Political Behavior

, Volume 34, Issue 4, pp 653–688

Personality and the Strength and Direction of Partisan Identification

Authors

  • Alan S. Gerber
    • Department of Political Science, Institution for Social and Policy StudiesYale University
    • Department of Political Science, Institution for Social and Policy StudiesYale University
  • David Doherty
    • Department of Political ScienceLoyola University Chicago
  • Conor M. Dowling
    • Department of Political Science, Institution for Social and Policy StudiesYale University
Original Paper

DOI: 10.1007/s11109-011-9178-5

Cite this article as:
Gerber, A.S., Huber, G.A., Doherty, D. et al. Polit Behav (2012) 34: 653. doi:10.1007/s11109-011-9178-5

Abstract

We examine the associations between personality traits and the strength and direction of partisan identification using a large national sample. We theorize that the relationships between Big Five personality traits and which party a person affiliates with should mirror those between the Big Five and ideology, which we find to be the case. This suggests that the associations between the Big Five and the direction of partisan identification are largely mediated by ideology. Our more novel finding is that personality traits substantially affect whether individuals affiliate with any party as well as the strength of those affiliations, effects that we theorize stem from affective and cognitive benefits of affiliation. In particular, we find that three personality traits (Extraversion, Agreeableness, and Openness) predict strength of partisan identification (p < 0.05). This result holds even after controlling for ideology and a variety of issue positions. These findings contribute to our understanding of the psychological antecedents of partisan identification.

Keywords

PersonalityParty identificationPartisan strengthBig Five

How do individuals make sense of the political world? Many individuals do so by forming partisan identities, which provide citizens with both a simplified framework for interpreting the political world (i.e., a “standing decision” regarding which candidate to support; see, e.g., Key 1966; Key and Munger 1959) and an “affective orientation to an important group-object in [their] environment” (Campbell et al. 1960, p. 121). [Prior research suggests that party identification affects a range of political outcomes including how people process political communications (Lodge and Taber 2000; Redlawsk 2002; Westen et al. 2006), evaluate and engage in the economy (Gerber and Huber 2009, 2010; Rudolph 2003), and make judgments about political actors (Gerber et al. 2010b; Goren 2002; Green et al. 2002).] Importantly, however, substantial portions of citizens do not identify with either party, and there is also wide variation in the strength of partisan affiliations among those who do identify with a party. What accounts for these differences? Why do some people strongly identify with a political party, whereas others identify only weakly or not at all?

We explore one factor contributing to the strength and direction of party identification: individuals’ core, dispositional personality traits. After a period of relative neglect, recent developments in personality research—perhaps most notably broad agreement that core personality traits can be conceived of and measured along five dimensions (the Big Five)—have led to a growing body of research linking personality traits to a diverse set of attitudes and behaviors. This research finds that Big Five personality traits predict political ideology, issue attitudes, and political behaviors (e.g., Gerber et al. 2010a, 2011; Mondak 2010; Mondak and Halperin 2008; Mondak et al. 2010). The Big Five also predict an array of non-political outcomes such as academic performance (Borg and Shapiro 1996; Diseth 2002; O’Brien et al. 1998; Wolk and Nikolai 1997; Ziegert 2000), health and longevity (Goodwin and Friedman 2006; Roberts and Bogg 2004), and earnings (Borghans et al. 2008).

Psychologists argue that Big Five personality traits serve as dispositional signatures that shape individuals’ responses to stimuli (e.g., Denissen and Penke 2008). That is, they are core aspects of what people are like that affect what ideas and experiences they find appealing and which they see as less attractive in both political and non-political settings. Research finds that these personality traits are strongly influenced by genetics and are highly stable through the life cycle (Bouchard 1997; Caspi et al. 2005; Costa and McCrae 1992; Gosling et al. 2003; Plomin et al. 1990; Van Gestel and Van Broeckhoven 2003). Thus, in contrast to most prior work on the origins of partisan attachments, which focuses on how parental socialization and the political environment affect partisanship, we examine how essential aspects of “who people are” shape these partisan attachments.1

We analyze how personality, as measured by the Big Five traits of Agreeableness, Openness to Experience, Emotional Stability (sometimes referred to by its inverse, Neuroticism), Conscientiousness, and Extraversion, affects the direction and strength of partisan attachments in the United States. There are two reasons to expect that personality is linked to party affiliation. First, as extensive prior research has documented, personality is associated with policy preferences and political ideology. For example, research consistently finds that those higher on the trait of Openness are more likely to be political liberals, whereas those higher on Conscientiousness are more likely to be political conservatives (Carney et al. 2008; Gerber et al. 2010a; Gosling et al. 2003; Jost et al. 2003; McCrae 1996; Mondak and Halperin 2008; Riemann et al. 1993; Van Hiel et al. 2000; Van Hiel and Mervielde 2004; but see Alford and Hibbing 2007; Mehrabian 1996). Thus, all else equal, people with personality traits that are associated with more extreme (liberal or conservative) ideological affinities should be more likely to also have adopted partisan labels, with the particular label chosen reflecting the ideological differences between the parties.

Second, personality is also likely to directly shape whether one is drawn to affiliate with a party for reasons unrelated to ideological proximity. This relationship has not been examined in prior work. Previous work (e.g., Burden and Klofstad 2005; Greene 1999, 2000) shows that cognitive (how we “think” about ourselves within the political world) and affective (how we “feel” about ourselves within the political world) factors, as well as political attitudes and ideology, help shape partisan attachments. Certain Big Five traits, we hypothesize, are likely to make those affective and cognitive benefits of partisan affiliation more attractive. Affiliating with a party—particularly a major party—provides a simplified cognitive framework for thinking about politics that is likely to appeal to individuals with certain personality traits. Affiliating also yields affective (social) benefits, such as a feeling of belonging, that those with particular personality traits are likely to find appealing. We develop hypotheses about specific Big Five traits and these types of rewards and assess whether they predict strength of partisan identification.

Our empirical investigation of the effect of personality on party identification makes two contributions to our understanding of the roots of partisan identities. First, we confirm previous work on how personality traits affect directional partisan identification—whether people identify with the Republican or Democratic Party. Prior work on Big Five traits and political outcomes has focused primarily on how these traits affect political ideology and, to a lesser extent, issue attitudes. Given that the direction of partisan attachment is likely to be closely tied to these attitudes, we predict and find that the associations between Big Five traits and decisions about which party to identify with are similar to those between Big Five traits and ideology.2

Second, we present new evidence on the relationship between personality and whether and how much people identify with a political party. We find evidence that Big Five traits are statistically significant (and often substantial) predictors of both whether people identify with a political party and the strength of their partisan identities. For example, a two standard deviation increase in Extraversion is associated with a six percentage point increase in the likelihood of identifying with a major party, whereas a similar increase in Openness is associated with an eight percentage point decrease in the likelihood of identifying with a major party. These associations persist even after accounting for individuals’ ideological and policy preferences, which suggests they are not simply a function of more ideological extreme individuals (as caused by the Big Five) being more likely to identify as partisans. Overall, our findings confirm ongoing research showing that Big Five traits shape political activities that vary in their social and political context and further unifies work on the relationship between the Big Five and political and non-political behaviors.

We begin by discussing how the Big Five traits and partisanship fit in the broader concept of personality. We also discuss how the contemporary political landscape in the United States may shape peoples’ understanding of what it means to identify as a Democrat, Republican, or Independent. After discussing our theoretical expectations and our data and methods, we report our analysis of the relationships between personality traits and the direction of partisan identification. Next we report our findings regarding whether personality traits affect the decision to identify with any political party, as well as the strength of that identification. Then we discuss some extensions of our research. We also present preliminary findings from analysis assessing the relationships between Big Five traits and (1) evaluations of an alternative political movement (the Tea Party) and (2) stability of partisanship over time. We conclude by discussing the implications of our findings.

Dispositional Personality Traits and Partisanship

Big Five Traits

In recent years, psychologists have reached a working consensus around the Five-Factor Model (the Big Five) as a way of conceptualizing and measuring dispositional traits (John and Srivastava 1999). The Big Five traits are: (1) Extraversion (associated with gregariousness and self-confidence); (2) Agreeableness (associated with harmonious relations with others); (3) Conscientiousness (associated with dutifulness and adherence to social norms); (4) Emotional Stability (associated with low levels of negative emotions, e.g., low anxiety); and (5) Openness to Experience (associated with intellectual curiosity and aesthetic appreciation).

Big Five traits are “the most stable and recognizable aspect[s] of psychological individuality” (McAdams and Pals 2006, p. 207), and they affect how people respond to the vast array of stimuli they encounter in their lives. Research finds that dispositional traits are quite stable through the life cycle (Caspi et al. 2005; but see Srivastava et al. 2003) and are strongly affected by genetics (Bouchard 1997; Plomin et al. 1990; Van Gestel and Van Broeckhoven 2003). Furthermore, the notion that dispositional traits can be conceptualized and measured according to a Five-Factor Model has been supported by studies conducted across a wide variety of populations (John and Srivastava 1999; Yang et al. 1999). As discussed above, research shows that these traits affect a vast array of behavioral and attitudinal outcomes.

Partisanship

The breadth and stability of Big Five traits suggests that they can be thought of as important “building blocks” of individuality. However, few would argue that these traits provide a complete picture of what distinguishes individuals from one another. People also vary in their values, habits, attitudes, and in the groups they identify with. These individual-level differences—which personality psychologists refer to as “characteristic adaptations”—are also aspects of personality. However, unlike basic dispositional traits which are relatively invariant, characteristic adaptations are “concrete manifestations of basic tendencies” (McCrae and Costa 1996, p. 69). In other words, they are a product of the interactions between people’s dispositional traits (e.g., the Big Five) and their experiences in the world. Thus, in contrast to dispositional traits, characteristic adaptations are more prone to vary in response to external influences such as peer socialization, education, cultural changes, and other environmental factors.3

Partisanship may be thought of as an example of a particular type of characteristic adaptation: a self-concept (Green et al. 2002; Greene 1999, 2000, 2002). Self-concepts are individuals’ understanding of their own identities. They are answers to the question “Who am I?” and may include religious affiliations, family roles (e.g., “mother”), occupations (“accountant”), and other aspects of a person’s identity (McCrae and Costa 2003, p. 215). These identities “provide social validation and a framework for interpreting the world” (Settles 2004, p. 488; also see Turner 1991). Identifying with a group, like a political party, therefore provides affective social benefits (see, e.g., Burden and Klofstad 2005; Campbell et al. 1960; Fiorina 1981; Greene 1999, 2000) as well as cognitive benefits (see, e.g., Burden and Klofstad 2005; Greene 1999, 2000; Fiorina 1981; Key 1966) that include “scripts or guides about how to behave” (Settles 2004, p. 488). As we discuss below, an individual’s personality is likely to affect the attractiveness of each of these aspects of partisan self-concepts. Importantly, these affective and cognitive benefits are in addition to the simple categorization process by which liberals (conservatives) would choose to identify as Democrats (Republicans) on policy grounds alone.

In formulating expectations about how dispositional traits affect party identification, it is important to consider the nature of the party system people can engage in. Affiliating with a party in the United Kingdom or France, where three (or more) major parties as well an array of minor parties are regularly on the ballot, may be quite different from affiliating in a two-party system like that found in the United States.4 In the United States, identifying as a Democrat or Republican entails siding with one of two “teams” engaged in sometimes pitched political battles. Thus, parties take clear issue stands on many positions, and individuals can choose a party on relative clear ideological divisions. Identifying as an Independent implies a lack of interest in such division or an explicit refusal to side with either “team” (e.g., Greene 1999, 2000). This refusal to take sides may stem from a sense that neither party represents one’s preferences, reluctance to take sides in conflicts between partisans, or a more basic commitment to maintaining one’s independence from any group. It is our argument that personality may shape strength of partisan identification for all three of these reasons.

Big Five Traits and Party Identification in the United States

In this section we present our theoretical expectations for how personality affects partisan affiliations. First, we discuss the relationship between Big Five traits and the direction of major party identification. Previous work has linked personality with ideology and we hypothesize that similar relationships exist between personality and directional party identification. Second, we discuss the effect of personality on the decision to identify with a party, including both the binary decision to identify with a major party and the strength of that identification.

Table 1 provides a brief summary of previous research on these topics. As this table makes clear, most of the questions we examine have not, to our knowledge, been addressed in previous research. For example, no published work has examined the relationships between Big Five traits and whether people adopt a partisan identification. Some work, however, has measured the relationships between dispositional traits and directional identification (or related outcomes such as candidate choice). In the remainder of this section we present our expectations regarding the relationships between Big Five traits and these outcomes. Our hypotheses are summarized in Table 2.
Table 1

Summary of prior literature on the relationships between Big Five personality traits and the direction and strength of party identification

Outcome

Directional party identification (right–left)

Strength of party identification

Studies using national US sample

Mondak (2010)US1

No studies

Studies using other US samples

Mondak and Halperin (2008)US2; Barbaranelli et al. (2007)US3; Rentfrow et al. (2009)US4

No studies

Studies using non-US samples

Caprara et al. (1999)EUR1; Schoen and Schumann (2007)EUR2

No studies

Prior findings by trait

 Extraversion

EUR1, US3; +US4

No studies

 Agreeableness

+US1, US2, US3, EUR1, EUR2

No studies

 Conscientiousness

US1, US2, US3, US4, EUR1, EUR2

No studies

 Emotional stability

US1, US2, US3, EUR2

No studies

 Openness to experience

+US1, US2, US3, US4, EUR1, EUR2

No studies

Note For prior literature, table entries are published work

+, Positive and statistically significant (p < 0.05) relationship in at least one sample

−, Negative and statistically significant (p < 0.05) relationship in at least one sample

Outcome variables Reported partisanshipUS1, US2; reported vote choiceEUR1, EUR2; vote intentionUS3; statewide vote (1996–2004)US4

Table 2

Theoretical expectations about the relationships between Big Five traits and the strength and direction of party identification

Outcome

Directional party identification (Republican to Democratic)

Strength of party identification

Extraversion

+

Agreeableness

None

+

Conscientiousness

+

Emotional stability

Openness to experience

+

Note “+” indicates positive association expected between trait and outcome; “−” indicates negative association expected; “none” indicates no association expected

Personality and the Direction of Party Identification

Politics in the United States is dominated by two parties that are closely identified with liberal (Democratic) and conservative (Republican) ideologies across social and economic policy domains. As prior research makes clear, people may choose (“sort into”) a party that shares their policy preferences (Abramowitz and Saunders 1998, 2006; Levendusky 2009). Once adopted, party identification may also lead people to adjust (“convert”) their policy attitudes to be congruent with the party line (Carsey and Layman 2006; Goren 2005; Jacoby 1988; Van Houweling and Sniderman 2005). These dynamics create a situation where an individual’s ideological preferences are likely to be well-aligned with the ideology of his or her preferred party.

Given the strong relationship between ideology and partisan identification, the extensive previous research on the effects of Big Five traits on ideology provides clear guidance about the expected relationship between personality and which party a person affiliates with. The most consistent findings from this work are that Openness to Experience is associated with liberalism and Conscientiousness is associated with conservatism. However, recent work has also found that Extraversion and Emotional Stability are associated with conservatism (Gerber et al. 2010a; Mondak and Halperin 2008). Agreeableness does not appear to be related to self-reported ideology (but see Mondak 2010, who finds a statistically significant relationship between Agreeableness and liberalism in one sample); however, analysis examining the relationships between Big Five traits and the economic and social dimensions of ideology finds that Agreeableness is associated with economic liberalism while simultaneously being associated with social conservatism—a finding that may explain the net null association between this trait and overall ideology (Gerber et al. 2010a). That analysis also finds that the positive associations between both Extraversion and Emotional Stability and economic conservatism are particularly strong relative to the weak associations between those traits and social conservatism. We expect that Big Five traits that are associated with liberalism will also be associated with Democratic identification, whereas those associated with conservatism will be associated with Republican identification.

To our knowledge, the relationship between the Big Five and directional partisan identification has only been examined in two U.S. studies. Using a sample of those called for jury duty in 19 U.S. counties, Mondak and Halperin (2008) find that those scoring higher on Openness are more likely to identify as Democrats, and those scoring higher on Conscientiousness are more likely to identify as Republicans (also see Mondak 2010, in a national sample). In the same sample, they find Emotional Stability is associated with Republican identification. In two separate samples from Tallahassee, Florida they also find a positive relationship between Agreeableness and Democratic identification (also see Mondak 2010, in a national sample).5

Overall, we expect that the associations between Big Five traits and directional partisanship will track closely with the relationships between these traits and ideology identified in prior work. We expect that Openness will be associated with Democratic identification and Conscientiousness with Republican identification. We also expect to find that Extraversion and Emotional Stability are associated with Republican identification. Because the association between Agreeableness and ideology varies across issue domains (i.e., it is associated with economic liberalism and social conservatism), it is possible that the nature of the relationship between this trait and partisan identification depends on whether the parties’ economic or social positions were more salient in the minds of respondents. Although economic concerns loomed large in the 2008 election, suggesting that Agreeableness may be associated with Democratic identification in this context, we offer no specific predictions regarding the relationship between this trait and partisan identification.6

Personality and the Strength of Party Identification

We also examine whether some kinds of people are more likely than others to affiliate with a political party for reasons other than ideological proximity. In addition to allowing one to choose a group that represents one’s policy preferences, party identification, like other group identities, provides individuals two other important benefits: (1) a sense of belonging and (2) a structure for interpreting the world. The first factor is often referred to as a “social” or “affective” benefit of affiliation and the second factor a “cognitive” benefit (see, e.g., Burden and Klofstad 2005; Greene 1999, 2000). We expect Extraversion and Agreeableness to draw people to the social component of partisan identities, and Conscientiousness, Emotional Stability, and Openness to shape the attractiveness of the simplified structure partisanship provides for interpreting the political world.

Previous research suggests that affective considerations shape individuals’ willingness to identify with a party. For example, although independent leaners and weak identifiers tend to be quite similar in their voting behavior and policy attitudes (Keith et al. 1992), Greene (2000) finds that relative to partisan leaners, weak partisan identifiers’ evaluations of their preferred party are more closely tied to affective considerations (e.g., how Democrats make them feel). This difference, with weak identifiers more motivated by affective concerns, suggests a psychological difference between leaners and weak identifiers.

Among the Big Five traits, Extraversion and Agreeableness are most clearly related to these affective concerns. Extraverted individuals tend to be more sociable, gregarious, and assertive. Previous research finds that this trait is associated with increased political participation. Furthermore, this relationship is particularly strong for forms of participation that involve social interactions, such as attending a rally or local meeting (Gerber et al. 2011; Vecchione and Caprara 2009). We therefore expect individuals high on this trait to be drawn to the affective, social benefits of party identification and, thus, to be more likely to identify with a party.

Agreeable individuals are characterized by a prosocial and communal orientation. Outside of the political arena, research finds that Agreeable individuals are more likely to volunteer to help others and engage in solidary behavior (Carlo et al. 2005; Okun et al. 2007). We expect that people high on this trait are therefore also more likely to be attracted to the communal and cooperative components of joining a political “team.” Consequently, we expect that Agreeableness will be positively associated with strong identification with a major political party.

The cognitive appeal of partisan affiliations is also likely to depend on personality dispositions. Prior research suggests that “need for structure” (the appeal of simplified conceptual structures) is associated with the formation and use of stereotypes in decision making (Schaller et al. 1995). Partisan identification is a similar decision-making heuristic because it provides a simplified framework for interpreting political events (see, e.g., Fiorina 1981; Rahn 1993). Certain Big Five traits are related to this need for structure (Neuberg and Newsom 1993): Conscientiousness is positively correlated with need for structure, whereas Emotional Stability and Openness are negatively correlated with this characteristic. We therefore expect to find a pattern of relationships between dispositional traits and affiliating with a party that mirrors the relationships between these traits and need for structure. Specifically, we hypothesize that Conscientiousness will be positively associated with affiliating with a major party and identifying with it strongly, whereas Openness and Emotional Stability will be negatively associated with those outcomes.

To foreshadow what follows in our empirical analysis, the central difficulty with discerning the effect of the Big Five—as factors that shape cognitive and affective concerns—on strength of partisan identification is that the Big Five also affect ideology, which independently shapes partisan attachments for the sorting reasons discussed above. This requires us to try and account for such policy and ideological considerations when trying to discern whether the Big Five traits independently (dis)incline individuals to identify with any party.

Data and Analysis

Our data come from the “baseline” wave of the 2007–2008 Cooperative Campaign Analysis Project (CCAP: Jackman and Vavreck 2009) survey, conducted between December 17, 2007 and January 3, 2008 by YouGov/Polimetrix. The CCAP is an opt-in internet-based survey that uses a combination of sampling and matching techniques to approximate a random digit dialing sample of self-identified registered voters.7 We employ sampling weights to generate estimates that are nationally representative.

The survey included the Ten-Item Personality Inventory (TIPI) developed by Gosling et al. (2003). The TIPI asks respondents to report whether “I see myself as” characterized by a series of 10 trait pairs using a seven-point scale ranging from Disagree Strongly to Agree Strongly. Each Big Five trait is captured by responses to two trait pairs. Responses to these 10 questions are used to score a respondent’s personality on each of the Big Five traits. Gosling et al. compared the performance of the 10 question battery to longer tests and found that scores obtained from the TIPI are highly correlated with those obtained from longer instruments (2003, Tables 6 and 9) and show good test–retest and self-other (self-rating vs. ratings provided by a peer) reliabilities.8

The CCAP also included measures of ideological self-placement, several questions about respondents’ policy preferences, a variety of demographic measures (including state of residence), and a standard branching party identification question. We restrict our analysis to respondents for whom all measures were available and, thus, could be included in all of the analysis. Even with these restrictions, our sample is quite large (over 12,000 individuals). We note that large samples tend to reduce the size of standard errors. This allows us to identify weaker statistically significant relationships than would be possible in a smaller survey. We address this in our analysis by adopting a conservative approach to our hypothesis testing. Although we have clear expectations regarding the direction of the relationships between Big Five traits and our outcomes of interest, we use two-tailed hypothesis tests. Additionally, we cluster our standard errors at the state level and use sample weights. Both of these techniques tend to inflate the size of standard errors. Summary statistics for the variables used in our analysis are presented in Table 3. Full question wording and coding rules are presented in the Appendix.
Table 3

Summary statistics for model variables

Variable

(1)

Extraversion (0–1)

0.517

[0.2476]

Agreeableness (0–1)

0.707

[0.1935]

Conscientiousness (0–1)

0.766

[0.2008]

Emotional stability (0–1)

0.681

[0.2251]

Openness (0–1)

0.702

[0.1961]

7-Point party ID (−3 = strong Republican; +3 = strong Democrat)

0.170

[2.2918]

Self-reported ideology (−2 = very cons.; 2 = very lib., 0 = not sure)

−0.155

[1.1789]

4-Point strength of party ID (0 = “true” independents; 1 = leaning partisans; 2 = weak partisans; 3 = strong partisans)

2.038

[1.0624]

Affiliated w/major party? (1 = yes) [strong or weak]

0.688

[0.4632]

3-Point strength of party ID (0–2) [leaners/weak collapsed]

1.349

[0.6796]

3-Point strength of party ID (0–2) [ind/leaners collapsed]

1.245

[0.7928]

3-Point strength of ideology (0 = moderate; 1 = liberal/conservative; 2 = very liberal/conservative)

0.908

[0.7679]

Female = 1

0.495

[0.5]

Black = 1

0.115

[0.3195]

Hispanic = 1

0.078

[0.2685]

Other (native American, Asian, mixed, other) = 1

0.043

[0.2031]

Age (years)

47.666

[15.7519]

Income (0–1; 1 = refused)

0.542

[0.2749]

Income refused = 1

0.085

[0.2784]

Education (1 = no high school; 2 = high school graduate; 3 = some college; 4 = two year degree; 5 = college graduate; 6 = post-grad)

3.417

[1.5271]

Observations

12,407

Note Cell entries are weighted means. Standard deviations in brackets. See the Appendix for complete question wording and coding details

Source 2008 CCAP

Big Five Traits and Directional Party Identification

We begin by briefly assessing the relationships between dispositional traits and directional party identification. In this and each subsequent case, we present estimates using this general equation:
$$ {\text{DV}} = {\text{B}}0 + {\text{C}}*{\text{Personality}} + {\text{D}}*{\text{Controls}} + {\text{F}}*{\text{State}}\;{\text{Fixed}}\;{\text{Effects}} + {\text{e}}, $$
(1)
where Personality is a vector of Big Five traits, State Fixed Effects is a vector of state of residence indicators, and standard errors are clustered at the state level to allow for interdependence of observations in a given geographic area. The inclusion of state fixed effects ensures that our results are not generated by some correlation between personality and other factors that might affect the expression of partisan sentiments in a state [e.g., state political culture (Elazar 1984) or legal rules affecting party registration (Burden and Greene 2000; Norrander 1989)].9Controls always include gender, race (indicators for Black, Hispanic, Other non-White), age (and age-squared to allow for non-linearity in the effects of age), income (measured as a series of indicators with a separate indicator for income refused), and educational attainment (measured as a series of indicators for response categories). Because we have a large sample, we employ the flexible approach of including category indicator variables rather than using scales that would require us to assume linear effects are correct.10

As we explain above, prior research regarding the link between ideology and party identification, as well as the fact that ideology and party identification are highly correlated (r = 0.699 in our sample), suggests that the relationships between dispositional traits and directional party identification will track the relationships between these traits and ideology. As a reminder, these predictions are summarized in Table 2.

Column (1) of Table 4 presents an ordered logit model where the dependent variable is 7-Point Party Identification, scaled so that Democratic affiliation is the positive direction.11 In column (2) we present an ordered logit model where 5-Point Ideology, scaled so that positive values are more liberal, is the dependent variable. Our expectation was that the coefficients on the Big Five traits in columns (1) and (2) would be similar, which is what we find. The relationships between the Big Five measures and party identification in column (1) are similar in sign and significance to those found in the ideology model in column (2). Specifically, individuals higher on Conscientiousness and Emotional Stability tend to report being more Republican (conservative), whereas respondents higher on Openness report being more Democratic (liberal). The coefficient on Extraversion is negative in both models, but falls short of conventional levels of statistical significance in the party identification model (p = 0.079; all p-values are two-tailed). We also note that the relative magnitude of the coefficients on the Big Five traits follows a similar pattern in each model: Openness, Conscientiousness, and Emotional Stability are all strong predictors of these outcomes, whereas Extraversion is a much weaker predictor. Finally, the relationships between Agreeableness and both Democratic identification and liberalism are modest in size and statistically indistinguishable from zero at conventional levels.
Table 4

The effect of personality on the direction of partisan identification

 

7-Point party ID (−3 = SR; +3 = SD)

Self-reported ideology (−2 = very cons.; 2 = very lib., 0 = not sure)

7-Point party ID (−3 = SR; +3 = SD)

(1)

(2)

(3)

Extraversion (0–1)

−0.146

[0.083]

−0.232

[0.101]*

0.017

[0.062]

Agreeableness (0–1)

0.162

[0.117]

0.079

[0.164]

0.165

[0.145]

Conscientiousness (0–1)

−0.759

[0.137]**

−1.086

[0.124]**

−0.106

[0.111]

Emotional stability (0–1)

−0.885

[0.112]**

−0.754

[0.101]**

−0.478

[0.137]**

Openness (0–1)

1.341

[0.110]**

2.152

[0.096]**

−0.070

[0.101]

Female = 1

0.341

[0.035]**

0.413

[0.039]**

0.066

[0.043]

Black = 1

1.526

[0.114]**

0.573

[0.072]**

1.412

[0.132]**

Hispanic = 1

0.704

[0.142]**

0.417

[0.139]**

0.558

[0.125]**

Other (native American, Asian, mixed, other) = 1

0.235

[0.095]*

0.130

[0.100]

0.222

[0.133]

Age (years)

−0.006

[0.008]

−0.033

[0.009]**

0.012

[0.009]

Age-squared/100

0.003

[0.008]

0.022

[0.009]*

−0.005

[0.009]

Ideology indicators: very conservative, conservative, liberal, and very liberal, with moderate as the omitted category?

No

No

Yes

Education and income indicators?

Yes

Yes

Yes

State fixed effects?

Yes

Yes

Yes

Observations

12,407

12,407

12,407

F-test: joint significance of Big Five

0.000

0.000

0.002

Note Ordered logit coefficients. Indicators for education and income categories, state fixed effects, and cutpoint thresholds not reported to save space. Robust standard errors (clustered by state) in brackets. See the Appendix for complete question wording and coding details

Source 2008 CCAP

* Significant at 5%; ** significant at 1%, two-tailed tests

In column (3) we examine whether ideology mediates the relationships between Big Five traits and directional party identification. Specifically, we re-estimate the model presented in column (1) with controls for ideology (specified as a set of indicators for each response to the ideology question).12 We find that controlling for ideological self-placement dramatically reduces the magnitude of the effects of the Big Five on directional party identification—only the coefficient for Emotional Stability (at approximately half its original size) is statistically significant. In additional analysis (available upon request), we conduct more formal Sobel tests to assess whether a linear measure of ideology significantly mediates the relationships between Big Five traits and directional partisanship. (More specifically, we treat directional party identification as the outcome, Big Five traits as independent variables, and directional ideology as the mediator.) Consistent with the analysis presented in column (3) of Table 4, we find clear evidence that accounting for the mediating effect of ideology in this fashion reduces the magnitude of the relationships between Extraversion, Conscientiousness, Emotional Stability, and Openness and directional partisanship (differences in magnitudes when not controlling for ideology and when controlling for ideology are significant at p < 0.05).

Big Five Traits and Strength of Partisan Identification

We next consider the relationship between personality and the decision to adopt any partisan identity. Table 5 presents the results of a series of models employing Eq. 1 for four measures of partisan affiliation. We begin by examining the relationships between personality traits and the respondent’s party affiliation measured using a four-point strength of partisanship scale, which is coded 0 for “true” independents, 1 for “leaning” partisans, 2 for “weak” partisans, and 3 for “strong” partisans. The advantage of this measure is that it allows a full range of expressions of partisan affinity.
Table 5

The effect of personality on the strength of partisan identification

 

4-Point strength of party ID (0–3)

Affiliated w/major party? (1 = yes)

3-Point strength of party ID (0–2) [leaners/weak collapsed]

3-Point strength of party ID (0–2) [ind/leaners collapsed]

(1)

(2)

(3)

(4)

(5)

(6)

(7)

Extraversion (0–1)

0.531

[0.087]**

0.513

[0.093]**

0.490

[0.087]**

0.500

[0.092]**

0.521

[0.121]**

0.479

[0.087]**

0.551

[0.089]**

Agreeableness (0–1)

0.479

[0.116]**

0.503

[0.130]**

0.500

[0.110]**

0.496

[0.131]**

0.594

[0.126]**

0.492

[0.148]**

0.471

[0.130]**

Conscientiousness (0–1)

0.327

[0.133]*

0.268

[0.142]

0.249

[0.135]

0.241

[0.140]

0.234

[0.144]

0.221

[0.149]

0.333

[0.140]*

Emotional stability (0–1)

−0.026

[0.173]

−0.165

[0.176]

−0.131

[0.178]

−0.153

[0.179]

−0.321

[0.174]

−0.085

[0.192]

−0.123

[0.172]

Openness (0–1)

−0.649

[0.121]**

−0.586

[0.125]**

−0.589

[0.121]**

−0.584

[0.123]**

−0.847

[0.174]**

−0.486

[0.123]**

−0.658

[0.117]**

Female = 1

0.229

[0.055]**

0.277

[0.061]**

0.301

[0.060]**

0.292

[0.063]**

0.454

[0.075]**

0.220

[0.063]**

0.288

[0.067]**

Black = 1

0.401

[0.091]**

0.659

[0.087]**

0.469

[0.093]**

0.641

[0.090]**

0.660

[0.097]**

0.660

[0.097]**

0.634

[0.090]**

Hispanic = 1

0.052

[0.087]

0.178

[0.125]

0.110

[0.090]

0.183

[0.130]

0.056

[0.156]

0.261

[0.118]*

0.152

[0.150]

Other (native American, Asian, mixed, other) = 1

−0.404

[0.096]**

−0.439

[0.107]**

−0.427

[0.107]**

−0.455

[0.110]**

−0.442

[0.127]**

−0.457

[0.118]**

−0.412

[0.112]**

Age (years)

−0.018

[0.007]**

−0.010

[0.007]

−0.020

[0.007]**

−0.013

[0.008]

−0.022

[0.009]*

−0.012

[0.008]

−0.006

[0.009]

Age-squared/100

0.022

[0.007]**

0.014

[0.007]

0.023

[0.007]**

0.017

[0.008]*

0.019

[0.009]*

0.020

[0.008]*

0.010

[0.009]

Ideology indicators: very conservative, conservative, liberal, and very liberal, with moderate as the omitted category?

No

Yes

No

Yes

Yes

Yes

Yes

Eleven policy opinion indicators: abortion (3 indicators); civil unions (3 indicators); health care (2 indicators); taxing the rich (3 indicators)?

No

No

Yes

Yes

Yes

Yes

Yes

Education and income indicators?

Yes

Yes

Yes

Yes

Yes

Yes

Yes

State fixed effects?

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Observations

12,407

12,407

12,407

12,407

12,407

12,407

12,407

F-test: Big Five

0.000

0.000

0.000

0.000

0.000

0.000

0.000

Note Ordered logit coefficients, except for logit coefficients in column (5). Indicators for education and income categories, state fixed effects, and constant/cutpoint thresholds not reported to save space. Robust standard errors (clustered by state) in brackets. See the Appendix for complete question wording and coding details

Source 2008 CCAP

* Significant at 5%; ** significant at 1%, two-tailed tests

In column (1) we present a specification that controls for only the same demographic characteristics included in the Table 4, column (1) analysis. The results provide clear support for our hypotheses. We posited that Extraversion and Agreeableness would encourage affiliating for affective reasons and find positive and statistically significant relationships between each of these traits and strength of partisan identification. We also predicted that the remaining three Big Five traits, which are associated with the need for structure, would be associated with variation in the attractiveness of holding a partisan identity because of its usefulness in interpreting political stimuli. Consistent with our expectations we find a positive association between Conscientiousness and strength of partisan identification and a negative association between Openness and this outcome. We do not find support for our expectation that those higher on Emotional Stability would be negatively associated with strength of partisanship.

One concern with this specification is that because personality is associated with issue positions, this model may attribute effects to the Big Five factors that are only operating through their effects on ideology and issue attitudes. In other words, the relationships between Big Five traits and strength of partisanship may be mediated by the strength of political attitudes.13 Therefore, in column (2) we add controls for ideology, and in column (3) we include measures of issue positions on four salient policies during the 2008 campaign (abortion, civil unions, health care, and taxes). Finally, in column (4) we simultaneously include both ideology and issue positions. These models provide a robust way to examine whether political attitudes mediate the relationships between Big Five traits and strength of partisanship.

Two findings emerge across the first four models shown in Table 5.14 First, personality traits have a statistically significant effect on affiliation; three traits (Extraversion, Agreeableness, and Openness) are statistically significant (p < 0.01) in the predicted direction across all four models. The relationship between Conscientiousness and strength of party identification is in the predicted direction, but falls short of conventional levels of statistical significance in the specifications presented in columns (2–4; p = 0.059, 0.065, and 0.086, respectively). Second, with the exception of the coefficient on Conscientiousness, the addition of controls for ideology and issue positions does not substantially affect the results. This suggests that the relationships between Big Five traits and strength of partisanship are not mediated by a relationship between these traits and a tendency to adopt strong attitudes.15 Given the small differences across the columns and the fact that the model that includes controls for ideology and issue positions is probably more conservative, we focus on the results from column (4) in the discussion that follows.

To make interpreting the magnitude of these effects more transparent, Table 6 displays the marginal effects for a modal respondent (a White male living in California who is 48 years old, earns about $50,000, and has some college education) of a two-standard deviation shift in each Big Five trait, income [which is a significant predictor in the column (4) specification], and gender [which is a significant predictor in the column (4) specification and previous work finds is associated with willingness to adopt a partisan identity (e.g., Norrander 1997)]. Policy attitudes are set to their medians, which are the excluded categories, and ideology is set to moderate. The baseline probability of strongly identifying with a party for these individuals is 23.4%; the baseline probability of affiliating with a major party is 49.0%.
Table 6

Marginal effects of personality on the strength of partisan identification

 

(1)

(2)

Measure

4-Point strength of party ID (0–3)

Affiliated w/major party? (1 = yes)

Model estimates appear in

Table 5, column (4)

Table 5, column (5)

Outcome of interest

3 = Strong identifier

1 = Yes

Baseline predicted probability of outcome

23.4%

49.0%

Marginal effect of two standard deviation shift in variable

 Extraversion

4.4%**

6.4%**

 Agreeableness

3.4%**

5.7%**

 Conscientiousness

1.7%

2.3%

 Emotional stability

−1.2%

−3.6%

 Openness

−4.1%**

−8.3%**

 Income

5.7%*

12.0%**

 Gender

4.8%**

11.3%**

Note Estimated effects are for White male age 48 living in California with some college education and an income of about $50,000. All specifications include policy attitudes (set to medians, which are excluded categories) and ideology (set to moderate). Reported effects are absolute percentage point increases/decreases in the probability of the outcome of interest (relative to all other outcomes), not proportional increases/decreases relative to baseline probabilities

Source 2008 CCAP

* Significant at 5%; ** significant at 1%, two-tailed tests

For the four-point strength of partisan identification measure, we find support for our expectation that the Big Five traits likely to be associated with the affective appeal of group attachments—Extraversion and Agreeableness—would be associated with the decision to affiliate with a political party. A two standard deviation increase in Extraversion increases the probability of being a strong identifier (rather than an independent, leaner, or weak identifier) by 4.4 percentage points [see column (1) of Table 6; a proportional increase of 19% over the baseline probability], whereas a similar increase in Agreeableness increases that probability by 3.4 percentage points (a proportional increase of 14.7%). By comparison, an approximately two standard deviation increase in income (from $27,000 to $90,000 per year) is associated with a 5.7 percentage point increase (a proportional increase of 24.4%) in the likelihood of strongly identifying with a party, while being female (rather than male) is associated with a 4.8 percentage point increase in this probability (a proportional increase of 20.7%).

Our results also provide some support for our expectations about the relationships between partisanship and the traits that are likely to affect the cognitive appeal of the simplifying structure party identification provides. Those who are higher on Openness—a trait associated with low need for structure—are 4.1 percentage points less likely to strongly affiliate (a proportional decrease of 17.5% from the baseline probability). We also posited that those higher on Conscientiousness—a trait associated with high need for structure—would find partisan affiliations more attractive. As noted above, the coefficient on this trait is positive, but falls short of conventional levels of statistical significance in the column (4) specification. Estimating the magnitude of this insignificant relationship, we find that a two standard deviation increase in Conscientiousness is associated with being 1.7 percentage points more likely to be a strong identifier.

In columns (5)–(7) of Table 5 we examine the robustness of these findings to different measures of strength of partisan identification. For each measure we repeat the column (4) specification that controls for ideological self-placement and issue attitudes. In column (5) the dependent variable is a binary measure of partisan identification, which is coded 1 if the respondent identified as a Republican or Democrat in the stem question of the party identification measure and 0 if the respondent said they were an Independent, Other, or Not Sure (partisan leaners are therefore coded as 0). [The marginal effects associated with this model are reported in column (2) of Table 6 and are somewhat larger in absolute terms than the effects on being a strong identifier.] In column (6) the dependent variable is a three-point measure of strength of partisan identification, which collapses weak and leaning partisans. Given that weak and leaning partisans are often quite similar in their ideological positions and voting behavior, this analysis ensures our results are predictors of meaningful differences in strength of partisan identification. Similarly, in column (7) we present a different three-point measure of strength of partisan identification, this time pooling leaning partisans and pure independents, on the grounds that neither is willing to identify with a major party. We find little evidence that the choice of measures is consequential.

Assessments of whether a given relationship is substantively important are inherently subjective. However, we believe the effects we identify are important for three reasons. First, our analysis relies on rudimentary measures of Big Five traits that ask respondents to rate how well trait pairs that, on their face, appear to be entirely unrelated to politics (e.g., Agreeableness: “critical, quarrelsome” and “sympathetic, warm”) describe them. The fact that we find any relationships between Big Five traits and strength of partisanship is notable. Thus, we argue that the substantive importance of the analysis we present stems from the fact that we find that psychological traits that are seemingly unrelated to politics significantly predict partisan attachments. Second, this pattern persists both after accounting for the possibility that the relationships between Big Five traits and strength of party identification are mediated by a relationship between personality and strength of political attitudes and in spite of the fact that our personality measures are likely measured with error because we measure these traits with a brief battery of questions. Finally, the magnitude of these effects is comparable to those associated with other individual-level characteristics that predict partisan attachment.

To summarize, the analysis in Tables 5 and 6 provides evidence that personality traits shape both the decision to affiliate with a major political party and the strength of that affiliation. The results of this analysis are consistent with our expectation that the affective, social benefit of partisan affiliation would be greater for individuals higher on Extraversion and Agreeableness. Additionally, they support our expectation that the simplified structure for understanding the political world that partisan attachments provide are particularly appealing to individuals low on Openness and Emotional Stability and high on Conscientiousness (although the coefficients on Conscientiousness and Emotional Stability are not statistically significant at conventional levels across all models). Furthermore, as predicted, these relationships persist in models that account for the policy issues and ideological orientations that might incline one to hold more partisan views. These models therefore account for the possibility that certain personality traits may incline individuals toward ideological and policy positions that would independently push them toward either of the major parties. Thus, the relationships between traits and strength of party affiliation appear to reflect differences in the cognitive and affective appeal of affiliating with a political party.

Extensions/Future Research

The findings presented here suggest several avenues for future research. For example, although most of the relationships between the Big Five and directional party identification we find (when not controlling for ideology) are consistent with work using European samples (that use vote choice or vote intention as proxies for directional party identification, see Barbaranelli et al. 2007; Caprara et al. 1999; Schoen and Schumann 2007), we are not aware of any published work that examines the relationships between the Big Five and strength of partisan identification outside of the U.S. Assessing these relationships in a different political context with a larger number of viable parties could yield a broader understanding of how and why dispositional traits affect decisions about whether to identify with any political party.

Future research could also consider whether Big Five traits moderate responsiveness to party cues. We find that certain traits are associated with a proclivity to adopt a partisan identity. We posit that some of these relationships are driven by variation in the attractiveness of the social benefits of party affiliation, whereas others are a product of differences in the desirability of the simplifying structure party labels provide. One possibility is that traits that make the simplifying structure party identification provides attractive (e.g., low Openness) are associated with a strong reliance on, and responsiveness to, partisan cues. By contrast, individuals whose affiliation is a product of traits likely to make the social aspects of partisanship appealing (high Extraversion and Agreeableness) may be less responsive to party cues. Along these lines, a great deal of psychological research finds that a key consequence of group identities is that people tend to evaluate in-group members particularly favorably and out-group members particularly unfavorably (e.g., Brewer 1979; Brown et al. 1980; Tajfel et al. 1971). Examining how personality traits alter the extent to which partisans see their own party as superior to the opposing party is likely to be a fruitful line of research.

Another extension to the present line of inquiry is to bring together research on personality and politics with the growing body of research on the relationships between genetics and political attitudes and behaviors (e.g., Alford et al. 2005; Fowler et al. 2008; Hatemi et al. 2009; Settle et al. 2009). As discussed above, Big Five traits appear to be strongly heritable. Thus, our findings regarding the relationships between Big Five traits and strength of partisanship raise the question of whether these traits can help to account for the apparent association between genetics and the strength of an individual’s partisan identification (Hatemi et al. 2009; Settle et al. 2009).

Finally, we are able to present preliminary evidence regarding the relationships between Big Five traits and two additional outcomes of interest: (1) affiliation with third parties or political movements and (2) stability of partisan identification. We examine the relationships between Big Five traits and favorability ratings of the “Tea Party” movement16 using data from the 2010 Cooperative Congressional Election Study (CCES).17 This analysis is reported in Appendix Table 10. First, we re-estimate the model presented in column (1) of Table 4, specifying Tea Party favorability (−2 = very unfavorable; 2 = very favorable) as the dependent variable. Consistent with the notion that the Tea Party is broadly viewed as a conservative movement, we find positive associations between Extraversion, Conscientiousness, and Emotional Stability and favorability toward the Tea Party (p < 0.01) and negative associations between both Agreeableness and Openness and favorability toward the Tea Party (p = 0.089 and p < 0.01, respectively). When we control for respondent’s ideology [Table 10, column (2)], these relationships are drastically attenuated and none reach conventional levels of statistical significance (p-values range from 0.111 to 0.957). This suggests that, at least in the case of the Tea Party, the relationships between Big Five traits and support for this type of group depend on the ideological orientation of the movement. A notable limitation of this analysis is that the outcome we use is a favorability rating, rather than a measure of identification with that group. Future research should examine the relationships between Big Five traits and willingness to identify with third-parties or other non-mainstream political groups.

The CCAP data used for our primary analysis also allow us to examine whether Big Five traits are associated with the stability of partisan identification. Appendix Table 11 presents the results of model specifications that mirror those presented in columns (1)–(4) of Table 5 for a variety of measures of partisan stability. Specifically, we use three different measures: (1) the absolute amount of change in the 7-point party identification measure from the baseline to the post-election wave of the survey (0 = No change; 6 = Moved 6 points); (2) an indicator for whether there was any change between the baseline and post-election waves (0 = No change; 1 = Change); and (3) the standard deviation of reported party identification for respondents who completed all five waves of the CCAP—baseline, March, September, October, and post-election. For all specifications we restrict our analysis to the 8132 participants who completed all five waves of the CCAP.18 We do not find any evidence that Big Five traits significantly predict any of these measures of stability in partisanship. We note that this analysis does not rule out the possibility that the Big Five (or other personality traits) predict stability of partisan attachments over a longer time frame or over the life course (e.g., during important periods of socialization), a possibility future multi-year panel designs could explore.

Discussion

Because ideology and the direction of partisan identification are closely connected—particularly in a two-party system—we expected that the relationships between Big Five traits and the direction of peoples’ party identification (whether they identify with the Democratic or Republican Party) would be similar to the relationships found in previous research between these traits and ideology. Our analysis supported this expectation. Our data do not allow us to directly address longstanding questions about the direction of the causal relationship between ideology and partisan identity (but see Gerber et al. 2010b). However, the findings we present do suggest that the correlation between ideological and partisan leanings may stem from similar core psychological antecedents.

At the same time, our findings provide clear support for the novel claim that the appeal of partisan affiliation varies across individuals with different personality traits. These psychological characteristics play an important role in determining whether people are willing to adopt a self-concept that involves allegiance to a political party. The results presented in Table 5, which are generally consistent with our expectations, suggest that the affective social benefits of affiliating with a dominant social political group are particularly attractive to people high on Extraversion and Agreeableness. Our evidence also supports the notion that parties provide people with a simplified structure for understanding the complex political arena that appeals to those high on traits associated with need for structure (Conscientiousness) but is less appealing to those high on traits that are negatively related to need for structure (Openness). The negative relationship between Openness, a trait associated with eagerness to entertain new ideas, and strength of partisanship is particularly strong. Although individuals high on this trait tend to support liberal policies, they may be resistant to, or even repelled by, the notion that political engagement involves an uncomplicated decision to side with one of two parties.

Some caveats about the analysis presented here are in order. First, although the brief personality battery we use is appealing for use in political research, more extensive and robust measures of personality would be preferable. Of particular interest would be a measure that allowed us to distinguish between the six underlying components (or “facets”) of each Big Five trait. For example, the facets of Extraversion identified by personality psychologists are: Warmth, Gregariousness, Assertiveness, Activity, Excitement-Seeking, and Positive Emotions. It could be that people high on Warmth and Gregariousness are drawn to affiliate with a major party whereas those high on Assertiveness and Excitement Seeking gravitate to third parties. Unfortunately, batteries capable of measuring these facets are extremely long (e.g., the NEO-PI-R includes 240 items), making them difficult to use on a large scale. However, it is important to acknowledge that our analysis may miss such nuanced relationships between personality and partisanship. We also note that because our analysis is from a single cross-sectional sample, we may miss variation in the relationships between personality traits and partisanship that may occur over time due to factors such as fluctuations in the relative popularity of the parties or changes in public perceptions about the role parties play in the political arena (e.g., increased frustration with partisan gridlock).

The research presented here is part of a revival of interest in how individual-level differences affect political attitudes and behaviors. This line of inquiry is particularly compelling because it provides a way to identify how essential differences in “what people are like” shape their political lives. Our findings demonstrate that core personality traits not only affect which policy platforms people find most appealing, but also their inclination to adopt a partisan self-concept. One implication of our findings is that they suggest a different dimension by which partisan affinities are transmitted beyond pre-adult socialization and early adult experiences (e.g., Alwin and Krosnick 1991; Campbell et al. 1960; Green et al. 2002; Jennings and Niemi 1974; Niemi and Jennings 1991). More broadly, our findings provide further evidence that individual-level differences in personality affect political attitudes and behaviors, suggesting the value of integrating models of political behavior with models of human decision-making.

Footnotes
1

Scholars have considered some important psychological sources of partisan affiliation. For example, a relative lack of a “partisan social identity” compared to an “independent social identity” helps explain why partisan “leaners” are different from “true” partisans (Greene 2000). However, the present research is unique in that it focuses on dispositional traits, which develop independently of political environments.

 
2

Mondak (2010) and Mondak and Halperin (2008) also find evidence that this is the case.

 
3

For example, Gerber et al. (2010a) find that the relationships between Big Five traits and political attitudes vary across racial groups. Similarly, Mondak et al. (2010) find that the relationships between Big Five traits and political participation depend on characteristics of the political environment. Each of these studies suggests that the attitudinal and behavioral consequences of dispositional traits depend on contextual factors.

 
4

Elections in the United States are dominated by the Democratic and Republican parties. In the 2008 U.S. elections, 97.3% of votes cast in House races went to candidates from these parties, and a similar proportion went to candidates in gubernatorial (97.6%), Senate (96.6%), and presidential (98.5%) elections. From 1961 to 2008, politicians from parties other than the two major parties have only served 12 out of the 10,859 representative/terms in the House of Representatives and 5 of approximately 828 senator/terms in the Senate.

 
5

Other studies in the U.S. and Europe have examined the associations between Big Five traits and vote choice or vote intention (Barbaranelli et al. 2007; Caprara et al. 1999; Schoen and Schumann 2007) or statewide vote returns (Rentfrow et al. 2009) and find similar relationships. In a German sample, for example, Schoen and Schumann find that those high on Openness and Emotional Stability tend to vote for liberal parties, whereas those high on Conscientiousness and Agreeableness tend to vote for conservative parties. We note that Schoen and Schumann’s finding that Emotional Stability is associated with affiliating with a liberal party conflicts with Mondak and Halperin’s (2008) finding of the inverse relationship.

 
6

We have also investigated the direct effects of personality on directional party identification controlling for ideology. We find that once ideology is controlled for, the effects of Big Five personality traits on which party one chooses to identify with are greatly attenuated. See below for further details.

 
7

The survey sample is constructed by first drawing a target population sample. This sample is based on the 2005–2007 American Community Study (ACS), November 2008 Current Population Survey Supplement, and the 2007 Pew Religious Life Survey. Thus, this target sample is representative of the general population on a broad range of characteristics including a variety of geographic (state, region, metropolitan statistical area), demographic (age, race, income, education, gender), and other measures (born-again status, employment, interest in news, party identification, ideology, and turnout). Polimetrix invited a sample of their opt-in panel of 1.4 million survey respondents to participate in the study. Invitations were stratified based on race, gender, and battleground status, with an oversample of nine battleground and early primary states (Florida, Iowa, Minnesota, New Hampshire, New Mexico, Nevada, Ohio, Pennsylvania, and Wisconsin). Those who completed the survey (approximately 2.5 times the target sample) were then matched to the target sample using nearest-neighbor matching based on the variables listed in parentheses above. Finally, weights were calculated to adjust the final sample to reflect the national public on these demographic and other characteristics (including correcting for the oversampling of battleground states). For more detailed information on this type of survey and sampling technique see Vavreck and Rivers (2008). In concrete terms, the weighted CCAP sample we use in our analysis appears similar in levels of political interest to that found in the weighted 2008 American National Election Study (ANES) time-series survey. In the September wave of the CCAP we find that 55% of respondents are “very much” interested in politics (variable = scap813, “How interested are you in politics?”). In the ANES pre-election survey, the comparable figure is 58% (variable = V0830001b, “How interested are you in information about what’s going on in government and politics?” = Extremely or very interested, restricted to reported registered voters).

 
8
Trait pairs for each trait. Observed correlations in brackets; (R) indicates reverse scoring:
  • Extraversion: Extraverted, enthusiastic; Reserved, quiet (R) [r = 0.432]

  • Agreeableness: Sympathetic, warm; Critical, quarrelsome (R) [r = 0.221]

  • Conscientiousness: Dependable, self-disciplined; Disorganized, careless (R) [r = 0.395]

  • Emotional Stability: Calm, emotionally stable; Anxious, easily upset (R) [r = 0.467]

  • Openness: Open to new experiences, complex; Conventional, uncreative (R) [r = 0.267].

We note that the TIPI was not designed with the intent of achieving high inter-item correlations. Rather, it was designed to (1) be brief; (2) achieve high test–retest reliability (as well as reliability between self- and peer-administered ratings); and (3) yield measures that are highly correlated with those obtained using much longer batteries (the correlations between TIPI measures and the 44-item Big Five Inventory range from 0.65 to 0.87; correlations with measures from the much longer, 240-item NEO PI-R range from 0.56 to 0.68). Therefore, because each question in the TIPI is designed to measure part of a broader Big Five trait, inter-item correlations between the two items used to measure each trait are less informative of the items’ reliability (Gosling 2009; more generally, see Kline 2000; Woods and Hampson 2005 on the misleading nature of alphas calculated on scales with only a small number of items). Table 7 in the Appendix reports sample correlations between the Big Five traits.

 
9

The inclusion of state fixed effects is not material to the results we present in Tables 4, 5, and 6. That is, if state fixed effects are excluded, the size and statistical significance of the personality coefficients are largely unchanged. Results are available upon request.

 
10

The inclusion in our analysis of income and education, which unlike gender, age, and race are not immutable characteristics, deserves special attention. We noted above research that finds that personality traits predict these outcomes. However, we believe that including them in the reported analysis is a conservative strategy for two reasons. First, it demonstrates that the effects of personality on the different outcomes we study are not due to the indirect effect of personality on earnings and educational attainment. Second, in practical terms, including income and education in the estimated models tends to yield more conservative estimates of the effect of personality on our outcomes of interest. (Parallel analysis excluding these measures appears in Appendix Tables 8 and 9.) We suppress the education and income indicators from the estimates presented in the body of the text for space reasons. Full results are available upon request.

 
11

We have also repeated the analysis in column (1) using a 5-Point Party Identification measure that pools weak and leaning partisans. The results (available upon request) are qualitatively similar to those reported in column (1). This demonstrates that the pattern of results we observe in column (1) is not driven by differences between weak and leaning partisans, whose voting behavior is often quite similar.

 
12

We note that this analysis rests on two assumptions. First, we assume that Big Five traits affect directional partisanship through ideology rather than affecting ideology through partisanship. Second, we assume that it is ideology that mediates the relationship between Big Five traits and partisanship rather than an omitted variable that is correlated with ideology, partisanship, and personality.

 
13

We conducted additional analysis (available upon request), replacing strength of partisanship with strength of ideology as the dependent variable in the model specification presented in Table 5, column (1). In contrast to the similar relationships between Big Five traits and both ideology and directional partisanship we report in Table 4, we find that the relationships between Big Five traits and strength of ideology are quite different from the relationships between these traits and strength of partisanship. Conscientiousness is positively associated with both of these outcomes; however, the only other statistically significant relationship we find is a positive association between Emotional Stability and strength of ideology, which is in contrast to the negative, but statistically insignificant association between this trait and strength of partisanship.

 
14

We note that the “independence gap” in which women are more likely to (strongly) affiliate with a party than men (see Norrander 1997; Burden 2008) persists in our models in Table 5 that include personality traits.

 
15

In more formal tests (available upon request), we conducted Sobel tests to assess whether a linear measure of strength of ideology significantly mediates the relationships between Big Five traits and strength of partisanship. As the findings in Table 5 suggest, including a measure of strength of ideology significantly reduces the magnitude of the coefficient on Conscientiousness. We also find evidence that strength of ideology affects the (statistically insignificant) relationship between Emotional Stability and strength of partisanship. However, in this case the pattern of relationships suggests that excluding strength of ideology from the model suppresses the relationship between these measures (i.e., including the strength of ideology measure strengthens the negative relationship between Emotional Stability and strength of partisanship).

 
16

The Tea Party movement emerged following the 2008 general election and is broadly viewed to be ideologically conservative.

 
17

The CCES is also administered over the Internet by YouGov/Polimetrix. On the post-election CCES participants were asked: “What is your view of the Tea Party movement—would you say it is very positive, somewhat positive, neutral, somewhat negative, or very negative, or don’t you know enough about the Tea Party movement to say?” (Very Positive; Somewhat Positive; Neutral; Somewhat negative; Very Negative; Don’t know enough to say; No opinion). The analysis reported in Table 10 excludes responses of “don’t know enough to say” and “no opinion”.

 
18

The analysis is further restricted by the loss of 31 cases from Washington, DC and Hawaii because no respondents from those areas changed their responses to the party identification question.

 

Acknowledgments

We thank Barry Burden, Casey Klofstad, Matthew Levendusky, Paul Goren, the anonymous reviewers, and the editors for comments on earlier versions. This research was funded by Yale’s Center for the Study of American Politics and Institution for Social and Policy Studies.

Copyright information

© Springer Science+Business Media, LLC 2011