Personality Traits and Participation in Political Processes

Personality Traits and Participation in Political Processes Alan S. Gerber Yale University Gregory A. Huber Yale University David Doherty Loyola University Chicago Conor M. Dowling Yale University Connor Raso Stanford University Shang E. Ha Brooklyn College-CUNY Using data from two recent surveys, we analyze the relationship between Big Five personality traits and political participation. We examine forms of participation that differ in domain (local politics vs. national campaigns) as well as in the amount of conflict involved, whether they are likely to yield instrumental benefits, and whether they are likely to be viewed as a duty—characteristics that may affect the relationships between dispositional personality traits and political activity. We find relationships between personality traits and: (1) both self-reported and actual turnout (measured using administrative records), (2) overreporting of turnout, and (3) a variety of other modes of participation. The effect of personality on political participation is often comparable to the effects of factors that are central in earlier models of turnout, such as education and income. Consistent with our theoretical expectations, these relationships vary depending on personality-relevant characteristics of each participatory act. I t is evident to almost everyone that there are different types of personalities. At the most basic level, people commonly ask: ‘‘what sort of person is she?’’ and find the answer illuminating. In other words, people intuitively understand what psychologists have We note that there are many dimensions along which one could identify personality differences, and prior scholarship has considered the role of, among others, personality as measured using Right-wing Authoritari- anism (e.g., Hetherington and Weiler 2009; Stenner demonstrated empirically—that an individual’s behav- 2005), altruism (e.g., Fowler 2006), self-esteem (e.g., iors and attitudes show consistency across seemingly Sniderman 1975), conflict avoidance (e.g., Mutz 2002; unrelated domains (Gosling 2008). Psychologists have Ulbig and Funk 1999), and racial resentment (e.g., identified a small number of personality dimensions Feldman and Huddy 2005). Analysis using the Big Five that reduce the complexities of personality to a handful complements this earlier work because the Big Five are of basic traits. These traits (the ‘‘Big Five’’) capture seen in psychological theory as ‘‘core dispositional broad and enduring dispositions that shape how traits’’ that are causally prior to midlevel psychological people respond to the stimuli they encounter in the constructs, like Right-wing Authoritarianism, that are world. Research finds that these traits predict a wide products of both dispositional traits and the environ- range of behavioral outcomes (Gosling 2008) and are ment (e.g., McAdams and Pals 2006). An additional also highly stable over time and appear to be shaped by distinction is that, relative to other psychological biological (genetic) factors (e.g., Plomin et al. 1990). constructs, the Big Five are measured with minimal In this article, we examine the relationships between references to political content, and are therefore less political participation and the Big Five traits identified likely to be confounded by the political outcomes they in the Five-Factor Model (FFM) of personality.1 may predict. 1 An online appendix with supplementary material for this article is available at https://journals.cambridge.org/jop. Data and supporting materials necessary to reproduce the numerical results will be made available at http://huber.research.yale.edu/ upon publication. This research was funded by Yale’s Center for the Study of American Politics and Institution for Social and Policy Studies. The Journal of Politics, Vol. 73, No. 3, July 2011, Pp. 692–706 doi:10.1017/S0022381611000399 Ó Southern Political Science Association, 2011 ISSN 0022-3816 692 personality and political participation 693 We make three contributions to existing research. findings have a variety of implications, which we First, using two datasets—one from a national survey discuss in the conclusion. and one from a survey of Connecticut residents—that were both matched to public voter rolls, we examine the relationships between Big Five traits and validated The Five-Factor Model of voter turnout in general elections. While previous Personality research has examined the relationships between a number of individual and contextual factors that In psychology a working consensus has emerged that predict validated turnout (e.g., Katosh and Traugott personality traits can be measured using a Five-Factor 1981), to our knowledge, this is the first study to Model (FFM). The FFM, ‘‘the most widely used and validate the relationship between Big Five Traits and extensively researched measure of personality’’ (Gosling, political behavior. Our results indicate that people Rentfrow, and Swann 2003, 506), emerged from anal- high on Extraversion (a trait associated with asser- ysis of natural language. Researchers have found that tiveness and enthusiasm) and Emotional Stability this trait structure is consistent across different types of (associated with low anxiety) are more likely to vote, samples, languages, raters (including self versus peer while those high on Conscientiousness (associated ratings), and methodological variations (John and with achievement striving) are less likely to vote. The Srivastava 1999, 106–109). These five traits are de- data from the Connecticut survey, which include self- scribed by John and Srivastava (1999, 121) as follows: reports of turnout in 2004 and 2006, also allow us to Extraversion implies an energetic approach to the determine whether personality traits are associated social and material world and includes traits such as with misreporting of turnout. sociability, activity, assertiveness, and positive emo- Second, we examine how the relationships be- tionality. Agreeableness contrasts a prosocial and tween Big Five traits and other forms of participation communal orientation toward others with antagonism and includes traits such as altruism, tender-minded- compare with the relationships between these traits ness, trust, and modesty. Conscientiousness de- and turnout. The outcomes we examine include scribes socially prescribed impulse control that summary measures of participation in national cam- facilitates task- and goal-directed behavior, such as paigns and participation in local affairs. We also thinking before acting, delaying gratification, follow- assess the relationships between personality traits and ing norms and rules, and planning, organizing, and engagement in specific participatory acts that differ in prioritizing tasks. [Emotional Stability describes even-temperedness and] contrasts . . . with negative the amount of interpersonal interaction and conflict emotionality, such as feeling anxious, nervous, sad, they are likely to involve (e.g., Ulbig and Funk 1999), and tense . . . Openness to Experience (versus the instrumental benefits they are likely to yield (e.g., closed-mindedness) describes the breadth, depth, Gerber, Green, and Larimer 2008; Riker and Ordes- originality, and complexity of an individual’s mental hook 1968), and the extent to which participation is and experiential life (bolded personality traits added for emphasis; italics in original). likely to be seen as a duty (e.g., Dalton 2008; Riker and Ordeshook 1968). We posit that these differences These traits are variations in basic individual level are likely to make some modes of participation partic- tendencies (McCrae and Costa 1996). They are ularly attractive to individuals with certain personality largely heritable (e.g., Bouchard 1997; Plomin et al. traits while making other modes less appealing. We find 1990; Van Gestel and Van Broeckhoven 2003) and are that some traits predispose people to engage in a wide remarkably stable through life (e.g., Costa and array of participatory acts. However, our findings also McCrae 1992; Gosling et al. 2003). Because of this, support our expectation that the relationships between scholars refer to Big Five traits as ‘‘core’’ (Asendorpf Big Five traits and political participation vary substan- and van Aken 2003) or ‘‘dispositional’’ (McAdams and tially across participatory acts. Pals 2006) traits. Dispositional traits are theorized to Third, we assess the relative importance of be causally prior to both (1) midlevel aspects of personality compared to two variables traditionally personality (‘‘characteristic adaptations,’’ McAdams at the heart of analysis of political participation, and Pals 2006)—such as political ideology, Right-wing education and income (e.g., Rosenstone and Hansen Authoritarianism, Social Dominance Orientation, and 1993). We find that even after controlling for these values—and (2) specific attitudes and behaviors and other demographic variables, personality varia- (McCrae and Costa 1996). bles are frequently comparable in importance to Research finds that Big Five traits predict a wide those of canonical predictors that have been the range of behaviors, including job performance, focus of numerous studies of participation. Our school performance, juvenile delinquency, overall 694 alan gerber et al. health, musical tastes, dress, and a variety of other attending local meetings and signing petitions. No behaviors and attitudes (e.g., Gosling 2008; Ozer and other studies find a relationship between Agreeable- Benet-Martı´nez 2006; Paunonen and Ashton 2001). ness and political participation. Mondak and Halperin These traits also predict a number of political out- (2008) find a positive association between Conscien- comes. Most notably, there is a great deal of evidence tiousness and attending local meetings and contacting that Big Five traits, particularly Openness and Con- local officials; however, in later work Mondak and his scientiousness, are associated with political ideology colleagues (Mondak et al. 2010) do not find a statisti- (for recent work in the U.S. context, see, e.g., Carney cally significant relationship between Conscientious- et al. 2008; Gerber et al. 2010). ness and contacting elected (not necessarily local) Although scholars have devoted extensive atten- officials. This later study does find that Conscientious- tion to the relationships between Big Five personality ness is negatively associated with both working for and traits and political attitudes and ideology, relatively contributing money to a party or candidate. little work has examined the relationships between Mondak and Halperin (2008) also find a positive these traits and political participation (see Gerber relationship between Emotional Stability and both et al. 2011 for a review). Only one previous study has attending rallies and working for national parties or examined these relationships using a national sample candidates, but Mondak et al. (2010) do not. While (Mondak et al. 2010), and, in general, the findings the 2008 article does not find any relationships reported in previous published work have been mixed. between Emotional Stability and other forms of local (These findings are summarized in Table A1 of the and national participation (including reported turn- online appendix.). These differences may be the prod- out), the 2010 article reports negative associations uct of the samples used or the historical political between Emotional Stability and contributing money context in which the studies were conducted (see to a political party or candidate, contacting elected Gerber et al. 2010). They may also stem from variation officials, and reported turnout.2 Finally, there is some in the particular behaviors examined or how Big Five evidence that Openness is positively associated with a traits are measured. In the remainder of this section we wide variety of participatory acts, including reported review the basic contours of these findings. turnout (Mondak et al. 2010). While these findings The Big Five trait most consistently associated with are not replicated in other U.S. samples, they are political participation is Extraversion. In the United largely consistent with research using non-U.S. sam- States, individuals scoring high on Extraversion are more ples that finds that Openness is associated with some likely to attend campaign events and local meetings and forms of political participation and community express their views through petitions, letters to the editor, engagement (Vecchione and Caprara 2009; Mondak and contact with elected officials (Mondak et al. 2010; et al. 2011). Mondak and Halperin 2008). Similar findings emerge Before proceeding, we note that although Big Five in non-U.S. samples (in Uruguay and Venezuela, traits are broadly accepted as the best way to compre- Mondak et al. 2011; and in Italy, Vecchione and hensively measure dispositional traits, they are not the Caprara 2009). However, while previous research finds only way to conceive of personality. Previous research a number of statistically significant relationships be- has examined the relationships between other psycho- tween Extraversion and participation, in some cases this logical characteristics and political participation (e.g., trait does not significantly predict participation. Most Blais and Labbe´ -St-Vincent 2011; Denny and Doyle notably, none of the analyses of the relationships 2008; Mussen and Wyszynski 1952). For example, between Big Five traits and reported turnout find a there is evidence that individuals high on altruism significant relationship between Extraversion and this (a specific component or ‘‘facet’’ of Agreeableness) form of participation (Anderson 2009; Mondak et al. are more likely to vote in the United States and 2010; Mondak and Halperin 2008). Prior research has also found a number of signifi- cant relationships between the other Big Five traits and 2 Anderson (2009) also finds a negative and statistically significant participation, but these findings are also mixed. For association between Emotional Stability and turnout in local elections in one of the models she reports. It is important to note example, Mondak and Halperin (2008) find a negative that Anderson includes controls that Mondak and his colleagues association between Agreeableness and reported (and we, see below) do not. Mondak and Halperin (2008) control turnout in one sample, but not in another. They also for gender, age, race, and education; Mondak et al. (2010) control find several positive relationships between this trait for gender, age, and race; Anderson (2009) includes a variety of controls, such as internal efficacy and political knowledge, that and a variety of forms of local political participation are likely endogenous to Big Five personality traits (see Mondak (but not national campaign participation), including and Halperin 2008). personality and political participation 695 Canada (Blais and Labbe´ -St-Vincent 2011; Fowler speaking at a local meeting necessarily involves taking 2006; Fowler and Kam 2007). There is also some a stand that others may challenge. Participation in evidence that the conflict avoidant, as measured by a local politics is more likely to yield direct personal self-expressed distaste for contentious (sometimes benefits than participation in a national campaign. explicitly political) discussion, are less likely to Turning out to vote is more likely to be seen as a civic participate in politics (Blais and Labbe´ -St-Vincent duty than volunteering for a candidate (Dalton 2008). 2011; Mutz 2002; Ulbig and Funk 1999). This The top portion of Table 1 displays variation research provides valuable insight into how individ- along these three personality-relevant dimensions for ual-level characteristics affect participatory behavior each general mode of participation we analyze (vot- and informs our hypotheses about the likely relation- ing, national campaign participation, and participa- ships between Big Five traits and participation, both tion in local affairs): interpersonal interaction (and generally and across modes of participation. the accompanying possibility of conflict), norms, and instrumental outcomes. (We consider further dis- tinctions among participatory acts in each category Personality and Political below.) Each ranking along those dimensions is relative. So, for example, interpersonal interaction is Participation: Theory and lowest for voting relative to participation in national Hypotheses campaigns or local politics. We use this description of the characteristics of different forms of participation Big Five traits shape the attractiveness of different to formulate predictions about the likely effects of Big forms of stimuli. Determining exactly what sort of Five traits on each mode of participation, realizing stimuli political participation constitutes is there- that in some cases traits are likely to have counter- fore a necessary step in forming expectations about vailing effects across the different characteristics of the relationship between personality and those the mode of participation. activities. We begin by specifying which forms of We begin with Extraversion. People high on this political participation we consider in our analysis trait are assertive and sociable. As such, they are likely and then discuss how differences in the nature of to be drawn to the social engagement aspects of each mode of participation may suggest variation in political participation and to be eager to advocate for the relationships between Big Five traits and under- their preferences. Thus, in line with previous work taking each type of action. We focus on three broad (Mondak et al. 2010; Mondak and Halperin 2008), categories of participation: (1) voting in general we expect that Extraversion will be a particularly elections, (2) participating in national political cam- strong predictor of participation that involves inter- paigns, and (3) participating in local community personal interactions, like participating in national affairs and politics. Within the latter two categories campaigns and local politics. Across forms of partic- we distinguish among different types of activities. ipation in campaigns, individuals low on this trait Three characteristics of these different participatory may be willing to send a check to a candidate, but acts are likely to be relevant to the relationships between may be unwilling to participate in more active, social personality and participation: (1) interpersonal inter- forms of participation such as attending a meeting or action and the accompanying potential for exposure to rally. Previous research provides some support for conflict (e.g., Ulbig and Funk 1999), (2) social and civic this expectation. Both Mondak et al. (2010) and norms and expectations concerning behavior (e.g., Mondak and Halperin (2008) find relationships Dalton 2008; Gerber et al. 2008), and (3) the weak between Extraversion and a variety of socially engag- relationship between political participation and in- ing forms of participation, but do not find a relation- strumental outcomes (e.g., Riker and Ordeshook ship between this trait and the relatively private act of 1968). It is immediately obvious that virtually all turning out to vote. forms of political participation involve more inter- We also expect that Agreeableness will affect personal interaction and potential for conflict than political participation. Those high on this trait tend watching television at home; that each may be affected to be altruistic, modest, and sympathetic. Findings by social norms; and that each mode of participation is regarding the relationship between this trait and less likely to yield instrumental benefits than showing political participation have been mixed. Agreeable- up for work. However, modes of participation also ness is associated with nonpolitical volunteering (e.g., clearly vary along these three dimensions. Donating Bekkers 2005) and one aspect of Agreeableness— money involves little interaction with others while altruism—is associated with higher levels of turnout 696 alan gerber et al. T ABLE 1 Summary of Expectations Vote National Campaigns Local Politics Interpersonal – + + Norm + – + Instrumental Benefit – – + Extraversion + ++ ++ Agreeableness None – – (+ for less conflictual activities) Conscientiousness None (2 if instrumental None (may vary across None (may vary across benefits dominate, specific activities: specific activities: + if norms dominate) see text for details) see text for details) Emotional Stability + ++ ++ Openness to None + + Experience Note: In the top half of the table, rankings are relative. + indicates participatory activity is higher on characteristic than for activities denoted with a 2. In the bottom half of the table, + indicates an expected positive relationship; – indicates an expected negative relationship. (Blais and Labbe´ -St-Vincent 2011; Fowler 2006). than to engage in other forms of participation, such However, research also finds that conflict avoidance as attending a rally, that are unlikely to be viewed as (another characteristic likely to be associated with civic duties. However, Conscientiousness is also Agreeableness) is associated with lower levels of associated with a focus on instrumental benefits— political participation (Blais and Labbe´ -St-Vincent benefits that are unlikely to be garnered by voting or 2011; Mutz 2002; Ulbig and Funk 1999). For this donating money to a national candidate. For this reason, we expect the relationship between Agree- reason, Conscientious individuals may eschew polit- ableness and participation to vary depending on the ical participation in favor of more practical activities nature of the participatory act. Individuals high on (including perhaps participation in local politics— Agreeableness are likely to be repelled by (and thus behavior that is more likely to lead to concrete unlikely to participate in) forms of participation that personal payoffs than participation in national cam- may involve conflictual interactions (see Antonioni paigns). Prior research offers some support for each 1998), such as speaking at a local meeting or attend- of these offsetting predictions, particularly Mondak ing a rally. However, other forms of participation, et al.’s finding that Conscientiousness is more likely such as voting (and perhaps some forms of local to be associated with political participation when the community decision making) involve less conflict. individual perceives the campaign activity to be Thus, Agreeableness may be less negatively (or even ‘‘important’’ (2010, 96–98). positively) associated with these types of participa- Emotional Stability is associated with self- tion. Mondak and Halperin (2008) find mixed sup- assuredness and an absence of anxiety, depression, port for these predictions, reporting a number of and other negative emotionality. We expect the self- statistically significant and positive relationships be- assuredness and lack of anxiety that characterize tween Agreeableness and local participation, but a Emotional Stability to lead to greater willingness to negative relationship between this trait and turnout participate in the conflictual realm of politics. As in national elections. discussed above, however, previous findings regard- Conscientious individuals are characterized by ing the relationship between this trait and political dutifulness, norm compliance, and achievement participation have been mixed (Anderson 2009; striving. As with Agreeableness, some aspects of Mondak et al. 2010; Mondak and Halperin 2008). political participation may be attractive to those high Last, Openness is associated with curiosity and a on this trait while other aspects may be less appealing. willingness to entertain novel ideas. The most recent For example, to the extent that political participation work to examine the relationships between the Big is viewed as a civic duty, Conscientious people may Five and political participation finds that Openness be likely to participate as a way of adhering to social is associated with a variety of political activities norms. Individuals high on this trait may therefore be (Mondak et al. 2010), but this stands in contrast to more likely to fulfill a perceived obligation to vote the earlier work that typically found no relationship personality and political participation 697 (Anderson 2009; Mondak and Halperin 2008). We chief advantage of the CCAP is that it is a large expect that individuals high on Openness will be sample that, after applying weights, is nationally particularly drawn to participatory activities where representative of registered voters. The chief advan- they are likely to be exposed to a variety of ideas, tages of the CT Survey are that it is drawn from a such as local meetings. However, we do not have telephone directory sample and, because respondents clear expectations about how this trait will be related to this survey were also asked to report whether they to other forms of participation. voted in 2004 and 2006, we can examine whether Big We summarize our expectations on a trait by Five traits predict the extent to which respondents trait basis in the bottom panel of Table 1. In those misreport turnout behavior. Moreover, similar ques- cases where we identify salient countervailing forces, tions were asked on both and we obtain similar most notably regarding the associations between results (exceptions are noted), increasing our con- Conscientiousness and political participation, we also fidence in the inferences we make. note the ambiguity in our predictions. Measuring the Big Five. Our surveys use the Ten Item Personality Inventory (TIPI), developed by Gosling et al. (2003), to measure the Big Five person- Data Sources (CT Survey and CCAP) ality dimensions. This battery is ideal in the survey context because its length and speed of administra- tion make it feasible where longer batteries are not. The data for our analysis come from two surveys. The The TIPI asks respondents to report how well ten first is the 2007–2008 Cooperative Campaign Analysis pairs of traits (e.g., ‘‘extraverted, enthusiastic’’) de- Project (hereafter ‘‘CCAP’’: Jackman and Vavreck scribe themselves. Gosling et al. compared the per- 2009). The CCAP is an Internet-based panel survey of formance of the 10-question battery to much longer 20,000 registered voters that uses a combination of tests and find that scores obtained from the TIPI are sampling and matching techniques to approximate a highly correlated with those obtained from longer random digit dialing sample.3 We employ sampling instruments (2003, see Tables 6 and 9). We present a weights to approximate a nationally representative more complete discussion of the reliability, robust- sample in our analysis. Demographic measures were ness, and use of the TIPI in the online appendix. collected in December 2007 and measures of non- voting participation were collected in September and October of 2008. Additionally, we have a partially Analysis overlapping sample of 3,367 CCAP respondents, of whom 2,447 (73%) were successfully matched to Validated Turnout. We begin by analyzing the publicly available voter rolls by the survey firm. relationships between Big Five traits and validated The second data source is a telephone survey of a turnout in both surveys. In each case we have records random sample of approximately 1,800 Connecticut of turnout in the four even-year general elections residents with listed phone numbers (hereafter ‘‘CT from 2000 to 2006. Validated Turnout Count is Survey’’) fielded in June 2008. To facilitate matching therefore the number of these four elections in which with Connecticut voter records, which list name and a respondent voted that ranges from zero (voted in address for all registered voters, Survey Sampling Inc. none of the elections) to four (voted in all four drew a random sample from a residential phone elections). We present results for this turnout index directory of Connecticut households with accurate because accounting for turnout across multiple elec- mailable addresses. In addition to a personality battery tions reduces measurement error associated with (see below) and demographic items, survey respondents three factors: idiosyncratic reasons why one would were asked about their political participation. Respond- vote in any given election, potential random error in ents who completed the survey were then matched to the recording of voting in any given election, and Connecticut voter file to obtain validated turnout data unobserved contextual factors that might affect turn- from 2000 to 2006 (see online appendix). out in any given election. In addition to this measure Both surveys permit us to verify participation of average or typical behavior, we also report analysis using administrative records of actual behavior. The for each election separately in the online appendix. 3 (Turnout is measured slightly differently in the two Details about the construction of all samples used in our analysis datasets because the CCAP sample is restricted to (including full question wording, coding rules, and summary statistics) and supplemental analysis appear in the online cases of [self-reported] registered voters successfully appendix. matched to the voter file while the CT sample 698 alan gerber et al. includes unregistered individuals and therefore un- strong positive associations between both Extraver- matched cases are coded as zero turnout.) Because sion and Emotional Stability and Validated Turnout they were not eligible to vote for the full period for Count. These findings support our theoretical claim which we examine turnout behavior, we exclude that Extraverts are drawn to the interpersonal com- respondents who were not of voting age in 2000 (less ponents of political participation and the more than 26 years old at the time the surveys were Emotionally Stable are more confident in the face fielded—approximately 5% of each weighted sample) of the contestation of the political realm and, thus, from all analyses. more willing to participate. Results on an election- In our analysis of turnout, as well as our analysis by-election basis are similar, although indications of of other modes of participation, we present specifi- statistical significance and magnitudes of effects vary cations using this general equation: from year to year (see online appendix). To demonstrate the relative importance of these DV 5 B0 þ C*Personality þ D*Controls associations, in columns (1) and (2) of Table 3 we þ F*State Fixed Effects þ e; ð1Þ present the estimated marginal effect of a two- standard-deviation increase in each of the Big Five where Personality is a vector of Big Five traits and traits as well as in income (a substantial increase from Controls includes gender, race, age, and age-squared approximately $25,000 to $100,000 per year in each (to allow for nonlinearity in the effects of age), sample) and education (a shift from being a high income (measured as a linear scale with a separate school to a college graduate) on the likelihood of a indicator for income refused), and educational at- respondent being an above-average turnout voter tainment (measured using indicators for each educa- (voting in three or four elections in the CCAP; voting tion category [the excluded category is high school in two, three, or four elections in the CT sample) graduate]).4,5 (Summary statistics are presented in rather than a below average turnout voter.6 Given the the online appendix.) In CCAP analyses we also similarity of the Validated Turnout Count and the include State Fixed Effects (a vector of state of election-by-election results, these marginal effects residence indicators) and cluster standard errors at also provide a sense of the average effect of each trait the state level to allow for interdependence of ob- on the probability of turning out in any given servations in a given geographic area. The inclusion election relative to staying home (see online appendix of state fixed effects ensures that our CCAP results are for marginal effects by election). In the CCAP and CT not generated by some correlation between person- Surveys, respectively, a two-standard-deviation in- ality and other factors that might affect the propen- crease in Extraversion is associated with a 7.5% and sity to participate in political activities (e.g., state 9.8% increase in the likelihood of a respondent being political culture or legal rules affecting registration). a high-turnout voter (relative to the baseline proba- We present the results of our ordered logit bilities of 60.2% and 46.7%). A similar increase in analysis of the relationships between Big Five traits Emotional Stability is associated with 14.1 and 8.9% and validated turnout using multielection indices in increases in the likelihood of being a high-turnout Table 2. We did not expect Openness to affect voter in the CCAP and CT Surveys. These magni- turnout and find disparate and statistically insignif- tudes are comparable to the estimated effects of a icant relationships across datasets. Consistent with two-standard-deviation increase in income (18.0 and our expectations, across the two surveys we find 10.9%) and education (8.7 and 9.9%)—canonical predictors of participation. 4 We cannot control for gender in our analysis of the CT survey Our expectations regarding the relationship be- because this variable was not recorded. We note that only about 11% of respondents in the CT survey identified as nonwhite, with tween Conscientiousness and participation were mixed. fewer than 3% in any specific nonwhite racial category. For this We posited that individuals high on this trait may be reason, we include an indicator for ‘‘nonwhite’’ in our analysis of more likely to turn out because they see voting as a these data rather than the more detailed set of indicators we use in the CCAP analysis. social norm to be followed. Alternately, we proposed 5 that this trait could be associated with lower turnout We also report in the online appendix models without measures of income and education, which yield similar results, because due to the minimal instrumental benefits associated both of those characteristics have been shown to be at least with voting in national elections. The negative and partially endogenous to personality (e.g., Borghans et al. 2008; Paunonen and Ashton 2001). We view models that include these 6 controls as quite conservative because the indirect effects of All estimated marginal effects are for a 51-year-old white female personality on participation as mediated through income and from California, with personality traits, education, and income education will be absorbed by those variables. set to their sample means. personality and political participation 699 T ABLE 2 Validated and Overreporting Turnout in General Elections (3) (4) (5) (1) (2) CCAP CT Survey CT Survey General Election Turnout General Election Turnout 2000-2006 (4 elections) 2004 and 2006 (2 elections) Validated Reported Overreport Turnout Turnout Turnout Validated Turnout Count (0-4) Count (0-2) Count (0-2) Count (0-2) Extraversion (0-1) 0.390 [0.234]* 0.319 [0.152]** 0.328 [0.160]** 0.605 [0.190]*** –0.096 [0.162] Agreeableness (0-1) –0.471 [0.293] 0.129 [0.217] 0.292 [0.220] –0.197 [0.270] –0.435 [0.226]* Conscientiousness (0-1) –0.615 [0.366]* –0.383 [0.232]* –0.402 [0.239]* 0.050 [0.270] 0.327 [0.236] Emotional Stability (0-1) 0.793 [0.261]*** 0.350 [0.188]* 0.399 [0.195]** 0.437 [0.227]* –0.225 [0.198] Openness (0-1) 0.242 [0.272] –0.302 [0.189] –0.097 [0.194] 0.239 [0.235] 0.221 [0.204] Female 5 1 –0.072 [0.094] Black 5 1 –0.202 [0.146] Hispanic 5 1 0.231 [0.273] Other (Native American, 0.012 [0.363] Asian, Mixed, Other) 5 1 Nonwhite 5 1 –0.380 [0.151]** –0.334 [0.151]** –0.136 [0.170] 0.162 [0.145] Age (Years) 0.105 [0.026]*** 0.117 [0.018]*** 0.104 [0.020]*** 0.128 [0.022]*** –0.020 [0.019] Age2/100 –0.063 [0.025]** –0.075 [0.015]*** –0.066 [0.017]*** –0.076 [0.019]*** 0.017 [0.016] Income (0-1, 15Refused) 0.934 [0.346]*** 0.402 [0.203]** 0.364 [0.218]* 0.981 [0.271]*** 0.151 [0.213] Income Refused –0.182 [0.271] –0.157 [0.173] –0.005 [0.191] –0.711 [0.234]*** –0.372 [0.193]* Educ , HS –0.470 [0.341] –0.920 [0.262]*** –0.826 [0.248]*** –0.978 [0.255]*** –0.058 [0.250] Educ5some college –0.058 [0.188] 0.103 [0.145] 0.115 [0.145] 0.275 [0.170] 0.088 [0.146] Educ52 year college –0.007 [0.204] 0.084 [0.140] 0.232 [0.146] 0.487 [0.173]*** 0.117 [0.149] Educ5College 0.223 [0.198] 0.188 [0.132] 0.269 [0.139]* 0.685 [0.166]*** 0.245 [0.138]* Educ5Post Grad 0.048 [0.213] 0.389 [0.133]*** 0.460 [0.136]*** 0.938 [0.166]*** 0.049 [0.142] Indicators for state? Yes No No No No Observations 2147 1924 1909 1909 1909 F-test: Big Five 0.041 0.034 0.021 0.002 0.117 Mean 2.429 1.591 1.009 1.611 0.666 Note: See text for coding details. Ordered logit coefficients with robust standard errors (clustered by state in CCAP models) in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. Two-tailed tests. statistically significant coefficients on Conscientious- the variation in electoral environments across states, ness in the models presented in columns (1) and (2) of it does suggest that Agreeableness may have offsetting Table 2 suggest that the latter mechanism dominates. effects. For example, people high on this trait may be The marginal effects are also relatively large: A two- more likely to register to vote but, conditional on standard-deviation increase in this trait is associated having done so, be less likely to vote than other with a 9.3 and 7.9% decrease in the likelihood a registered voters. This would be consistent with the respondent is a high-turnout voter in the CCAP and notion that Agreeable individuals seek to avoid the CT samples, respectively. conflictual milieu of politics, but because they are The relationship between Agreeableness and also communal and prosocial in orientation, end up turnout is inconsistent across samples. In the national engaging in the collective (and nonconfrontational) CCAP sample of registered voters, we find that more act of registering to vote (an act which increases the Agreeable individuals are less likely to turn out to likelihood of voting, thereby attenuating the negative vote. However, in the CT sample including unregis- relationship between this trait and turnout). tered voters, the association is slightly positive, but Reported Turnout and Overreporting Turnout. statistically indistinguishable from 0. While we hesi- The CT Survey included items asking respondents tate to infer too much from these differences given whether they voted in the 2004 and 2006 general 700 alan gerber et al. T ABLE 3 Marginal Effects for Table 2 Results (1) (2) (3) Data Source: CCAP CT Survey CT Survey Overreport General General Election Turnout Election Turnout 2004 Measure: 2000-2006 (4 elections) and 2006 (2 elections) Turned out in 3 or 4 Turned out in 2, 3 or 4 Overreported Voting in 1 or 2 Marginal Effect for Outcome: (out of 4) Elections (out of 4) Elections (out of 2) Elections Column in Table 2: (1) (2) (5) Baseline Probability 60.2% 46.7% 44.4% Extraversion 7.5% 9.8% –3.1% Agreeableness –7.0% 2.9% –10.1% Conscientiousness –9.3% –7.9% 7.0% Emotional Stability 14.1% 8.9% –6.0% Openness 3.8% –7.9% 6.0% Income 18.0% 10.9% 4.3% Education 8.7% 9.9% 13.7% Note: See text for details of marginal effects specifications. Table entries are proportional changes relative to baseline probability for two-standard-deviation increase in each item. For income this corresponds to a change from approximately $25,000/year to $100,000/ year. For education this is a change from high school graduate to college graduate. elections. In combination with the matched voter file (3) of Table 3 indicate, a two-standard-deviation records, these data allow us to assess whether person- increase in Agreeableness reduces the probability of ality is related to the overreporting of turnout. In overreporting turnout in one or two elections by columns (3) through (5) of Table 2 we examine this 10.1%. The magnitude of this effect is comparable to possibility by presenting ordered logit models for three a change in education (a variable found to be asso- outcomes: Validated Turnout Count in both 2004 and ciated with overreporting turnout in previous work, 2006 (column 3), Reported Turnout Count in these two Vavreck 2007) from high-school graduate to college elections (column 4), and finally, Overreport Turnout graduate, which increases overreporting by 13.7%. The Count in these two elections (column 5), which is the negative relationship between Agreeableness and over- number of elections the respondent reported voting in reporting may stem from the fact that Agreeable that the voter rolls indicate she did not (we did not find individuals tend to be modest and, thus, may be less any evidence that personality predicted the under- inclined to represent (or remember) their behavior in reporting of turnout). Each measure ranges from 0 to an excessively favorable light. 2, with average validated turnout equal to about 1 and Emotional Stability is also associated with being average reported turnout about 1.6. less likely to overreport turnout, although the coef- We begin by noting the similarity between the ficient is not statistically significant (p-value5.26, column (2) specification—validated turnout in the two-tailed test). This is consistent with the notion four general elections from 2000-2006—and the col- that Emotionally Stable people are not emotionally umn (3) results—validated turnout in the two general reactive and are therefore less likely to feel emotional elections in 2004 and 2006. Column (4) displays results urgings to provide socially acceptable responses. By for reported turnout. We focus, however, on the contrast, Conscientiousness is positively associated column (5) specification, which is the measure of with overreporting turnout, although again this overreporting of turnout. Here, we see several interest- coefficient falls short of statistical significance at ing associations, although only the coefficient on conventional levels (p5.16, two-tailed test). This Agreeableness is statistically significant at conventional association, however, is consistent with the idea that levels. Extraverts do not appear to misreport their Conscientious people are aware of social norms turnout (the coefficient is negative but relatively small), concerning turnout, but are unwilling to alter their but less Agreeable individuals are more likely to over- real (rather than reported) behavior to achieve them. report voting. As the marginal effects shown in column Last, we find that Openness is associated with personality and political participation 701 overreporting turnout, but the coefficient is both and both participation indices that are substantively relatively small and not statistically significant. large and statistically significant across specifications. In Cumulatively, these findings add to the growing the CCAP, a two-standard-deviation increase in Extra- literature on which individual-level characteristics are version is associated with a 45.3% increase in the most likely to incline individuals to misrepresent their likelihood that a respondent engaged in at least one political behavior (e.g., Vavreck 2007). Not all of these of the three forms of campaign participation; in the CT relationships reach conventional levels of statistical sample this estimate is 32.8%. For the CT sample’s local significance, but the magnitudes of the relationships participation index the marginal effect is 26.5%. The are large, suggesting the value of using larger samples in magnitudes of these effects are comparable to similar future research. We note these findings suggest some increases in income (44.0, 34.5, 20.3%) and education caution about interpreting reported behavior measures. (62.4, 41.9, 50.8%). Nonvoting Modes of Participation. Next we Also consistent with our expectations and with the examine the relationships between Big Five traits and findings from the turnout models, we find a uniformly nonvoting forms of political participation. We begin by positive relationship between Emotional Stability and assessing the relationships between these traits and two participation (although this relationship falls short of summary measures of participation. The first summary conventional levels of statistical significance in the CT measure is a (national) campaign participation index. Survey models). The results also provide support for This measure is coded slightly differently in the CCAP our expectation that individuals high on Openness and CT surveys (see the online appendix for complete would be drawn to the diversity of ideas encountered details). For the CCAP survey, the participatory acts in interpersonal political interactions and therefore were: donating to a candidate, wearing a button or participate more frequently. Although the coefficient sticker in support of a candidate, and attending a on this trait is not statistically significant for the political rally. For each form of participation respond- campaign participation index outcome in the CT survey ents were assigned a 1 if they reported that they engaged model, the sign is positive across all models and reaches in the act ‘‘yesterday’’ in either the September or conventional levels of statistical significance in the October wave of the survey. We then created an other two estimated models. The marginal effect of additive scale of the number of reported modes of an increase in this trait is also relatively large, ranging participation (ranging from 0 to 3). (Because the scale from 9.4 to 29.4% across outcomes and datasets. was created based on whether a respondent did some- For the remaining two traits—Agreeableness and thing ‘‘yesterday,’’ we also include indicators for the day Conscientiousness—results differ across the two data- of week the respondent completed the September and sets (our findings regarding the relationship between October waves.) In the CT Survey, respondents were Agreeableness and turnout were also inconsistent across asked whether they had participated in a variety of ways datasets). In the CT survey we find a negative and in the previous two years. We focus on three measures significant or borderline statistically significant relation- of participation comparable to those used in the CCAP: ship between Agreeableness and both participation donating to a candidate, volunteering for a candidate or indices. This is consistent with our earlier hypotheses party, and attending a political rally. Again, we created that Agreeable individuals will avoid political activities an additive index of the number of acts the respondent that draw them into potentially conflictual situations. reported participating in (ranging from 0 to 3). We do not find this effect in the CCAP sample of The second measure of participation we employ registered voters for activities that took place ‘‘yester- is a local participation index, which draws only on the day,’’ however, where the estimated coefficient is near 0. CT survey, in which respondents were also asked This pattern is reversed for Conscientiousness. We about their participation in local politics in the past find positive, but not statistically significant, relation- two years. Respondents were asked whether they had ships between Conscientiousness and both forms of contacted a local official about a political matter, participation in the CT survey, but a negative and attended a meeting about a local issue, or spoken at a statistically significant relationship between this trait local meeting. We created an additive scale ranging and participation in the CCAP sample. The magnitude from 0 to 3 based on these three measures. of this effect in the CCAP sample is also relatively Table 4 presents the results of our ordered logit large—the marginal effect of a two-standard-deviation analysis of the relationships between Big Five traits increase in Conscientiousness is to decrease the prob- and each of the participation indices; marginal effects ability of reporting any participation by 20.2%. This appear in Table 5. Consistent with our expectations, we finding is consistent with our expectation that Con- find robust positive associations between Extraversion scientious individuals are likely to devote their energies 702 alan gerber et al. T ABLE 4 Nonvoting Political Participation: Indices (1) (2) (3) CCAP CT Survey Campaign Participation Campaign Participation Local Participation Index (0-3) Index (0-3) Index (0-3) Extraversion (0-1) 1.044 [0.084]*** 0.729 [0.187]*** 0.804 [0.156]*** Agreeableness (0-1) 0.049 [0.233] 20.331 [0.257] 20.435 [0.226]* Conscientiousness (0-1) –0.566 [0.155]*** 0.253 [0.286] 0.220 [0.243] Emotional Stability (0-1) 0.262 [0.142]* 0.181 [0.242] 0.248 [0.192] Openness (0-1) 0.838 [0.183]*** 0.248 [0.238] 0.454 [0.195]** Female 5 1 0.063 [0.046] Black 5 1 0.469 [0.095]*** Hispanic 5 1 0.233 [0.146] Other (Native American, 0.178 [0.124] Asian, Mixed, Other) 5 1 Non-White 5 1 0.453 [0.160]*** 0.133 [0.151] Age (Years) 20.015 [0.012] 0.060 [0.025]** 0.090 [0.021]*** Age2/100 0.030 [0.011]*** 20.031 [0.021] 20.074 [0.017]*** Income (0-1, 15Refused) 0.992 [0.175]*** 0.872 [0.249]*** 0.699 [0.218]*** Income Refused 20.559 [0.122]*** 20.845 [0.212]*** 20.527 [0.191]*** Educ , HS 0.211 [0.236] 20.376 [0.358] 0.065 [0.265] Educ5some college 0.454 [0.072]*** 0.509 [0.185]*** 0.378 [0.151]** Educ52 year college 0.319 [0.111]*** 0.494 [0.187]*** 0.643 [0.145]*** Educ5College 0.564 [0.081]*** 0.671 [0.171]*** 0.890 [0.143]*** Educ5Post Grad 0.740 [0.086]*** 1.085 [0.164]*** 0.903 [0.140]*** Indicators for state and Yes No No day of week of surveys? Observations 11362 1924 1924 F-test: Big Five 0.000 0.000 0.000 Mean 0.254 0.402 0.852 Note: CCAP Campaign Participation Index includes three items: donated to a candidate, wore a button, and attended a rally; CT Survey Campaign Participation Index includes three items: donated to a candidate, volunteered, and attended a rally; CT Survey Local Participation Index includes three items: contacted a local official, attended a local meeting, and spoke at a local meeting. See text for coding details. See online appendix for item question wording. Ordered logit coefficients with robust standard errors (clustered by state in CCAP models) in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. Two-tailed tests. to activities other than participation in national poli- the level of interpersonal interaction and conflict tics. What accounts for the different results in the CT associated with various forms of participation. Using and CCAP samples? As before, it may simply be due to the individual participation items from the CT Survey, the fact that the CCAP sample is restricted to registered we report in the online appendix the relationship voters, or because the CT sample is restricted to between personality and different forms of participation. Connecticut residents. In light of the evidence pre- Briefly, we find that Extraversion is consistently associ- sented above about the positive relationship between ated with higher levels of participation across all of the Conscientiousness and overreporting turnout, how- outcomes, but, as predicted, this relationship is partic- ever, another possibility is that in the CT sample, ularly pronounced for forms of participation that involve where respondents were asked to recollect about interacting with others, such as attending a rally (also see participation over the previous two years, Conscien- Mondak et al. 2010). We also find support for our tious people were more likely to misreport their expectation that the negative association between behavior than Conscientious respondents in the CCAP Agreeableness and participation would be strongest who were asked about their behavior yesterday (for for forms of participation that are likely to involve which norms would seem much weaker). conflict. For example, we find that Agreeableness is Above we also discussed how the relationship unrelated to attending a local meeting, but that there is between personality and participation may depend on a strong negative and statistically significant relationship personality and political participation 703 T ABLE 5 Marginal Effects for Table 4 Results (1) (2) (3) Data Source: CCAP CT Survey CT Survey Campaign Participation Measure: Index (0-3) Local Participation Index (0-3) .0 .0 Marginal Effect for Outcome: Column in Table 4: (1) (2) (3) Baseline Probability 10.9% 22.5% 42.8% Extraversion 45.3% 32.8% 26.5% Agreeableness 1.7% –10.7% –10.4% Conscientiousness –20.2% 7.6% 4.9% Emotional Stability 10.6% 6.7% 6.7% Openness 29.4% 9.4% 12.7% Income 44.0% 34.5% 20.3% Education 62.4% 41.9% 50.8% Note: See text for details of marginal effects specifications. Table entries are proportional changes relative to baseline probability for two- standard-deviation increase in each item. For income this corresponds to a change from approximately $25,000/year to $100,000/year. For education this is a change from high school graduate to college graduate. between this trait and more conflictual forms of reationships between personality traits and participa- participation, such as speaking at a local meeting. These tion must carefully consider the essential character- findings, which we elaborate on in the online appendix, istics of different modes of participation. Speaking at suggest the value of theorizing about the essential a local meeting, by its nature, involves social inter- characteristics of participatory acts in order to more action and a potential for conflict, while writing a fully understand the relations between personality and check to a political candidate does not. Beyond these political behavior. differences, however, the meaning of participation may also vary across contexts (see Gerber et al. 2010). Promising evidence of this sort of contingent rela- tionship between personality and context is provided Discussion by Mondak and his colleagues (2010), who find that Conscientious individuals are more likely to partic- The results presented here demonstrate that there are ipate when they believe it is important to be involved statistically significant and behaviorally important rela- in election campaigns. tionships between personality traits and key measures of While many of our findings are consistent with political participation. Our analysis indicates that Ex- the prior literature on the associations between Big traversion and Emotional Stability are associated with Five personality traits and political participation in higher levels of participation in a broad range of the United States, others are not. For Extraversion political activities. In many cases the magnitudes of and Openness, our findings are largely consistent these associations are comparable to those for canonical with those reported by Mondak and his colleagues, predictors of participation such as education and although our data reveal a positive and statistically income. We also find that the relationships between significant association between Extraversion and other Big Five traits and participation vary across modes turnout (including validated turnout) whereas pre- of participation. For example, although Agreeableness is vious work does not (although the coefficient is associated with lower levels of participation across a uniformly positive). Our findings for Agreeableness variety of participatory acts, this negative relationship is and Conscientiousness are also largely consistent with most pronounced for modes of participation likely to prior work, as they demonstrate that, for these traits involve conflict (e.g., speaking at a local meeting). in particular, the type of participatory act matters a The results, in concert with the theoretical frame- great deal for whether an individual participates. The work we propose, suggest that examinations of the findings we report about Emotional Stability are the 704 alan gerber et al. most inconsistent with prior work. Most notably, we online appendix and allows us to compare the effects find strong positive relationships between Emotional of personality on these forms of participation with Stability and turnout (both validated and reported) and their corresponding effects on attitudes as reported in donating money to a political candidate, while Mondak prior research (Gerber et al. 2010). In particular, et al. (2010) find negative associations. One potential given that Emotional Stability is associated with source of the inconsistent findings may be the different holding conservative economic views and a variety batteries used to measure the Big Five traits, which may of forms of political participation, our results suggest measure more or less specific aspects (or, ‘‘facets’’) of that conservative economic preferences may be over- each trait. While an advantage of the battery we use is represented on Election Day and at local meetings. that it has been validated against more extensive As is the case with all research, the present study batteries, analysis using batteries capable of measuring has its shortcomings. Perhaps its most significant the facets of the Big Five may help to explain some of limitation is that the outcomes we examine occurred the differences between our results and previous work. during a fairly narrow window of time. It is possible The apparent inconsistencies in findings across studies that some unmeasured contextual variable correlated and different samples may also reflect a failure to with personality and the various dependent variables identify differences both in what various modes of produced spurious correlations. Thus, extending participation entail and contextual factors, a promising our work to other electoral contexts is warranted. area for subsequent work. Overall, however, there are A related difficulty is the hazard to inference posed many areas of agreement about the relationship be- by measurement error. We are able to address any tween personality and political participation. reporting error concerns directly in our analysis of Our findings provide further evidence that turnout by using validated measures. However, we also individual-level differences in personality affect polit- found suggestive relationships between Big Five traits ical behavior. This influence of the core, broad aspects and the likelihood of overreporting turnout. This of individuality on participation, and the robustness of suggests that our estimates of the associations between personality effects on behavior in many other domains personality traits and other forms of participation may of life, suggests the value of integrating models of be somewhat biased. Additionally, regression analysis political behavior with models of human decision does not demonstrate a causal relationship between making. Politics, by this account, is just one domain variables. We have elected to interpret the demon- in which individual-level differences shape how we strated link between personality and political partic- behave. Models of participation would benefit from ipation as evidence that having a certain personality attention to sources of those differences that originate type has an effect on the individual’s political activity. in events earlier in the ‘‘funnel of causality’’ than has From this we posit that if a person’s personality was traditionally been considered. somehow changed then, holding other factors fixed, The associations we find between Big Five traits she would behave differently. It is difficult to imagine and political participation may also have significant an experimental manipulation of personality, and so consequences for the process of representation. of necessity we rely on statistical associations. The Previous work has identified a number of important direction of causality is a generic concern, but in this relationships between these traits and political atti- context there is no evidence that causality flows in the tudes. For example, Conscientiousness is consis- opposite direction from that hypothesized. tently found to be associated with conservatism. We find that individuals high on this trait are also less likely to turn out to vote. Other work finds Acknowledgments that Extraversion and Emotional Stability—traits we find are associated with higher levels of political We thank John Bullock, James Fowler, Sam Gosling, participation—are associated with holding conserva- Jeffery Mondak, the anonymous reviewers, and the tive economic policy attitudes (Gerber et al. 2010). editors for comments on earlier versions. These dynamics suggest that political participation may attract individuals with distinctive political attitudes, creating a politically engaged citizenry References whose views are not representative of the broader public. More detailed analysis examining differences Anderson, Mary R. 2009. ‘‘Beyond Membership: A Sense of in two activities as measured using the CT Survey, Community and Political Behavior.’’ Political Behavior 31 (4): voting and speaking at a local meeting, appears in the 603–27. personality and political participation 705 Antonioni, David. 1998. ‘‘Relationship between the Big Five John, Oliver P., and Sanjay Srivastava. 1999. ‘‘The Big Five Trait Personality Factors and Conflict Management Styles.’’ Inter- Taxonomy: History, Measurement, and Theoretical Perspec- national Journal of Conflict Management 9 (4): 336–55. tives.’’ In Handbook of Personality: Theory and Research, eds. Asendorpf, Jens B., and Marcel A. G. van Aken. 2003. ‘‘Personality- Lawrence A. Pervin and Oliver P. John. New York: Guilford Relationship Transaction in Adolescence: Core versus Sur- Press, 102–38. face Personality Characteristics.’’ Journal of Personality 71 (4): Katosh, John P., and Michael W. Traugott. 1981. ‘‘The Con- 629–66. sequences of Validated and Self-Reported Voting Measures.’’ Bekkers, Ren´e. 2005. ‘‘Participation in Voluntary Associations: Public Opinion Quarterly 45 (4): 519–35. Relations with Resources, Personality, and Political Values. ’’ McAdams, Dan P., and Jennifer L. Pals. 2006. ‘‘A New Big Five: Political Psychology 26 (3): 439–54. Fundamental Principles for an Integrative Science of Person- Blais, Andr´e, and Simon Labb´e-St-Vincent. 2011. ‘‘Personality ality.’’ American Psychologist 61 (3): 204–17. Traits, Political Attitudes, and the Propensity to Vote.’’ European McCrae, Robert R., and Paul T. Costa, Jr. 1996. ‘‘Toward a New Journal of Political Research 50 (3): 395–417. Generation of Personality Theories: Theoretical Contexts for Borghans, Lex, Angela Lee Duckworth, James J. Heckman, and Bas the Five-Factor Model.’’ In The Five-Factor Model of Personality: ter Weel. 2008. ‘‘The Economics and Psychology of Personality Theoretical Perspectives, ed. Jerry S. Wiggins. New York: The Traits.’’ Journal of Human Resources 43 (4): 972–1059. Guilford Press, 51–87. Bouchard, Thomas J., Jr. 1997. ‘‘The Genetics of Personality.’’ In Mondak, Jeffery J., Damarys Canache, Mitchell A. Seligson, and Handbook of Psychiatric Genetics, eds. Kenneth Blum Matthew V. Hibbing. 2011. ‘‘The Participatory Personality: and Ernest P. Noble. Boca Raton, FL: CRC Press, Inc., 273–96. Evidence from Latin America.’’ British Journal of Political Science 41 (1): 211–21. Carney, Dana R., John T. Jost, Samuel D. Gosling, and Jeff Potter. 2008. ‘‘The Secret Lives of Liberals and Conservatives: Person- Mondak, Jeffery J., and Karen D. Halperin. 2008. ‘‘A Framework ality Profiles, Interaction Styles, and the Things They Leave for the Study of Personality and Political Behaviour.’’ British Behind.’’ Political Psychology 29 (6): 807–40. Journal of Political Science 38 (2): 335–62. Costa, Paul T., and Robert R. McCrae. 1992. NEO PI-R. Mondak, Jeffery J., Matthew V. Hibbing, Damarys Canache, Professional Manual. Odessa, FL: Psychological Assessment Mitchell A. Seligson, and Mary R. Anderson. 2010. ‘‘Person- Resources, Inc. ality and Civic Engagement: An Integrative Framework for the Study of Trait Effects on Political Behavior.’’ American Dalton, Russell J. 2008. ‘‘Citizenship Norms and the Expansion of Political Science Review 104 (1): 85–110. Political Participation.’’ Political Studies 56 (1): 76–98. Mussen, Paul H., and Anne B. Wyszynski. 1952. ‘‘Personality and Denny, Kevin, and Orla Doyle. 2008. ‘‘Political Interest, Political Participation.’’ Human Relations 5 (1): 65–82. Cognitive Ability and Personality: Determinants of Voter Turnout in Britain.’’ British Journal of Political Science 38 Mutz, Diana C. 2002. ‘‘The Consequences of Cross-Cutting (2): 291–310. Networks for Political Participation.’’ American Journal of Political Science 46 (4): 838–55. Feldman, Stanley, and Leonie Huddy. 2005. ‘‘Racial Resentment and White Opposition to Race-Conscious Programs: Principles or Ozer, Daniel J., and Vero´nica Benet-Martinez. 2006. ‘‘Personality Prejudice?’’ American Journal of Political Science 49 (1): 168–83. and the Prediction of Consequential Outcomes.’’ Annual Review of Psychology 57 (1): 401–21. Fowler, James H. 2006. ‘‘Altruism and Turnout.’’ Journal of Politics 68 (3): 674–83. Paunonen, Sampo V., and Michael C. Ashton. 2001. ‘‘Big Five Factors and Facets and the Prediction of Behavior.’’ Journal of Fowler, James H., and Cindy Kam. 2007. ‘‘Beyond the Self: Personality and Social Psychology 81 (3): 524–39. Altruism, Social Identity, and Political Participation.’’ Journal of Politics 69 (3): 813–27. Plomin, Robert, John C. DeFries, Gerald E. McClearn, and Peter McGuffin. 1990. Behavioral Genetics: A Primer. New York: Gerber, Alan S., Donald P. Green, and Christopher W. Larimer. W.H. Freeman & Company. 2008. ‘‘Social Pressure and Voter Turnout: Evidence from a Large-Scale Field Experiment.’’ American Political Science Riker, William H., and Peter C. Ordeshook. 1968. ‘‘A Theory of Review 102 (1): 33–48. the Calculus of Voting.’’ American Political Science Review 62 (1): 25–42. Gerber, Alan S., Gregory A. Huber, David Doherty, Conor M. Dowling, and Shang E. Ha. 2010. ‘‘Personality and Political Rosenstone, Steven J., and John Mark Hansen. 1993. Mobilization, Attitudes: Relationships across Issue Domains and Political Participation, and Democracy in America. New York: MacMillan. Contexts.’’ American Political Science Review 104 (1): Sniderman, Paul M. 1975. Personality and Democratic Politics 111–33. Berkeley: University of California Press. Gerber, Alan S., Gregory A. Huber, David Doherty, Conor M. Stenner, Karen. 2005. The Authoritarian Dynamic. Cambridge: Dowling. 2011. ‘‘Personality Traits in the Political Arena.’’ Cambridge University Press. Annual Review of Political Science 14: 265–87. Ulbig, Stacy G., and Carolyn L. Funk. 1999. ‘‘Conflict Avoidance Gosling, Samuel D. 2008. Snoop: What Your Stuff Says about You. and Political Participation.’’ Political Behavior 21 (3): 265–82. New York: Basic Books. Van Gestel, Sofie, and Christine Van Broeckhoven. 2003. ‘‘Ge- Gosling, Samuel D., Peter J. Rentfrow, and William B. Swann, Jr. netics of Personality: Are We Making Progress?’’ Molecular 2003. ‘‘A Very Brief Measure of the Big-Five Personality Psychiatry 8 (10): 840–52. Domains.’’ Journal of Research in Personality 37 (6): 504–28. Vavreck, Lynn. 2007. ‘‘The Exaggerated Effects of Advertising on Hetherington, Marc J., and Jonathan D. Weiler. 2009. Authori- Turnout: The Dangers of Self-Reports.’’ Quarterly Journal of tarianism and Polarization in American Politics. Cambridge: Political Science 2 (4): 325–43. Cambridge University Press. Vecchione, Michele, and Gian Vittorio Caprara. 2009. ‘‘Person- Jackman, Simon, and Lynn Vavreck. 2009. The 2008 Cooperative ality Determinants of Political Participation: The Contribu- Campaign Analysis Project, Release 2.1. Palo Alto, CA: tion of Traits and Self-Efficacy Beliefs.’’ Personality and YouGov/Polimetrix, Inc., [producer and distributor]. Individual Differences 46 (4): 487–92.  706 alan gerber et al. Alan S. Gerber is the Charles C. and Dorathea S. Conor M. Dowling is a Postdoctoral Associate at Dilley Professor of Political Science and Faculty in the Institution for Social and Policy Studies, Yale Residence, Institution for Social and Policy Studies, University, New Haven, CT 06520. Yale University, New Haven, CT 06520. Connor Raso is a Law Clerk, United States Court Gregory A. Huber is a Professor of Political Science of Appeals for the Second Circuit, New York, and Faculty in Residence, Institution for Social and NY 10007. Policy Studies, Yale University, New Haven, CT 06520. Shang E. Ha is an Assistant Professor of Political David Doherty is an Assistant Professor of Political Science at Brooklyn College, City University of New Science at Loyola University Chicago, Chicago, IL 60660. York, Brooklyn, NY 11210. View publication stats