It’s Who’s on the Inside that Counts: Campaign Practitioner Personality and Campaign Electoral Integrity

Nai’s (2018) article comes the closest to evaluating the effect of personality on campaign tactics. In his research, Nai shows that expert evaluations of both the personality of candidates in 57 countries and the negativity of campaigns are correlated. While a positive step forward, it is difficult to know how well these are actually related to each other given the reliance on outsiders’ opinions of both campaign negativity and personality. Given the reliance on expert classifications for both measures, it is difficult to identify the causal relationship between these two variables as it is possible that experts’ evaluations of personality are influenced by the campaigns’ perceived negativity. Moreover, previous research has shown that there is a lack of convergence between citizens’ perceptions of campaign negativity and social science-style classifications of these campaigns (Sigelman and Kugler 2003).

There is also weak evidence that agreeableness decreases the propensity of an individual to become involved in campaign activity outside of being a candidate (Gerber et al. 2011).

Most scholars would not argue that political elites involved have no differences in their approaches to political situations, and particularly political campaigns. Rather, part of the motivation for this particular approach has been the difficulty in surveying political elites to better measure individual factors that might have a strong influence on decision making. As a result, the focus of work on political campaigns has largely been that of understanding the structural and institutional factors and incentives that affect decision making.

Some recent work has suggested that neuroticism, extraversion, conscientiousness, and openness decline slightly as an individual ages (Graham et al. 2017).

It should be noted that Gerber et al. (2011) look at an aggregated measure that consists of donating to campaigns, volunteering for campaigns, wearing buttons or stickers, and attending rallies. With the exception of volunteering for campaigns, these items are slightly different from working on a campaign as a paid or volunteer staffer. However, it is not unlikely that the propensity to engage in these behaviors is likely correlated with interest in working on a campaign. Moreover, the personality traits of candidates, professional staff, and volunteer staff are not significantly different from each other with the exception of agreeableness where candidates actually have significantly higher levels of agreeableness (2.99) than paid campaign staffers (2.47) (see Table A.1 in the online appendix for more details). If, as previous research has shown, agreeableness is negatively correlated with political ambition, it appears that negative correlation is even stronger for working on a campaign.

In their study of members of Congress, Ramey et al. (2017) find that agreeableness is correlated with higher levels of partisanship and argue that agreeableness reflects the orientation of agreeableness to the fellow members of their party. However, the orientation of members of Congress towards their caucus is different from candidates and political operatives in the campaign environment who are thinking more about the campaign, their opponent, and voters. Regardless, analysis of an interaction of agreeableness by an individual’s role in the campaign (staff or candidate) shows no differences across groups from what is presented below.

Nai (2018) finds that experts who evaluate a candidate as higher in openness to experience are also likely to indicate that the campaign which that candidate ran was higher in negativity, but provides no theoretical expectations behind such a finding. Moreover, given the difficulty in extracting causality (i.e. it is possible that candidates who are more negative are viewed as having certain personality traits) it is difficult to determine the validity of these arguments.

The survey of congressional campaigns conducted by Druckman et al. (2009) has a sample of 31% incumbents, 53% challengers, and 15% open seats. The differences between the incumbents and open seat candidates in this sample and Druckman et al. sample are likely the result of when during the election cycle the surveys were fielded. This survey was in the field before the date of the primary election when more challenger candidates were still competing, while the survey conducted by Druckman et al. was fielded in October after the primary had eliminated a large number of open seat candidates and challengers.

The percentage of responses given by candidates (as opposed to paid and volunteer staffers) may seem high, but it is not surprising given the large number of challengers in the data set and given that the survey was conducted during the primary election campaign period. According to Herrnson (2012), up to 41% of challenger campaigns in the general election are managed by the candidate.

One might note that a large portion of these responses come from third party candidates who may be fundamentally different from major party candidates. Indeed, roughly 1/6 of the challengers are third party challengers and just under half of all responses that were completed by candidates themselves are from third party candidates. This raises obvious questions as to whether the results may be skewed because of the presence of their party candidates. On the one hand, third party candidates appear to run for office for different reasons than major party candidates (Canon 1990). On the other hand, previous surveys show that many of these candidates were active participants in major party politics previously, and that third party candidate have a wealth of experience (about 10 years on average) in major party practical politics (Collet et al. 1996). Indeed, in this sample, 50% of respondents working in political campaigns indicated they had previous campaign experience. Regardless, as noted in footnote 17, the personality profile of respondents from these campaigns is not significantly different from respondents who work on major party candidates’ campaigns (see Table A.2 in the online appendix). Moreover, as noted below, if we exclude responses that came from individuals working on third party campaigns, we find little significant differences on the conclusions we are able to draw here.

Once possible concern with self-reported personality tests taken by elected officials is social desirability bias in their responses (Remmel 2016). Individuals, especially candidates, may be reluctant to admit to negative traits. Remmel (2016) examines this concern by conducting a Big Five personality test on Vermont state legislators and then comparing these self-reports to peer-reports of those same state legislators completed by their friends and family and finds a strong correlation between the self-reports and peer-reports with state legislators more likely to agree with the negative statements than their peers.

Although some research has suggested that observer-ratings of agreeableness (Marcus and Schütz 2005) are correlated with higher response rates, the majority of research suggests limited or no impact of personality on non-response bias (Gershen and McCreary 1983; Johnson and Mowrer 2000; Uriell et al. 2007).

The questions the survey used are based on the questions Theilmann and Wilhite (1998) used negative advertising. One major difference, however, is that while some of Theilmann and Wilhite’s (1998) scenarios were dependent on previous scenarios [i.e. the small lead scenario was described to respondents as “a substantial lead but smaller than before” (1055) referring to the large lead described in the previous scenario and some respondents even explicitly indicated they were considering the changing dynamics of the campaign rather than just the lead or deficit of the candidate had in the scenario as “five respondents said they would stop spending money or would have already been fired if they had managed to blow such a big lead” (1055)], these are not. These questions also varied in that they included variation in the number of undecided voters for the scenarios where the candidate had a big lead or was far behind in the polls.

Analyzing each scenario individually does not have an effect on the substance of the conclusions about the effect of personality on negativity.

While almost all respondents recognized that “removing opponent yard signs and replacing them with your own” was unethical (average response of 0.15 on a zero to four scale), there was more variation on other responses. “Recruiting individuals to post positive comments about the candidate on newspaper articles” was deemed most acceptable with an average rating of 2.58 on the same zero to four scale. The questions about recruiting individuals to do something for the campaign were the three scenarios that respondents viewed as most ethical.

Because negativity is correlated with challenger status and losing campaigns (Druckman et al. 2009; Hassell and Oeltjenbruns 2016), one concern is that the results that we see are the result of certain types of campaigns being more likely to attract certain personality traits. Analysis of the responses suggests that this is not the case. There are no significant differences in the personality traits of those individuals working on campaigns that lost the primary (often by a sizeable amount) and those working on campaigns that won their primary election. There are also no significant differences in personality traits between campaign practitioners who work for incumbent candidates and those who work for non-incumbent candidates (although agreeableness approaches significance (p < 0.09) with political practitioners who work for non-incumbents being slightly more agreeable). Regardless, however, running models for just incumbents or just non-incumbents shows the same effect of agreeableness on the willingness to engage in campaign negativity (although the small sample size among incumbents limits the ability to find statistical significance).

Similar to the point made in the previous footnote, there are also no significant differences in the personality traits of individuals working for third party candidates and those working for major party candidates (although both agreeableness and conscientiousness approach significance (p < 0.12). These results are available in Table A.2 in the online appendix.

Analyzing the two different scenarios separately shows no significant or substantive differences in the relationship between the traits and respondents’ support for negative campaigning. The relationship between the traits does not vary across the two situations.

As with negativity, there may be concerns that certain types of campaigns (which may be more risk-averse ore risk-accepting) attract certain types of individuals. As mentioned in footnote 16, analysis of the responses suggests that this is not the case. However, re-running the models with only incumbents or only non-incumbents results in nearly identical outcomes to those shown in Tables 6 and 7.