Political Behavior

, Volume 37, Issue 1, pp 67–86 | Cite as

The Role of Indifference in Split-Ticket Voting

Original Paper

Abstract

Previous studies of split-ticket voting have demonstrated that partisan ambivalence—countervailing affective cross-pressure that decreases preference stability—is positively related to an individual’s likelihood of casting a split ballot. While these findings are intuitive, recent methodological work regarding the measurement of ambivalence hints that indifference—i.e. the complete absence of affective political attachments—should produce a stronger positive effect on split-ticket voting than ambivalence. If partisan considerations are not central to the self-image of indifferent voters—who have little cognitive or emotional attachments from which they draw politically-relevant information—then they should be very likely to cast split ballots given that their nominal partisan attachments are only tentatively related to electoral choice. Drawing upon this distinction, I disaggregate indifferent individuals (i.e. those voters who are neither positively nor negatively oriented towards the parties) from ambivalent ones (i.e. those voters who possess mixed or conflicting affective attachments to both parties) and demonstrate that indifference has a greater positive effect on an individual’s propensity to engage in split-ticketing. I then show how the prevailing interval-level operationalization of ambivalence underestimates the true effect of indifference on split-ticketing.

Keywords

Spit-ticket voting Indifference Ambivalence 

Studies examining split-ticket voting—the phenomenon whereby individuals split their vote between the two major parties for different offices on the same ballot—have enjoyed consistent popularity within the voting literature (Campbell and Miller 1957; Fiorina 1996; Beck et al. 1992; Garand and Lichtl 2000; Burden 2002; Mulligan 2011). Although a number of theoretical frameworks have been tendered to explain this phenomenon, scholars often employ one of two “motivated” explanations for why voters may split their ballots: (1) policy-balancing, where voters intentionally split their votes to achieve an ideal balance of preferred government ideology, and (2) weak partisanship theory, where voters without strong partisan identities are simply less likely to cast a straight-ticket vote.

It may be the case, however, that these two theories undersell a key psychological process that drives ticket-splitting behavior. As contemporary work on partisan cross-pressures suggests, ambivalence—internalized conflict between countervailing partisan affect—influences how decisional criteria are processed, which generates instability in political preferences (Basinger and Lavine 2005; Lavine et al. 2012). When this affective cross-pressure is present, research demonstrates that ambivalent voters possess a strong propensity to cast a split-ticket, whether in an attempt to satisfy a sense of internal dissonance (Mulligan 2011) or because party labels are simply less useful to the ambivalent voter (Basinger and Lavine 2005).

However, although ambivalence appears to induce heightened levels of split-ticketing, recent methodological work regarding the measurement and operationalization of ambivalence provides a theoretical framework that hints that indifference, not ambivalence, should be the stronger motivator of split-ticketing behavior (Yoo 2010; Thornton 2011). Drawing upon the conceptual and empirical distinction between these two attitudinal states proposed in this research program, I use American National Election Studies (ANES) data from six elections to demonstrate that indifferent voters—who possess no affective partisan attachments from which they draw politically-relevant information—are more likely to cast a split-ticket than their ambivalent counterparts (who, by definition, possess at least a modest degree of emotional attachments). Further, I illustrate that the effect of indifference upon an individual’s likelihood to cast a split-ticket is underestimated when one utilizes the prevailing interval-level measurement of ambivalence.

The Psychological Motivations of Split-Ticketing

In their early research on straight and split-ticket voting, Campbell and Miller (1957) argued that under conditions of “goal-directed motivation,” electoral choice reflects the realization of an individual’s preferences as a function of the strength of their motivations. Within the split-ticket voting literature, these motivations often take one of two forms. First, Fiorina (1996) and a host of others (Alesina and Rosenthal 1995; Lewis-Beck and Nadeau 2004; Carsey and Layman 2004) have argued that split-ticket voting is the result of a purposive effort of a specific portion of the electorate. Here, electoral choice is depicted as a proximity-based decision where prospective voters conceptualize and place themselves along an ideological continuum and then select a combination of political actors that will most closely approximate their own spatial positioning (Fiorina 1996). In other words, voters explicitly compare their personal ideological preferences to the ideological locations of the parties and take midpoint between those two positions; based on this calculation, voters then decide whether divided or unified party control of government is preferable to realizing their overarching ideological preferences.

A second theoretical approach to the split-ticket phenomenon treats split ballots as the outcome of an absence of strong ideological or partisan affiliations (i.e. a “lack” of goal-directed motivation). Here, the likelihood of casting a straight ticket is very high for an individual who strongly identifies as a partisan or ideologue. Alternatively, because “conflict between motives reduces the strength of goal-directed motivation and produces ambiguities in behavior” (Campbell and Miller 1957, p. 310), the likelihood of casting a split-ballot rises sharply for those individuals who are unsure of or independent in their partisan affiliation and ideological alignment (Beck et al. 1992).

While these two theories have engendered varying degrees of support in the ticket-splitting literature, neither is without its shortcomings. Although the “cognitive Madisonianism” inherent in the policy-balancing thesis appears to be solidly grounded in theory, the empirical evidence that links these strategic preferences to ticket-splitting has been mixed (Beck et al. 1992; Garand and Lichtl 2000). Because studies of mass public opinion suggest that individuals do not possess particularly “constrained” attitudes (Zaller and Feldman 1992, p. 579), citizens do not appear to be uniformly well-equipped to evaluate candidates on the basis of such ideological criteria (Lodge et al. 1995). Moreover, as Brunell and Grofman (2009) illustrate, Fiorina’s expectation that ticket-splitting should increase as the parties have polarized has not been borne out (but see Garand and Lichtl 2000).1

If policy-balancing is an unsatisfactory explanation, what of partisanship theory? Here, weak-partisanship theory may underspecify the causal motivations of the moderate or conflicted voter.2 Moderation, in fact, need not reflect active indifference, a lack of motivation, or the presence of “non-attitudes” (Mulligan 2011). As Lazarsfeld and colleagues (Lazarsfeld et al. 1944; Berelson, Lazarsfeld and McPhee 1954) argue, voters are often caught between conflicting partisan or ideological pressures. These cross-pressured, prospective voters may be “attracted to each party by one set of opinions and repelled by another” (Berelson et al. 1954, 190), which leads the conflicted voter to effectively plant one foot in both groups’ camp (Green et al. 2002). As a result, this cross-pressured, or ambivalent, voter may try to satisfy this dissonance by casting a vote for candidates of both parties.

Affective Attachments and Split-Ticket Voting

The Role of Ambivalence

As an alternative to these well-known theories of ticket-splitting, Mulligan (2011) and Lavine et al. (2012) have argued that ambivalence plays a direct, causal role in producing split ballots. Here, partisan ambivalence is described as a disjuncture between long-term identification with a political party and the short-run evaluations of the parties’ capacity to engage in meeting a voter’s expectations (Basinger and Lavine 2005).3 Under conditions where a party does not behave as expected (e.g. a party is enveloped in scandal, fields poor candidates, or is associated with protracted economic instability), disequilibrium of attachments may occur, resulting in deeply unstable preference orientations (Basinger and Lavine 2005). These cross-pressured, prospective voters may be “attracted to each party by one set of opinions and repelled by another” because policies that satisfy one particular value may require the individual to sacrifice another (Berelson et al. 1954, p. 190; Tetlock 1986). Further, because political party platforms may lack an internally-consistent rigor of logic, the prospective voter cannot help but face myriad cross-pressures when casting a series of votes.

As a consequence of this cross pressure, concurrent positive and/or negative summary evaluations toward the parties may result in unstable, unreliable, and ambivalent attitudes that are tentative, weakly rooted, and easily changed (Mulligan 2011). In the context of an election, these unstable attitudes may further reduce stability in preferences, leading to a reduction in goal-directed behavior; when such disequilibrium occurs, we would expect that ambivalent voters would be more likely to split their tickets, whether to satisfy this sense of internal cognitive dissonance (Campbell and Miller 1957) or because party labels are simply less useful as heuristics (Basinger and Lavine 2005). At present, this is exactly what Mulligan (2011) and Lavine et al. (2012) observe: ambivalent voters have a much higher propensity to split their tickets as these cross-pressures increase.

The Role of Indifference

However, if ambivalent voters are fundamentally those individuals who have mixed feelings or connections to the parties, then these voters have at least some affective connection to the parties. Because these individuals, cross-pressured though they may be, still have a partisan component to their identities that is consequential for processing decisional criteria (Nir and Druckman 2008; Basinger and Lavine 2005), they are far less likely, for example, to defect to out-party candidates (Thornton 2013a). Thus, there may be good reason to suspect that ambivalent voters should behave differently than those indifferent voters who have no affective connection to the parties at all. For indifferent individuals, “who are neither one-sided nor conflicted, but who have only a nominal attachment to one of the two parties” (Thornton 2013a, p. 5), I argue that the likelihood of splitting a ticket should be much greater because these voters lack any meaningful affective ties to the parties that might function as determinative voting criteria.

In fact, there is a developed narrative within the literature on voting behavior that underscores this prediction that indifferent voters should be more likely to split their ballots. If partisan considerations are not central to the self-image of indifferent voters (Campbell and Miller 1957)—i.e. indifferent individuals have little or comparatively less cognitive or emotional attachments from which they draw politically-relevant information—then they should deviate from nominal partisan attachments when considering their electoral options. This comports well with Fiorina’s (1977, 1981) conceptualization of partisanship, where an individual’s partisan identity is the combination of political experience and socialization: if one-sided (or polarized) individuals have repeated and positive past experience dominated by explicit connections to a party, indifferent individuals have little more than surface-level experience that governs their decisional criteria. As Thornton (2013a) finds, this is a recipe for increased partisan defection: indifferent individuals have little in the way of positive affective attachments that bind them to a particular party and, thus, defect to out-party candidates with greater regularity then their ambivalent counterparts.

Here, an illustration is helpful to more fully appreciate the unique differences that exist between indifferent and ambivalent individuals across a range of politically-relevant attitudinal orientations and behaviors. Using American National Elections Studies (ANES) survey data taken from the period 1984–2004, Fig. 1 demonstrates that indifferent individuals are less politically knowledgeable, pay less attention to national news generally and election-related media specifically, and are less likely to possess strong partisan attachments than their ambivalent counterparts. Without robust levels of political interest and a distinct lack of affective partisan attachments to anchor choice, I argue that indifferent individuals’ ability—or desire—to cast a straight ticket should be quite muted, that the indifferent individual appears to make a “selection among candidates on a capricious, quasi-random basis (Campbell and Miller 1957, p. 300).4 As a result, we should expect indifferent voters to be prime candidates to split their tickets. Moreover, I expect that after disentangling indifferent from ambivalent individuals, the true effect of “ambivalence” will be muted (though not necessarily non-existent), while indifference should be a much stronger predictor of engaging in split-ticketing. Before testing these predictions, however, I first turn to a brief discussion of how we might extricate these two affective states from one another.
Fig. 1

One-sided, indifferent, and ambivalent individuals mean scores across range of political behaviors and attitudinal orientations. Source 1984-2004 Cumulative ANES. Note All variables have been rescaled from 0 to 1. The nominal categories of one-sided, indifferent, and ambivalence are derived from the Comparative Ambivalence scale (c.f. Basinger and Lavine 2005), where “one-sided” individuals are aggregated together from those persons who score between -2.5 and -0.01, “indifferent” individuals are those persons who only score 0, and “ambivalent” individuals are those persons aggregated together who score from 0.01 to 5

Disentangling the Indifferent Voter from the Ambivalent One

In its most widely-utilized form, the degree to which an individual is ambivalent about political objects is operationalized through a two-part equation that taps both the similarity and intensity of affect (Basinger and Lavine 2005), which are the necessary conditions required to elicit the arousal of ambivalence (Thompson, Zanna, and Griffin 1995).5 This formula—termed “comparative ambivalence”—produces two formal expectations: (1) as the affective evaluation of one object becomes stronger than another other—in this case as the affective reactions to the parties become unequal—an individual’s attitude should polarize, reducing ambivalence; and (2) the strength of an individual’s degree of ambivalence should be greater when these positive and negative components are of moderate intensity.6 Practically, comparative ambivalence is most often constructed from open-ended response evaluations—i.e. likes and dislikes about the parties—where the first term measures similarity and the second term the intensity of affective preferences:
$${\text{Ambivalence}}_{\text{comp}} = \frac{{{\text{Democrat}} + {\text{Republican}}}}{2} - \left| {{\text{Democrat }} - {\text{ Republican}}} \right|,$$
where
$${\text{Democrat}} = \frac{{{\text{PositiveD}} + {\text{NegativeR}}}}{2}\;{\text{and Republican}} = \frac{{{\text{PositiveR}} + {\text{NegativeD}}}}{2}$$

This equation produces an interval-level variable whose values range from −2.5 to 5, where negative scores represent the presence of polarized or one-sided attitudes and positive scores represent the presence of ambivalence. There is mounting evidence, however, that this interval-level operationalization of ambivalence may be problematic (Rudolph 2005), that linear scales of ambivalence may fail to fully differentiate between “indifferent” and “ambivalent” individuals (Meffert et al. 2004; Thornton 2011). Although these criticisms are fully outlined elsewhere (see Thornton 2011), I briefly summarize the argument against using an interval-level operationalization of ambivalence because this has significant bearing on how we conceptualize and operationalize indifference’s effect on split-ticketing behavior.

First, because a majority of comparative ambivalence scores cluster between a very short range, the prevailing, interval-level measurement scheme may fail to empirically differentiate between the observed variation in light of theoretical expectations. Looking at the Cumulative ANES data, for example, more than 85 % of all possible ambivalence scores fall between −1 and 1, while fewer than two % of individuals score higher than “positive two” (i.e. moderate ambivalence). In other words, it may be the case that scores at the upper and lower bounds of the comparative ambivalence scale are capturing more noise than substantive information (Thornton 2011).7

Second, there appears to be a fundamental problem with the way in which the comparative ambivalence scale treats similar scores insofar as there are a variety of routes for two individuals to achieve the same score—routes which should inform different conclusions about the individuals, even though their scores are identical. Here, Thornton (2011) describes a scenario in which a respondent who dislikes both parties a great deal actually receives the same ambivalence score as an individual who is appears slightly ambivalent on the face of it; calculating the two individuals’ level of ambivalence, however, yields the same score. In this case, variation may be artificially established through the measurement technique rather than as the result of actual substantive differentiation.

Finally, and most relevant to the analyses at hand, comparative ambivalence doesn’t handle nonresponses well. From the period of 1980–2004, the modal category for all ambivalence scores, 0, is overwhelmingly populated by individuals who simply couldn’t offer a single like or dislike about the parties. In a series of previous analyses, Thornton (2011) demonstrates that these “indifferent” individuals appear substantively distinct from ambivalent and one-sided (polarized) respondents on the strength of their ideological attachments, their interest in politics, and their knowledge of basic economic information. This underscores the descriptive presentation of the differences between indifferent and ambivalent individuals in Fig. 1.

Taken collectively, these criticisms suggest that a more appropriate operationalization of ambivalence—and, therefore by extension, indifference—occurs under a nominal coding of these affective states. This has important theoretical and empirical consequences for our conceptualization of indifference. Because a number of studies hint that treating indifference as a midpoint within comparative ambivalence is inappropriate insofar as this obscures important substantive distinctions between ambivalent and indifferent individuals (Meffert et al. 2004; Thornton 2011; Rudolph 2005; Burden and Summary 2012), there is good reason to suggest that this current measurement practice may underestimate the true effect that indifference has on split-ticketing. My primary argument, then, is that by parceling out indifferent individuals from ambivalent ones, we should observe that indifferent individuals cast split-tickets at greater rates than their ambivalent counterparts—an effect that has been otherwise undersold in previous research.

Data and Measures

To demonstrate the difference between the effects of indifference and ambivalence on an individual’s propensity to engage in split-ticketing, I utilize data from American National Election Studies (ANES) Time-Series surveys for the period 1984 through 2004.8

The Split-Ticket

The dependent variable for the forthcoming analyses is the self-reported House-President split-ticket, where a respondent identified that they cast a vote for either a Republican Presidential candidate and a Democratic House candidate or a Democratic Presidential candidate and a Republican House candidate. These split-tickets are coded 1, and otherwise 0.

Decomposing Comparative Ambivalence: Ambivalence, Indifference, and One-Sided Individuals

Comparative ambivalence is calculated using the Basinger and Lavine (2005) equation discussed in the previous section (where possible scores range from −2.5, very polarized or one-sided individuals, to 5, very ambivalent respondents). To decompose this interval-level scale into nominal categories, I follow Thornton’s (2011) coding scheme, where individuals with negative scores on comparative ambivalence are coded as “one-sided,” those with positive scores are coded as “ambivalent,” and those who made no positive or negative statements about the parties are coded “indifferent.”9 Under this coding scheme, indifference and ambivalence are each coded as dummy variables, where the effect of being indifferent is coded 1 and otherwise 0, and the effect of being ambivalent is coded 1 and otherwise 0. This leaves the group of polarized individuals as the excluded category of interest.

Control Variables: Known Correlates of Split-Ticketing

In order to provide a robust test of the effects of ambivalence and indifference on split-ticket voting, I include a number of variables that have been previously demonstrated to influence the likelihood of an individual to split their ticket—if either ambivalence or indifference retains its significance while controlling for these alternative explanations, then we can be confident in the role that they play in producing split-tickets.

Campbell and Miller’s (1957) early work on voting behavior suggests that competing theories of split-ticketing might be broken down into motivated and unmotivated explanations. In the first case, motivated theories of split-ticketing are related to characteristics or attitudinal orientations of the prospective voter, which includes weak-partisanship and policy-balancing theories. Using data available within the cumulative ANES dataset, I control for two motivated theories of split-ticketing: partisan strength and policy-balancing. Here, the strength of partisan attachment is captured by folding the party identification scale in half, where Independents are located at the lowest value of the scale, coded 0, and strong partisans are located at the highest value, coded 3.

In addition to controlling for partisan strength, I also employ Carsey and Layman’s (2004) version of policy balancing that captures strategic balancing through the use of ideological self- and party-placement scales. This formula is specified as:
$${\text{Party balancing}} = \left| {{\text{R}}_{\text{ideo}} {-}{\text{C}}_{\text{ideo}} } \right|{-}\left| {{\text{R}}_{\text{ideo}} {-}\frac{{{\text{GOPideo}} + {\text{Dideo }}}}{2}} \right|,$$
where Rideo is the respondent’s self-placement within ideological space; Dideo is the respondent’s placement of Democratic Party within ideological space; GOPideo is the respondent’s placement of Republican Party within ideological space; Cideo is the the value of Dideo or GOPideo that most closely approximates the respondents ideological self-placement, Rideo.

This measure of ideal ideological preferences ranges from -5 to +5, where respondents at the negative pole (−5) perceive the parties as fully polarized and also locate themselves at one of the ideological extremes. These voters, obviously, would be very unlikely to cast a split-ticket given that their ideal preferences are for uniform party control. Conversely, individuals at the positive pole (+5) are those persons who see the parties as both fully polarized, yet locate themselves at the midpoint between the two groups. These prospective voters, then, would be much more likely to support divided government (Carsey and Layman 2004).

However, because roughly 25% of individuals within the cumulative ANES are unable to place themselves within ideological space—an omission that precludes the creation of a policy-balancing variable for a substantially large portion of individuals—I adjust this measure of policy-balancing by coding these missing data at 0 on the policy-balancing scale. In order to account for the effect of individuals who could not place themselves on this scale, I then create a dummy variable “no ideology” that is coded 1 for individuals who could not place themselves on the 7-point ideology scale and 0 for otherwise. To interpret a score of 0 on the policy-balancing variable (i.e. an individual cannot ideologically differentiate between the parties and sees themselves as a moderate) one needs to hold the balancing item at 0 and the “no ideology” variable at 0. In order to appropriately interpret the effect of “no ideology,” the balancing variable should again be held at 0 and the “no ideology” variable at 1. This arrangement allows for the maximization of potential data, and is a conservative test of the effects of indifference insofar as both the policy-balancing and no ideology variables should be positively related to split-tickets.

In addition to these controls, I also include variables that control for both voters’ level of knowledge and the effects of (out-party) incumbency. Unfortunately, because the requisite variables needed to construct a political knowledge scale where respondents attempt to identify political figures are unavailable before 1988, I employ the use of another type of political knowledge item. Here, individuals are asked to correctly place both the Republicans and Democrats on an ideology scale, where correctly identifying that Republicans are conservative and Democrats are liberal will result in a score of 1, where identifying one party correctly is coded 0.5, and incorrectly identifying both parties is coded 0. Finally, following previous literature, out-party incumbency is employed as a control because voters are often likely to cross party lines to vote for an incumbent member of Congress (Burden 2002); I code the presence of an out-party incumbent 1 and otherwise 0.

Analytical Strategy

The results are broken down into two series of analyses. Table 1 presents a pooled model that depicts House-President split-ticketing as a function of ambivalence, indifference, and controls. Both marginal effects and predicted probabilities are presented for each of the variables in order to assist in the interpretation of the logistic regression coefficients. Table 2 illustrates a series of 12 models that span the six presidential elections of interest. Here, the first column under each year “(1)” presents a model in which comparative ambivalence is employed, while a second column “(2)” depicts a model that utilizes nominal categories of ambivalence and indifference (where one-sided is the excluded category). Table 3 provides a series of predicted probabilities that convert these analyses into easily interpretable results that juxtapose indifferent relative ambivalent rates of split-ticket voting. Finally Table 4 compares estimates of indifferent split-ticketing across linear and nominal codings of indifference.
Table 1

Pooled, cross-sectional logit estimates of the effects of ambivalence and indifference on house-president split-ticketing, 1984–2004

 

b

Marginal effect

Predicted prob. of ticket-split at

Min. value

Max. valu

Ambivalence [+]

0.27**

0.04

20.27%

24.35%

(0.09)

(0.01)

(0.01)

(0.01)

Indifference [+]

0.52**

0.09

20.33%

28.57%

(0.11)

(0.02)

(0.01)

(0.02)

Out-party incumbent [+]

1.48

0.26

13.94%

39.11%

(0.08)**

(0.01)

(0.01)

(0.01)

Policy-balancing [+]

0.07

0.01

16.32%

27.5%

(0.02)**

(0.003)

(0.01)

(0.02)

No ideology [+]

0.22*

0.03

21.19%

24.57%

(0.11)

(0.018)

(0.01)

(0.02)

Partisan strength [−]

−0.46**

−0.07

37.58%

15.08%

(0.04)

(0.006)

(0.02)

(0.01)

Knowledge [−]

−0.15

−0.02

23.35%

21.06%

(0.11)

(0.02)

(0.01)

(0.01)

1988

0.05

   

(0.11)

1992

−0.05

   

(0.11)

1996

−0.68**

   

(0.16)

2000

−0.52**

   

(0.16)

2004

−0.44**

   

(0.13)

Constant

−0.97**

   

(0.15)

Likelihood ratio x2

562.92

   

Pseudo R2

0.120

   

N

4463

   

Source 1984–2004 ANES cumulative file

p ≤ 0.05, ** p ≤ 0.01 (one-tailed tests). Logistic regression coefficients with standard errors in parentheses; expected direction of coefficients in brackets. Predicted probabilities of Min/Max values on independent variables calculated while holding other variables at means

Table 2

Logit estimates for house-president split-ticketing, 1984–2004

  

1984

1988

1992

1996

2000

2004

(1)

(2)

(1)

(2)

(1)

(2)

(1)

(2)

(1)

(2)

(1)

(2)

Comparative Ambivalence (interval)

[+]

0.20**

0.17**

0.07

0.09

0.33**

 

–0.19*

(0.09)

 

(0.08)

 

(0.09)

 

(0.14)

 

(0.14)

 

(0.11)

 

Ambivalence (dummy)

[+]

0.37**

0.45**

0.18

0.30

0.04

0.13

 

(0.18)

 

(0.19)

 

(0.19)

 

(0.32)

 

(0.30)

 

(0.25)

Indifference (dummy)

[+]

0.64**

0.53**

0.53**

0.38

-0.37

0.69*

(0.20)

(0.26)

(0.22)

(0.42)

(0.52)

(0.32)

Out-party Incumbent

[+]

1.42**

1.41**

1.83**

1.85**

1.35**

1.34**

1.24**

1.24**

1.59**

1.56**

1.45**

1.47**

(0.16)

(0.16)

(0.18)

(0.18)

(0.17)

(0.17)

(0.28)

(0.28)

(0.29)

(0.29)

(0.22)

(0.23)

Policy-balancing

[+]

0.06

0.05

0.10*

0.10*

0.07

0.06

0.18

0.18

0.03

0.05

0.09*

0.08

(0.04)

(0.04)

(0.04)

(0.04)

(0.04)

(0.04)

(0.07)

(0.07)

(0.06)

(0.06)

(0.05)

(0.05)

No ideology

[+]

0.06

0.01

0.46*

0.42*

0.33

0.29*

0.42

0.35

0.22

0.18

0.39*

0.30

(0.21)

(0.19)

(0.24)

(0.24)

(0.22)

(0.22)

(0.40)

(0.42)

(0.42)

(0.42)

(0.31)

(0.32)

Partisan strength

[–]

−0.53**

−0.51**

−0.40**

−0.38**

−0.47**

−0.43**

−0.41**

−0.38**

−0.56**

−0.62**

−0.51**

−0.50*

(0.09)

(0.09)

(0.09)

(0.09)

(0.09)

(0.09)

(0.15)

(0.16)

(0.16)

(0.16)

(0.12)

(0.12)

Knowledge

[–]

−0.19**

−0.09*

−0.34

−0.33

−0.48*

−0.38

0.17

0.21

−0.12*

−0.10*

0.15

0.25

(0.20)

(0.20)

(0.23)

(0.23)

(0.21)

(0.22)

(0.39)

(0.39)

(0.41)

(0.41)

(0.31)

(0.32)

Constant

 

−0.53*

−0.91**

−0.91**

−1.23**

−0.53

−0.83**

−1.71**

−1.96**

−1.24

−1.05

−1.37**

−1.60**

(0.22)

(0.26)

(0.26)

(0.29)

(0.24)

(0.28)

(0.47)

(0.53)

(0.49)

(0.53)

(0.35)

(0.39)

Likelihood Ratio

χ2

138.8

143.8

144.5

144.9

98.8

103.7

37.70

38.5

50.6

45.3

69.2

70.89

Pseudo R2

0.115

0.121

0.140

0.144

0.093

0.097

0.100

0.102

0.134

0.120

0.113

0.116

N

1,040

1,040

899

899

1,007

1,007

433

433

430

410

674

674

Source 1984–2004 ANES Cumulative File

 p ≤ 0.10, * p ≤ 0.05, ** p ≤ 0.01 (one-tailed tests). Logistic regression coefficients with standard errors in parentheses; expected direction of coefficients in brackets

Table 3

Predicted probability of likelihood of split-ticketing for one-sided, indifferent, and ambivalent individuals, house-president voting 1984–2004

Likelihood of split-ticket…

 

1984 (%)

1988 (%)

1992 (%)

1996 (%)

2000 (%)

2004 (%)

One-sided

21.15

20.97

19.64

13.52

17.57

14.77

Indifferent

34.88

32.56

29.42

19.69

13.50

25.19

Ambivalent

30.29

29.42

24.04

17.98

17.54

17.87

Source 1984–2004 ANES cumulative file

Note Predicted values calculated using respective-year “Model 2” from Table 2, while holding other variables at their means; italic values are significant at p ≤ 0.10 (one-tail test), bolded values are significant at p ≤ 0.05 (one-tail test)

Table 4

Predicted probability of ticket-splitting for “indifferent” individuals under 2 specifications of indifference, house-president voting 1984–2004

Likelihood of split-ticket…

 

1984 (%)

1988 (%)

1992 (%)

1996 (%)

2000 (%)

2004 (%)

Model 1 specification

 Comparative ambivalence Indifference (ambivalence = 0)

26.10

24.94

22.30

15.63

16.34

16.60

Model 2 specification

 Nominal-level ambivalence Indifference

34.88

32.56

29.42

19.69

13.50

25.19

 Actual proportion of indifferent individuals in sample who cast split ticket

34.75

33.13

34.26

24.00

17.14

26.26

Source 1984–2004 ANES cumulative file

Note Model 1 and Model 2 correspond to the models (1) and (2) found in Table 2. Predicted values calculated holding other variables at their means; italic values are significant at p ≤ 0.10 (one-tail test), bolded values are significant at p ≤ 0.05 (one-tail test)

Evidence from the Pooled Models

Ambivalence and Indifference

As the model in Table 1 illustrates, I find that ambivalence has a positive and moderately strong effect upon split-ticketing (b = 0.27, s.e. = 0.09). Simply put, for individuals who possess conflicting affective evaluations of the parties, the likelihood of casting a split-ticket increases by a little more than four percentage points. However, consistent with my expectations, I find that indifference has a much stronger, positive effect upon the likelihood of casting a split ticket than ambivalence (b = 0.52, s.e. = 0.11), and a Wald test suggests the difference between these two coefficients is significant (χ2 = 4.70, p = 0.03). Here, feeling indifferent about the parties increases the likelihood of casting a split-ticket by almost nine percentage points. If we compare the predicted probabilities of ambivalent relative indifferent voters to cast a split-ticket, we find that, holding all else equal, ambivalent individuals are predicted to cast a split-ticket about 24.35 % of the time compared to indifferent individuals at 28.57 %.

According to the extant literature on comparative ambivalence, this finding is unexpected: in fact, according to previous scholarship, we would expect more split-ticketing from ambivalent individuals than indifferent ones. This discrepant observation, then, serves as an initial starting point for questioning the propriety of utilizing prevailing interval-level measures of ambivalence when predicting split-ticketing behavior. If these estimates are any indication, it appears that there is something systematically unique about indifferent individuals that make them more likely to engage in split-ticketing than their ambivalent counterparts.

Control variables

According to the pooled model presented in Table 1, I find that each of the correlates of split-ticketing behave as previous literature predicts. First, policy-balancing has a positive, modest effect upon the likelihood of an individual to cast a split ballot (b = 0.07, s.e. = 0.02). For individuals who see the parties as fully polarized, yet locate their own ideological preferences in the middle (represented by a positive score on the balancing scale), the likelihood of casting a split-ticket increases. For those individuals who are unable to locate themselves within ideological space, the “no ideology” variable captures the effect of a lack of ideological consciousness. Here, the inability to place oneself within the liberal-conservative scale results in an increase in the likelihood of casting a split-ticket (b = 0.22, s.e. = 0.11). Second, the strength of psychological (group) attachments has a strong effect on split-ticketing. Here, partisan strength has a strong negative and significant effect on the likelihood of casting a split-ticket (b = −0.46, s.e. = 0.04). As an individual’s partisan identity becomes stronger, the likelihood of casting a split ballot drops precipitously. Third, transitioning from “motivational” theories of ticket-splitting to structural or “unmotivated” ones, I observe that out-party incumbency is a major driving factor for split-ticketing. Here, the coefficient for out-party incumbency is positive, highly significant, and substantively consequential in light of its impressive marginal effect (b = 1.48, s.e. = 0.08): when an opposite-party candidate is an incumbent, individuals are highly likely to vote for that candidate, even at the expense of voting for an in-party alternative. Fourth and, finally, political knowledge, operationalized as the ability to correctly place Republicans as conservatives and Democrats as liberals is insignificance, though its effect is negative as expected (b = −0.15, s.e. = 0.11).

Ambivalence, Indifference, and the Split-Ticket Over Time

Turning, now, to a fuller discussion of the effects of ambivalence and indifference on split-ticketing over time, it is worth examining proportional trends in these groups’ split-ticketing during the twenty-year period dating from 1984 to 2004. Because the theoretical framework put forth here contends that indifference should be considered a separate psychological state apart from ambivalence, Fig. 2 presents group proportions of “ambivalent,” “indifferent,” and “one-sided” voters who carried split House-President tickets.
Fig. 2

Proportion of one-sided, indifferent, and ambivalent house-president split-ticketing, 1984–2004 ANES

Looking closely at this figure, two immediate conclusions can be drawn. First, over time and throughout each group, split-ticketing has slowly declined. This corresponds to a voluminous research program that suggests American voters have become more polarized (McCarty et al. 2006) and, therefore, less likely to cast split-ballots (Burden and Kimball 2004). Second, there appears to be a clear ordering among the groups in the proportion of individuals who cast split ballots: on balance, one-sided individuals were the least unlikely individuals to cast a split-ballot, while indifferent voters were more likely to cast a split ticket than ambivalent individuals for every election save 2000, a presidential election year in which indifferent individuals contributed historically low levels of split-ticketing.

In order to more closely examine the difference between the effects of ambivalence and indifference (and how these effects may vary from election to election), I now turn to Table 2. Here, according to each Model 1 in Table 2 comparative ambivalence has a positive and significant effect in three of six elections, with the exceptions occurring in 1992, 1996 and 2000.10 In general, these results correspond with previous findings, although Mulligan’s (2011) statement that the effects of ambivalence are relatively static over time—“the effects of partisan ambivalence on ticket splitting over the years is not much different from its effect in 2004” (Mulligan 2011: 518)—seems to run counter to both the observational data presented in Fig. 2 and this election-by-election analysis offered in Table 2. In fact, over time, it does appear as if the effect of comparative ambivalence not only weakens, but is highly variable as well.

Transitioning to the second set of models “(2)” within each election in Table 2, I find that although ambivalence is again variable in its significance, indifference appears to be a more consistent predictor of positive split-ticketing. In fact, here, I find that while ambivalence behaves in the expected way for 1984 and 1988, indifference has a positive and significant effect on the likelihood of casting a split-ticket in four of six elections. Here, the two aberrations occur in 1996, where indifference is correctly signed but insignificant, and 2000, a highly polarized election in which very little split-ticketing occurred in general.11 Thus, while ambivalence has uneven influence on split-ticketing in these election-by-election analyses, indifference has a consistently stronger, positive effect than ambivalence; moreover, the models with nominal codings of ambivalence and indifference slightly fit the data better than the models that contain the interval-level operationalization.

To provide a clearer portrait of this phenomenon, Table 3 provides predicted probabilities of ticket-split voting for one-sided, indifferent, and ambivalent voters over time. According to Table 3, it is apparent that for 1984 and 1988, the likelihood of ambivalent relative indifferent voters to cast a split-ticket is fairly equitable, although we still observe that indifferent voters are modestly more likely to cast a split-ballot in both elections. In 1984, indifferent voters are about four percentage points more likely to cast a split ticket relative ambivalent voters (34.88% vs. 30.29% respectively), while indifferent voters are about three percentage points more likely to cast a split ballot in the 1988 election (32.56% vs. 29.42% respectively). In 1992, this difference widens further: indifferent voters are about five and a half percentage point more likely to cast a split-ticket than ambivalent individuals (29.42% vs. 24.04% respectively).

As we transition to the 1996 and 2000 elections, a slightly less clear picture emerges. Although indifference has a significant and positive bivariate effect in 1996, this effect fades in the full model. Here, both indifferent and ambivalent individuals are about equally-likely to cast a split-ticket, although we observe that the effect for indifference is at least correctly signed. In 2000, however, both indifference and ambivalence have statistically insignificant effects, and, in the case of indifference, the direction of the effect is in the opposite direction theorized.12 However, as we transition to the 2004 election, indifference, again, has a significant and strong positive effect on the likelihood of an individual to cast a split-ballot. Here, indifferent individuals are predicted to cast split-tickets about 25 % of the time; fascinatingly, the effect of ambivalence on casting a split-ticket is again statistically indistinguishable from zero.

If the story told in Table 3 is that indifference is a more consistent predictor of an individual’s likelihood of casting a split ticket, then “indifference” within a linear specification of ambivalence may be underestimated since it is treated as a midpoint between one-sided and ambivalent individuals (i.e. its true effect may be artificially constrained within the linear scale). Turning to the results found in Table 4, I demonstrate just how much comparative ambivalence underspecifies the “true” effect of indifference. In each election year, Model 1 incorrectly specifies the effect of indifference, with the most pronounced difference occurring in the 1992 election, where comparative ambivalence underestimates the effect of indifference by about 8 percentage points. Over time, comparative ambivalence’s predictions of indifference’s effect on split-ticketing are considerably lower than those of the nominal coding; across all six elections, comparative ambivalence under-estimates the effect of indifference on casting a split-ticket by an average of five and half percentage points. Moreover, if we ignore the election in 2000 where indifference had a negative effect, then this number jumps to over seven percentage points. This is not an insignificant set of errors: in the 2004 election alone, comparative ambivalence underspecifies the effect of a nominal coding of indifference by more than 30 %.

The true test of any set of predictions, however, is how well they match observed reality. To this end, we might contextualize the predicted values in Table 4 with the actual proportion of indifferent split-ticket voting in the sample population (also in Table 4). Comparing these values, the effect of indifference when coded as a nominal measure is clearly preferable to the set of estimates derived from the linear operationalization of ambivalence. Although the predictions of the nominal coding in Model 2 are not perfect, these estimates are remarkably cleaner and more closely approximate reality, even after controlling for the strongest possible competing theories of split-ticketing.

Discussion and Conclusions

The study of split-ticket voting has engendered a substantial amount of interest from scholars, and the suspicion that affective partisan cross-pressures may induce split-tickets dates back to Berelson et al. (1954) early work. The results presented in this paper should not be read as a complete repudiation of previous work on the connection between ambivalence and split-ticket voting; rather, the primary objective of this analysis is to demonstrate the that indifference is an extraordinarily consequential attitudinal state. A second objective naturally stems from the first: to carefully scrutinize the intersection of measurement and theory as these relate to our understanding of the empirical effects that ambivalence and indifference have on political behavior.

The results of the preceding analyses indicate that although comparative ambivalence generally has a statistically significant effect on split-ticketing, this effect is highly variable over time and underestimates the true behavior of indifferent individuals. Although extant literature has persuasively chronicled ambivalence’s effect on split-ticketing (Mulligan 2011), this previous research leads us to expect to far greater split-ticketing for ambivalent voters—and, subsequently, a much reduced likelihood of split-ticketing for indifferent individuals—than we actually observe in practice.13 While the results of this analysis don’t explicitly preclude a graduated effect of ambivalence at the causal level—surely some individuals may be more ambivalent than others, the consequences of which being potentially different—my findings demonstrate that indifference is a more consistent and powerful predictor of split-ticket behavior. Moreover, these findings fit into a growing literature that suggests that treating indifference as a midpoint or threshold within the linear specification of comparative ambivalence may be misguided (Thornton 2013a)—that, “when such linear constraints are not justified given the data, there is an increased risk of drawing erroneous substantive inferences” (Rudolph 2005, p. 924) . If the preceding analyses are any indication, using the comparative ambivalence index to predict split-ticket voting does, in fact, lead the researcher to draw the erroneous conclusion that ambivalence has a stronger positive effect on split-ticketing than indifference. In reality, it is the indifferent voter who possesses the stronger propensity to cast a split-ticket.

That indifference should more strongly induce split-ticket voting than ambivalence should not surprise us. Partisan ambivalence is, after all, a pull between conflicting affective attachments to the parties—the idea that “one may vote for a Republican candidate and yet feel part of a Democratic team” (Green et al. 2002, p. 8). This sense of basic identity—conflicted though it may be—appears to be generally lacking for indifferent individuals who pay less attention to election campaigns, possess the lowest levels of political knowledge, and are less likely to hold constrained ideological preferences. And even though the proportion of indifferent individuals has fallen over time—suggesting that even indifferent individuals may not be immune from the pull of extreme partisan pressures of political elites (Thornton 2013b)—the absence of affective partisan connections is still highly consequential. Without the strong bonds or even weak, conflicted ties of such psychological partisan attachments, indifferent individuals are prime candidates for splitting their electoral decisions.

Footnotes

  1. 1.

    One further alternative that bears mentioning is a kind of sincere ticket-splitting that occurs when perceptions of an opposite-party candidate’s ideology is similar to an in-party candidate’s (Frymer et al. 1996).

  2. 2.

    A further problem with weak partisanship theory is that those voters who identify as independent, yet who “lean” in one partisan direction or another actually behave significantly more partisan than they confess to interviewers (Petrocik 1974).

  3. 3.

    This definition has slightly changed in Lavine et al.’s (2012, p. 54) recent work on ambivalence, where ambivalence is defined as “a disjuncture between the identification and evaluative components of partisanship.” Conceptually, this places more emphasis on the role of identity, which has some empirical implications for conceptualizing “ambivalence” that are discussed later in the text.

  4. 4.

    It’s worth noting, however, that the decision to split one’s ticket need not be absolutely governed by complete disinterest, ignorance, or even a lack of identity. Recent research on the motivations of political independents suggests that these individuals may possess a conscious “independence” which functions, to them, as a brand of partisan identification (Klar 2013). Under these circumstances, then, a conscious split-ticket by an indifferent individual may be a purposeful attempt at fostering a certain detachment from the parties. Still, the portrait of indifference presented here is that indifferent individuals do seem truly less attached to politics generally.

  5. 5.

    As mentioned previously, a more recent operationalization of ambivalence by Lavine et al. (2012) recasts ambivalence explicitly in terms of identity-consistent and identity-conflicting evaluations. This departure from previous work computes ambivalence by considering different configurations of these two evaluations, which are treated as independent sets of affective appraisals insofar as they are entered as separate variables in models rather than combined within an index. Unfortunately, this measurement strategy purges those individuals who identify as independents from analysis because they do not have a partisan identity that would allow the construction of “conflicting” or “consistent” party evaluations. Because independents are an important population to consider (especially so in studies of split-ticketing), I rely on the prevailing comparative ambivalence index to construct the nominal categories of indifference and ambivalence.

  6. 6.

    To clarify, I use the term “comparative ambivalence” in the forthcoming analyses when I refer to the original interval-level ambivalence scale, which is the parlance originally used by Lavine (2001) to describe this measure. I believe this strategy helpfully differentiates between the measurement instrument and the actual affective state it purports to measure.

  7. 7.

    Moreover, the “additive” process of aggregating “likes” and “dislikes” within the two components of the comparative ambivalence equation may misrepresent how responses are weighted by the individual insofar as subsequent affective reactions after an initial response may be given differential consideration by the individual (Priester and Petty 1996). In this case, the strength—or extremity—of reported affect may not be systematic.

  8. 8.

    Unfortunately the results of the present analysis are constrained to this time period because, even though data from additional years (2008, 2012) is available, the party likes/dislikes necessary for the construction of ambivalence/indifference have not yet been formally released by the principle ANES investigation team.

  9. 9.

    Because a very few individuals may score as “indifferent” by virtue of the computation of their likes and dislikes, only individuals with non-response 0 s are counted as indifferent. Conversely, individuals who score 0 as a function of the combination of their likes and dislikes are counted as ambivalent.

  10. 10.

    The sample size of the populations for each of these models is slightly smaller than one might expect for two reasons related to data collection. First, the amount of individuals who are able to recall who they voted for as Representative varies over time. Second, the number of individuals responding to the likes/dislikes item used to construct the ambivalence/indifference measures varies as a function of the introduction of split-survey designs in later years, where the open-ended survey items used to construct the likes/dislikes are not asked of all respondents.

  11. 11.

    Unfortunately, a split-survey design in 1996 results in a particularly low sample size compared to other years which may be partially responsible for the observed insignificance of both ambivalence and indifference in the multivariate model; still, in bivariate tests we observe that indifference has a significant and slightly larger effect than ambivalence.

  12. 12.

    In the case of 2000, the negative and insignificant effect for indifference actually resembles what occurred in reality: 2000 was a historical low-point for indifferent individuals to cast split-tickets. In part this may be the result of a particularly polarizing election between Bush and Gore, where individuals were much more likely to profess having some affective attachments to one of the two parties.

  13. 13.

    That we do not observe these trends raises some questions about the effect of ambivalence on split-ticket voting. Most obviously, what are we to make of highly-ambivalent individuals? Although I have demonstrated that indifference plays a stronger role in inducing split-tickets, the argument could be made that by aggregating together all voters who express varying degrees of ambivalence, we have simply constrained the true variable effect that high levels of ambivalence should have on an individual’s likelihood of casting a split ballot. In other words, we might ask, “Do “higher” ambivalence scores actually yield more split-ticket voting?” The short answer is a qualified no. Of the two individuals in the entire sample of six elections who score the maximum value on comparative ambivalence, neither splits their ticket. Moreover, the likelihood of splitting an individual’s ticket crests around scores of 2 and fall precipitously as scores increase (note that 90 % of all scores are less than 2). This effectively demonstrates prior research’s concern about the propriety of considering ambivalence’s effect on political behavior to be interval in nature (Thornton 2011), rendering it difficult to make the case that by considering ambivalence as a nominal state we have inappropriately constrained its effects.

Notes

Acknowledgements

The author would like to thank Judd Thornton for his encouragement and advice on this manuscript, as well as Chris Weber and Jim Garand for their generous comments on previous drafts. I would also like to thank the anonymous reviewers who delivered substantial feedback in a thoughtful manner; their careful reading of this manuscript greatly improved the final product. Any remaining errors are the author’s own.

References

  1. Alesina, A., & Rosenthal, R. (1995). Partisan politics, divided government, and the economy. Cambridge, UK: Cambridge University Press.CrossRefGoogle Scholar
  2. Basinger, S. J., & Lavine, H. (2005). Ambivalence, information, and electoral choice. American Political Science Review, 99(2), 169–184.CrossRefGoogle Scholar
  3. Beck, P. A., Baum, L., Clausen, A. R., & Smith, C. E. (1992). Patterns and sources of ticket splitting in subpresidential voting. American Political Science Review, 86, 916–928.CrossRefGoogle Scholar
  4. Berelson, B. R., Lazarsfeld, P. F., & McPhee, W. N. (1954). Voting: A study of opinion formation in a presidential campaign. Chicago: University of Chicago Press.Google Scholar
  5. Brunell, T. L., & Grofman, B. (2009). Testing sincere versus strategic split-ticket voting at the aggregate level: Evidence from split house-president outcomes, 1900–2004. Electoral Studies, 28(1), 62–69.CrossRefGoogle Scholar
  6. Burden, B. C. (Ed.). (2002). Uncertainty in American politics. Cambridge: Cambridge University Press.Google Scholar
  7. Burden, B. C., & Kimball, D. C. (2004). Why Americans split their tickets: Campaigns, competition, and divided government. Ann Arbor: University of Michigan Press.Google Scholar
  8. Campbell, A., & Miller, W. E. (1957). The motivational basis of straight and split-ticket voting. American Political Science Review, 51, 293–312.CrossRefGoogle Scholar
  9. Carsey, T. M., & Layman, G. C. (2004). Policy balancing and preferences for party control of government. Political Research Quarterly, 57, 541–550.CrossRefGoogle Scholar
  10. Fiorina, M. P. (1977). An outline for a model of party choice. American Journal of Political Science, 21(3), 601–625.CrossRefGoogle Scholar
  11. Fiorina, M. P. (1981). Retrospective voting in American national elections. New Haven, CT: Yale University Press.Google Scholar
  12. Fiorina, M. P. (1996). Divided government (2nd ed.). Needham Heights, MA: Allyn and Bacon.Google Scholar
  13. Garand, J. C., & Lichtal, M. G. (2000). Explaining divided government in the United States: Testing an intentional model of split-ticket voting. British Journal of Political Science, 30, 173–191.CrossRefGoogle Scholar
  14. Green, D., Palmquist, B., & Schickler, E. (2002). Partisan hearts and minds: Political parties and the social identities of voters. New Haven, CT: Yale University Press.Google Scholar
  15. Klar, S. (2013). Identity and engagement among political independents in America. Political Psychology. doi:10.1111/pops.12036.
  16. Lavine, H. (2001). The electoral consequences of ambivalence toward presidential candidates. American Journal of Political Science, 45(4), 915–929.CrossRefGoogle Scholar
  17. Lavine, H., Johnson, C. D., & Steenbergen, M. R. (2012). The ambivalent partisan: How critical loyalty promotes democracy. Oxford: Oxford University Press.CrossRefGoogle Scholar
  18. Lazarsfeld, P. F., Berelson, B. R., & Gaudet, H. (1944). The people’s choice. New York: Duell, Sloane, and Pearce.Google Scholar
  19. Lewis-Beck, M. S., & Nadeau, R. (2004). Split-ticket voting: The effects of cognitive Madisonianism. The Journal of Politics, 66, 97–112.CrossRefGoogle Scholar
  20. Lodge, M., Steenbergen, M. R., & Brau, S. (1995). The responsive voter: Campaign information and the dynamics of candidate evaluation. American Political Science Review, 89(2), 309–326.CrossRefGoogle Scholar
  21. McCarty, N., Poole, K. T., & Rosenthal, H. (2006). Polarized America: The dance of ideology and unequal riches. MIT Press.Google Scholar
  22. Meffert, M. F., Guge, M., & Lodge, M. (2004). Good, bad, indifferent, and ambivalent: The consequence of multidimensional political attitudes. In W. E. Saris & P. M. Sniderman (Eds.), Studies in public opinion: Attitudes, nonattitudes, measurement error, and change (pp. 60–100). Princeton, NJ: Princeton University Press.Google Scholar
  23. Mulligan, K. (2011). Partisan ambivalence, split-ticket voting, and divided government. Political Psychology, 32(3), 505–530.CrossRefGoogle Scholar
  24. Nir, L., & Druckman, J. N. (2008). Campaign mixed-message flows and timing of vote decision. International Journal of Public Opinion Research, 20(3), 326–346.CrossRefGoogle Scholar
  25. Petrocik, J. R. (1974). An analysis of intransitivities in the index of party identification. Political Methodology, 1, 31–48.Google Scholar
  26. Priester, J. R., & Petty, R. E. (1996). The gradual threshold model of ambivalence: Relating the positive and negative bases of attitudes to subjective ambivalence. Journal of Personality and Social Psychology, 71(3), 431–449.CrossRefGoogle Scholar
  27. Rudolph, T. J. (2005). Group attachment and the reduction of value-driven ambivalence. Political Psychology, 27(1), 99–122.CrossRefGoogle Scholar
  28. Tetlock, P. E. (1986). A value pluralism model of ideological reasoning. Journal of Personality and Social Psychology, 50(4), 819–827.CrossRefGoogle Scholar
  29. Thompson, M., Zanna, M., & Griffin, D. (1995). Let’s not be indifferent about (attitudinal) ambivalence. In J. Krosnick (Ed.), Attitude strength: Antecedents and consequences (pp. 361–386). Hillsdale, NJ: Lawrence Erlbaum.Google Scholar
  30. Thornton, J. R. (2011). Ambivalent or indifferent: Examining the properties of measures of partisan ambivalence. Political Psychology, 32(5), 863–884.CrossRefGoogle Scholar
  31. Thornton, J. R. (2013a). Getting lost on the way to the party: Ambivalence, indifference, and defection with evidence from two presidential elections. Social Science Quarterly. doi:10.1111/j.1540-6237.2012.00940.x.
  32. Thornton, J. R. (2013b). The impact of elite polarization on partisan ambivalence and indifference. Political Behavior, 35(2), 409–428.CrossRefGoogle Scholar
  33. Yoo, S. (2010). Two types of neutrality: Ambivalence versus indifference and political participation. Journal of Politics, 72(1), 163–177.CrossRefGoogle Scholar
  34. Zaller, J., & Feldman, S. (1992). A simple theory of the survey response: Answering questions versus revealing preferences. American Journal of Political Science, 36, 579–616.CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2014

Authors and Affiliations

  1. 1.Department of Political ScienceLouisiana State UniversityBaton RougeUSA

Personalised recommendations