![[PDF]](/pb-assets/Icons/adobe-pdf-icon-1498649896243.png){kind=icon}
Abstract
Using an experimental study fielded before the U.S. 2016 presidential election, we test one potential mechanism to explain the outcome of the election: threatened gender identity. Building on masculine overcompensation literature, we test whether threat to masculinity can explain differential support for Donald Trump and Hillary Clinton among men, and adjudicate between two mediators: desire for a male president and desire for a masculine president. As predicted, we find that masculinity threat increases desire for a masculine president (but not desire for a male president), which in turn increases support for Trump and decreases support for Clinton among men. This study empirically documents the role masculinity threat may have played in the 2016 presidential election and politics more generally. This study also contributes to theory by providing evidence that masculine overcompensation works symbolically to reassert the status of masculinity over femininity rather than to simply emphasize maleness over femaleness.
The U.S. 2016 presidential election was marked by gender and not just because of Hillary Clinton’s historic nomination or Donald Trump’s now-infamous 2005 Access Hollywood tape (Fahrenthold 2016). Clinton and Trump also represented distinct visions about the current state of gender and its future in the United States. For instance, according to one poll, Trump supporters were more likely to believe men are discriminated against and that society has become “too soft and feminine” (Khazan 2016). For voters who feel men’s status is declining, social scientists and popular commentators speculated (e.g., Katz 2016), Trump—and in particular, Trump head-to-head against Clinton—served as a powerful symbol for their desire to return to a more gender-conventional society in which masculinity is publicly considered higher status than femininity. In this study, we ask whether threat to masculinity can be causally linked to support for Donald Trump for president. We further differentiate between two causal pathways of the effect of masculine threat on support for Trump: desire for a male president and desire for a masculine president.
Theoretical Framework: Masculine Overcompensation and the Gender Hierarchy
A growing body of literature demonstrates the lengths to which men will go to reassert their masculinity when it is called into question. In response to such gender identity challenges, researchers have found that men are more likely to sexually harass a fictitious female research participant (Maass et al. 2003), show increased support for the Iraq war and homophobic views (Willer et al. 2013), place greater blame on a female date-rape victim (Munsch and Willer 2012), and become physically aggressive (Bosson et al. 2009). Most studies have not found this overcompensation effect among women when their femininity has been called into question (cf. Munsch and Willer 2012), suggesting that masculinity is precarious and easily lost, whereas femininity is not (Vandello et al. 2008).
The fact that the overcompensation effect is found only among men aligns with both Ridgeway’s (2011) assertion that masculinity is regarded as more desirable and respectable than femininity and Connell’s ([1995] 2005) conceptualization of gender relations as hierarchical. Because femininity is subordinate to masculinity (Carrigan, Connell, and Lee 1985) and they feel they have more to lose, men feel more compelled to overcompensate in the face of gender identity threat than women. Arguably, men’s actions after experiencing gender identity threat offer researchers the opportunity to capture the behaviors associated with hegemonic masculinity—the most idealized form of masculinity within a given culture—which sits at the top of the larger gender order in terms of status (Connell [1995] 2005; Connell and Messerschmidt 2005). As such, men’s responses to threats to their masculinity can largely be understood as attempts to regain their status by realigning themselves with what they believe to be this socially idealized form of masculinity.
Hegemonic masculinity can take on different characteristics and features across different groups of men (Connell [1995] 2005). However, employment and being able to provide for one’s family remain pervasive components of widespread constructions of masculinity in the United States (Connell and Messerschmidt 2005; Gerson 2011; Townsend 2002). This would suggest that unemployed men, or men whose employment status is threatened, may seek ways to reassert their masculinity. Indeed, Brines (1994) found that unemployed men often attempt to recover or preserve some of their masculinity and status by doing less housework. Yet the causal link between threats to employment and masculine overcompensation has not been tested.
Current literature on masculinity threat has largely explored only the effect of gender identity threat on men’s personal masculinity. For example, Willer et al. (2013) used false feedback on a gender identity questionnaire to tell men that they were either masculine or feminine. However, threatening the employment of men as a group helps to expand the masculine overcompensation literature by offering insight into the group dynamics that underpin masculinity. More specifically, masculinity is relational, as hegemonic masculinity gains its status through the subordination of femininity and other forms of masculinity, and many men are complicit in its preservation at the top of the gender hierarchy (Connell [1995] 2005). Given this, we expect a threat to the employment status of men as a group, rather than threats to an individual man’s masculinity, to prime concern over greater shifts in the gender order and lead to an overcompensation effect among men. This concern would also be consistent with group position theory (Blumer 1958; Bobo 1999), which claims that group members will adopt pro-in-group policy preferences in the face of a perceived threat to their group’s position. In the following section, we draw upon this rationale and situate it in relation to the 2016 U.S. presidential election to provide empirical predictions that guide our experiment.
Empirical Predictions
From claiming to kiss and grope women without their consent (Fahrenthold 2016) to mocking a handicapped male journalist (Carmon 2016), Trump’s actions and rhetoric work toward the subordination of women and particular groups of men. As such, the actions and statements of Donald Trump epitomize a form of masculinity that is consistent with the dynamics that underpin Connell’s ([1995] 2005) conceptualization of hegemonic masculinity. Whether or not they explicitly support these actions, those who support Trump are complicit in upholding an oppressive form of masculinity over femininity and other forms of masculinity. From this, we hypothesize that if employment is central to masculinity, threats to employment status will lead men, but not women, to show increased support for Trump.
Hypothesis 1: Among men, threat to employment status will lead to increased support for Donald Trump.
As described above, we do not expect a similar response from women in the face of threat to their employment because employment is not central to femininity and femininity is not as status worthy as masculinity (Ridgeway 2011; Connell [1995] 2005).
At the same time, threats to men’s employment status may also drive them to distance themselves from the feminine to a greater extent than they usually would (Brines 1994). This suggests that threatening men’s employment status will lead them to show decreased support for Hillary Clinton.
Hypothesis 2: Among men, threat to employment status will lead to decreased support for Hillary Clinton.
Hypotheses 1 and 2 are derived from the masculine overcompensation literature, which is, in part, built upon identity control theory. According to identity control theory, individuals are motivated to act in ways that uphold particular identity standards that are central to their self-concepts (Stets and Burke 2005). Thus, threats to an identity that are central to individuals’ self-concepts often produce exaggerated and extreme behaviors as they seek to restore their alignment with that identity. Yet, confirmation of an identity is not expected to lead to such extreme behaviors. As a result, we do not expect improvements in employment outlook to lead to a significant decrease in support for Trump or an increase in support for Clinton among men or women.
Hypothesis 3: Improvements in employment outlook will not lead to decreased support for Donald Trump among men or women.
Hypothesis 4: Improvements in employment outlook will not lead to increased support for Hillary Clinton among men or women.
Two potential mediators could be driving the threat response among men. On the one hand, consistent with group position theory (Blumer 1958; Bobo 1999), threatening the employment status of men may generate concern about the status of men as a group, relative to women. Thus, we might expect increased support for Trump to be driven primarily by an increased desire for a male president. On the other hand, masculinity theory suggests that threat to men’s employment might be met with a more symbolic concern over the status of masculinity itself. Thus, threatened men might increase their support for a masculine president to reassert the value of masculinity over femininity. In this study, we differentiate between these as two possible and distinct causal pathways between threat to masculinity and candidate preference.
Hypothesis 5a: Increased desire for a male president will mediate the effect of employment threat on men’s support for Trump and Clinton.
Hypothesis 5b: Increased desire for a masculine president will mediate the effect of employment threat on men’s support for Trump and Clinton.
We predict that desire for a masculine president, not desire for a male president, will mediate support for Trump among men whose employment is threatened. We argue that group position theory produces too simplistic an explanation in the case of masculinity threat: voting for any male will not alleviate the symbolic threat posed to men by prospective unemployment, but voting for a masculine man will. Thus, we expect to find support for Hypothesis 5b, not Hypothesis 5a.
Methods
Participants
In total, 311 participants (168 men, 143 women) were recruited through Amazon Mechanical Turk (MTurk) to participate in a three-condition between-subjects experiment. MTurk is a crowdsourcing website that allows researchers to access samples for experimental research. While MTurk samples are not nationally representative, they are more representative than convenience samples (Berinsky, Huber, and Lenz 2012) and have been found to produce results similar to population-based samples (Weinberg, Freese, and McElhattan 2014).
Data were dropped from 38 participants due to failure of a manipulation check question (i.e., if the rate of unemployment was increasing or decreasing in the previous graph), from 1 participant for failing to complete all dependent measures, and from 13 participants who indicated that they were not intending to vote, leaving a total of 259 participants (147 men, 112 women) in the final sample. The average age of participants was 36 years (SD = 10.91), 57.52 percent reported having a bachelor’s degree or higher, 85.32 percent identified as heterosexual, and 47.88 percent reported being married. When asked to indicate which race-ethnicity they most identified as, 76.06 percent selected white/Caucasian, 5.02 percent selected black/African American, 3.47 percent selected Latino/Hispanic, 13.13 percent selected Asian/Asian American, and 2.32 percent selected other.
Dependent Measurements
The primary dependent variables were composites of support for Donald Trump (α = .98) and Hillary Clinton (α = .95). More specifically, in randomized order, participants were asked the following questions: “Whether or not you intend to vote in the 2016 election, to what extent would you say you support Hillary Clinton/Donald Trump?” (1 = not at all; 7 = extremely); “If the 2016 general elections were held today, how likely would you be to vote for Hillary Clinton/Donald Trump for president?” (1 = extremely unlikely; 7 = extremely likely); and “Would you say your overall opinion of Hillary Clinton/Donald Trump is . . . ?” (1 = extremely unfavorable; 7 = extremely favorable).
Mediators
We tested two mediators: desire for a male president and desire for a masculine president. On a 7-point Likert scale (1 = strongly disagree; 7 = strongly agree), participants were asked to indicate the extent to which they agreed with the following statements: “It is important that a man is the president of this country” (which we refer to as male president) and “All other things being equal, I would prefer a president with masculine qualities” (which we refer to as masculine president).
Procedure
Participants were recruited to take part in a study regarding their attitudes and opinions and randomly assigned to one of the following three conditions: control, masculine affirm, or masculine threat. After completing a few demographic questions, participants assigned to the control condition were presented with a graph that indicated that unemployment rates would remain stable in the participant’s area. Those assigned to the treatment conditions were presented with one of two graphs that indicated that unemployment rates in their area were predicted to either (1) increase for men and decrease for women (masculine threat) or (2) decrease for men and increase for women (masculine affirm). All participants were then asked to briefly describe what might be the effect of the projected unemployment rate on their life. Next, participants completed the dependent and mediating measures. Finally, participants completed some demographic questions and the study ended.
Results
Table 1 shows the results of t tests comparing support for Trump and Clinton by condition among men and women. Hypotheses 1 and 2 predicted that men in the masculine-threat condition would be more likely to support Trump and less likely to support Clinton. Table 1 shows that there was no direct effect of masculine threat on support for Trump or Clinton among men (or women). Thus, we do not find support for Hypotheses 1 or 2. On the other hand, there was also no direct effect of masculine affirm on support for Trump or Clinton among men (or women), which supports Hypotheses 3 and 4.
|
Table 1. Means and Standard Deviations (in parentheses) of Trump and Clinton Support by Condition and Participant Gender.
Despite the lack of a significant direct effect, we analyzed the effect of condition on the mediating variables, first through pairwise t tests and then through formal mediation analysis. Table 2 shows the means for male president and masculine president for each experimental condition by gender. Neither condition significantly affected desire for a male president among men or women. Thus, we find no support for Hypothesis 5a, as expected. However, we find that men in the masculine-threat condition were significantly more likely to support a president with masculine qualities (masculine president). As hypothesized, this was not the case among women in any condition.
|
Table 2. Means and Standard Deviations (in parentheses) of Desire for Masculine President and Desire for Male President by Condition and Participant Gender.
Given the significant effect of masculine threat on masculine president, we conducted a formal mediation analysis. Such an analysis without a direct effect has been used in previous research (Rucker et al. 2011). We used Preacher and Hayes’ (2004) bootstrapping procedures with 5,000 samples to find the indirect effect of masculinity threat on Trump and Clinton support. Figures 1 and 2 show these results. Figure 1 reveals that desire for a masculine president has an indirect effect of 0.43 (0.61*0.70) on support for Trump for men in the masculine-threat condition, with the 90 percent confidence interval ranging from 0.07 to 0.86. Thus, the indirect effect is marginally significant. Figure 2 shows that desire for a masculine president has an indirect effect of −0.37 (0.63*–0.58) on support for Clinton for men in the masculine-threat condition, with the 90 percent confidence interval ranging from −0.69 to −0.05. Like for Trump support, the indirect effect is marginally significant. Thus, we find support for Hypothesis 5b, which stated that desire for a masculine president would mediate the effect of masculine threat on increased support for Trump and decreased support for Clinton.
Discussion
While we did not find the expected direct effect of masculinity threat on support for Donald Trump and Hillary Clinton, the study differentiates between two possible pathways of masculinity threat on presidential candidate preference. We found that men in the masculine-threat condition were significantly more likely to support a masculine president but were not more likely to support a male president. Using Preacher and Hayes’ (2004) bootstrapped test of mediation, we find that desire for a masculine president indirectly mediates support for Trump and support for Clinton among men in the masculine-threat condition. This finding suggests that masculine overcompensation works symbolically to reassert the value of masculinity over femininity, as suggested by masculinity theory, rather than to reassert the status of men over women, as suggested by group position theory. Instead of attempting to emphasize maleness, men in the masculine-threat condition attempted to reassert their group’s symbolic status by emphasizing the value of masculinity.
There are several potential reasons why we did not find a direct effect of masculine threat on increased support for Donald Trump or decreased support for Hillary Clinton in our study. The experiment was fielded less than a month before the election. Before that time, the Washington Post had already released the 2005 Access Hollywood tape in which Trump states, “When you’re a star, [you can] grab them [women] by the pussy” (Fahrenthold 2016). This was just one scandal in a long list that polarized voters’ thoughts about Trump and Clinton. By the time the experiment was fielded, participants’ views about each of them may have already concretized. Thus, while men in the masculine-threat condition were more likely to desire a masculine president, the manipulation may not have been strong enough to alter their preexisting views about the Republican or Democratic candidate. A simpler explanation might be that male participants did not see Trump as sufficiently masculine; in fact, they rated him as a 4.86 on a scale of 1 (not at all) to 7 (extremely) when asked how masculine they perceived him. While they did rate him as more masculine than Clinton (who was rated on average 3.01), a score of 4.86 indicates that they rated him as less than 1 point more masculine than the neutral midpoint (4). Similarly, we may not have found a direct negative effect of masculine threat on Clinton support among men because they did not see her as particularly feminine (on average she was rated a 4.24 on a scale of 1 to 7) even though she was seen as more feminine than Trump (who was rated on average 1.96). Regardless, this study identifies an important link between threat to masculinity and candidate preference.
Conclusion
The study makes an important link between masculine overcompensation and candidate preference. We draw upon the literature on masculine overcompensation, group position theory, and identity control theory to offer one mechanism to explain the outcome of the 2016 presidential election: threatened gender identity. We find that men who perceived a threat to men’s group position were significantly more likely to desire a masculine president, which indirectly explained increased support for Donald Trump and decreased support for Hillary Clinton. As Katz (2016) has argued, our findings suggest that a presidential candidate’s performance of masculinity can be an avenue for political support. In sum, we find that the symbolic precariousness of masculinity certainly plays an important yet complicated role in candidate preference.
Acknowledgements
The authors would like to thank Robb Willer, Chrystal Redekopp, Shelley Correll, and the Laboratory for Social Research at Stanford University for their comments, guidance, and support.
Authors’ Note
Both authors contributed equally.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was made possible by funding through the Lab for Social Research at Stanford University.
References
|
Berinsky, Adam J., Huber, Gregory A., Lenz, Gabriel S. 2012. “Evaluating Online Labor Markets for Experimental Research: Amazon.com’s Mechanical Turk.” Political Analysis 20(3):351–68. Google Scholar | |
|
Blumer, Herbert . 1958. “Race Prejudice as a Sense of Group Position.” Pacific Sociological Review 1(1):3–7. Google Scholar | SAGE Journals | |
|
Bobo, Lawrence D. 1999. “Prejudice as Group Position: Microfoundations of a Sociological Approach to Racism and Race Relations.” Journal of Social Issues 55(3):445–72. Google Scholar | Crossref | ISI | |
|
Bosson, Jennifer K., Vandello, Joseph A., Burnaford, Rochelle M., Weaver, Jonathan R., Wasti, S. Arzu. 2009. “Precarious Manhood and Displays of Physical Aggression.” Personality and Social Psychology Bulletin 35(5):623–34. Google Scholar | SAGE Journals | ISI | |
|
Brines, Julie . 1994. “Economic Dependency, Gender, and the Division of Labor at Home.” American Journal of Sociology 100(3):652–88. Google Scholar | Crossref | ISI | |
|
Carmon, Irin . 2016. “Trump’s Worst Offense? Mocking Disabled Reporter, Poll Finds.” NBC News. Retrieved June 13, 2017 (http://www.nbcnews.com/politics/2016-election/trump-s-worst-offense-mocking-disabled-reporter-poll-finds-n627736). Google Scholar | |
|
Carrigan, Tim, Connell, Bob, Lee, John. 1985. “Toward a New Sociology of Masculinity.” Theory and Society 14(5):551–604. Google Scholar | Crossref | ISI | |
|
Connell, Raewyn. [1995] 2005. Masculinities. 2nd ed. Berkeley: University of California Press. Google Scholar | |
|
Connell, Raewyn, Messerschmidt, James W. 2005. “Hegemonic Masculinity: Rethinking the Concept.” Gender & Society 19(6):829–59. Google Scholar | SAGE Journals | ISI | |
|
Fahrenthold, David A. 2016. “Trump Recorded Having Extremely Lewd Conversation about Women in 2005.” Washington Post, October 8. Retrieved June 12, 2017 (https://www.washingtonpost.com/politics/trump-recorded-having-extremely-lewd-conversation-about-women-in-2005/2016/10/07/3b9ce776-8cb4-11e6-bf8a-3d26847eeed4_story.html?utm_term=.9c5194c8e721). Google Scholar | |
|
Gerson, Kathleen . 2011. The Unfinished Revolution: Coming of Age in a New Era of Gender, Work, and Family. 1st ed. New York: Oxford University Press. Google Scholar | |
|
Katz, Jackson . 2016. Man Enough? Donald Trump, Hillary Clinton, and the Politics of Presidential Masculinity. Northampton, MA: Interlink. Google Scholar | |
|
Khazan, Olga . 2016. “The Precarious Masculinity of 2016 Voters.” The Atlantic, October 12. Retrieved June 12, 2017 (https://www.theatlantic.com/politics/archive/2016/10/male-trump-voters-masculinity/503741/). Google Scholar | |
|
Maass, Anne, Cadinu, Mara, Guarnieri, Gaia, Grasselli, Annalisa. 2003. “Sexual Harassment under Social Identity Threat: The Computer Harassment Paradigm.” Journal of Personality and Social Psychology 85(5):853–70. Google Scholar | Crossref | Medline | ISI | |
|
Munsch, Christin L., Willer, Robb. 2012. “The Role of Gender Identity Threat in Perceptions of Date Rape and Sexual Coercion.” Violence Against Women 18(10):1125–46. Google Scholar | SAGE Journals | ISI | |
|
Preacher, Kristopher J., Hayes, Andrew F. 2004. “SPSS and SAS Procedures for Estimating Indirect Effects in Simple Mediation Models.” Behavior Research Methods, Instruments, & Computers 36(4):717–31. Google Scholar | Crossref | Medline | |
|
Ridgeway, Cecilia L. 2011. Framed by Gender: How Gender Inequality Persists in the Modern World. New York: Oxford University Press. Google Scholar | Crossref | |
|
Rucker, Derek D., Preacher, Kristopher J., Tormala, Zakary L., Petty, Richard E. 2011. “Mediation Analysis in Social Psychology: Current Practices and New Recommendations.” Social and Personality Psychology Compass 5(6):359–71. Google Scholar | Crossref | |
|
Stets, Jan E., Burke, Peter J. 2005. “New Directions in Identity Control Theory.” Advances in Group Processes 22:43–64. Google Scholar | Crossref | |
|
Townsend, Nicholas . 2002. Package Deal: Marriage, Work And Fatherhood In Men’s Lives. 1st ed. Philadelphia: Temple University Press. Google Scholar | |
|
Vandello, Joseph A., Bosson, Jennifer K., Cohen, Dov, Burnaford, Rochelle M., Weaver, Jonathan R. 2008. “Precarious Manhood.” Journal of Personality and Social Psychology 95(6):1325–39. Google Scholar | Crossref | Medline | ISI | |
|
Weinberg, Jill D., Freese, Jeremy, McElhattan, David. 2014. “Comparing Data Characteristics and Results of an Online Factorial Survey between a Population-based and a Crowdsource-recruited Sample.” Sociological Science 1:292–310. Google Scholar | Crossref | |
|
Willer, Robb, Rogalin, Christabel L., Conlon, Bridget, Wojnowicz, Michael T. 2013. “Overdoing Gender: A Test of the Masculine Overcompensation Thesis.” American Journal of Sociology 118(4):980–1022. Google Scholar | Crossref | ISI |
Author Biographies
Emily K. Carian is a PhD candidate in sociology at Stanford University. Her research explores gender, masculinity, and gender inequality from a social movements and social psychological perspective.
Tagart Cain Sobotka is a PhD candidate in sociology at Stanford University. His research interests include gender, masculinity, social psychology, and medical sociology. His current research uses social psychological frameworks to better understand men’s role in upholding gender inequality across various domains.
