Sex, Stereotypes, and Security: A Study of the Effects of Terrorist Threat on Assessments of Female Leadership

Abstract

This project examines the extent to which terrorist threat shapes affect assessments of female political leadership and whether these assessments are affected by the characteristics of the evaluated leader. Using experimental and survey data, we find that conditions of terrorist threat increase the expression of certain gender-trait stereotypes and negatively affect evaluations of female Democratic leaders. Our analysis further suggests that Republican women and/or those with strong national security experience may be able to overcome the otherwise negative implications of security threats and gender-trait stereotypes.

Keywords:

• threat
• terrorism
• gender
• sex
• stereotypes
• leadership

Notes

1. At the aggregate level, studies have demonstrated that certain factors can disadvantage women who are running for elected office, such as particular institutional arrangements, a small pool of female candidates in many states, and running in more conservative and Republican states. Meanwhile, states with a history of electing women to office are generally more open to female leaders (e.g., Arceneaux 2001; Fox and Oxley 2003; 2004; Nechemias 1987; Norrander and Wilcox 1998; Rule 1981; 1990; Wilcox and Norrander 2005). Other research has found that even when women overcome challenges, they still may be disadvantaged by their gender (e.g., Duerst-Lahti 2005 and chapter 4 in Seltzer, Newman, and Leighton 1997). At the individual level, scholars have identified a number of variables that increase the likelihood of voting for female candidates: for example, being a woman, a Democrat, a liberal, and having liberal opinions on women's issues (e.g., Cook 1994; Dolan 1998; Huddy and Terkildsen 1993a; 1993b; McDermott 1997; Paolino 1995; Schroedel and Snyder 1994).

2. Belief stereotypes refer to the ideology and policy preferences that individuals ascribe to males and females, while trait stereotypes refer to the personal qualities and characteristics that people infer about men and women (Huddy and Terkildsen 1993b).

3. Male duties are also generally perceived as more important than female duties, which results in a larger effect at higher levels of office (Rosenwasser and Dean 1989).

4. Scholars also examine factors that decrease these stereotypes such as females emphasizing masculine traits and issues, differences by partisanship, type of office, and the information environment (e.g., Fox and Oxley 2003; Hedlund et al. 1979; Huddy and Terkildsen 1993a; Kahn 1992; Leeper 1991; Little et al. 2007; Matland and King 2002; McDermott 1997; Rosenwasser and Dean 1989).

5. In related work on the importance of masculine traits in times of war, Hansen and Otero (2007) argue that the trait of strong leadership should become more consequential to presidential voting decisions. However, in their analysis of ANES data from 1984 to 2004, they do not find a stronger effect for leadership on voting decisions in 2004 compared to earlier years, a result that may be due to not separately analyzing those concerned about war or terrorism. Other scholars have found that leadership becomes more consequential to presidential voting decisions among those who think a future attack is more likely (Merolla and Zechmeister 2009).

6. We refer to the group that did not receive a treatment as both the Control and the Status Quo group, to acknowledge that subjects entering the lab carried with them understandings of current security threat levels, perceptions which we increased with the Terror Threat condition and decreased through the Good Times condition.

7. There were no significant differences in the distribution of subjects’ potentially relevant variables (party identification, ideology, political sophistication, gender, education, and race) across groups.

8. Those in the Terror Threat treatment expressed higher levels of worry about a terror attack than those in the control group (Pr (T > t) = 0.0262) and those in the Good Times treatment (Pr (T > t) = 0.0264).

9. We coded the data so that a higher value indicates a higher level of agreement, on the seven-point scales, with the statement; thus, on the combined measure higher values indicate higher gender-trait stereotypes.

10. Hanushek and Jackson (1977, 68–69), among others, note that one-tailed hypothesis tests are appropriate when one has a strong theory pointing to a unidirectional hypothesis, as we do in this case. The results are the same if we run a multivariate regression with good times as the baseline (see Appendix A) or with the control group as the baseline.

11. We tested for, but did not find, interactions between gender and the conditions enhanced the model for either female political leader. If we use the control group as the baseline instead of Good Times, we also find support for a mediating relationship.

12. Refer to note 11.

13. We ran tests to see if partisanship moderated the effects of the treatments and found this to be the case only for John Kerry. In Kerry's model, the inclusion of party identification and interactions between party and the treatments results in the Terror Threat treatment achieving significance, as well as the gender-trait stereotype and the party identification variable. In this model, strong Democrats appraise Kerry more favorably in response to the terror threat condition, while Republicans in the condition have a more negative evaluation of Kerry.

14. This finding occurs once the gender stereotype measure is at 4.0 for Rice and at 5.0 for Bush.

15. The ANES sample averages a 2.9 (in 2002) and a 3.0 (in 2004) on the seven-point partisan scale, which is more conservative than the mean of 2.01 in the experimental sample.