition governing record: It will not have an incentive to emphasize positive corruption-related messages (as it has not or will not have an opportunity to gain a record for integrity in office), but it consistently has an incentive to emphasize negative corruption-related messages. Similarly, we have argued that in a multiparty system, ceteris paribus, parties’ incentive for negative campaigning could be considered less important than that for positive campaigning as in the former case a party cannot completely internalize the electoral benefits of such a move. All these hypotheses are an obvious consequence of the previous theoretical discussion. Unfortunately, they cannot be tested using the CMP data set, given its inability to distinguish between positive and negative statements based on corruption claims.

From the entire range of observations contained in the CMP data set (Volkens, Lacewell, Regel, Schultze, & Werner, 2009), our sample includes countries graded as “free” by Freedom House in the year in which the manifesto was coded by the CMP. This subset of countries was selected because we intend to examine the ideological incentives of parties under fair competitive contexts. This leaves us with 36 countries, covering on average 11.5 elections per country. The total number of observations (i.e., the number of manifestos) included in our sample is 2,346 (see the Appendix).

As noted above, we employ the Political Corruption category in the CMP data set as our dependent variable, which we have labeled CORRUPTION. The average emphasis that parties place on political corruption is far from irrelevant. In our entire sample, the overall average value of CORRUPTION is .012 (i.e., 1.12% of all quasi-sentences in manifestos are devoted to political corruption: see the Appendix), a value that exceeds those recorded in other categories such as Economic planning (per404) or Military (per105) and close to the average emphasis parties placed on the European Community/Union (per108).

To measure parties’ ideological positions, we applied the scaling method proposed by Lowe, Benoit, Mikhaylov, and Laver (2011) to the 26 categories identified in Laver and Budge (1992) as relevant to assess the left to right placement of parties over time. A logit scale avoids some flaws of the traditional approaches employed to extract parties’ positions from the CMP data set, both from a theoretical and an empirical perspective (see also Benoit, Laver, Lowe, & Mikhaylov, 2012).⁶ The scale generally contrasts emphasis on strong defense, free enterprise and traditional morality on the Right with the prioritization of peaceful internationalism, welfare and government intervention on the Left.⁷

Estimating parties’ policy positions is a necessary preliminary step to assess the ideological distance between each Party i and its adjacent party(ies). Excluding the extreme parties (i.e., parties by definition exhibiting only one adjacent party) all other inner parties will always present two nearest neighbors (one to their right, one to their left). Our main independent variable according to H1 is therefore the (sum of the) reciprocal of the distance(s) separating one party from its nearest neighbor(s). Given the extremely skewed distribution of the measure so estimated, we have taken the log of it to normalize the measure and labeled this new variable PROXINCENTIVE. In our data set, the average value of PROXINCENTIVE is 1.88, with a standard deviation of 1.56. As stated in H1, we expect a positive relationship between this variable and a party’s incentive to campaign on valence issues.

According to H2 and H3, however, the magnitude of this positive relationship can be conditional on a number of intervening variables. We have therefore included two interaction terms in the analysis (see below). The first term involves PROXINCENTIVE and a variable labeled NEAREST STATUS. This latter variable assumes the value 0 if both neighbor parties at the time of the election of Party i share its status (i.e., both cabinet parties, if Party i is a cabinet party; both non-cabinet parties, if Party i is a non-cabinet party), 1 if at least one neighbor has the opposite status, and value 2 if both neighbors (or the only neighbor in case of extreme parties) have the opposite status.

The second interaction term involves PROXINCENTIVE and a variable labeled VOTE at STAKE, which is simply the sum of the vote share obtained at time t by Party i and its nearest neighbor(s). As VOTE at STAKE increases, the density/mass of voters available in the ideological space surrounding Party i also rises.⁸

We expect ∂CORRUPTION / ∂PROXINCENTIVE to be positive but increasing as NEAREST STATUS increases (H2). Similarly, we expect ∂CORRUPTION / ∂PROXINCENTIVE to be positive but increasing as VOTE at STAKE rises (H3).

Empirical Analysis

Our dependent variable CORRUPTION comprises fractional response data, bounded between 0 (a party makes no reference to this category) and 1 (the entire manifesto is dedicated to this topic). Using standard linear models may raise problems such as non-normality in the distribution of errors (Wooldridge, 2002). Thus, we prefer to follow Papke and Wooldridge (2008), who propose a fractional logit model. We also clustered the standard errors on party programs to obtain heteroskedasticity- and autocorrelation-consistent standard errors (Rogers, 1993).

Following the above discussion, we argue that, under favorable ideological circumstances, according greater prominence to CORRUPTION in a manifesto can benefit a party. However, certain caveats should be kept in mind. For example, a party with a reputation for corruption may wish to avoid the topic altogether because increasing the salience of corruption may then be detrimental. Similarly, merely mentioning the word corruption does not immediately enhance a party’s reputation for honesty; this likely depends on how credible the party is on this issue meaning that discussing corruption may be effective for some parties but ineffective or even detrimental for others.

In this respect, we have included various control variables in our analysis in an attempt to capture the impact of several factors that can differentiate one party from another with respect to CORRUPTION beyond the ideological considerations discussed in the section “The (Conditional) Ideological Incentive for Character Valence Campaigning.”

First, to distinguish between extreme and inner parties, we estimated a variable labeled EXTREME, coded 1 for the former and 0 for the latter. This variable allows us to control for the possibility that extreme parties could have a stronger incentive to compete on ideological grounds rather than on character valence issues ceteris paribus, as their constituencies tend to be more ideologically committed than supporters of mainstream parties (Adams, Clark, Ezrow, & Glasgow, 2006).

Second, one can argue that a party’s opportunities to campaign on corruption issues are not independent of its status as governing or opposition party. Governing parties, being able to control the public economy and influence the distribution of property rights, are more likely to be blamed for corruption (Kunicova & Rose-Ackerman, 2005). We therefore assume that non-incumbent parties would place greater emphasis on political corruption than incumbent parties, ceteris paribus, and included a dummy variable labeled INCUMBENT PARTY that takes the value 1 for governing parties and 0 otherwise.⁹

Third, within each country, we counted the number of consecutive legislatures during which a major party was included in the cabinet and termed this number CABINET CONTINUITY. CABINET CONTINUITY begins with the value 1 and increases to 2 if at election time t, the major cabinet party does not change between times t − 1 and t. Similarly, it increases to 3 if there is no change in its cabinet membership during t − 2 and t − 1, and t − 1 and t. We expect that as CABINET CONTINUITY increases, the incentive to highlight political corruption increases within the party system as a whole. Fewer alternations in office often imply more deeply entrenched connections between government and business and larger numbers of individuals seeking privileged treatment by the state (Chang, 2005). Corruption issues should be more prominent under such circumstances irrespective of other considerations.

Fourth, to control for the possible impact of different party system formats across time and countries, we included the number of effective electoral parties in each election t (variable name: ENEP) and a measure of party system polarization labeled POLARIZATION. The latter is calculated using the index advanced by Dalton (2008) that, except for a constant, mathematically corresponds to the weighted population standard deviation of party positions in a given election.

Fifth, to account for any period effects, we introduced a dummy for each decade analyzed, from the 1940s until the 2000s. We take the 1970s as our baseline category, as the Lockheed scandal affected several countries during that decade (Sampson, 1977). Finally, we included a dummy for each country to control for all other relevant aspects that are idiosyncratic to our statistical units.¹⁰ By doing so, the resulting model is estimated based on variation within each country only, while the (remaining) variance in the CORRUPTION level (i.e., the difference in the average value of CORRUPTION between countries) is fully captured by the fixed effects. This is in line with our theory, as our hypotheses are devoted to understanding CORRUPTION changes within each party system according to spatial considerations.¹¹

Table 1 reports the results of the models we have estimated.

Table 1.

The Determinants of the Emphasis on CORRUPTION in Party Manifestos.

	(1)	(2)	(3)	(4)
PROXINCENTIVE	0.066*	−0.037	−0.135	−0.179
PROXINCENTIVE	(0.031)	(0.057)	(0.090)	(0.097)
NEAREST STATUS	—	−0.277	—	−0.152
NEAREST STATUS		(0.152)		(0.137)
PROXINCENTIVE × NEAREST STATUS	—	0.118*	—	0.092*
PROXINCENTIVE × NEAREST STATUS		(0.050)		(0.046)
VOTE at STAKE	—	—	−0.016**	−0.014**
VOTE at STAKE			(0.006)	(0.005)
PROXINCENTIVE × VOTE at STAKE	—	—	0.004*	0.004*
PROXINCENTIVE × VOTE at STAKE			(0.002)	(0.002)
EXTREME	0.012	−0.010	−0.163	−0.163
EXTREME	(0.116)	(0.110)	(0.115)	(0.114)
INCUMBENT PARTY	−0.577**	−0.558**	−0.509**	−0.528**
INCUMBENT PARTY	(0.106)	(0.109)	(0.109)	(0.110)
CABINET CONTINUITY	0.088**	0.086**	0.089**	0.086**
CABINET CONTINUITY	(0.017)	(0.017)	(0.018)	(0.018)
POLARIZATION	0.318**	0.362**	0.339**	0.364**
POLARIZATION	(0.087)	(0.092)	(0.090)	(0.091)
ENEP	0.100**	0.096*	0.084*	0.083*
ENEP	(0.039)	(0.038)	(0.040)	(0.039)
1940s	0.238	0.215	0.202	0.194
1940s	(0.283)	(0.284)	(0.274)	(0.273)
1950s	0.348	0.353	0.353	0.355
1950s	(0.256)	(0.249)	(0.256)	(0.245)
1960s	−0.261	−0.229	−0.222	−0.208
1960s	(0.205)	(0.212)	(0.213)	(0.218)
1980s	−0.112	−0.074	−0.096	−0.078
1980s	(0.149)	(0.142)	(0.156)	(0.150)
1990s	−0.206	−0.200	−0.202	−0.202
1990s	(0.186)	(0.183)	(0.187)	(0.185)
2000s	−0.816**	−0.809**	−0.796**	−0.787**
2000s	(0.247)	(0.243)	(0.242)	(0.240)
Constant	−6.546**	−6.348**	−5.560**	−5.522**
Constant	(0.529)	(0.545)	(0.602)	(0.594)
Observations	2,346	2,346	2,346	2,346
Log pseudo likelihood	−99.849	−99.560	−99.294	−99.129
F test that all country effects = 0	764.69**	734.07**	791.46**	786.64**

Standard errors adjusted for clusters in party program in parentheses. Country fixed effects suppressed to conserve space (available on request).

p < .05. **p < .01.

Model 1 is our basic model, with which we tested H1, while the following models allow us to test H2 and H3. As seen from Model 1, an increase in PROXINCENTIVE produces a significant increase in the emphasis that parties place on CORRUPTION, as predicted by H1. The impact of ideological considerations on parties’ incentives to invest in valence campaigning centering on corruption is far from negligible. The first difference¹² in CORRUPTION when, for example, PROXINCENTIVE passes from its median to its 90th percentile produces a 14% expected increase in the average emphasis a party places on CORRUPTION in its election manifesto.

Regarding the control variables, INCUMBENT PARTY significantly decreases the emphasis on CORRUPTION. Conversely, CORRUPTION increases with CABINET CONTINUITY, as well as with the two properties of a party system: ENEP and POLARIZATION. In particular, the sign of POLARIZATION indicates that parties are more inclined to emphasize CORRUPTION in more polarized party systems. The result that the two ideological measures that we employ in our analysis (i.e., PROXINCENTIVE and POLARIZATION) exhibit opposite impacts on CORRUPTION¹³ is interesting. With respect to PROXINCENTIVE, the fact that we observe a significant impact on CORRUPTION in the analysis, even after controlling for POLARIZATION, indicates that the spatial relationship among neighboring parties is relevant in explaining a party’s decision to campaign on character valence issues regardless of the ideological distribution of the entire party system. With respect to POLARIZATION, our result is coherent with the finding reported in Clark and Leiter (2013) who note that the “electoral impact” of valence issues tends to be higher given more ideologically dispersed parties. According to these authors, this occurs because voters, faced with parties offering more distinct policy positions, give more weight to parties’ valence attributes. This could induce parties to invest more in CORRUPTION, therefore producing the positive sign of POLARIZATION reported in Table 1.

Finally, compared with the 1970s (our omitted temporal category), parties in subsequent decades do not seem to devote systematically more (or less) attention to political corruption in their manifestos, with the exception of the last decade. Moreover, EXTREME does not seem to exert any significant impact on CORRUPTION, while the country dummies always appear highly relevant (as can be observed in the results of the F test reported in Table 1). This indicates the utility of including them in the analysis to obtain more efficient estimators.

Model 2 introduces the interaction term implied by H2. The results confirm our expectation. As shown in the upper panel of Figure 5, the marginal impact of a one-unit increase in PROXINCENTIVE (i.e., corresponding roughly to half of a one-standard-deviation change) on CORRUPTION is significant but only when NEAREST STATUS is different from 0. In particular, consistent with H2, the ideological inducement to invest in valence campaigning reaches its maximum when a party is surrounded by neighbors with a different status than its own (i.e., when NEAREST STATUS = 2, corresponding to a situation in which either neighbors of Party i are cabinet parties, if Party i is a non-cabinet party, or both neighbors of Party i are non-cabinet parties, if Party i is a cabinet party).

Figure 5.

Marginal impact of PROXINCENTIVE as NEAREST STATUS (left panel) or VOTE at STAKE (right panel) increases.

Symmetrically, the interaction term involving PROXINCENTIVE and VOTE at STAKE is highly significant and in the direction assumed by H3 (see Model 3). In particular, as seen in the lower panel of Figure 5, the marginal impact of PROXINCENTIVE increases substantially as VOTE at STAKE increases. Moreover, for low values of VOTE at STAKE (i.e., lower than 35% of vote share at the 90% confidence interval), this marginal impact becomes statistically indistinguishable from zero implying that a party must foresee a relative large electoral gain from a (possibly costly) valence campaign to be driven to invest in such a strategy by ideological considerations.¹⁴

Finally, Model 4 combines Models 2 and 3 without implying any substantial difference in the analysis.

Conclusions

In the present article, we investigated under which ideological conditions a party can have a stronger incentive to campaign on character valence issues. In the case of two-party competition, we have shown that the more the parties resemble one another ideologically, the more a party is induced to employ character valence issues as a tactic to gain votes. We then generalized our theoretical finding to systems with any number of parties. This led us to the hypothesis that this inverse relationship between distance and the incentive for valence campaigning only concerns spatially adjacent parties (what we have called “neighbor parties”), irrespective of the ideological positions of the remaining parties.

This conclusion can also be appreciated from a complementary perspective: When adjacent parties are near one another, praising one’s own policy position or criticizing that of one’s adversary is tantamount to praising the adversary’s position and criticizing one’s own, as the two programs are so similar. Thus, the more similar parties are ideologically, the more they need to find a different way to distinguish themselves before the electorate.¹⁵ Highlighting character valence issues provides them with that opportunity. We have also shown that this hypothesis applies irrespective of the “direction” of the valence campaign, that is, negative or positive alike.

The empirical analysis devoted to understanding the determinants of the emphasis parties place on political corruption in their manifestos (a particularly relevant character valence issue) confirms our hypothesis. In this sense, the ideological positions of parties vis-à-vis one another have important implications for the types of issues parties opt to advertise during an electoral confrontation; this produces, as a consequence, campaigns more or less centered on character valence issues.

Second, we have shown how a number of intervening variables, at both the institutional and electoral levels, are important in mediating the relationship between the ideological placement of parties and the incentive to emphasize character valence issues. In particular, valence considerations seem more important to a party when that party is spatially “squeezed” between two competitors (i.e., when the incumbent status of its nearest neighbors is the opposite of its own) and when there is more to gain from engaging in valence campaigning (i.e., when the share of votes at stake for a party involved in valence campaigning is larger). In this sense, integrating “institutional or behaviouralist constraints” (Debus, 2008, p. 520) into spatial theoretical considerations definitively improves our understanding of political dynamics.

Our results suggest several directions for future research. In particular, while focusing on parties’ manifestos presents certain advantages,¹⁶ a more systematic examination of various possible implications of our model (such as the focus on positive and negative valence campaigning) would require a larger set of data that go beyond party manifestos. Further information on the characteristics of parties’ campaigns such as published and broadcasted interviews with party leaders would both be worthy of examination. The collection of such a data set across countries would also allow for a systematic exploration of when character valence campaigning can be more effective in terms of votes. As noted above, Clark and Leiter (2013) demonstrate that the electoral consequences of character valence issues tend to be greater as party system polarization increases. Focusing on the “micro” spatial relationship among (adjacent) parties, as we have done here, rather than “macro” properties of the entire party system alone represents another interesting research topic.

Footnotes

Appendix

Acknowledgements

Earlier version of this article was presented at the New Developments in Modeling Party Competition Conference, Berlin, July 2012. The author acknowledges helpful comments from the seminar’s participants, and in particular from Samuel Merrill III, Bernard Grofman and James Adams. A special thank to Paolo Martelli for having discussed with the author several times the theoretical framework of the paper.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Japan Society for the Promotion of Science Grant S-12123 and the Italian Ministry for Research and Higher Education, Prin 2009–prot. 2009 TPW4NL_002.

Notes

Author Biography

Luigi Curini is associate professor of political science at Università degli Studi of Milan, Italy. His research areas include electoral competition, legislative behaviour, and social media analysis. Recent publications include articles in European Journal of Political Research, British Journal of Political Science, Journal of Politics, West European Politics, Electoral Studies, European Political Science Review, International Political Science Review, and Party Politics.

References

Abney

Adams

Clark

Easton

Ezrow

Kosmidis

Neundorf

(2013). When does valence matter? Heightened valence effects for governing parties during election campaigns. Party Politics, 19, 61-82.

Adams

Clark

Ezrow

Glasgow

(2006). Are Niche Parties fundamentally different from mainstream parties? American Journal of Political Science, 50, 513-529.

Adams

Merrill

(2008). Policy-seeking parties in a parliamentary democracy with proportional representation: A valence-uncertainty model. British Journal of Political Science, 39, 539-558.

Adams

Merrill

(2013). Policy-seeking candidates who value the valence attributes of the winner. Public Choice, 155, 139-161.

Adams

Merrill

Simas

Stone

(2011). When candidates value good character: A spatial model with applications to congressional elections. Journal of Politics, 73, 17-30.

Ansolabehere

Snyder

J. M.

(2000). Valence politics and equilibrium in spatial election models. Public Choice, 103, 327-336.

Aragones

Palfrey

(2002). Mixed Equilibrium in a Downsian Model with a Favored Candidate. Journal of Economic Theory, 103: 131-161.

Ashworth

Bueno de Mesquita

(2009). Elections with platform and valence competition. Games and Economic Behavior, 67, 191-216.

Beck

Clarke

Groff

Keefer

Walsh

(2001). New tools in comparative political economy: The database of political institutions. World Bank Economic Review, 15, 165-176.

10.

Benoit

Laver

Lowe

Mikhaylov

(2012). How to scale coded text units without bias: A response to Gemenis. Electoral Studies, 31, 605-608.

11.

Budge

Farlie

D. J.

(1983). Explaining and predicting elections: Issue effects and party strategies in twenty-three democracies. London, England: Allen & Unwin.

12.

Budge

Klingemann

H. D.

Volkens

Bara

Tanenbaum

(Eds.). (2001). Mapping policy preferences. Oxford, UK: Oxford University Press.

13.

Buttice

M. K.

Stone

W. J.

(2012). Candidates matter: Policy and quality differences in congressional elections. Journal of Politics, 74, 870-887.

14.

Chang

E. C.

(2005). Electoral incentives for political corruption under open-list proportional representation. Journal of Politics, 67, 716-730.

15.

Chiru

Gherghina

(2012). When voter loyalty fails: Party performance and corruption in Bulgaria and Romania. European Political Science Review, 4(1), 29-49.

16.

Chong

De La O

A. L.

Karlan

Wantchekon

(2011). Looking beyond the incumbent: The effects of exposing corruption on electoral outcomes (NBER Working Paper No. 17679). Retrieved from http://www.nber.org/papers/w17679

17.

Clark

(2009). Valence and electoral outcomes in Western Europe, 1976-1998. Electoral Studies, 28, 111-122.

18.

Clark

(2014). Does public opinion respond to shifts in party valence? A cross-national analysis of Western Europe, 1976-2002. West European Politics, 31, 91-112.

19.

Clark

Leiter

(2013). Does the ideological dispersion of parties mediate the electoral impact of valence? A cross-national study of party support in nine Western European democracies. Comparative Political Studies. Advance online publication. doi:10.1177/0010414013488537

20.

Clarke

H. D.

Sanders

M. S.

Stewart

M. C.

Whiteley

(2009). The American voter’s British cousin. Electoral Studies, 28, 632-641.

21.

Cox

(1990). Centripetal and centrifugal incentives in electoral systems. American Journal of Political Science, 34, 903-935.

22.

Curini

Martelli

(2010). Ideological proximity and valence competition: Negative campaigning through allegation of corruption in the Italian legislative arena from 1946 to 1994. Electoral Studies, 29, 636-647.

23.

Dalton

R. J.

(2008). The quantity and the quality of party systems. Comparative Political Studies, 20(10), 1-22.

24.

Damore

D. F.

(2002). Candidate Strategy and the Decision to Go Negative. Political Research Quarterly, 55(3), 669-686.

25.

Debus

(2008). Office and policy payoffs in coalition governments. Party Politics, 14, 515-538.

26.

De Winter

(2002). Parties and government formation, portfolio allocation, and policy definition. In Luther

K. R.

Müller-Rommel

(Eds.), Political parties in the New Europe: Political and analytical challenges (pp. 171-206). Oxford, UK: Oxford University Press.

27.

Elmelund-Præstekær

(2010). Beyond American negativity: Toward a general understanding of the determinants of negative campaigning. European Political Science Review, 2, 137-156.

28.

Enelow

Hinich

(1982). Nonspatial candidate characteristics and electoral competition. Journal of Politics, 44, 115-130.

29.

Ensley

M. J.

(2007). Candidate divergence, ideology, and vote choice in U.S. Senate elections. American Politics Research, 35, 103-122.

30.

Green

(2007). When voters and parties agree: Valence issues and party competition. Political Studies, 55, 629-655.

31.

Green

Hobolt

S. B.

(2008). Owning the issue agenda: Party strategies and vote choices in British elections. Electoral Studies, 27, 460-476.

32.

Green

Jennings

(2012). Valence as macro-competence: An analysis of mood in party competence evaluations in Great Britain. British Journal of Political Science, 42, 311-343.

33.

Groseclose

(2001). A model of candidate location when one candidate has a valence advantage. American Journal of Political Science, 45, 862-886.

34.

Hansen

K. M.

Pedersen

R. T.

(2008). Negative campaigning in a multiparty system. Scandinavian Political Studies, 31, 408-427.

35.

Kunicova

Rose-Ackerman

(2005). Electoral rules and constitutional structures as constraints on corruption. British Journal of Political Science, 35, 573-606.

36.

Laver

Budge

(Eds.). (1992). Party policy and government coalitions. Basingstoke, UK: Macmillan.

37.

Laver

Sergenti

(2012). Party competition: An agent-based model. Princeton, NJ: Princeton University Press.

38.

Lowe

Benoit

Mikhaylov

Laver

(2011). Scaling policy preferences from coded political texts. Legislative Studies Quarterly, 26, 123-155.

39.

McDonald

Budge

(2005). Elections, parties, and democracy: Conferring the median mandate. Oxford, UK: Oxford University Press.

40.

Meirowitz

(2008). Electoral contests, incumbency advantages and campaign Finance. Journal of Politics, 27, 681-699.

41.

Papke

L. E.

Wooldridge

J. M.

(2008). Panel data methods for fractional response variables with an application to test pass rates. Journal of Econometrics, 145, 121-133.

42.

Plümper

Troeger

V. E.

(2007). The estimation of time-invariant and rarely changing variables in finite sample panel analyses with unit effects. Political Analysis, 15, 124-139.

43.

Rogers

W. H.

(1993). Regression standard errors in clustered samples. Stata Technical Bulletin, 13, 19-23.

44.

Sampson

(1977). The Arms Bazaar: From Lebanon to Lockheed. New York: Viking Press.

45.

Schofield

(2003). Valence competition in the spatial stochastic model. Journal of Theoretical Politics, 15, 371-383.

46.

Schofield

Sened

(2006). Multiparty democracy. Cambridge, UK: Cambridge University Press.

47.

Serra

(2010). Polarization of what? A model of elections with endogenous valence. Journal of Politics, 72, 426-437.

48.

Sides

(2006). The origins of campaign agendas. British Journal of Political Science, 36, 407-436.

49.

Sigelman

Buell

E. H.

(2004). Avoidance or engagement? Issue convergence in U.S. presidential campaigns, 1960-2000. American Journal of Political Science, 48, 650-661.

50.

Skaperdas

Grofman

(1995). Modeling negative campaigning. American Political Science Review, 89(1), 49-61.

51.

Stokes

D. E.

(1963). Spatial models of party competition. American Political Science Review, 57, 368-377.

52.

Stokes

D. E.

(1992). Valence politics. In Kavanagh

(Ed.), Electoral politics (pp. 141-162). Oxford, UK: Clarendon Press.

53.

Stone

W. J.

Simas

E. N.

(2010). Candidate valence and ideological positions in U.S. House Elections. American Journal of Political Science, 54, 371-388.

54.

Volkens

Lacewell

Regel

Schultze

Werner

(2009). The Manifesto data collection: Manifesto project. Berlin, Germany: Wissenschaftszentrum Berlin für Sozialforschung.

55.

Woldendorp

Keman

Budge

(2000). Party government in 48 democracies. London, England: Kluwer Academic Publishers.

56.

Wooldridge

J. M.

(2002). Econometric analysis of cross section and panel data. Boston, MA: MIT Press.

57.

Zakharov

A. V.

(2009). A model of candidate location with endogenous valence. Public Choice, 138, 347-366.

58.

Zakharova

Warwick

P. V.

(2014). The sources of valence judgments: The role of policy distance and the structure of the left-right spectrum. Comparative Political Studies. Advance online publication. doi:10.1177/0010414013516928

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB