Terrorism, Gender, and the 2016 Presidential Election (Tables)

Terrorism, Gender, and the 2016 Presidential Election Abstract: The 2016 U.S. election provides the opportunity to assess how gender, party, and experience shape candidate evaluations when terrorist threat is elevated. Hillary Clinton was the first woman major party nominee; Donald Trump was the first major party nominee without political experience; and terrorism was salient. Our core thesis is that security threats dampen public confidence in Demo-cratic female political leaders, yet an experience advantage could countervail against those tenden-cies. We test expectations using the 2016 ANES and two experimental studies. We first affirm that individuals worried about terrorism held lower evaluations of Clinton and higher evaluations of Trump. We then test an active manipulation of the salience of national security experience and find that it mitigates Clinton’s disadvantage, but only in the absence of a counter-message. The results underscore the difficulty that Democratic female candidates face in overcoming the nega-tive influence of party and gender stereotypes when running for office in times of terrorism threat. 1.1. Introduction Terrorism punctuated the 2016 U.S. presidential election, from the primaries through the general election. Attacks at home (e.g., San Bernardino and Orlando) and abroad (e.g., Brussels, London, and Paris) unsettled the public. Political rhetoric cemented terrorist threat as a key electoral issue on which the candidates sought to distinguish themselves. Democratic candidate Hillary Clinton argued that “…defeating global terrorist networks and protecting the homeland takes… a real plan, real experience and real leadership. Donald Trump lacks all three.”1 Speaking to the issue at the Republican National Convention, Rudy Giuliani stated that Donald Trump would “keep us safe and help us achieve and embrace our greatness.”2 How do heightened conditions of terrorist threat affect candidate evaluations? Females tend to fare worse under elevated national security threats, because of stereotypes that diminish confidence in their competency (e.g., Falk and Kenski 2006; Gordon 2001; Little et al. 2007; Holman, Merolla, and Zechmeister 2011, 2016; but see Schneider and Bos 2014). Yet, certain factors can countervail or amplify that outcome. One is partisanship; Republican party “ownership” of national security issues (Petrocik 1996) can leave female Republican leaders comparatively immunized against decreases in the public’s evaluations of their qualities, while Democratic females can be doubly disadvantaged (Holman, Merolla, and Zechmeister 2011, 2016; Bauer 2016). Theoretically, another relevant factor is experience. Extant scholarship suggests that voters evaluate candidates based on information about their occupations and relevant past performance, although it is not always the case that experience is weighted to a significant degree in candidate evaluations (see Funk 1997; Fridkin and Kenney 2011). 1 http://time.com/4355797/hillary-clinton-donald-trump-foreign-policy-speech-transcript/ 2 https://www.bizjournals.com/newyork/news/2016/07/19/rudy-giuliani-storms-the-rnc-praises-good-man.html 1 The 2016 election provides the opportunity to advance research on how terrorist threat, candidate gender, and experience interact: Hillary Clinton was the first woman to compete as a major (Democratic) party nominee and Donald Trump was the first modern major party (Republican) nominee without political experience. During the election, some suggested that terrorist threat could favor Clinton because raising public anxiety could put a premium on threat-relevant experience, which ought to have improved evaluations of someone like former Secretary of State Hillary Clinton (Albertson and Gadarian 2016; Hillin 2016). Indeed, a Washington post poll during the primary showed Clinton with a 54% to 40% advantage over Trump in whom people trust to deal with terrorism.3 Further, a survey we conducted using SSI panelists during the general election showed a large gap in perceived experience between the two candidates in dealing with national security issues – even Republicans perceived Clinton as having more experience.4,5 3 http://wapo.st/1qaeej9?tid=ss_tw-bottom&utm_term=.73be6a896e2f A poll conducted in June showed a similar gap: http://abcn.ws/28Xy4Fl. 4 The data was collected in October of 2016 with panelists from Survey Sampling International. Respondents rated Clinton as having high levels of experience when asked: “Regardless of whether you believe they performed well or not, how much experience do the following leaders have in dealing with national security issues.” They responded on a four-point scale with higher values indicating more experience. Mean experience evaluations for Clinton were high, 3.36, and did not vary greatly between Democrats (mean=3.61) and Republicans (mean=3.07). Meanwhile, mean experience evaluations were much lower for Trump, 1.66, and again varied only slightly between Democrats (mean=1.53) and Republicans (mean=2.09). 5 One might argue that Trump’s hyper-masculinity advantaged him on the issue of terrorism, but we consider this counter-acted to at least some degree by his position as a political outsider, a fact that may have curtailed the extent to which he benefited from the Republican label to the degree that party establishment leaders, such as George W. Bush, have benefited in times of terrorist threat (see Merolla and Zechmeister 2009). 2 But how much did Clinton’s experience matter in evaluations of her and Trump? We assess the relevance of experience by investigating whether experience protected Clinton from the negative public opinion consequences that terrorist threat normally produces for female Democratic candidates. And, given Trump’s lack of experience, we consider whether the Republican male candidate’s lack of experience undercut the normally positive assessments made under the specter of terrorist threat. We consider the influence of experience in two ways: first, as a mere characteristic of the electoral environment and, second, as a manipulated feature of campaign rhetoric captured by the media. As a baseline test of the relevance of gender, party, and experience under heighted concerns about terrorist threat, we look at the correlation between worry about terrorism and candidate evaluations using data from the 2016 American National Election Study (ANES). We find a negative association between worry about terrorism and evaluations of Clinton and a positive association for Trump. Then, with an experiment run via Mechanical Turk, we show that a causal relationship undergirds the correlations in the survey data: priming individuals to think about terrorism decreased evaluations of Clinton and increased evaluations of Trump. These results stand as evidence that Clinton’s comparative experience advantage could not overcome the extent to which her gender and partisanship hampered her candidacy in the terrorist threat-laden electoral context. We then ask whether there is a way to more actively counteract these negative effects by highlighting traits that work to a candidate’s advantage; in this case, Clinton’s experience in national security compared to her opponent’s lack of experience. Via an experimental study conducted online with a national sample in October 2016, we demonstrate that a focus on Clinton’s national security experience and Trump’s weakness on the issue leads to a positive uptick in Clinton’s evaluations, though the effect is slight and dissipates in the presence of counter information. Taken as a whole, these findings speak to factors – experience and terrorist threat – that have received little attention in post-mortems on the election among pundits and academics alike, even though terrorism was salient in the 2016 presidential election and was a focus of both Trump and 3 Clinton’s campaigns. The findings also have broader implications for understanding how candidate gender, partisanship, experience, and the policy environment interact to shape voter evaluations. Specifically, we conclude that Democratic females may be able to shed some of the disadvantage of gender and party when terrorism is salient and they have the requisite experience, but they need to have a strict comparative advantage in campaign messages highlighting that experience. 2.1. Evaluating Leaders when Terrorism is Salient When news of terrorism is salient, individuals seek strong leaders. This tendency is especially consequential for elections: it causes individuals to change their assessments of politicians and place a greater premium on strong leadership qualities in voting decisions (Merolla, Ramos, and Zechmeister 2007; Merolla and Zechmeister 2009a, 2009b). But, how do we know if a particular candidate will be advantaged or disadvantaged when terrorism is salient? Extant scholarship points to the comparative advantage bestowed on politicians to the extent that they are incumbents, male, and running on the Republican ticket (Merolla and Zechmeister 2013). There are also reasons to consider that national security experience could matter, yet this factor has been under-examined in scholarship on the intersection of terrorist threat and candidate evaluations.6 All else equal, female candidates tend to be disadvantaged in a context of terrorist threat. Male political leaders are typically stereotyped as stronger leaders and as more competent on national security issues than females (Bauer 2016, 2017; Cassese and Holman 2017); in line with that tendency, the U.S. voting public favors male leadership during times of national security threat (e.g., Falk and Kenski 2006; Holman, Merolla, and Zechmeister 2011, 2016; Little et al. 2007). Importantly, politician partisanship also matters. The public generally views the Republican Party as more capable on national security issues (Pope and Woon 2009; Petrocik 1996), and thus 6 Incumbency also matters in times of threat (Mueller 1970; Berinsky 2009; Merolla and Zechmeister 2009, 2013), yet this characteristic was not pertinent to the 2016 race. 4 shows a preference for Republican leadership when terrorism is salient (Merolla and Zechmeister 2009, 2013). In theory, the comparative disadvantage of Democrats on national security combines with gender biases to leave Democratic male candidates worse off than Republican male candidates in times of threat, and Democratic female candidates comparatively more disadvantaged (Holman, Merolla, and Zechmeister 2016).7 To the extent that gender and party are dominant in shaping candidate evaluations when terrorist threat is salient, Donald Trump entered the 2016 campaign with two advantages, and Hillary Clinton with two disadvantages. Another factor that could matter, though, is a candidate’s experience in national security affairs. In theory, when terrorism is salient, the public should prefer a candidate with experience in handling national security to one without experience, since experience should signal relevant competency (Albertson and Gadarian 2016; Merolla and Zechmeister 2009). While the public’s ability to assign relevant foreign security experience is yet unexamined to our knowledge, women in office often position themselves to overcompensate on issues of national security to alleviate voters’ concerns (Swers 2013; Windett 2014). Further, the public pays attention to campaigns and can correctly assign relevant experience to the appropriate candidates and news of candidates’ past performance in a relevant domain feeds into assessments of competence (Funk 1997). At the same time, not all scholarship finds a robust link between candidate experience and evaluations (Fridkin and Kenney 2011). If experience in national security affairs matters more than gender or partisanship, the “penalty” that befalls female Democratic leaders would be minimized for experienced leaders. In fact, a study conducted after Clinton took on the role of secretary of state found that evaluations of Clinton were diminished by terrorist threat to a lesser degree than were evaluations of Nancy Pelosi, a female Democratic leader without similar experience in foreign affairs (Holman, Merolla, and Zechmeister 7 The effect on Republican women is less clear, but theoretically they should be protected to some extent by their party affiliation (Holman, Merolla, and Zechmeister 2011, 2016; King and Matland 2003). 5 2017). This suggests that experience might counteract the negative effect of terrorist crises on evaluations of Democratic female politicians. If experience matters, then a lack of experience could also undercut the boost that doubly-advantaged (male, Republican) candidates receive. This discussion leads to two different sets of hypotheses about the linkages between worry about terrorism and candidate evaluations. Absent Clinton’s experience and Trump’s total lack thereof, theoretical priors are that Hillary Clinton (Democratic female) was significantly disadvantaged by the issue of terrorism in the 2016 election (H1a) and Donald Trump (Republican male) was advantaged (H2a). If experience matters and effectively counter-acts sex and partisanship, it ought to wash away the public opinion penalty exerted by terrorist threat on Clinton (H1b), and remove the boost that the issue of terrorism afforded Trump (H2b).8 Is it possible for a Democratic female with significant foreign policy experience to actively combat the tendency toward negative evaluations in times of terrorist threat? Some scholars argue that female candidates act to reduce gender-based biases (Bauer 2014; Dittmar 2015; Windett 2014), including on security issues (Swers 2013).9 Further, information on candidate qualifications can bolster against tendencies to devalue female leadership (Mo 2015). This may be particularly relevant in contexts of terrorist threat, when the public ought to prioritize experienced, competent leadership (Albertson and Gadarian 2015). In short, trumpeting experience in national security could leave a female political leader less susceptible to the biases that privilege masculine leadership when terrorist threat is salient. If so, priming Clinton’s experience in foreign policy will lead to more positive assessments of her (H3). 8 Theoretically, it is possible that experience diminishes the penalty/boost without removing it entirely (Holman, Merolla and Zechmeister 2017); however, the study design is limited in its ability to test for this type of effect. 9 Brooks (2013) argues that the very fact that these individuals are actively positioned in public roles may counteract gender biases because the public comes to value them as “leaders, not ladies.” 6 3.1. Concerns about Terrorism and Candidate Evaluations in 2016 The 2016 election was contested by a female Democratic candidate with experience in national security and a male Republican candidate with no experience in office. The profile of these traits across the candidates provides the opportunity to test the relevance of experience vis-à-vis traits (gender and partisanship) that have been examined more completely in past scholarship. To do so, we turn first to the 2016 American National Elections Study (ANES), which asked 3,640 individuals in the post-election wave, “How worried are you that the United States will experience a terrorist attack in the near future?”10 Respondents could indicate extremely worried (selected by 18%), very worried (23%), moderately worried (34%), slightly worried (19%), or not at all worried (7%). We use this five-point worry about terrorism measure as the primary independent variable in the analyses. The dependent variables are feeling thermometers and leadership evaluations11 and we control for education, income, race, gender, and party identification. We use OLS for the feeling thermometer analyses and leadership analyses.12 The analyses reveal that worry about terrorism is associated with cooler evaluations of Clinton and warmer evaluations of Trump. With respect to feeling thermometer ratings, the effect of worry about terrorism on evaluations of Clinton is negative but not statistically significant, while the effect is significant and positive for Trump (p<0.001); with a maximum effect of almost 11 points on the 100-point thermometer (see Appendix Table 1 for the regression output). For evaluations of strong leadership, worry about terrorism is associated with higher evaluations of Trump’s leadership and lower evaluations of Clinton’s leadership, and both of these effects are statistically significant (see 10 We find comparable results if we use the question, “During the next 12 months, how likely is it that there will be a terrorist attack in the United States? These results are available in the appendix Table 2. 11 ANES respondents were asked how well “she/he provides strong leadership” described Hillary Clinton/ Donald Trump, with five response options: extremely well, very well, moderately well, slightly well, or not well at all. 12 The findings are robust to using ordered logit. See Appendix Table 3. 7 Appendix Table 1). As Figure 1 shows, moving from the lowest to highest level of worry about terrorism increases the leadership evaluation of Trump by .40 points on the five-point scale and decreases the evaluation by .28 points for Clinton. Taking stock of both sets of results, we have mixed findings for Clinton. Worry about terrorism is associated with more negative leadership evaluations, which supports H1a, but is not associated with feeling thermometer ratings, which supports H1b. We have more consistent results for Trump, who does not appear to be penalized for lack of experience, given the positive association between worry about terrorism and both evaluations, which is more consistent with H2a. Figure 1. Predicted Effect of Worry about Terrorism on Candidate Leadership Evaluations Note: Figure values are calculated from post estimation predicted values from OLS models including worry about terrorism and control variables, including party identification. See appendix for full results. 3.2. Experimental evaluation of Terrorism and Candidate Evaluations in 2016 To assess whether there is a causal connection between the salience of terror threat and candidate evaluations, we ran a simple experiment with an adult convenience sample. In April 2017, we conducted a study with 541 Mechanical Turk (MTurk) participants, who were randomly assigned to receive a question about their level of worry about terrorism either before or after reporting their feelings 8 toward the candidates (using the same feeling thermometer as the ANES).13 These questions were included in a longer question battery on a variety of social and political attitudes. The sample’s mean age is 35.5. Like other MTurk samples, the respondents are more male (45.4% female), more Democratic (45.2%; with a strong preference for Bernie Sanders), more educated (modal education is 4-yearcollege), and less diverse (79%) than the general population. The worry treatment group (n = 179) was asked “How worried are you that there could be a violent attack by terrorists in the United States in the next 12 months? Are you very, somewhat, a little, or not at all worried, or would you say that you have not thought much about this?” Following this question, they were asked to rate Bernie Sanders, Hillary Clinton, Elizabeth Warren, Ted Cruz, and Donald Trump on a feeling thermometer question. The control group (n = 362) received the worry question after the candidate thermometer questions. We find that this simple question wording experiment reduced feeling thermometer evaluations for Clinton and increased them for Trump. Asking about worry about terrorism first decreased evaluations of Clinton on the feeling thermometer measure by 7.11 points, as compared to the control group (p=0.017), and increased evaluations of Trump by 5.64 points (p=0.054). In short, results from survey and experiment data are generally more consistent with theoretical predictions and findings from past scholarship on this topic (H1a and H2a): Democratic women are disadvantaged by terrorist threat, while male Republicans are advantaged. Clinton’s experience and her association with the incumbent party do not appear sufficient, in general, to have shielded her from that disadvantage. Furthermore, 13 We had an additional experimental condition that used the word angry instead of worried. We focus on worry in the main text since it overlaps with the ANES question, but the findings are similar for the anger condition (see Appendix Table 4). 9 Trump does not seem to suffer from his lack of experience.14 Clearly, the question of whether experience shapes candidate evaluations warrants additional investigation. What happens when experience is made salient? 4.1. Making Experience Salient: A National Experiment To assess whether highlighting experience in foreign policy shapes candidate evaluations when terrorism is made salient, we fielded an experimental study during the 2016 campaign. 4.2. Experiment Design The study ran online, October 20-21, 2016, with a near-nationally representative sample of 1,192 Survey Sampling International panelists. Participants filled out a brief pre-treatment survey indicating their age, gender, race/ethnicity, and region in which they live. They were then randomly assigned to read one of five news stories, and then filled out a post-treatment survey in which they answered questions about the articles, their perceptions of the two candidates running for office, and for whom they would vote if the election were being held at that point. Descriptive statistics for the sample, by experimental condition, are provided in Appendix Table 5. The sample was drawn to be comparable to the national population on gender, race, and region. The sample leans just left of center. Statistical tests affirm balance across conditions (See Appendix Table 6). Treatment articles were designed with an eye toward external validity by taking content from real news. A non-threat control article presented a story about a dog that had been lost and then found after gaining notoriety on Facebook; the intention of this story was to have a condition that read a news story that did not contain any element of threat (see Appendix for treatment text). The remaining four articles addressed international terrorism. The first, baseline terrorism article reported that terrorist threat looms large. The article highlighted attacks by ISIS overseas in Paris, Nice and Brussels, as well 14 However, we do not have the counterfactual of whether the positive association between terrorism worry and eval- uations of Trump is lower than it would be for a candidate Trump with foreign policy experience. 10 as ISIS-inspired attacks in San Bernardino and Orlando. It also referenced survey data that the U.S. public is worried about terrorism. The treatment mimics those used in the extant literature (Merolla and Zechmeister 2009; Holman, Merolla, and Zechmeister 2016). A threat with Clinton experience article started with the same paragraph on attacks and then added information about Clinton’s plans for confronting terrorism, her experience as Secretary of State, and attacks on Trump’s inexperience. For example, the article quoted Clinton stating “I was involved in a number of efforts to take out Al Qaeda leadership when I was secretary of state, including of course, taking out Bin Laden.” The article discussed Clinton’s claims against Trump: “Donald has consistently insulted Muslims abroad, Muslims at home.... They need to have close working cooperation with law enforcement in these communities, not be alienated and pushed away as some of Donald’s rhetoric unfortunately has led to.” A threat with Trump experience article had the same structure but instead included information from Trump, including his aggressive stance on terrorists and endorsements from military officials, as well as his attacks on Clinton’s experience as ineffective. Importantly, Trump put a negative spin on Clinton’s experience: “Trump agreed that Hillary Clinton has experience, but described it as “bad experience”, going on to state: “Well, President Obama and Secretary Clinton created a vacuum the way they got out of Iraq because they got out -- they shouldn’t have been in but once they got in, the way they got out was a disaster. And ISIS was formed. But they wouldn’t have even been formed if they left some troops behind.” Last, a threat with both experience article had the same introductory paragraph and contained information about both candidates’ experience and approach to terrorism, and condensed versions of their critiques of each other.15 15 The content of the articles was obtained from news about the election and actual quotes. As we wanted to make the treatments externally valid, the treatments contained multiple pieces of information. In particular, all of the treatments 11 What are our expectations? In a general threat condition (baseline threat), we expect negative results (vs. control) for Clinton’s evaluations, and an uptick in Trump’s evaluations according to H1a and H2a, or null effects per H1b and H2b. If priming is effective in bringing certain considerations more to the forefront of one’s mind, then making Clinton’s experience particularly salient (threat with Clinton experience) should boost her evaluations (H3), and may even downgrade assessments of Trump. A corollary to this is that highlighting Trump’s strengths in the area of national security while undercutting Clinton’s experience (threat with Trump experience) should leave her devoid of any countervailing positive effect for experience while boosting evaluations of Trump. Finally, in the case that presents positives and negatives on both candidates ((threat with both experience), we do not have a clear expectation for whether this treatment will increase, decrease, or have null effects on evaluations of Clinton or Trump. One potential feature of the timing of our study to bear in mind is that there is likely a slight bias toward null effects given that it was taking place during the peak of the general election; thus, participants were likely already inundated with campaign information. 4.3. Compliance and Manipulation Checks Immediately following the treatments, we asked about the content of the article to assess compliance. More specifically, those in the control group were asked which animal was discussed in the article, while those in the remaining conditions were asked which issue was discussed. Over 80% of respondents correctly answered the question; therefore, participants were paying attention to the general gist put information on candidate experience into context by providing the candidate’s approach to handling the issue of terrorism. Given that the candidates take different approaches to foreign policy, it may be that this factor, rather than experience alone, is driving reactions to the treatments. That said, the policy stances of both candidates are more similar here than on the experience dimension: both promoted hawkish policies against ISIS. 12 of the articles (see Appendix Table 7).16 Given the high levels of compliance, we analyze the results for the full sample. To assess the effectiveness of the terrorism articles in evoking negative reactions, post-treatment we asked individuals the extent to which they were feeling each of eleven emotions on a fivepoint scale. A factor analysis yields two factors with eigenvalues over one, a positive emotions factor (enthusiastic, hopeful, and proud), and a negative emotions factor (angry, afraid, anxious, worried, hatred, contempt, bitterness and resentful). As expected, individuals in the terrorism conditions report significantly higher negative emotions and lower positive emotions compared to those in the control (see Appendix Figure 1). 4.4. Direct Effects of the Treatments on Leadership and Experience Evaluations We first assess whether the treatments had direct effects on perceptions of leadership and experience. In the post-treatment survey, we asked participants for their level of agreement (on a seven-point scale) with the following statement for each candidate: Hillary Clinton/Donald Trump will provide strong leadership for the nation. We begin with this question as past scholarship has shown that terrorist threat influences leadership evaluations (Merolla and Zechmeister 2009, 2013), and because the measure was assessed in our analyses of the 2016 ANES data. 16 We had one other question to assess compliance, but realized that the questions we asked for the candidate experi- ence conditions were a bit ambiguous. The control group and terrorism only conditions were asked to recall very clearly presented facts from each article (the dog’s favorite song and results from a poll from PEW about terrorism concerns), and over 70% answered correctly. For the candidate experience conditions, we asked: “Which allies did Clinton/Trump say the U.S. needs to work with to defeat ISIS?” Only about 40% correctly answered the question. However, the response options did not match up perfectly with the content in the treatments, which may have confused participants. If we use a looser standard of identifying at least one of the allies mentioned, we find XXX. 13 In the control group, the mean perception of Clinton’s strong leadership qualities is above the neutral point, at 4.31, whereas the mean evaluation of Trump’s leadership is slightly lower, closer to the somewhat disagree stance, at 3.15. In short, the sample as a whole perceived Clinton slightly more favorably on this dimension, which may reflect the more left-leaning sample.17 We regress each seven-point leadership evaluation on dummy variables for the treatment, with the control group serving as the baseline. We illustrate the results in Figure 2. Reading about terrorism alone has little effect on either candidate’s leadership evaluations. This result differs from that found in the two earlier studies (ANES and MTurk), and is more consistent with H2a and H2b, that knowledge of Clinton’s experience and Trump’s lack of experience neutralized any penalty or boost that they may have received from terrorist threat due to gender and partisanship. However, the null result could be due to the timing of our study. By mid-October, terrorism had become very salient in the campaign and it is possible that the saturated nature of the actual campaign environment minimized effects in an experimental setting since the control group was also being “treated” by the ongoing campaign.18 As expected, those who read the condition that featured a discussion of terrorism and Clinton’s experience (threat with Clinton experience) are more likely to perceive her as a strong leader compared to respondents in the control group. On the other hand, reading about Trump’s credentials and his critique of Clinton’s experience (threat with Trump experience) has no effect on Clinton’s evaluations; and, finally, the condition that combines information about the experience of both candidates also has no effect. Considering evaluations of Trump’s leadership qualities, we find no significant effects of the treatments. He is not harmed by the Clinton critique, nor is he helped by information that frames his 17 This varies by partisanship: mean Clinton leadership evaluations among Democrats are higher, 5.53; and lower among Republicans, 2.82. We find the opposite pattern for Trump evaluations: Democrats = 2.29, Republicans = 4.87. 18 Another possibility is that we are underpowered to detect the type of modest effects that we anticipate. 14 candidacy in a more positive light (though the coefficients are positive and close to significant in the Trump experience and both conditions). We find similar effects when we use feeling thermometers as the dependent variable (Appendix Table 8). Figure 2. Effects of Terrorism News on Evaluations of Clinton and Trump, SSI Study Note: Based on an OLS regression with the control (non-threat) group serving as the baseline. The figure plots the coefficients on each treatment condition. While outside of our main hypotheses and exploratory, we analyzed the data for heterogeneous treatment effects by gender. Our rationale is that women on both sides of the aisle may have been more attuned to information about Clinton prior to entering the study, and thus less likely to be affected by the treatments. In fact, we find essentially no effect for the experience priming conditions on females in either party, while Republican males exposed to the Clinton experience condition come to see Clinton as a stronger leader and Trump as a weaker leader (see Appendix Tables 8A and 8B). We view 15 these results merely as suggestive, given that the hypotheses were not set a priori and the study is underpowered when it comes to leverage over the detection of effects by party and gender.19 In sum, while there is some evidence that perceptions of Clinton’s leadership can increase modestly with more active priming of her experience, it is only effective when that message is not counter-balanced by one’s opponent. This type of situation seems quite unlikely in an electoral context. It may be more likely that highlighting experience diminishes the extent to which terrorist threat takes a negative toll on Democratic female candidates like Clinton (more akin to the null effects we see in the Trump and Both conditions). 5.1. Conclusion The 2016 U.S. presidential election was unique for a variety of factors, including that Hillary Clinton was the first woman to run on a major ticket and Donald Trump was the first modern day major party candidate without significant political experience. Further, few elections have featured such a sustained and focused attention on terrorism driven by events and candidate rhetoric. Extant theory and evidence suggests that Democratic female candidates are doubly-disadvantaged while Republican male candidates are doubly-advantage when it comes to assessments made of them under contexts of terrorist threat. The unique experience gap between Clinton and Trump provides the opportunity to see whether experience (or lack of) can effectively countervail against those tendencies. Across multiple sets of data and analyses, we find that Clinton tends to be harmed by terrorism, while Trump is not. In some cases, the effect of terrorist threat on evaluations of Clinton is 19 The study also had a question asking respondents how well each candidate would handle issues of foreign policy and national defense. We did not find any significant main treatment effects, though men from both parties in the Clinton experience condition have higher evaluations of her ability to handle these issues than their counterparts in the control group, while Republican men in this condition perceive Trump as less able to handle these issues. 16 negative (ANES data (leadership); MTurk study (feeling thermometers) or null (ANES (feeling thermometer); terrorism condition in SSI study). Only when we isolate her experience in a targeted prime does Clinton receive an evaluation boost. We conclude that Democratic female leaders with experience may be able to somewhat diminish the negative effects of terrorist threat by touting their experience, but the effectiveness of this strategy is likely to vary across the course of the campaign, across subgroups, and across information contexts. Experience, in other words, is a salve, but not a salvation for Democratic female candidates running under conditions marked by terrorism threat. In a parallel manner, we find evidence that Trump’s lack of experience may have only mattered at the margins when it came to public evaluations as the issue of terrorism inserted itself into the campaign. In both the ANES and MTurk studies, he received more positive evaluations when terrorism worries were heightened, even though he had little foreign policy experience. Only in the SSI data do we see null effects rather than positive ones. That may be because experience was put front and center and worked to his disadvantage, or because of the timing of our study. The fact that Clinton touted her experience, and Trump touted his friendly relations with high-ranking military officials, suggests that both candidates felt they could leverage this in the course of the 2016 election. We agree their strategies may have helped public evaluations of their candidacies at the margins, which interestingly enough, is where the race was won. 17