150817 the big five, party identification, and voting behavior in … · 2015-08-17 · 1 the big...

25
1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr Center for Military History and Social Sciences Zeppelinstrasse 127/128 D-14471 Potsdam [email protected] Paper prepared for delivery at the 9 th ECPR General Conference, Université de Montréal, Montreal, Canada, August 26-29, 2015. Panel: Personality and Voting.

Upload: others

Post on 21-Jul-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

1  

The Big Five, party identification, and voting behavior in Germany

Markus Steinbrecher

Bundeswehr Center for Military History and Social Sciences

Zeppelinstrasse 127/128

D-14471 Potsdam

[email protected]

Paper prepared for delivery at the 9th ECPR General Conference, Université de Montréal, Montreal, Canada, August 26-29, 2015.

Panel: Personality and Voting.

Page 2: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

2  

Abstract:

The Michigan Model is still the most important model for explaining voting behavior in many

countries. This paper will look at the link between personality (measured by the Big Five), party

identification, the short-term components of the Michigan Model and voting behavior in a German

federal election. It will thus analyze the psychological base of partisan attachments, candidate, and

issue orientations as well as the direct and indirect impact (via all three attitudes) of personality traits

on voting in a multi-party system. The analysis accordingly will focus on a causal relationship already

suggested by the “funnel of causality” in “The American Voter”. It will also address the short-comings

of present research in this area which has more or less ignored the complex causal interplay between

personality, political attitudes, and voting behavior.

Page 3: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

3  

1. Introduction

The Michigan Model is still the most important model for explaining voting behavior in many

countries. Germany is no exception (e.g., Rattinger et al. 2011; Schmitt-Beck et al. 2014; Weßels et al.

2014). Party identification, candidate, and issue orientation are usually among the strongest predictors

of party choice. “The American Voter” introduced the concept of the “funnel of causality” to

summarize the factors influencing the three essential components of the Michigan Model (Campbell et

al. 1960). The funnel includes predictors like socio-demographic characteristics, political

communication, socialization processes, and personality traits. However, Campbell and his colleagues

stated that “it has apparently not been easy to demonstrate that personality factors discriminate

between Democrats and Republicans. […] In sum, there is little evidence to make us expect much

continuity between deeper personality factors and partisan choice” (Campbell et al. 1960, 506-507). In

contrast to this quite pessimistic perspective from the 1960s, recent research on the impact of

personality traits on political behavior and political attitudes has surged, mainly due to the

establishment of reliable and valid instruments to gauge the Big Five personality traits in large

population surveys and promising results using them for research problems in political science.

Accordingly, researchers have looked both at the importance of personality for party identification

(e.g., Bakker et al. 2015; Gerber et al. 2011, 2012; Mondak 2010; Mößner 2005; Schoen & Schumann

2005) or party evaluations (e.g., Schoen & Schumann 2007; Schumann 2001, 2002), the short-term

orientations of the Michigan Model (e.g., Mondak & Halperin 2008), and electoral behavior (e.g.,

Gerber et al. 2009; Schoen & Schumann 2007; Schumann 1990, 2014). The available research on

these issues suffers from several shortcomings though: First, most of these analyses are based on the

US, a polity with a presidential and a two-party system. Citizens in a parliamentary multi-party system

with a different political culture like Germany will likely react differently to external stimuli and thus

personality traits might exhibit different effects on the attitudes of the Michigan Model or voting

behavior. Second, the available research just focuses on the importance of personality for party

identification or the impact of the Big Five on voting behavior, hence just looking at the direct effects

of personality traits. Accordingly, these approaches usually ignore the exact causal interplay between

personality, party identification, the short-term components of the Ann Arbor Model and voting

behavior. It is highly plausible that personality will not have a direct impact on the decision to vote for

a party, but that it will rather influence it indirectly via party identification, issue, and candidate

orientations. This paper will thus provide an added value for research on the importance of personality

characteristics for political attitudes and behavior. It will look at the link between personality

(measured by the Big Five), party identification, the short-term components of the Michigan Model

and voting behavior in German federal elections and will hence analyze the psychological base of

partisan attachments, candidate, and issue orientations as well as the direct and indirect impact (via all

Page 4: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

4  

three attitudes) of personality traits on voting in a multi-party system against the background of the

2013 federal election in Germany.

The paper will be structured as follows. The second section will provide the theoretical background

and the hypotheses for the analysis. It will start with a short summary of the Big Five, continue with a

look at the three components of the Michigan Model with a special focus on party identification, and

will then come up with hypotheses regarding the total, direct, and indirect effects of the Big Five

personality traits on voting behavior (via party identification, candidate orientation, and issue

orientation). The third section will focus on data from the GLES cross section survey 2013 and will

shortly discuss the measurement and operationalization of the relevant variables. The analysis in

section four will provide evidence regarding the different kinds of causal effects and will mainly

concentrate on results from a path model that both allows for direct and indirect effects (via the three

central attitudes) of the Big Five on voting behavior. The last section will summarize the findings and

discuss further implications for future research on personality, partisan attachments, and voting

behavior in Germany and beyond.

2. Theory and hypotheses

Since the 1990s, the trait-based paradigm has emerged as the leading approach in personality

psychology. Methodologically-diverse research led to an agreement that the Five-Factor-Model, also

known as the Big Five, is the most appropriate concept to describe personality in various cultures (e.g.,

Goldberg 1993, John et al. 1988; John 1990; Saucier & Goldberg 1996; McCrae & Costa 1997).

Personality traits are not merely descriptive dimensions, but have explanatory power. This conclusion

rests on the finding that the five factors have a biophysical basis (e.g., McCrae & Costa 1995, 238,

248) and have motivational implications. They play a role in influencing the stimuli a person perceives

as relevant in her environment, the goals a person pursues and how she responds to external stimuli

(e.g., Costa & McCrae 1988; Luk & Bond 1993; Jost et al. 2003). They are thus causally prior to

values and attitudes that emerge from and are shaped by the interaction of personality traits and

environmental stimuli (McAdams & Pals 2006; McCrae & Costa 1996). This means personality traits

will influence which ideas and experiences individuals will find more or less attractive in all kinds of

settings, but also in politics (Gerber et al. 2012, 654).

The five personality factors are agreeableness, conscientiousness, extraversion, emotional stability (or

neuroticism), and openness. To begin with, agreeableness refers to trust, straightforwardness, altruism,

compliance, modesty, and tender-mindedness. High scorers on this trait are thus characterized as being

altruistic, trusting, generous, soft-hearted, and sympathetic, while low scorers are suspicious, hard-

hearted, and demanding. Conscientiousness mainly refers to impulse control that is socially prescribed

so that persons at the high end of this scale are thorough, organized, industrious, ambitious,

Page 5: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

5  

resourceful, and enterprising, whereas their counterparts at the lower end are immature, impatient,

lazy, careless, and moody. Extraversion comprises warmth, gregariousness, positive emotions, and

assertiveness. Thus, extraverts are upbeat, energetic, active, friendly, talkative, and assertive, while

introverts are reserved or even shy. Emotional stability chiefly refers to controlling negative emotions

like anxiety, depression, anger, discontent, and irritation. Finally, openness refers to tolerance of

diversity, broadness of one’s own cultural interest, and exploration of novelty. As a result, persons

who score high on this dimension are curious, imaginative and original, while persons who exhibit low

scores are mild, cautious, and conservative (e.g., Costa & McCrae 1989, 1992; Mondak 2010).

Before I turn to the relevance of personality traits for party identification, candidate, and issue

orientations, a short look at the Michigan Model of electoral behavior seems to be adequate: The

social-psychological model was developed by a research group around Angus Campbell at the

University of Michigan in Ann Arbor in the 1950s. In several steps (Campbell et al. 1954, 1960) they

arrived at a version of their model which has been used in the large majority of applications afterwards

and thus became known as the adopted version of the Michigan Model (Schoen & Weins 2014: 255-

262). In the adopted version the model comprises three core attitudes: party identification, candidate

orientations, and issue orientations. Party identification is a long-term affective connection to a

specific party which has a central position in individual attitude and belief systems. Party

identifications are a result of long-term socialization processes and are usually adopted during

childhood. There is a high likelihood that party identification is transferred from parents to their

offspring. Thus, socialization, not genetic transmission or personality, is the main mechanism behind

the development of party identification. One reason for this assumption by Campbell and his

colleagues (1960), however, might be that personality psychology was dominated by psycho-

analytical approaches and the authoritarian personality when “The American Voter” was published.

The established trait-based approach provides a much better representation of personality and

eventually offers new answers regarding the impact of personality on party identification (Mondak

2010). In addition, prior research shows that personality is associated with policy preferences, political

ideology (e.g., Gerber et al. 2011; Mondak 2010; Mondak & Halperin 2008), and party identification

(e.g., Bakker et al. 2015; Gerber et al. 2011, 2012; Mondak 2010; Mößner 2005). Thus, it makes sense

to expect personality effects on the core attitudes of the Michigan Model.

More arguments for the relevance of personality for party identifications might be gleaned from the

affective and cognitive functions of party identification for individuals. Party identification serves as a

perceptual screen and as a central heuristic that both influences other political attitudes (like the short-

term orientations in the Michigan Model) as well as information-processing and decision-making.

Following arguments by Gerber et al. (2012), Mößner (2005) as well as Schoen and Schumann (2005),

I assume that people with certain personality characteristics will find a party identification more or less

attractive. Some individuals will like that a party identification helps them to reduce complexity in the

Page 6: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

6  

political world, thus focusing on the cognitive benefits of a party identification. This might apply to

people low on openness or emotional stability. Others might find a party identification attractive for

affective reasons as it can provide a sense of belonging to a certain group. This aspect might also be

appealing for people with certain personalities. Also, individuals might assign certain characteristics to

a specific party that might be attractive for people with a certain personality, too. Thus, personality

should also influence which party citizens will feel attached to.

The same might be true for the short-term components, issue and candidate orientations. Regarding

candidates, citizens will assign certain personality characteristics to politicians or politicians will

present their personalities in a certain way in the media or during an electoral campaign. Following the

arguments of the attraction paradigm, people will find those candidates to be more attractive who

mirror their personality or whose personality they evaluate positively (Byrne 1971; Schoen &

Schumann 2005). Accordingly, personality will have an impact on candidate preferences. Regarding

issue orientations, personality might drive which issues or problems a citizen regards to be relevant or

most important, e.g., someone with a high level of conscientiousness will try to avoid uncertainty and

promote order, so that he might be more supportive of law and order politics (Schoen & Schumann

2007, 475). Since opinions among the electorate on the ability to solve these problems are quite stable

over time and parties are partly evaluated through the lense of an individual’s personality (see above),

personality traits will also have an impact on issue orientations.

Based on these considerations it can be expected that all three components of the Michigan Model will

somehow be influenced by personality. Since party identification as a central political attitude partly

shapes evaluations of candidates and political issues, the effect of personality on the short-term

orientations will be mediated by party identification. The same is true for voting behavior. It is highly

unlikely that personality will have a direct effect on the decision to vote for a party if strong attitudinal

antecedents of this decision are taken into account. Accordingly, I expect personality to have indirect

effects on voting behavior via party identification, candidate orientation, and issue orientation. The

exact causal interplay between the different variables and concepts in the analysis is reflected in Figure

1.

As indicated in the introduction there has only been some research on the impact of personality on the

components of the Michigan Model and voting behavior in Germany. Analyses on the US usually find

that conscientious and emotionally stable people tend to identify with the Republicans, while open and

agreeable citizens are more likely to identify with the Democrats (Gerber et al. 2009, 2011, 2012 (in

the two latter cases with the exception of agreeableness); Mondak 2010; Mondak & Halperin 2008).

The findings regarding candidate orientations are very similar: Conscientiousness and emotional

stability are positively correlated with higher evaluations of Republican candidates, while openness

and agreeableness correlate with preferences for Democratic candidates (Gerber et al. 2009; Mondak

& Halperin 2008). However, the extent and sometimes the statistical significance of these effects is

Page 7: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

7  

dependent on the candidates running in a specific election because of their particular profile.

Considering two-party vote as dependent variable, conscientious, emotionally stable and close-minded

people tend to vote for the Republicans (Gerber et al. 2009).

Figure 1: Causal connections between personality, party identification, short-term orientations, and voting behavior

The German party system is much more complex than the US system and this is already one reason

why findings for the US cannot be transferred directly to the German case. I will just focus on the five

most relevant parties here: CDU and CSU are Christian Democratic parties which form a joint faction

in the German parliament. They are center right parties which combine conservative, liberal (in the

European sense), but also social democratic positions. The FDP is a liberal party with respect to civil

rights and clearly prefers free market solutions for economic issues. The Greens share the FDP’s

position on civil rights, but have a clear preference for state intervention in the economy. The main

rationale behind that is their strong stance on environmental protection and related issues. The SPD is

a center-left social democratic party which favors state intervention and redistributional policies. The

Left is a result of several developments. First, they are the leftover of the former state party of the

GDR. Second, they were formed by left-leaning people who left the SPD out of dissatisfaction with

political decisions under chancellor Gerhard Schröder. Accordingly, they are very much in favor of

state intervention and dominance in economic and other policies.

Page 8: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

8  

The characteristics of the five personality traits and the German parties as well as the results of

previous analyses on Germany indicate the following impact of personality on identification with the

different parties (Table 1): Identification with the CDU should be more common among those who are

emotionally stable, conscientious, and less open. Since findings on extraversion are mixed and there is

no evidence on effects of agreeableness, there are no hypotheses regarding these traits (Bakker et al.

2015; Mößner 2005; Schoen & Schumann 2005). Identifiers with the FDP should show similar

personality traits like identifiers with the Christian Democrats on emotional stability and

conscientiousness. Liberal identifiers also seem to display higher levels of extraversion, while findings

for openness and agreeableness regarding the Liberals are mixed (Bakker et al. 2015; Mößner 2005;

Schoen & Schumann 2005). Thus, I do not assume a direct effect regarding the latter traits. Turning to

the Social Democrats, identifiers with this party should be less conscientious and more extraverted,

while findings for the other traits either do not appear to be consistent across studies and

operationalizations or there is no effect at all (Bakker et al. 2015; Mößner 2005; Schoen & Schumann

2005). Sympathizers with the Greens should show a clearer profile regarding their personality: They

can be expected to be more open, less conscientious and seem to have higher levels of emotional

stability (Bakker et al. 2015; Mößner 2005; Schoen & Schumann 2005). Finally, identifying with The

Left comes along with higher levels of openness, lower levels of emotional stability, and lower levels

of conscientiousness (but see Mößner 2005). Findings for the other traits vary between different

analyses (Bakker et al. 2015; Mößner 2005; Schoen & Schumann 2005). Schoen and Schumann

(2007) present relatively similar findings regarding partisan attitudes (based on feeling thermometers).

Analyses on the short-term orientations have been quite rare for Germany. Schoen and Schumann

(2005) report that conscientious and less open people sympathize more with the chancellor candidate

of the CDU in the federal election of 2002, while those low on conscientiousness and high on

agreeableness tend to support the SPD-chancellor candidate. However, this result is very likely

dependent on the specific chancellor candidates of both parties, Schröder and Stoiber, in the 2002

election. The top candidates of the five parties in the 2013 federal election also have a distinct

personality profile that should make them attractive to voters with similar personality characteristics.

The main problem in this case is that reliable information on the personality of the candidates in the

2013 federal election using the Big Five is not available. For the sake of simplicity, I thus assume that

voters will more or less attribute similar characteristics to parties and the respective candidates.

Accordingly, the hypotheses and expectations for party identification and candidate orientations will

be identical. The same might be true with respect to issue orientations. This is even more plausible,

because issue orientations are strongly based on long-lasting issue ownership and perceptions about

the parties in general. If this simplifying assumption is true, party identification should be the central

mediator of personality effects on voting behavior.

Page 9: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

9  

Table 1: Hypotheses regarding direct and indirect effects of personality on attitudes of the Michigan Model and voting behavior

Emotional stability

Extraversion Conscien-tiousness

Agreeable-ness

Openness

Direct effects on party identification CDU + 0 + 0 - SPD 0 + - 0 0 FDP + + + 0 0 Greens + 0 - 0 + The Left - 0 - 0 + Direct effects on candidate orientation CDU + 0 + 0 - SPD 0 + - 0 0 FDP + + + 0 0 Greens + 0 - 0 + The Left - 0 - 0 + Direct effects on issue orientation CDU + 0 + 0 - SPD 0 + - 0 0 FDP + + + 0 0 Greens + 0 - 0 + The Left - 0 - 0 + Direct effects on voting behavior (if controlled for party identification, candidate, and issue orientation All parties 0 0 0 0 0 Indirect effects on voting behavior via party identification CDU + 0 + 0 - SPD 0 + - 0 0 FDP + + + 0 0 Greens + 0 - 0 + The Left - 0 - 0 + Indirect effects on voting behavior via candidate orientation CDU + 0 + 0 - SPD 0 + - 0 0 FDP + + + 0 0 Greens + 0 - 0 + The Left - 0 - 0 + Indirect effects on voting behavior via issue orientation CDU + 0 + 0 - SPD 0 + - 0 0 FDP + + + 0 0 Greens + 0 - 0 + The Left - 0 - 0 +

Moving on to voting there has been even less research on the impact of personality on this central

behavioral aspect in Germany, with Schoen and Schumann’s contributions being the only exceptions

(2005, 2007). The main finding, however is, that personality almost has no direct effect on voting

behavior once central attitudinal predictors of voting are taken into account. This result is not

surprising at all if the high level of predictive power of the Ann Arbor Model and the long causal

distance between personality and political behavior in the funnel of causality are considered.

Page 10: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

10  

Accordingly, I hypothesize both from a theoretical and an empirical perspective that personality

should not exert a direct effect on the decision to vote for a party. Rather, personality will indirectly

influence voting via party identification and the short-term attitudes. Based on the above reasoning I

expect these indirect effects via the central elements of the Michigan Model to be similar to the direct

effects of personality on the three attitudes.

3. Data and operationalization

The analyses will be based on data collected around the 2013 German federal election. I will use the

pre-and-post-election cross-section survey which is part of the German Longitudinal Election Study

(GLES).1 Fieldwork lasted from July 29 until September 21, 2013 (pre-election) and from September

23 until December 23, 2013 (post-election). For the present purpose this cross-section data set

comprises 3,722 respondents (out of 3,911 overall interviews). It is based on a household survey for

which households were selected by random-route. Within households, the respondents were chosen

according to the Kish-selection grid and then interviewed via CAPI. The analyses combine pre- and

post-election participants and ensure for proportional representation of East and West Germans

according to their share in the German population. The weight used additionally corrects for different

selection probabilities due to household size.

The dependent variable of interest in this paper is the decision to vote in the 2013 German federal

election. For pre-election respondents this is the intention to vote for a party, for post-election

respondents it is the respondents’ recall of their electoral behavior after the election. Party

identification and its strength are measured with the German standard instrument, first asking for a

general affiliation with a specific party and then asking for the affiliation’s intensity on a five-point

scale. Candidate orientations are based on 11-point feeling thermometers towards the chancellor or the

top candidate of the respective party which are collapsed to a scale between -1 and +1. Issue

orientations are operationalized using an index on a party’s perceived ability to solve the most

important and the second most important problem in Germany.

Given the time constraints of multi-purpose election surveys, it is impossible to utilize full-fledged

trait inventories, like the NEO-FFI (Five Factor Inventory), to capture the Big Five. Instead,

researchers have to employ short measures of the five factors (Gosling et al. 2003; Rammstedt & John

2007; Rammstedt 2007). In the GLES survey respondents were presented one item per trait only (see

Appendix for wording in German and English). This is rather unusual as a ten-item version of the Big

Five has been established as a standard instrument in research on effects of personality traits.

However, analyses by Gosling et al. (2003) and Woods and Hampson (2005) show that the Big Five                                                             1 Rattinger, Hans; Roßteutscher, Sigrid; Schmitt-Beck, Rüdiger; Weßels, Bernhard; Wolf, Christof 2014. Pre- and post-election cross-section (cumulation) (GLES 2013). GESIS Data Archive, Cologne: ZA 5702 Data File Version 1.0.0, doi: 10.4232/1.11891.

Page 11: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

11  

can be measured adequately with data on only one trait factor per dimension. Despite these

conclusions, I have to allow for measurement issues when it cannot be ruled out, particularly because

Gerber and his colleagues (2011) have shown that the impact of the Big Five on attitudinal and

behavioral variables depends on the Big Five measures applied. Since the personality traits just cover

specific facets of the respective dimension of personality, it has to be kept in mind for the

interpretation of the results that deviations of the presented findings from those of other researchers

might be related to the suboptimal operationalization of the core variables.2 Table 2 shows descriptive

statistics for all relevant variables in the analysis.

Table 2: Descriptive statistics on variables in the analysis

Variable Mean/% sd Min Max N CDU-vote 27.02 - 0 1 3,911 SPD-vote 19.87 - 0 1 3,911 FDP-vote 2.71 - 0 1 3,911 Greens-vote 9.94 - 0 1 3,911 The Left-vote 6.22 - 0 1 3,911 Strength PID CDU 0.19 0.33 0 1 3,911 Strength PID SPD  0.14 0.28 0 1 3,911 Strength PID FDP  0.01 0.08 0 1 3,911 Strength PID Greens  0.06 0.20 0 1 3,911 Strength PID The Left  0.04 0.16 0 1 3,911 Candidate orientation CDU 0.24 0.59 -1 1 3,849 Candidate orientation SPD  0.05 0.50 -1 1 3,777 Candidate orientation FDP  -0.22 0.47 -1 1 3,562 Candidate orientation Greens  -0.13 0.51 -1 1 3,674 Candidate orientation The Left  -0.07 0.54 -1 1 3,719 Issue orientation CDU 0.28 0.41 0 1 3,911 Issue orientation SPD  0.21 0.37 0 1 3,911 Issue orientation FDP  0.01 0.08 0 1 3,911 Issue orientation Greens 0.07 0.21 0 1 3,911 Issue orientation The Left  0.07 0.23 0 1 3,911 Emotional stability 0.69 0.25 0 1 3,870 Extraversion 0.54 0.27 0 1 3,884 Conscientiousness 0.83 0.19 0 1 3,879 Agreeableness 0.56 0.25 0 1 3,884 Openness 0.68 0.24 0 1 3,833 Age 49.84 19.04 18 99 3,911 Education 1.88 0.83 1 3 3,841 Male 0.49 - 0 1 3,911 East 0.21 - 0 1 3,911

                                                            2 As all variables in the analysis are measured in the same survey, readers might criticize that the analysis does not demonstrate the temporal antecedence of the independent variable and correlations might be inflated due to respondents’ inclination to give consonant answers. However, personality is assumed to be stable over time and the survey items that tap personality traits have no apparent political content. Accordingly, inflated estimates due to consistency effects are not particularly likely.

Page 12: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

12  

In the analysis, I will estimate total, indirect, and direct effects of personality traits on the individual

decision to vote for a party via (strength of) party identification, candidate, and issue orientation, while

controlling for social structural variables, including gender, age, education (measured by an ordinal

scale), and place of residence (East/West). Different methods like logistic, ordinal logistic, and OLS-

regressions as well as path models were used to disentangle total, direct, and indirect effects of the Big

Five. All models were calculated using Stata 13.

Some comments have to be provided regarding assumptions and limitations of the analysis. The

arguments follow those by Schoen and Steinbrecher (2013) very closely: In particular, I utilize the so-

called Baron-Kenny (1986) method to examine whether potential mediators serve as intervening

variables between personality traits and voting behavior. This method of mediation analysis rests on

several strong assumptions (Bullock et al. 2010; Bullock & Ha 2011; Coffman 2011; Green et al.

2010; Imai et al. 2011). Concerning the causal order, personality traits are assumed to shape political

attitudes that affect the decision to vote. Building on the above discussion of the notions that attitudes

serve as characteristic adaptations and that personality traits do not affect behavior directly, but via

attitudes, this assumption appears to be reasonable. Moreover, to interpret findings causally I assume

that there is neither an unobserved variable nor random measurement that drives variation in

personality traits, attitudes, and turnout. By using the Baron-Kenny method, the assumption is that

there are no unobserved causes of voting behavior which are positively correlated with the mediators

included in the analysis. Otherwise, the results of the analysis will be biased in favor of the hypothesis

that the included potential mediators serve as intervening variables (e.g., Bullock et al. 2010; Green et

al. 2010). This kind of bias cannot be completely ruled out.

4. Analysis

The following empirical analysis will move forward stepwise, from simpler to more complex models.

I will start with models that try to explain the different endogenous variables in the model. These

models will use some kind of regression technique to predict the decision to vote for a specific party,

strength of party identification with a specific party, candidate orientations, and issue orientations. The

final model will be a path model which takes account of the causal interrelations between personality,

political attitudes, and voting behavior and thus looks at the mediation effect of political attitudes. All

models will be run for the five most important German parties CDU/CSU, SPD, FDP, Greens, and The

Left.

Starting with the models on voting behavior (Table 3), I differentiate between a base and a full model.

While the base model just includes personality traits and some relevant socio-demographic

characteristics, the full model additionally includes the three components of the Michigan Model, thus,

making up for a strong test of the relevance of personality traits in explaining the decision to vote. If

Page 13: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

13  

the mediation hypothesis is correct, all direct effects of personality traits should disappear in the full

model. Regarding the base model, personality is particularly relevant for predicting a vote for the

Christian Democrats: As expected before, emotionally stable, conscientious, and close-minded people

are more likely to cast a ballot for CDU and CSU. Introverted people are also more likely to vote for

the CDU. Looking at the other parties, people high on extraversion are more likely to elect the Social

Democrats, while personality traits are not relevant at all for explaining a vote for the FDP, even in the

base model. The Greens are more attractive for people with high levels of agreeableness and openness,

as expected before. The base model for The Left only shows one significant effect of personality traits:

A high level of extraversion is beneficial for casting a vote for this party. Moving on to the full model,

there are still three significant effects of personality traits among the five party-specific analyses.

Thus, although controlling for strong predictors of party choice, extraversion and agreeableness are

still relevant explanations for voting behavior and not all direct effects of personality are mediated by

the three attitudinal variables. Interestingly, the effects of extraversion on voting for SPD and the

Greens even get stronger in the full model. While extraverted people are more likely to cast a vote for

the Social Democrats, the effect of this personality trait is the other way round in the model for the

Green party. In addition, Green voters also stick out because of their high level of agreeableness.

As Tables 4 to 6 show, personality is a relevant predictor of all three attitudinal components of the

Michigan Model among all five parties. A stronger party identification with the Christian Democrats is

more likely among introverted, conscientious, and close-minded people (Table 4), thus supporting the

hypotheses on conscientiousness and openness. People high on extraversion and agreeableness are to

be found among strong SPD-identifiers. Stronger identification with the Liberals can be characterized

by low levels of agreeableness and high levels of openness. These results do not match my ex-ante

expectations at all. The effect of openness is positive for the Greens. At the same time, these people

are more agreeable. A higher level of extraversion once again is relevant for developing a stronger

affiliation with The Left. All in all, only about 50 percent of the hypotheses for party identification

with the five parties are supported by the results.

Page 14: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

14  

Table 3: Explanation of voting behavior in 2013 German federal election

Variable Base model Full model CDU SPD FDP Greens The Left CDU SPD FDP Greens The Left Strength PID - - - - - 3.31c 3.67c 4.64c 3.60c 3.28c Candidate orientation - - - - - 1.72c 1.21c 2.05c 1.63c 2.93c Issue orientation - - - - - 1.49c 1.67c 2.30b 2.26c 2.96c Emotional stability 0.41a -0.18 -0.26 0.45 -0.22 0.50 -0.50 0.06 0.29 -0.07 Extraversion -0.39a 0.51a 0.22 -0.47 0.78a -0.20 0.62a 0.36 -0.83a 0.18 Conscientiousness 1.05c 0.36 0.14 -0.61 -0.28 -0.26 0.15 0.22 -0.26 -0.15 Agreeableness 0.25 0.29 -0.51 1.18c 0.28 0.34 -0.36 -0.61 0.93a 0.52 Openness -0.67b 0.07 -0.33 1.03b 0.34 -0.25 0.09 -0.60 0.61 -0.46 Age 0.02c 0.01c 0.02c -0.02c -0.01 0.01 0.01a 0.01 -0.01 -0.02a Education 0.21c -0.05 0.44b 0.41c 0.15 0.08 -0.08 0.20 0.21 0.30 Male -0.09 0.19 0.14 -0.25 0.11 -0.13 0.03 -0.06 0.01 0.00 East 0.10 -0.24a -0.46 -0.49b 1.38c 0.58c 0.27a -0.50 -0.62b 0.69b Constant -3.15c -2.70c -5.31c -2.73c -3.60c -4.19c -3.33c -4.55c -3.43c -4.53c N 3,722 3,722 3,722 3,722 3,722 3,669 3,608 3,432 3,522 3,558 Nagelkerke’s R2 0.07 0.03 0.03 0.11 0.08 0.65 0.56 0.30 0.51 0.64 Note: unstandardized coefficients from logistic regressions, levels of significance: a: p < 0.05, b: p < 0.01, c: p < 0.001.

Page 15: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

15  

Table 5 presents the results on candidate orientations. The patterns for this attitude are more complex

than for issue orientations and thus show that party-related heuristics seem to be less relevant when

forming opinions about politicians or top candidates, respectively: Introverted, conscientious,

agreeable, and close-minded individuals are more likely to evaluate Angela Merkel (the CDU-

candidate in 2013) better, while Peer Steinbrück, the candidate of the Social Democrats, is rated better

by those individuals who are higher on agreeableness. Rainer Brüderle of the FDP gets better

evaluations from those who are more conscientious, more agreeable, and less open. Agreeableness is

the only personality trait which has a significant impact on ratings of Jürgen Trittin, the top candidate

of the Greens, who is evaluated more positively by those who are more agreeable. Evaluations of

Gregor Gysi, top candidate of The Left, are dependent on four out of five personality traits:

Extraverted, less conscientious, agreeable, and open people are more likely to provide better

evaluations for him.

Explanatory patterns look very similar for the models on issue orientations (Table 6) when comparing

them with those on strength of party identification. The only difference is the size of the effects. This

results in a lack of statistical significance for the extraversion effect on SPD-issue preference and the

agreeableness effect on Greens-issue preference. The similarity of the results might indicate that the

processes which link personality and both party identification and issue orientation are very similar.

Considering the theoretical and empirical dependence of issue orientations on party identification a

host of indirect effects of personality on vote choice can be expected for the final path model from

these findings.

Table 4: Explanation of strength of party identification in 2013 German federal election

CDU SPD FDP Greens The Left Emotional stability 0.38 -0.02 -0.78 0.33 -0.50 Extraversion -0.49b 0.44a -0.34 -0.09 1.01b Conscientiousness 1.59c 0.25 -1.09 -0.85 -0.17 Agreeableness -0.03 0.70c -1.39b 0.71a -0.12 Openness -0.70c 0.09 1.55b 1.35c 0.49 Age 0.02c 0.02c 0.03b -0.03c 0.00 Education 0.12a -0.09 0.93c 0.58c -0.05 Male 0.00 0.26b 0.42 -0.55b 0.19 East -0.31c -0.60c -0.45 -0.19 1.47c N 3,722 3,722 3,722 3,722 3,722 Nagelkerke’s R2 0.08 0.04 0.09 0.12 0.07 Note: unstandardized coefficients from ordinal logistic regressions, levels of significance: a: p < 0.05, b: p < 0.01, c: p < 0.001.

Page 16: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

16  

Table 5: Explanation of candidate orientation in 2013 German federal election

CDU SPD FDP Greens The Left Emotional stability -0.06 -0.03 -0.04 -0.04 -0.09 Extraversion -0.14a -0.03 -0.05 0.03 0.09a Conscientiousness 0.22b 0.06 0.12a -0.01 -0.18b Agreeableness 0.16b 0.22c 0.12b 0.12b 0.15b Openness -0.18b -0.01 -0.14b 0.02 0.16b Age 0.01c 0.00 0.00 -0.004c 0.00 Education 0.12c 0.06c 0.01 0.04a 0.05b Male -0.05a 0.07c 0.01 -0.04 0.02 East 0.01 -0.03 -0.03 0.01 0.30c Constant -0.33c -0.26c -0.26c -0.02 -0.24b N 3,669 3,608 3,432 3,522 3,558 Adjusted R2 0.08 0.03 0.01 0.04 0.07 Note: unstandardized coefficients from OLS- regressions, levels of significance: a: p < 0.05, b: p < 0.01, c: p < 0.001.

Table 6: Explanation of issue orientation in 2013 German federal election

CDU SPD FDP Greens The Left Emotional stability 0.24 0.19 -0.65 0.13 -0.19 Extraversion -0.33a 0.26 -0.19 0.04 0.90b Conscientiousness 1.34c 0.02 -0.43 -0.63 0.06 Agreeableness -0.02 0.61c -1.22a 0.54 0.26 Openness -0.37a 0.08 1.38a 1.22c 0.46 Age 0.02c 0.01c 0.03b -0.03c 0.00 Education 0.08 -0.13a 0.81c 0.65c 0.00 Male 0.00 0.21a 0.27 -0.42b 0.31a East -0.47c -0.68c -0.58a -0.44b 1.62c N 3,722 3,722 3,722 3,722 3,722 Nagelkerke’s R2 0.07 0.05 0.07 0.13 0.09 Note: unstandardized coefficients from ordinal logistic regressions, levels of significance: a: p < 0.05, b: p < 0.01, c: p < 0.001.

All in all, up to this point, there has been a host of evidence for the relevance of personality in

explaining voting behavior, party identification, candidate, and issue orientations. However, the

evidence presented looked at direct effects of personality on the respective attitudes and political

behavior. The full model explaining the decision to vote already showed that personality effects on

voting are rather indirect than direct, because almost all effects of personality traits on the decision to

vote disappeared when controlling for the three essential attitudinal predictors of voting behavior.

The path models presented in Table 7 take into account this more complex causal interplay. The

models work very well when looking at CFI and SRMR, but are beyond acceptable levels when

focusing on the RMSEA. There are also big differences in explanatory power between the five parties

Page 17: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

17  

if one considers the different R2-statistics. Obviously, the model does not work well for explaining a

vote for the FDP in the 2013 German federal election.

Looking at the different sections in Table 7, there are almost no direct effects of personality on voting

behavior, as expected before the analysis. The only exceptions are found in the models for SPD and

the Greens: Extroverted people are more likely to vote for the SPD, while they are less likely to cast a

ballot for the Green Party. In addition, higher levels of agreeableness are beneficial for the latter party.

Personality traits exert a host of total effects on the vote, though. Considering the almost complete lack

of significant direct effects, this means that personality traits influence voting behavior first and

foremost indirectly via party identification and the two short-term attitudes. Regarding total effects on

voting behavior, I find the same patterns than in the base model for voting behavior displayed in Table

3. This means that voting for CDU, SPD, Greens, and The Left is indirectly influenced by personality

traits. High levels of emotional stability and conscientiousness as well as low levels of extraversion

and openness are on total beneficial for voting for the Christian Democrats. People scoring high on

extraversion are more likely to vote for Social Democrats and The Left on total. Finally, agreeableness

and openness have positive total effects on voting for the Greens.

Moving on to the part of Table 7 on strength of party identification, it becomes clear that party

identification itself is a major mediator for personality effects on voting behavior. This is particularly

true for the CDU/CSU-model where I find significant positive effects of emotional stability and

conscientiousness as well as negative effects of extraversion and openness. High levels of

agreeableness and extraversion are conducive to identifying with the Social Democrats, while there are

small significant effects of agreeableness (negative) and openness (positive) on FDP-identification.

Identifiers with the Greens are more agreeable and more open, while those who identify with The Left

score higher on extraversion.

Despite including party identification in the sub-models on both short-term orientations, there is still a

host of significant direct effects of personality on issue and candidate orientations. In case of candidate

orientations there is usually not a big difference between total and direct effects, meaning that

personality has a direct impact on candidate orientations and party identification is not an important

mediator for personality effects in that case. The only exception is the sub-model for Angela Merkel.

For all personality traits except for agreeableness there is a big difference between total and direct

effects. This implies that party identification with the Christian Democrats is a more important

mediator for the impact of personality traits on evaluations of Angela Merkel than for the other

candidates and parties. This might also explain the much better model fit for CDU and CSU compared

to the other parties (R2=.32).

Moving to issue orientations, findings are similar to the previous sub-models on party identification

and candidate orientations regarding conscientiousness and openness effects in the model for the

Page 18: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

18  

Christian Democrats. Both personality traits have a rather indirect impact via party identification,

indicated by the difference between total and direct effects. The same is true for the effect of

agreeableness on issue orientations towards the Social Democrats, and the impact of openness on the

issue variable for the Greens. In addition, the very limited number of significant effects of personality

traits on issue orientations for all parties shows the high relevance of party identification as predictor

of issue orientations. All in all, party identification is definitely the essential mediator of effects by

personality traits on voting behavior in the 2013 German federal election. Accordingly, the statement

by Campbell et al. on the relevance of personality cited above (1960) cannot be supported for

Germany in 2013.

Page 19: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

19  

Table 7: Explanation of voting behavior and Michigan attitudes in path models by party for German federal election 2013

CDU SPD FDP Greens The Left Total Direct Total Direct Total Direct Total Direct Total Direct Vote Strength PID 0.94c 0.64c 0.93c 0.68c 0.85c 0.69c 0.93c 0.66c 0.96c 0.65c Candidate orientation - 0.10c 0.09c 0.04c 0.07c 0.05c Issue orientation - 0.25c 0.24c 0.29b 0.33c 0.33c Emotional stability 0.08a 0.05 -0.03 -0.04 -0.01 0.01 0.04 0.03 -0.01 0.01 Extraversion -0.08a -0.01 0.08a 0.05a 0.01 0.01 -0.04 -0.04a 0.05a 0.00 Conscientiousness 0.19c -0.03 0.07 0.02 0.00 0.00 -0.05 -0.01 -0.02 0.01 Agreeableness 0.05 0.03 0.05 -0.04 -0.01 -0.01 0.10c 0.05a 0.02 0.03 Openness -0.14c -0.02 0.01 0.01 -0.01 -0.01 0.08b 0.03 0.02 -0.01 Age 0.004c 0.00 0.002c 0.001a 0.0007b 0.0003a -0.002c 0.00 0.00 0.00 Education 0.04b 0.01 -0.01 -0.01 0.01a 0.00 0.03b 0.01 0.01 0.01 Male -0.02 -0.01 0.03a 0.00 0.00 0.00 -0.03 0.00 0.00 0.00 East 0.02 0.06c -0.04b 0.02a -0.01a -0.01 -0.04b -0.03b 0.11c 0.03b Strength PID Emotional stability -  0.06a 0.00 -0.01 0.02 -0.02 Extraversion -  -0.07a 0.05a 0.00 -0.01 0.04b Conscientiousness -  0.22c 0.04 -0.01 -0.05 0.00 Agreeableness -  -0.01 0.08b -0.01a 0.04a -0.01 Openness -  -0.10b 0.01 0.02b 0.06b 0.02 Age -  0.003c 0.002c 0.0004b -0.001c 0.00 Education -  0.02a -0.01 0.01c 0.02c 0.00 Male -  0.00 0.03a 0.00 -0.03b 0.01 East -  -0.05c -0.06c 0.00 -0.01 0.07c Candidate orientation Strength PID - 0.91c 0.66c 0.86c 0.75c 0.97c Emotional stability -0.06 -0.12a -0.03 -0.03 -0.04 -0.03 -0.04 -0.05 -0.09 -0.08 Extraversion -0.14b -0.08 -0.03 -0.06 -0.05 -0.05 0.03 0.04 0.09a 0.05 Conscientiousness 0.22b 0.02 0.06 0.03 0.12a 0.13a -0.01 0.02 -0.18b -0.18b

Page 20: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

20  

Agreeableness 0.16b 0.17c 0.22c 0.17c 0.12b 0.13b 0.12a 0.09a 0.15b 0.16c Openness -0.18b -0.09 -0.01 -0.01 -0.14b -0.16c 0.02 -0.02 0.16b 0.14b Age 0.006c 0.004c 0.00 0.00 0.00 0.00 -0.004c -0.003c 0.00 0.00 Education 0.12c 0.10c 0.06c 0.07c 0.01 0.01 0.04a 0.02 0.05b 0.05c Male -0.05a -0.05a 0.07c 0.05b 0.01 0.01 -0.04 -0.02 0.02 0.01 East 0.01 0.05b -0.03 0.01 -0.03 -0.03 0.01 0.01 0.30c 0.24c Issue orientation Strength PID - 0.83c 0.77c 0.41c 0.64c 0.80c Emotional stability 0.01 -0.03 0.02 0.01 -0.01a -0.01 0.02 0.01 -0.01 0.01 Extraversion -0.05 0.01 0.03 -0.01 0.01 0.01 0.00 0.00 0.05b 0.03 Conscientiousness 0.23c 0.05 0.04 0.01 -0.01 0.00 -0.05 -0.02 -0.05 -0.06a Agreeableness 0.02 0.04 0.07a 0.01 0.00 0.00 0.06c 0.04b -0.03 -0.02 Openness -0.14c -0.06a 0.01 0.01 0.00 -0.01 0.05a 0.01 0.03 0.02 Age 0.003c 0.00 0.00 -0.001b 0.00 0.00 -0.002c -0.0008b 0.00 0.00 Education 0.04b 0.02a -0.01 0.00 0.01a 0.00 0.01 -0.01 -0.01 0.00 Male 0.00 0.00 0.02 0.00 0.01a 0.00 -0.01 0.00 0.00 -0.01 East -0.01 0.03b -0.07c -0.03a 0.00 0.00 -0.03b -0.02b 0.06c 0.01 N 3,669 3,608 3,432 3,522 3,558 R2 overall 0.14 0.07 0.03 0.10 0.10 R2 vote 0.54 0.48 0.20 0.42 0.51 R2 strength PID 0.07 0.03 0.02 0.06 0.03 R2 candidate orientation 0.32 0.17 0.03 0.12 0.16 R2 issue orientation 0.45 0.35 0.18 0.37 0.32 RMSEA 0.32 0.22 0.08 0.13 0.22 CFI 0.95 0.97 0.99 0.99 0.97 SRMR 0.02 0.02 0.01 0.01 0.02 Note: unstandardized coefficients from path model, levels of significance: a: p < 0.05, b: p < 0.01, c: p < 0.001.

Page 21: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

21  

5. Conclusion and discussion

The goal of this paper was to look at the link between personality (measured by the Big Five), party

identification, the short-term components of the Michigan Model and voting behavior in the 2013

German federal election. By using a stepwise approach, first looking at the direct effects of personality

traits on the different attitudinal and behavioral variables and then moving on to a final path model

including total, direct, and indirect effects of personality, the results of the analyses show that

personality rather has an indirect effect on the decision to vote. Party identification serves as the main

mediator for effects of personality, because it affects both voting behavior and the short-term attitudes.

The analyses show that this mediation model works differently for the five most relevant German

parties. Voting for the FDP was not influenced at all by personality in the 2013 German federal

election, while a vote for all other parties was influenced by personality. Extraversion seems to be a

particularly relevant personality trait with either significant total or direct effects for all other parties.

Those who are more extraverted are less likely to vote for the CDU and the Greens. However, high

levels of extraversion have a positive effect on a decision to vote for the Social Democrats and The

Left. Openness is relevant for explaining voting behavior for at least two parties. People high on

openness are less likely to elect the CDU, but they are more likely to favor the Greens on Election

Day. The other traits are just relevant for one party. Those who are more conscientious and more

emotionally stable are more likely to cast a vote for the Christian Democrats, while those scoring high

on agreeableness have a higher probability to vote for the Greens. All in all, these results show that it

makes sense to look at personality as an important causal predictor of both voting behavior and

relevant political attitudes.

However, this paper just used data on a specific German federal election. Further empirical tests are

necessary to evaluate whether the findings presented in this paper are typical or are just dependent on

the specific election, the specific set of parties and candidates, the campaign, and the

operationalization of the Big Five. The latter point might be particularly relevant since using just one

item per trait implies a strong focus on specific facets of the respective dimension of personality. The

rather unorthodox measurement of personality in the GLES cross-section survey thus might be a

reason for results that contradict the hypotheses presented in section 2. Also, future analyses should

pay attention to the specific personality of candidates and characteristics of party programs in order to

come up with more specific hypotheses for candidate and issue orientations. Assuming the same

causal correlations for all three attitudes is probably too simplistic. Accordingly, there are many

potential avenues for future research on the relevance of the Big Five for political attitudes and

political behavior in Germany and beyond.

Page 22: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

22  

References

Bakker, Bert N., Hopmann, David N., and Persson, Mikael 2015. Personality traits and party identification over time. In: European Journal of Political Research 54, 197-215.

Baron, Reuben M. and Kenny, David A. 1986. The moderator-mediator variable distinction in social-psychological research: Conceptual, strategic, and statistical considerations. In: Journal of Personality and Social Psychology 51, 1173-82.

Bullock, John G., Green, Donald P., and Ha, Shang E. 2010. Yes, But What’s the Mechanism? (Don’t Expect an Easy Answer). In: Journal of Personality and Social Psychology 98, 550-58.

Bullock, John G. and Ha, Shang E. 2011. Mediation Analysis Is Harder than It Looks. In: Druckman, James N. et al. (eds.), Cambridge Handbook of Experimental Political Science. New York: Cambridge University Press, 508-521.

Byrne, Donn 1971. The attraction paradigm. New York: Academic Press.

Campbell, Angus, Gurin, Gerald, and Miller, Warren E. 1954. The voter decides. Evanston, IL: Row, Peterson, and Company.

Campbell, Angus et al. 1960. The American Voter. New York: John Wiley & Sons.

Coffman, Donna L. 2011. Estimating causal effects in mediation analysis using propensity scores. In: Structural Equation Modeling 18, 357-369.

Costa, Paul T., Jr. and McCrae Robert R. 1988. From catalog to classification: Murray’s needs and the five-factor model. In: Journal of Personality and Social Psychology 55, 258-265.

Costa, Paul T., Jr. & McCrae, Robert R. 1989. The NEO PI/FFI Manual Supplement. Odessa, Florida: Psychological Assessment Resources.

Costa, Paul T., Jr. & McCrae, Robert R. 1992. NEO PI-R. Professional manual. Odessa, FL: Psychological Assessment Resources.

Gerber, Alan S. et al. 2009. Personality and Political Behavior. Unpublished Typescript, Yale University, New Haven, CT.

Gerber, Alan S. et al. 2011. The Big Five Personality Traits in the Political Arena. In: Annual Review of Political Science 14, 265-87.

Gerber, Alan S. et al. 2012. Personality and the Strength and Direction of Partisan Identification. In: Political Behavior 34, 653-688.

Goldberg, Lewis R. 1993. The structure of phenotypic personality traits. In: American Psychologist 48, 26-34.

Gosling, Samuel D., Rentfrow, Peter J., and Swann, William B., Jr. 2003. A Very Brief Measure of the Big Five Personality Domains. In: Journal of Research in Personality 37, 504-528.

Green, Donald P., Ha, Shang E., and Bullock, John G. 2010. Enough Already about “Black Box” Experiments: Studying Mediation Is More Difficult than Most Scholars Suppose. In: Annals of the American Academy of Political and Social Science 628, 200-208.

Page 23: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

23  

Imai, Kosuke et al. 2011. Unpacking the Black Box of Causality: Learning about Causal Mechanisms from Experimental and Observational Studies. In: American Political Science Review 105, 765-789.

John, Oliver P. 1990. The "Big Five" Factor Taxonomy. Dimensions of Personality in the Natural Language and in Questionnaires. In: Pervin, Lawrence A. (ed.), Handbook of Personality. Theory and Research. New York: Guildford Press, 66-100.

John, Oliver P., Angleitner, Alois, and Ostendorf, Fritz 1988. The lexical approach to personality: A historical review of trait taxonomic research. In: European Journal of Personality 2, 171-203.

Jost, John T. et al. 2003. Political conservatism as motivated social cognition. In: Psychological Bulletin 129, 339-375.

Luk, Chung L. and Bond, Michael H. 1993. Personality variation and values endorsement in Chinese university students. In: Personality and Individual Differences 14, 429-437.

McAdams, Dan P. and Pals, Jennifer L. 2006. A new Big Five: Fundamental principles for an integrative science of personality. In: American Psychologist 61, 204-217.

McCrae, Robert R. and Costa, Paul T., Jr. 1995. Trait explanations in personality psychology. In: European Journal of Personality 9, 231-252.

McCrae Robert R. and Costa Paul T., Jr. 1996. Toward a new generation of personality theories: Theoretical contexts for the five-factor model. In: Wiggins, Jerry S. (ed.), The five-factor model of personality: Theoretical perspectives. New York: Guildford Press, 51-87.

McCrae, Robert R. and Costa, Paul T., Jr. 1997. Personality trait structure as a human universal. In: American Psychologist 52, 509-516.

Mondak, Jeffery J. 2010. Personality and the Foundations of Political Behavior. Cambridge et al.: Cambridge University Press.

Mondak, Jeffery J. and Halperin, Karen .D. 2008. A Framework for the Study of Personality and Political Behaviour. In: British Journal of Political Science 38, 335-62.

Mößner, Alexandra 2005. Typisch Parteiidentifizierer? Parteiidentifikation und Persönlichkeit. In: Schumann, Siegfried and Schoen, Harald (eds.) 2005, Persönlichkeit. Eine vergessene Größe der empirischen Wahlforschung. Wiesbaden: VS – Verlag für Sozialwissenschaften, 77-91.

Rammstedt, Beatrice 2007. The 10-Item Big Five Inventory. Norm Values and Investigation of Sociodemographic Effects Based on a German Population Representative Sample. European Journal of Psychological Assessment 23, 193-201.

Rammstedt, Beatrice & John, Oliver P. 2007. Measuring personality in one minute or less: A 10-item short version of the Big Five Inventory in English and German. In: Journal of Research in Personality 41, 203-212.

Rattinger, Hans et al. 2011. Zwischen Langeweile und Extremen: Die Bundestagswahl 2009. Baden-Baden: Nomos.

Saucier, Gerard and Goldberg, Lewis R. 1996. The Language of Personality. Lexical Perspectives on the Five-Factor Model. In: Wiggins, Jerry S. (ed.), The Five-Factor Model of Personality: Theoretical Perspectives. New York, London: Guildford Press, 21-50.

Page 24: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

24  

Schmitt-Beck, Rüdiger et al. 2014. Zwischen Fragmentierung und Konzentration: Die Bundestagswahl 2013. Baden-Baden: Nomos.

Schoen, Harald and Schumann, Siegfried 2005. Missing Links? Zur Position von Persönlichkeitsfaktoren in Modellen zur Erklärung von Wahlverhalten. In: Falter, Jürgen W., Gabriel, Oscar W., and Weßels, Bernhard (eds.), Wahlen und Wähler. Analysen aus Anlass der Bundestagswahl 2002. Wiesbaden: VS Verlag für Sozialwissenschaften, 388-413.

Schoen, Harald and Schumann, Siegfried 2007. Personality Traits, Partisan Attitudes, and Voting Behavior. Evidence from Germany. In: Political Psychology 28, 471-497.

Schoen, Harald and Steinbrecher, Markus 2013. Beyond total effects: Exploring the interplay of personality and attitudes in affecting turnout in the 2009 German federal election. In: Political Psychology 34, 533-552.

Schoen, Harald and Weins, Cornelia 2014. Der sozialpsychologische Ansatz zur Erklärung von Wahlverhalten. In: Falter, Jürgen W. and Schoen, Harald (eds.), Handbuch Wahlforschung, 2nd revised edition. Wiesbaden: Springer VS, 241-329.

Schumann, Siegfried 1990. Wahlverhalten und Persönlichkeit. Opladen: Westdeutscher Verlag.

Schumann, Siegfried 2001. Persönlichkeitsbedingte Einstellungen zu Parteien. Der Einfluss von Persönlichkeitseigenschaften auf Einstellungen zu politischen Parteien. München, Wien: Oldenbourg.

Schumann, Siegfried 2002. Prägen Persönlichkeitseigenschaften Einstellungen zu Parteien? In: Kölner Zeitschrift für Soziologie und Sozialpsychologie 54, 64-84.

Schumann, Siegfried 2014. Persönlichkeit und Wahlverhalten. In: Falter, Jürgen W. and Schoen, Harald (eds.), Handbuch Wahlforschung, 2nd revised edition. Wiesbaden: Springer VS, 591-624.

Weßels, Bernhard et al. (eds.) 2014. Voters on the Move or on the Run? Oxford et al.: Oxford University Press.

Woods, Stephen A. and Hampson, Sarah E. 2005. Measuring the Big Five with Single Items Using a Bipolar Response Scale. In: European Journal of Personality 19, 373-390.

Page 25: 150817 The Big Five, party identification, and voting behavior in … · 2015-08-17 · 1 The Big Five, party identification, and voting behavior in Germany Markus Steinbrecher Bundeswehr

25  

APPENDIX Wording of the Big Five in the GLES cross-section survey 2013:

German version (Rammstedt & John 2007):

Bitte sagen Sie mir für jede der folgenden Aussagen auf dieser Liste, inwieweit sie auf Sie zutrifft. Benutzen Sie dazu bitte die Skala.

(A) Ich bin eher zurückhaltend, reserviert.

(B) Ich schenke anderen leicht Vertrauen, glaube an das Gute im Menschen.

(C) Ich erledige Aufgaben gründlich.

(D) Ich habe eine aktive Vorstellungskraft, bin phantasievoll.

(E) Ich werde leicht nervös und unsicher.

(1) trifft überhaupt nicht zu, (2) trifft eher nicht zu, (3) teils/teils, (4) trifft eher zu, (5) trifft voll und ganz zu.

English translation (own literal translation):

Please tell me to what extent you think the statements in the following list accurately describe you. Please use the scale for this purpose.

(A) I am rather cautious and reserved.

(B) I trust others easily, I believe in the good in humans.

(C) I perform tasks very thoroughly.

(D) I have an active imagination, and am imaginative.

(E) I get nervously and insecure easily.

(1) strongly disagree, (2) tend to disagree, (3) neither agree nor disagree, (4) tend to agree, (5) strongly agree.

The English translation differs somewhat from Rammstedt & John’s (2007) English version of the items which are evidently not literal translations.